Alternatives for Environmental Valuation
Environmental cost–benefit analysis was developed by economists in the belief that monetary valuation of the environmental repercussions of economic activity is essential if the ‘environment’ stands any chance of being included in governmental and business decisions. This volume examines the limitations of this monetary approach, and considers the alternatives. Three broad angles from which to view environmental values are presented: applying social psychology concepts which challenge standard approaches; introducing multidimensional and non-monetary techniques; and examining vested interest group and citizen participation in processes of environmental valuation. Combinations of these approaches are also covered. The contributions presented are a valuable resource for both environmental and ecological economists. This authoritative book will also prove useful for those with a general interest in the environment, including policy-makers and nongovernmental organizations. Michael Getzner is Associate Professor of Economics at the University of Klagenfurt, Austria. Professor Clive L. Spash holds the Research Chair in Environmental and Rural Economics at the University of Aberdeen, UK. He is head of the Socio-Economic Research Programme at the Macaulay Institute and President of the European Society for Ecological Economics. Sigrid Stagl is Lecturer and coordinator of the Ecological Economics MA Programme at the School of the Environment, the University of Leeds, UK. She is Vice-President of the European Society for Ecological Economics.
Routledge Explorations in Environmental Economics Edited by Nick Hanley University of Glasgow
1 Greenhouse Economics Values and ethics Clive L. Spash 2 Oil Wealth and the Fate of Tropical Rainforests Sven Wunder 3 The Economics of Climate Change Edited by Anthony D. Owen and Nick Hanley 4 Alternatives for Environmental Valuation Edited by Michael Getzner, Clive L. Spash and Sigrid Stagl
Alternatives for Environmental Valuation
Edited by Michael Getzner, Clive L. Spash and Sigrid Stagl
First published 2005 by Routledge 2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN Simultaneously published in the USA and Canada by Routledge 270 Madison Ave, New York, NY 10016 Routledge is an imprint of the Taylor & Francis Group This edition published in the Taylor & Francis e-Library, 2005. “To purchase your own copy of this or any of Taylor & Francis or Routledge’s collection of thousands of eBooks please go to www.eBookstore.tandf.co.uk.” © 2005 Michael Getzner, Clive L. Spash and Sigrid Stagl for selection and editorial matter; the contributors for individual chapters All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. 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 0-203-41287-7 Master e-book ISBN
ISBN 0-203-34079-5 (Adobe eReader Format) ISBN 0–415–31012–1 (Print Edition)
Contents
v
Contents
List of figures List of tables List of contributors Preface Acknowledgements List of abbreviations 1
Exploring alternatives for environmental valuation
vii ix xi xiii xiv xv 1
C L I V E L . S PA S H , S I G R I D S TAG L A N D M I C H A E L G E T Z N E R
PART I
Extending the environmental valuation approach 2
A framework for valuing nature: regional biodiversity
21 23
MICHAEL GETZNER
3
Non-use values and attitudes: wetlands threatened by climate change
51
J Ü RG E N M E Y E R H O F F
4
Modelling environmental behaviour: socio-psychological simulation
69
H A N S - J OAC H I M M O S L ER
PART II
Taking multiple criteria into account 5
Assessing the quality of different MCDA methods A N D R E A D E M O N T I S, P A S QUA L E D E T O RO, B E RT D RO S T E - F R A N K E , I N E S O M A N N AND S I G R I D S TAGL
97 99
vi Contents 6
MCDA and stakeholder participation: valuing forest resources
134
WENDY P ROCTOR
7
Confronting risk with precaution: a multi-criteria mapping of genetically modified crops
159
A N DY S T I R L I N G A N D S U E M AY E R
PART III
Deliberation, participation and value expression 8
Consumer valuation and citizen deliberation: towards a comparison
185
187
JONATHAN ALDRE D
9
Three approaches to valuing nature: forest floodplain restoration
209
W E N DY K E N YO N A N D N I C K H A N L E Y
10
Deliberation and economic valuation: national park management
225
R O S E M A RY F. J A M E S A N D R U S S E L L K . B L A M E Y
11
The challenges of stakeholder participation: agri-environmental policy
244
K AT JA A R Z T
12
Preference transformation through deliberation: protecting world heritage
263
S I M O N N I E M E Y ER
Index
290
List of figures vii
Figures
2.1 3.1 3.2 3.3 4.1 4.2 4.3
The ‘standard’ and ‘extended’ model of valuation A conceptual model of WTP determinants Tree diagram for non-users Tree diagram for users Framework model of behaviour toward the environment Representation of the ‘processing of group influences’ Change in average attitute due to various group-oriented interventions 4.4 Representation of ‘processing of observed behaviour towards the environment’ 4.5 Average population behaviour with differing interventions based on observational learning 4.6 Representation of ‘processing of information on the collective use of a resource’ 4.7 Use of an environmental resource 4.8 Representation of ‘processing of communicative influencing under conditions of knowledge and environmental concern’ 4.9 Information campaign experiments without environmentally relevant communication 4.10 Experiments with multiplicators 6.1 Hierarchy of objectives, criteria and indicators for the Southern Region RFA 6.2 Preferences of criteria 6.3 Preferences of environmental criteria 6.4 Preferences of economic criteria 6.5 Preferences of social criteria 6.6 Overall ranking of options 6.7 Priorities of broad criteria 6.8 Range of individual priorities 6.9 Environmental preference 6.10 Economic preference 7.1 Risk, uncertainty, ambiguity and ignorance defined
25 57 62 62 73 77 79 80 82 84 86 88 90 92 145 146 147 147 148 153 153 154 154 155 163
viii 7.2 7.3 7.4 7.5 7.6 10.1 10.2 10.3 10.4 11.1 12.1 12.2 12.3 12.4 12.5 12.6 12.7 12.8 12.9
List of figures Ambiguity of ordering in risk assessments A model of the relationships between risk, science and precaution The multi-criteria mapping process Final individual rankings for the basic options Diverse options mixes favoured by individual participants CJs applied to non-market valuation Spending on parks excluding salaries Process for decision making under ‘charge 1’ Process for decision making under ‘charge 2’ Participants at the first six meetings of the AEF The Bloomfield Track Juror loadings on factor 1 ‘preservation’ Juror loadings on faction 2 ‘optimism’ Juror loadings on factor 3 ‘pragmatism’ Juror loadings on factor 4 ‘symbolism’ Changes to preference factor loadings before and after deliberation Changes to loadings on ‘preservationist’ and ‘close road’ factors Changes to loadings on ‘pragmatist’ and ‘status quo’ factors Changes to loadings on ‘optimist’ and ‘status quo’ factors
166 168 175 179 181 230 232 234 235 253 267 274 275 277 277 280 281 282 283
List of tables
ix
Tables
2.1 2.2 2.3 2.4 2.5 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 3.10 5.1 5.2 6.1 6.2 6.3 6.4 6.5 6.6 6.7 6.8 6.9 6.10 6.11 6.12 7.1 7.2 7.3
Consumer vs. citizen and students’ perspectives Epistemological problems of WTP surveys WTP regression model variables Logit model estimated coefficients for WTP1 Log-normal OLS estimated coefficients for WTP2 Distribution of interviews and in-principle WTP Summary statistics of the WTP (in DM) for users and non-users Attitudes towards nature Attitudes towards climate change Attitudes towards nature and WTP Attitudes towards climate change and WTP Relationship between attitudes towards money and mean WTP Potential explanatory variables OLS results of non-user group OLS results of user group List of quality criteria for MCDA methods Summary of method comparison The AHP intensity of importance scale Example pairwise comparison Normalized matrix example Pairwise comparison of environmental criteria Overall ranking example Impact table for forest use options Pairwise comparison of options under environmental criteria Priority of options in terms of environmental criteria Pairwise comparison of options under economic criteria Priority of options in terms of economic criteria Pairwise comparison of options under social criteria Priority of options in terms of social criteria The definitions of the ‘basic options’ appraised by all participants The participants Additional options defined by participants
36 37 38 39 41 55 56 58 58 59 59 60 61 64 64 101 126 141 143 143 143 143 149 150 150 151 151 152 152 173 174 176
x List of tables 7.4 8.1 9.1 9.2 9.3 9.4 9.5 9.6 9.7 9.8 10.1 10.2 11.1 11.2 12.1 12.2 12.3 12.4
Broad groupings of criteria defined by participants Possible advantages of discussion Descriptive statistics for WTP for the Ettrick project Positive issues identified with the forest floodplain project Jury suggestions measuring environmental project success Original and revised bids with reasons for change WTP descriptive statistics for initial valuation WTP descriptive statistics for final valuation Good aspects of the Ettrick project Project problems, solutions and the importance of bad aspects Charge 1: CJ group preference among three options Charge 2: introducing Option 4 Variables feeding into trust, reciprocity and reputation Proceedings at the agri-environmental forum Statement scores for Q attitudinal factors Rank ordering of preferences for management options Factor scores for preference ranking Correlation between preference and Q factors loading
177 191 214 216 217 218 219 219 219 220 231 231 251 254 272 278 279 279
List of contributors
xi
Contributors
Jonathan Aldred, Department of Land Economy, Cambridge University, UK, email:
[email protected] Katja Arzt, Department of Agricultural Economics and Social Science, Chair of Resource Economics, Humboldt University of Berlin, Germany, email:
[email protected] Russell K. Blamey, Programme Visitor, Research School of Social Sciences, Australian National University, Canberra, Australia, email: russell.blamey@ anu.edu.au Andrea De Montis, Dipartimento di Ingegneria del Territorio, Sezione di Urbanistica, Università di Cagliari, Cagliari, Italy, email: ademonti@ vaxca1.unica.it Pasquale De Toro, Dipartimento di Conservazione dei Beni Architettonici ed Ambientali, Università di Napoli Federico II, Naples, Italy, email:
[email protected] Bert Droste-Franke, Institute of Energy Economics and the Rational Use of Energy, University of Stuttgart, Stuttgart, Germany, email:
[email protected] Michael Getzner, Department of Economics, University of Klagenfurt, Klagenfurt, Austria, email:
[email protected] Nick Hanley, Department of Economics, Adam Smith Building, The University of Glasgow, Glasgow, UK, email:
[email protected] Rosemary F. James, Centre for Resource and Environmental Studies, Australian National University and CRC for Weed Management Systems, Campbell, Australia, email:
[email protected] Wendy Kenyon, Socio-Economic Research Programme, Macaulay Institute, Aberdeen, UK, email:
[email protected] Sue Mayer, GeneWatch UK, Tideswell, Buxton, Derbyshire, UK, email:
[email protected]
xii
List of contributors
Jürgen Meyerhoff, Institute of Management in Environmental Planning, Technical University Berlin, Berlin, Germany, email:
[email protected] Hans-Joachim Mosler, Department of Psychology, Division of Social Psychology, University of Zürich, Zürich, Switzerland, email: mosler@ sozpsy.unizh.ch Simon Niemeyer, Social and Political Theory Group, Research School of Social Sciences, Austalian National University, Canberra, Australia, email:
[email protected] Ines Omann, Department of Economics, Karl-Franzens University, Graz, Austria, email:
[email protected] Wendy Proctor, Centre for Resource and Environmental Studies, Australian National University, Canberra, Australia, email:
[email protected] Clive L. Spash, Socio-Economic Research Programme, Macaulay Institute, Aberdeen, UK, and Geography and Environment Department, University of Aberdeen UK, email:
[email protected] Sigrid Stagl, School of the Environment, University of Leeds, Leeds, UK, email:
[email protected] Andy Stirling, Science Policy Research Unit (SPRU), University of Sussex, Brighton, UK, email:
[email protected]
Preface
xiii
Preface
This book arose from on-going debates occurring amongst European ecological economists who have been reflecting upon extensions of the ‘standard’ economic model of environmental valuation. Such extensions try to account for citizens’ attitudes, institutional framing and social context as they affect any given environmental change being valued. Three broad approaches to valuation can then be identified and are pursued here. First, economists have been looking to learn from social psychology. Thus, for example, the concept of reasoned action has been one way in which qualitative and quantitative factors have begun to be linked in some contingent valuation studies. At the same time the process of learning from social psychology challenges standard approaches and requires that new alternatives be developed. A major set of alternatives to environmental cost–benefit analysis, which form the second broad approach, focus upon multiple criteria for understanding and improving decisions. The work reported here explores both the potential, as well as the pitfalls of multidimensional and non-monetary techniques. The final approach deals with innovation in the area of vested interest group and citizen participation in processes of environmental valuation. One focus of research in this area has been in terms of combining economic valuation and participatory deliberative approaches arising from political science. Work is presented comparing citizens’ juries and contingent valuation and reflecting upon potential combinations termed deliberative monetary valuation. Stakeholder or vested interest group participation is more common than trying to represent the general public but, as discussed, can also prove difficult. Overall, this book shows a range of approaches being explored by ecological economists and how they are open to interdisciplinary critiques, often dismissed by narrowly focused disciplines. In particular, European ecological economists have been engaging with traditional economics approaches while also learning from applied philosophy, geography, political theory, social psychology and sociology.
xiv Acknowledgements
Acknowledgements
Michael Getzner would like to pay tribute to Professor Egon Matzner (Vienna), who sadly passed away prematurely in October 2003, and who he will miss as both an outstanding teacher and a colleague. We all acknowledge our debt to the European Society for Ecological Economics which has provided an active forum for discussion on such matters as are covered in this volume. There is now a great deal of networking that takes place across Europe amongst those addressing the socio-economic aspects of ecological problems. This includes both those involved in policy processes and academics. The area of science and society interactions is a general and key concern with specific realisations in new technologies. Friends and colleagues who have stimulated our thoughts in these areas with respect to the topics in this volume include: Anders Biel, Jacquie Burgess, Claudia Carter, Mick Common, Silvio Funtowicz, Colin Green, Alan Holland, Sybille van den Hove, Joan Martinez-Alier, Martin O’Connor, John O’Neill, Ortwin Renn, Peter Söderbaum and Arild Vatn. There are also several people outside Europe with whom we interacted and acknowledge as active in these debates including: Edmundo Claro, John Gowdy, Jack Knetsch, Michael Lockwood, Dick Norgaard, Bryan Norton and Sabine O’Hara. Of course as editors we also thank all the contributors to this volume.
Abbreviations
AEF AHP CBA CJs CVM DMV EU Evamix GM GP MAUT MCA MCDA MCM MOP NAIADE NGO NOAA UK USA WTA WTP
agri-environmental forum analytic hierarchy process cost–benefit analysis citizens’ juries contingent valuation method deliberative monetary valuation European Union evaluation matrix genetically modified goal programming multiple attribute utility theory multiple criteria analysis multiple criteria decision analysis multi-criteria mapping multi-objective programming novel approach to imprecise assessment and decision environments non-governmental organisation National Oceanic and Atmospheric Administration United Kingdom United States of America willingness to accept willingness to pay
xvi List of abbreviations
Exploring alternatives for environmental valuation
1
1
Exploring alternatives for environmental valuation Clive L. Spash, Sigrid Stagl and Michael Getzner
Introduction The method of inclusion of environmental resources and ecosystem services in decision processes determines how far they are taken into account with results affecting the quality of our lives and those of future generations. A persistent argument has been that monetary valuation is essential if the ‘environment’ is to have any chance of being included in government and business decisions. Environmental cost–benefit analysis (CBA) was developed by environmental economists in order to achieve this monetisation of environmental entities so that the prices in market economies might be adjusted. A range of methods were developed including travel cost, hedonic pricing, production function analysis, contingent valuation and choice modelling (see Hanley and Spash 1993; Spash and Carter 2002). The overall aim has been to select project options on the basis of their welfare impacts and to support government taxes which reflect the social costs of environmental degradation. This approach has met with some success in that various national and international agencies have been interested in performing monetary valuation exercises as part of their overall assessment of projects. The idea of environmental taxation has also risen on the political agenda if remaining limited in practice.1 However, there has also been criticism of environmental CBA, or more specifically of some of the studies conducted under that guise. Critiques can be broadly grouped into those concerned with the theoretical foundations of economic values, and those looking at the validity of specific numbers being produced and the tools employed. In the former case, Kapp (1950) provides an isolated early example which shows the limits to monetary as opposed to other values in society. More recently, environmental philosophers have produced a whole body of literature showing the narrowness of value theory typically expressed by economists and the failure of economic training to place this theory in the context of wider social values (see O’Neill 1993). Reflecting upon this literature requires realising that most economics concerns consequentialism using a form of preference-based utilitarianism, which is a very specific philosophy of value rather than a generally accepted meta ethic which can be universally applied. In trying to address the environmental problems of the 20th century, economic value theory has shown
2
Clive L. Spash, Sigrid Stagl and Michael Getzner
itself lacking in several areas, such as the treatment of time, complexity, strong uncertainty, political systems, rights and social norms. The enhanced Greenhouse Effect raises all these issues and exposes the failures of discounting, partial equilibrium analysis, risk assessment and assuming that efficiency is the key social goal (Spash 2002b). The other area of critical analysis abstracts from the first set of problems and tends to be a debate on validity which has been largely internal to the economics profession (although implications of the value debate are also related via theoretical foundations). Validity concerns result in specific applications being criticised for their failure to respect microeconomic and welfare theoretic constraints. Those working within decision making government institutions, such as environmental protection agencies or the treasury, may see such theoretical restrictions as academic and going beyond what is required for practical policy making (for example, see comments by Burney 2000). However, once valuation becomes divorced from its theoretical roots, numbers can be produced which have little content or meaning, and are defensible only in terms of their political role rather than their theoretical basis. In this respect the popularisation of environmental CBA has been a problem as studies proliferate, numbers are transferred out of context and applications seek to meet policy desires. Excessive aggregation is exemplified by attempts to value regional, national and even the world’s ecosystems, although what is meant by a trade price in these contexts is mystifying, and asking who might be the ‘seller’ and ‘purchaser’ exposes the fallacious use of economic value theory. Furthermore, the range of reasonable numbers stemming from monetisation exercises is often very broad, which facilitates the arbitrary use of valuation results. The contingent valuation method (CVM) has become the most widely conducted CBA tool. The main advantage attracting this attention is the ability of CVM to estimate what are termed option, ‘existence’ and bequest values in addition to direct use values. These are amounts reflecting: the preservation of an option to use a resource in the future, the maintained existence of a resource or entity regardless of personal use, and the desire to pass the resource to one’s descendents. The sum of these indirect or passive use values can be large compared with the direct use values associated with non-market goods to which the other CBA methods are solely restricted. The popularity of the CVM is also based upon the apparent simplicity of asking members of the general public the maximum they are willing to pay (WTP) for an environmental improvement or, less commonly, the least they would be willing to accept (WTA) in compensation for environmental degradation. Considerable research energy has gone into refining CVM and related stated preference approaches. The result has been to increase the complexity and cost of conducting a CVM if it is to attain the highest standards of validity. In the United States of America this is the legally defensible standard which can cost millions of dollars for a nationally representative random sample including a range of pretests, focus groups and internal consistency and validity checks in the survey, as well as several variations on survey design. The litigation surrounding the oil spill from the tanker Exxon Valdez led to a set of such standards being produced by a panel of experts commissioned by the
Exploring alternatives for environmental valuation
3
National Oceanic and Atmospheric Administration (NOAA). This panel consisted of a number of famous economists, such as Nobel Prize winners Kenneth Arrow and Robert Solow. The panel was also asked to investigate the concept of ‘existence’ value. The panel’s report gave a qualified support for the CVM (NOAA 1993). The panel stated that ‘existence’ values are a theoretically meaningful aspect of value. As regards the ability of the CVM to estimate such values, the panel noted five main problem areas: inconsistency of responses with economic models of rational choice; mental account bias, especially when respondents are inadequately informed as to substitutes; aggregating benefits; providing information; and warm glow effects. However, the panel felt that so long as certain guidelines were followed, CVM results could be judged as both meaningful and useful indicators of natural resource damages. These guidelines are that (i) WTP should be used in preference to WTA; (ii) mail surveys should be avoided; (iii) respondents should be given full information on the resource change (including information on substitutes), and be asked how well they understand this information; (iv) open-ended responses should be rejected in favour of closed-ended referendum formats; (v) a random population sample is required; (vi) respondents should be reminded about the need to reduce expenditure on some item of their budget in order to be able to pay their stated bid; and (vii) careful pre-testing should be carried out. Such rules tend to be excessively prescriptive and ignore the specific context in which a valuation exercise must take place. There are certainly general design features in any contingent valuation study which are desirable. These include clear description of the institutional context, explaining the consequences and expected benefits of payment, being aware of various biases which can be introduced in design, such as use of a contentious bid vehicle, and generally producing a realistic scenario. Without addressing such issues any research which claims to be relevant to the economic debate will be seen as inadequate and may also be misleading. However, employing rules which prevent hypothesis testing and experimental design is more worrisome. The NOAA Panel holds an implicit belief in a single universal standard of best practice which ignores cultural, legal and institutional variation. People trade differently in different countries (e.g. haggling vs. accepting the stated price), property rights vary (WTP vs. WTA), areas of trade which are taboo differ, and truly random sampling is difficult at best and impossible in industrially developing countries. A serious concern here is the extent to which research designs which fail to conform to the rules are branded as poor practice when they may be valid research aiming to question the underlying tenets of the model. Blind use of highly specific guidelines can remove the room for original research, especially in the areas of economic psychology and human behaviour. CBA, along with mainstream microeconomics, has been increasingly criticised for building upon an unsound theoretical base in terms of human behaviour and the CVM has made this even more evident. That is, the microeconomic axioms of choice make assumptions which are at best inconsistent with theories of modern psychology and empirical evidence. In the valuation area, Kahneman and Tversky’s (1979) prospect theory revealed that people value gains and losses asymmetrically and this can explain the observed gap between WTP and WTA measures. Knetsch
4
Clive L. Spash, Sigrid Stagl and Michael Getzner
(1994, 1995, 1999; also Knetsch and Sinden 1984) has shown the frequent occurrence of behaviours which are inconsistent with accepted economic norms but commonly dismissed by economic models. The refusal to make trade-offs has been shown to arise in CVM studies both amongst those who protest against the use of monetary valuation of the environment and also those prepared to make a contribution (Spash 2000b, 2002a, 2002c). The results can be linked to rightsbased beliefs which contrast with those of the economic utilitarian (Spash and Simpson 1994; Spash 2000c). In terms of the economic model, the only way in which such behaviour can be characterised is as lexicographic preferences which are deemed, at best, to be held only by a strange minority (Spash 1998, 2000a; Spash et al. 2000). A range of techniques is therefore required to understand human psychology in terms of the attitudes, ethical beliefs and social norms which motivate behavioural responses. Concerns raised about the practice of attaching monetary values to all dimensions of socio-economic and biophysical systems have led to the call for alternative valuation methods (Martinez-Alier et al. 1998). There has been a powerful argument for shifting away from the sole focus on outcome towards the quality of decision processes (Funtowicz and Ravetz 1990; O’Connor et al. 1998). In particular, multiple criteria assessment has become a more widely used method, offering the potential for the integrated assessment of local, national and international policies and as a means for combining different perspectives associated with sustainability goals. Sustainability raises a set of issues based on civil rights of current and future generations as well as respect for ecological systems. Understanding that biophysical as well as human systems are complex, and will never be fully understood, has led to the development of approaches which favour adaptive behaviour and learning processes over optimal solutions. Closely linked to this view is the acknowledgement of uncertainty of future events and their impacts on human and non-human systems. All these factors redefine the role of experts, the meaning of knowledge and how decision processes need to be designed to make more effective policy. In the environmental policy arena, and elsewhere, there has been a push for greater public participation, e.g. in Europe the Aarhus Convention (European Commission 1998), and the inclusion of non-governmental stakeholders in project appraisal. Researchers are meeting this challenge both through interdisciplinary discourse and also by contributing to the formation of novel democratic institutions. There is both innovation in traditional environmental CBA and in terms of new political institutions for addressing valuation issues. In the former case are the range of techniques termed deliberative monetary valuation (Spash 2001), and in the latter such tools as the citizens’ jury. Research into the operational criteria for effective public participation is at an early stage of development. The search for validity in applying CBA methods has lead to a variety of appeals to interest groups and/or members of the public in an attempt to supplement the normal information content of prices placed on the environment. Thus, the travel cost method may be combined with an interview approach in order to sustain assumptions of how individuals behave, value time and relate to the environment.
Exploring alternatives for environmental valuation
5
Hedonic pricing has sought corroboration from estate agent valuations as representing ‘informed’ preferences. Focus groups have been used in conjunction with the CVM to test survey design on the basis that group deliberation could validate the information content and help identify design biases. This last approach is most clearly where deliberative practices have begun to enter. For example, the largest CVM study in the UK was conducted on environmental impacts associated with aggregates (Department of the Environment, Transport and the Regions 1999). An interesting feature of this work in the current context was the informal use of vested industrial interests (stakeholders) in the first part of the study and the use of public focus groups in the design stage of the second, although the feedback from the public proved problematic by diverging from economic assumptions (e.g. the expressed desire for community compensation unrelated to the individual), and neither process was formally reported. Thus, two broad approaches to combining deliberation and monetary valuation can be identified. The first regards monetary valuation as basically sound but being able to benefit from supplementary, and often informal, processes borrowing elements of deliberation. The second sees the use of deliberative approaches as a new method allowing the (collective or individual) production of a monetary valuation for environmental goods and services. Under the first approach a variety of alternatives exists, and monetary valuation may be either followed or preceded by some element of deliberation. Stakeholder participation, as mentioned above, may be employed to validate outcomes. The implication being that ex post deliberations can be used in some way to adjust valuation results or their presentation. Deliberative processes and environmental valuation may also be sequential, e.g. selecting a sub-sample of participants from a CVM survey for a citizens’ jury on the same environmental issue. Ex ante deliberation has been employed in designing CVM surveys with the use of focus groups to test the wording and respondents’ understanding of survey questions. Deliberation is then regarded as useful in providing insight into the processes by which respondents produce their WTA or WTP bid. This may be extended to allowing a deliberative process to determine the options or institutional context to be valued in the survey. The second approach is what Spash (2001) has termed deliberative monetary valuation (DMV) as advocated by, for example, Brown et al. (1995), Jacobs (1997), Ward (1999) and Kenyon et al. (2001). DMV is the use of formal deliberation concerning an environmental impact in order to express value in monetary terms for policy purposes, and more specifically as an input to CBA. For example, consider a proposal to build a new road through a wilderness area and so destroy the habitat of a number of rare or threatened species. A group of citizens would be selected and meet to discuss information about these environmental damages associated with the development. Known costs and benefits (discounted) would be presented, while those pertaining to environmental damages would be deliberated. The citizens would form a jury aiming to provide a monetary value for environmental damages which might be in terms of an individual WTA to allow the project to proceed. The result would then be incorporated into a net present value calculation to determine the viability of the project.
6
Clive L. Spash, Sigrid Stagl and Michael Getzner
The meaning of such values remains contentious, as they are mediated individual values. In order to address the increasingly evident fact that preferences are formed and that there is a lack of arbitrage in valuing environmental goods and services, economists are appealing to methods from political science. However, approaches such as citizens juries and contingent valuation, or more generally deliberative forums and monetary valuation, differ in fundamental ways. This has been pointed out by Niemeyer and Spash (2001) as involving the approach taken to theoretical factors (individual and social ontology, preference basis, rationality theory), practical factors (justification, framing, value representation, institutional setting), and political factors (manipulation, representation, social impact). The variation between approaches to each of these factors means the very concept of DMV is brought into question. In simplified terms, can DMV take the best of both monetary and deliberative methods as advocates hope or does it merely create a messy confusion as to the values being expressed? Clearly, research into valuation needs to span different disciplines such as social psychology, applied philosophy, political theory and economics. While interest in doing so seems to have increased, the quality of the applications can be questionable (Spash 2000b). Part of the problem is for economists to understand the requirements of other disciplines, while practitioners in those other disciplines must similarly have a good practical knowledge of what economists are trying to do. Thus, applying attitudinal measures from social psychology may prove insightful with respect to WTP, but only if researchers also understand welfare economics and CBA tools so that their work can address the economics literature. This is a challenging research agenda but one which, as several contributions in this volume show, is now being taken on at several different levels.
Contributions to the debate in this book The book is organised around three major drivers influencing the field of environmental valuation, namely attempts at: understanding the results from environmental CBA and the foundations of economic behaviour, taking multiple values into account, and exploring the role for inclusive deliberation in valuation processes. The chapters reflect the debates over method development which are on-going in the valuation community. Clearly, all methods have advantages and disadvantages and there is no pretence on the part of the editors that any one approach offers a panacea. What the reader will find is the presentation of a range of attempts to learn from different disciplines in order to achieve practical approaches which can improve our understanding and expression of environmental values. Economic values and the psychology of behaviour Getzner begins the book by exploring epistemological and other theoretical problems in understanding economic values with a specific focus upon the CVM. What might be regarded as specific validity issues are addressed (i.e. hypothetical
Exploring alternatives for environmental valuation
7
bias, framing of questions), but a different perspective is gained by placing these issues within the context which the social role respondents adopt when answering CVM surveys. The refusal to trade money for environmental change is raised as an important behaviour and related to the literature on lexicographic preferences (see for example Spash and Hanley 1995; Spash 1998; Spash et al. 2000). In addition, some of the literature on the supposed dichotomy of individuals as political citizens and economic consumers is discussed. This then provides the background to a range of variables for inclusion in bid curve analysis. The empirical test presented uses WTP for a nature preservation programme in Austria stated to be able to secure the survival of certain species and ecosystems. A convenience sample of 189 university students form the database. The results on answers consistent with lexicographic preferences are mixed. While 94 per cent of respondents claimed species protection should be undertaken regardless of the cost, only just over 7 per cent rejected WTP questions because they regarded species could not be valued in monetary terms. This implies the need for careful probing of interviewees in order to understand their responses (Spash 2000b). The explanatory power of the model is greatly enhanced by what can be regarded as non-economic variables or at least non-standard ones. These factors are interpreted by Getzner as showing the importance of social and institutional context. A strong influence of variables related to charitable giving is revealed, which supports claims by others that CVM fails to measure a trade price as is assumed by practitioners (see Spash 2000b). A clear implication of this type of work is the need to probe more deeply into the motives behind stated preference behaviour. Meyerhoff follows up on this challenge by focusing upon the importance of attitudes as explanatory motives for WTP to improve environmental quality. A key contention here has been the claim by some social psychologists that the results of CVM are merely poor measures of attitudes. One method of testing this hypothesis is to conduct bid curve analysis and probe which variables best explain WTP or WTA. Meyerhoff focuses on the difference between non-users and users with the expectation that non-users only express indirect or passive use values (sometimes erroneously termed non-use values) and that these are related to non-economic factors. Indeed, Westra is cited as showing the relationship of such values to attitudes. Meyerhoff uses WTP for protection of the nature and landscape of the German Wadden, a coastal wetlands ecosystem, against threats from climate change. A total of 1412 semi-random interviews were conducted face to face. The data analysis proceeds by excluding ‘protest bids’ which are zero bids for reasons regarded as inconsistent with placing a zero value on the environmental improvement, e.g. protesting against the use of taxes to achieve this. The practice is highly questionable as Meyerhoff notes but he proceeds to use follow-up questions to exclude 56 per cent of the sample. How this may have affected the results is unclear. Among the remaining sample a higher WTP is found amongst those accepting money as a measure of environmental value, although interestingly rejection of money as a measure does not mean all such respondents give a zero WTP. This is consistent with other findings where those rejecting an economic rationale for valuing the environment (e.g. rights based) have been found to have a
8
Clive L. Spash, Sigrid Stagl and Michael Getzner
positive WTP; although, in apparent contrast with the results here, WTP has also been found to be higher amongst such groups than amongst those favouring economic trade-off justifications for their WTP (Spash 2000a). A key finding by Meyerhoff is that non-users’ WTP is explained by attitudinal and economic variables while only attitudinal variables prove significant for the user group. This would seem to imply that users have a stronger non-economic value orientation. In discussing attitudes, Meyerhoff notes the difference between measuring general attitudes which are related to ethical beliefs, and basic values, and those attitudes which are specific to a behaviour. The latter category and correspondence between the level of attitude framing and behaviour are recommended if attitudes are to be expected to predict behavioural intentions. His measures of attitudes towards Nature and climate change fall closer to the former category in terms of their generality, i.e. they are related to but unspecific in terms of the WTP scenario being valued. The results show there are clearly strong statistical relationships between CVM results and underlying attitudes. What remains unclear is the extent to which these motives are driven by social norms and ethical beliefs and how far the type of environmental entities currently being subject to CVM surveys fall outside the economic realm of value. Mosler is also concerned with the mix of economic and non-economic motives for human behaviour. He pursues a model of behaviour which includes social and personal factors which he differentiates from those which are purely economic. However, his emphasis is on how the psychology of the individual can be modelled to explain social decisions rather than narrowly defined values. The agent based social simulation model he develops is explained carefully along with a series of definitions and feedback loops which aim to characterise human resource use and environmental impacts. Amongst the factors included as explanatory of human behaviour are attitudes, subjective norms, behavioural control, weighing costs and benefits, and sustainability motives (combining a value and norm orientation). The CBA component is regarded as encapsulating economic considerations via personal returns from a given action. The sustainability motive is strongest where an individual values the environment highly and the gap between actual and sustainable patterns of use is small (on the basis that a large gap is demotivating due to the difficulty of closing it). Mosler examines the conditions necessary for a collective reorientation towards environmentally sustainable behaviour. His simulation modelling shows the role which can be played by the environmentally friendly behaviour of some ‘pioneer’ individuals and how policy might be developed to increase the number of persons joining ranks with such pioneers. Sustainability requires such an increase to build its own momentum leading to a large-scale ‘turn-around’ of previous environmentally harmful patterns of behaviour. The aim of the simulation approach is to indicate means by which policy instruments, such as environmental campaigns, can be implemented most effectively. One aspect of the model is how people process information and whether they become engaged and find an argument persuasive. This has also been raised as an issue for CVM because central processing means deep evaluation while peripheral processing means being led by comparatively insignificant factors. Indeed, for example, ethical
Exploring alternatives for environmental valuation
9
beliefs have been cited as such a peripheral factor (Ajzen et al. 1996), but empirical evidence shows those entering central processing mode tend to rely on their ethical beliefs as well as scientific information (Spash 2002c). As Mosler notes, there is scope for the interaction of simulation modelling and empirical studies which would seem essential for increasing understanding of complex systems. Taking multiple values into account Multiple criteria analysis (MCA) was initially developed for production planning and methods were also applied in the field of financial management. The field has since expanded dramatically and a great number of algorithms are now available for weighting, comparing and combining information. Since they were developed with different problems in mind, researchers and policy-makers in the environmental field need some way to distinguish between the different algorithms. De Montis, De Toro, Droste, Omann and Stagl develop a framework for the comparison of MCA methods. They set out to identify a list of criteria referring to issues of particular relevance for the analysis of questions concerning sustainable development such as uncertainty, stakeholder involvement and non-substitutability between criteria. This framework is then employed to analyse and compare seven MCA methods frequently used for integrated assessment, namely: multiple attribute value theory (MAUT), analytic hierarchy process (AHP), evaluation matrix (Evamix), Electre III, Regime, Novel Approach to Imprecise Assessment and Decision Environments (NAIADE), and Multi-Objective-Programming (MOP)/ Goal Programming (GP). The results highlight the types of issues for which the respective methods are most suitable as well as problems that cannot be addressed by any MCA method. Thus, MAUT requires adherence to welfare theory, NAIADE and AHP are useful where there are value conflicts, MAUT and AHP are useful as learning tools; where constraints are important Electre III and MOP/GP are best, the latter also serve to address cases lacking discrete alternatives, and for ranking alternatives MAUT, AHP, Evamix and Regime are suitable. Each approach has its own context within which it can work best but the analyst is then responsible for selecting the most appropriate tool and justifying that choice. When first developed, MCA methods were meant as tools for optimising production processes and for identifying Pareto optimal outcomes of planning tasks. Since identifying an optimal solution is impossible for problems in complex systems, MCA methods have more recently been combined with processes of public and/or vested interest (stakeholder) participation. This allows the analyst to gain legitimacy for the analysis but also adds to MCA the aspect of a learning tool, i.e. allowing participants to explore the values and qualities of the problem. Proctor addresses the Comprehensive Regional Assessments of Australia’s forests which aimed to ensure sustainable management. Planners faced the task of integrating different forest values in order to designate areas as reserves. Proctor uses a case study, outside the official assessment, which shows how MCA can be applied to the problem. She makes use of official assessments, criteria and priorities placed on forest values by a stakeholder group which was involved in the assessment
10 Clive L. Spash, Sigrid Stagl and Michael Getzner of the Southern Forest Region of New South Wales. Both the preferred option and the disparity/agreement between group members’ evaluations are explored. Polarised opinions can thus be identified and discussions about compromises can be supported within the MCA framework. Proctor points also to new requirements for research processes of this kind, such as the need for a trained facilitator to aid interactive group decision-making. The outcome shows how values come to the fore in such processes. The two most favoured options reveal a choice which can be summarised as between environmental quality and employment. Sensitivity analysis can be used to show the robustness of the outcome and the failure of what might appear to be compromise options. This shows the way in which MCA can reveal value differences which inform policy. Stirling and Mayer are also concerned by the approach governments take to form regulations and implement legislation. Their aim is to explicitly address issues of uncertainty related to the regulation of environmental and health risks. The chapter explains the need for a set of precautionary approaches, and as a consequence, reveals the problems associated with narrow risk assessment techniques based on rational choice and probability theory. Stirling and Mayer develop a set of criteria by which any ‘regulatory appraisal’ (i.e. the way in which regulations are established) can be evaluated. The general approach is to select a single criteria for assessment and exclude much information relevant to, say, technological risks. Hence the criteria here concern the need for: greater humility, completeness, consideration of net benefits in different contexts, transparency, full engagement of interested and affected parties, mapping out different value judgements and framing assumptions, and allowing for divergent views by considering diverse option mixes. These requirements might sound like something of a wish list which is itself hard to implement but Stirling and Mayer then go on to show how such an appraisal can be performed. An area of considerable conflict in the UK has been the genetic modification of agricultural crops. Stirling and Mayer use this as a case study to apply a novel approach which they call multi-criteria mapping (MCM). Twelve individual high profile protagonists in the GM food debate were selected as participants. These individuals came from a variety of backgrounds, including academics and government advisers, environmental, consumer and religious organisations and representatives from the farming, food and biotechnology industries. Results are reported on each individual and under the categories of ‘academic’, ‘NGO’, ‘industry’ and ‘government’. Interviews were used to get each individual to add options, appraisal criteria, apply and scale the criteria against the options, and weight the criteria. In addition to the six ‘basic options’ a further 18 alternative agricultural strategies were identified; 117 appraisal criteria were defined by different participants. There are lessons for standard MCA as the authors state: ‘Participants adopt a variety of different “framing assumptions”, resulting in significant differences in the scores assigned by different participants under the same criteria. This has implications for conventional multi-criteria analysis, in which scoring is often conducted by a separate body of experts, with an assumption that different value judgements can be captured simply in the weightings.’ In
Exploring alternatives for environmental valuation
11
addition, the information gained on vested interest is often surprising: for example, the biotechnology industry representatives underemphasised the social, environmental and safety considerations which are prominent under all other perspectives. The approach provides specific information on the motives behind positions as well as allowing an overall appraisal of options. In the latter regard, the organic farming and integrated pest management options tend to perform significantly better overall. Participation, deliberation and valuation Aldred starts out with a comparison of the CVM with the CJ. The former is an individualistic preference based method while the latter is a community based form of social deliberation. The CVM and CJ answer to different institutional needs, cultural roles and social contexts and are based upon different explanations of rationality. This raises the question as to what extent they can in fact be compared. Aldred argues that an assessment of which consultation process is superior for policy formation can be undertaken on the basis of considering the added value of discussion. He specifies the characteristics of the policy process in terms of open and closed inputs (or task set) and outputs (or expected result). The former is principally classified according to control, while the latter according to agreement. The CVM has a closed input while the CJ is normally open but has the option of considering a closed input. Aldred argues that comparison requires he match the requirements so a CJ with closed input must be considered. In order to achieve the same for outputs he argues the complete decision process must be considered through to the final recommendation, i.e. a full CBA compared with the CJ best option. However, he does allow that these requirements only apply for direct comparison and that indirect comparison can also be made, although that is not the aim in his chapter. Aldred draws a distinction between the role of discussion versus deliberation in decision processes and prefers to use the former term. So his aim is to compare closed forms of CVM and CJs on the basis that they differ in term of the latter adding discussion prior to an aggregative decision, and his aim is to judge whether the content of decisions is better with that discussion. He goes on to show that arguments by participants in a CJ must appeal to widely shared principles and he discredits the oft-cited self-interested/selfish and strategic explanations for human behaviour in such contexts. He cites O’Neill (1993) on the Aristotelian view which encourages self-awareness on behalf of listeners so they recognise their inability to evaluate certain claims and instead concentrate on assessing the character of the speaker. In general, such an approach is applied when we judge a person’s motivation, their trustworthiness and their good judgement. CJs are then seen as having the potential to allow listeners to assess the credibility of speakers as they share arguments. Having negated the criticisms arising mainly from a rational choice theory perspective, Aldred finishes by exploring the implications of imposing a closed output on a CJ and hence moves into the area of DMV (discussed earlier in this chapter), which claims to combine the advantages of discussion with the merits of a valuation output. The concerns
12 Clive L. Spash, Sigrid Stagl and Michael Getzner raised here cover forcing a decision on the CJ, losing the trust of the jurors in the process, and whether quantitative or qualitative outputs are appropriate. In summary, where bounded rationality is likely to be severe, principles of public interest arise, or important private information is present, CJs appear more attractive, and while both CJs and the CVM have problems Aldred does not see DMV as the solution. He does, however, appeal for ‘a careful, empirical, case-bycase comparison’ to add to his abstract epistemic arguments. Kenyon and Hanley are concerned with the empirical application and comparison of all three methods (CJs, CVM, DMV) to the Ettrick Forest Floodplain Restoration Project in Scotland. The arguments are firmly based upon an economic model with, for example, CJs described as addressing ‘the information problem better than the CVM’, i.e. addressing the fact that preferences are formed and not merely informed during a valuation process and any information set is never ‘neutral’ (see Spash 2002c). The reason Kenyon and Hanley pursue DMV is because CJs ‘do not provide an economic estimate of the value of any particular project, nor whether it constitutes an efficient use of resources’. For the CVM a stratified sample of nine towns in the Borders Region of Scotland resulted in 336 responses to the survey including the request for a charitable donation. Of these, 29 per cent were classified as protest bids and removed from the data set and any further analysis. CJ participants were selected from the CVM questionnaire respondents. Kenyon and Hanley claim that the eleven jurors were selected ‘to be representative of the Borders population’, although what they represent is left unspecified. The jurors were concerned that they might need to set up a trust fund to cover underfunding (as implied by the CVM) but ‘after speaking with a member of the local community’ were reassured that the local community was fully involved in the project and the site would be well managed into the future. The CJ considered environmental and social elements important in judging the success of the project and in making management recommendations, but ‘did not seem to consider economic criteria important’. Selection for the DMV was on the basis of responses to a letter sent to 500 households. Two workshops were carried out in each of two towns in the Borders, giving a total of four workshops and 44 participants. The DMV started by administering the CVM, then took participants into small group (four to seven people) discussions on problems and management options, and finally asked them as individuals to answer some questions including whether they now wanted to revise their WTP and only 14 per cent did so. DMV initial bids included 5 per cent protest bids and 34 per cent ‘don’t know’ responses, and only two people moved from the latter after the discussion stage. An advantage of the DMV over the CVM was seen to be the information on positive and negative views of the project. When an aggregated WTP figure was presented to DMV participants there were three responses: the impossibility of putting a value on such a project, the poor economic situation meaning the need to spend money elsewhere, and wanting some other public fund to pay. Overall the authors conclude in favour of DMV as a middle path. James and Blamey concentrate on a DMV approach, which they describe as a CJ with the added task of determining societal WTP for a specified programme
Exploring alternatives for environmental valuation
13
involving environmental improvement. They open by addressing problems and issues relating to the theory and application of both the CVM and deliberative approaches. This aims to show the potential of a DMV approach. The empirical case study is the management of national parks in New South Wales, Australia. As in the chapter by Kenyon and Hanley, there was a concern that the participants be representative and in this case the criteria are given (gender, age, place of residence, ranking of the environment in relation to other social issues, occupation, income, income source and education). Such representation is a little strange in these studies as the only aim can be statistical significance but the sample sizes are so small as to make this irrelevant (for elaboration on the difference between political and statistical representation, see O’Neill 2001). There were thirteen participants (and one no show). The DMV panel was given three options developed by the researchers and a fourth developed by the researchers on the basis of answers given to a ‘straw preference poll’ (although what this involved remains unclear). Participants were limited to considering only these options. This approach can be seen as Aldred’s closed input and closed output case, making the DMV comparable (in those terms) to the CVM. After individual consideration of three options the panel was convened and the choices discussed. The aim of the researchers seems to have been for a consensus report and the panel gave a preferred option with qualifications covering the concerns of those who were initially against that option. Next, the valuation question was introduced as part of the fourth option. The task for the panel was to determine how high a park ‘levy’ would have to be before the NSW public would be no better off than under the status quo. That is, they were being asked to set the average annual WTP, rather than their own maximum WTP as in a CVM survey. Perhaps surprisingly, the panel was able to come up with two amounts and voted to decide which amount was to be recommended. The use of majority voting occurred at several stages in the discussions in order to close down dissent. In their conclusions the authors discuss a question which remains open: how to interpret the value obtained. The amount fails to relate to economic welfare theory and would seem hard to compare with other microeconomic welfare theoretic measures. At the same time, the thirteen panellists cannot claim to be statistically representative, which only leaves them with stating a political position as might be done more directly. The authors also raise several other issues including consensus formation, decision rules for polling or voting, equality of juror impact, and provision of information. Arzt picks up some of the same themes in addressing whether researchers should aim to reveal existing preferences or initiate and foster learning processes and thereby help form preferences. The contention is that environmental valuation is a subjective, cultural and contextual phenomenon. The approach for addressing policy needs here is to appeal to vested interest group representation in a framework termed an ‘interactive valuation process’, where individuals, in groups, construct values rather than express prior preferences. This approach is equated to a roundtable and cited as deriving from Ostrom’s (1998) ‘second generation models of rationality for social dilemma situations’ to understand when co-operation in small groups can take place. Much is made of the need to build trust over time
14 Clive L. Spash, Sigrid Stagl and Michael Getzner through repeated meetings, although this would seem to contrast significantly with the CJs used in other chapters of this volume. In this respect, the motives for behaviour described by Arzt are worth comparing with those discussed by Aldred. She also describes the roundtable as a consensus seeking exercise in which decisions are made without voting or other mechanisms, and this can be contrasted with the way in which James and Blamey ran their DMV panel. In this roundtable consensus seeking, some tools and techniques are advocated including MCA, visualisation techniques and professional moderation. The case study concerned a roundtable run in the Uckermark, Brandenburg county, region of northeast Germany with the topic of decentralising decisions about agri-environmental measures. Agri-environmental schemes are already run from the local government level (as opposed to, say, those in the UK) but decisions on targets are made by experts without consultation with farmers or conservationists. A community meeting was used to select the roundtable with 43 people out of 100 attending the meeting and 23 volunteering. Six meetings were held on a monthly basis with varying participation rates. The interdisciplinary research team from different institutions found their own problems before even starting the roundtable: ‘The different backgrounds of the project team made it difficult to find a general understanding of the definition of “value” and “learning processes” beforehand. The team could not create a common understanding about the institutional arrangement’. The aim had been to cover a range of five issues in sequence: reasons for the problem and expected longterm consequences, solutions including justifications, measures to achieve these solutions including monitoring, financing and action planning. However, after three meetings, the main interest of the stakeholders was still focused on measures and their financing. One farmer is reported as stating: ‘I would do anything, if I got enough cash.’ Attempts to discuss scientific uncertainties over a specific type of water body (Sölle) management, at the third meeting, also failed as the farmers/ land managers expected the expert scientists to tell them what was best to do. Only when the next (fourth) meeting followed a very structured approach to management of these water bodies was some headway achieved. The final two meetings switched to soil erosion: the fifth meeting was dominated by a soil expert and the sixth, where only three farmers attended, by the agricultural administration. The difficulty of running this process compared with the high expectations of the theoretical literature raises a range of good lessons for research and practice. Power is an important element and affected representation with small farmers unwilling to attend knowing that ‘consensus’ processes in the past were dominated by large cooperatives, and women being a minority and silent. This has obvious parallels with the ‘willingness to say’ problems as raised by O’Neill and discussed by Aldred, and this issue also arose directly with experts dominating some meetings. Representation problems also arose in a different form with one organisation alternating their representatives but with the two individuals holding very different opinions. The need for greater structure in the process might be met by use of an MCA type of process such as the MCM approach developed by Stirling and Mayer. However, there also may be some cultural differences
Exploring alternatives for environmental valuation
15
arising in this study (e.g. desire for structure and being told the best option, high esteem of scientists), so that how far the design features can be determined in any one country-specific study becomes a relevant research question. Indeed more generally the extent to which the benefits of participation and deliberation can be regarded as universal remains unanswered. Niemeyer shows how we might begin to address this last point with respect to CJs. Political science advocates of CJs have seen them as deliberative forums in which people can find truths and transform themselves away from more mundane daily preoccupations. Whether CJs are a process which can lead to the change of fundamental attitudes and preferences in relation to environmental policies is then a testable hypothesis. Niemeyer contends that preferences are shaped during deliberation by information, attitude and process. The chapter shows how to measure both changes to (rank ordered) preferences and the underlying subjectivity for those preferences (using Q method). The case study concerns a controversial ‘road’, the Bloomfield Track, constructed by politically influential developers through a rainforest in north Queensland, Australia, against the protests of conservationists. The area was later, after a change of regional government, designated the Wet Tropics World Heritage Area. What action should be taken with respect to the route remains a problem. Local and Aboriginal communities now use the track as a convenient means of communication while environmentalists regard it as a scar across a World Heritage site causing erosion and damage to forest and coastal reef ecosystems. A random-stratified sample of twelve people from the region was given the task of considering management options for the Track over a four-day period. Attitude information was obtained by surveying participants at the outset of deliberations, at the midway point, and immediately after proceedings. The combined data for all three Q sorts by the jurors produced four distinct factors: Preservation, Optimism, Pragmatism and Symbolism. Factorial analysis revealed four preference options: Close Road, Stabilize, Bituminize and Status Quo. This approach allows analysis of motive, and examples of correlations are Preservation with Close Road and Pragmatism with Status Quo. Preferences with regard to these options changed substantially during deliberation. Two jurors experienced a near complete reversal of their original orderings, and preferences among jurors converged, although they never reached consensus. Jurors were encouraged by the process to take a community perspective, rather than that of individual self interest. By the end of deliberation road closure had become the most preferred option. As Niemeyer notes, this CJ was a success in terms of the expectations from political theory but it is also only one example and others have had mixed results, as did Arzt in this volume. He concludes that the outcome of deliberative forums is highly sensitive to context and design, and a wider transformation of society towards deliberative democracy is required but that such institutional designs can help with that transformation.
16 Clive L. Spash, Sigrid Stagl and Michael Getzner
Drawing out some lessons There are significant gains to be made by further use of participatory approaches in environmental management, but maintaining the interest, involvement and commitment of civil society may prove a major challenge. Enthusiasm on behalf of researchers for new approaches needs to be qualified by awareness of the difficulties in applying techniques in different contexts and their potential for both failure and manipulation. At the same time consumer boycotts and support for non-violent direct action (e.g. anti-road demonstrations) do indicate public dissatisfaction with current institutions and a demand for greater say in policy, as opposed to an infrequent vote for a possible representative. The concern for the political legitimacy of government policy is particularly relevant in societies with low voter turn-out or where systems fail to represent minority parties. Both the UK and the USA, for example, have systems which increasingly seem inadequate as far as voters are concerned. This general problem takes on specific significance in the formation of environmental policy where the directly concerned political parties are small and lack representation via established institutions. Thus, for example, under the UK system, which has no proportional representation at the national level, a substantial minority of votes for the Green Party has in the past been unable to gain even one member in parliament. At the same time redressing the political imbalance through new institutions is no easy task. In this volume contributors show the difficulty in getting politically weak groups to voice their opinions in forums which are hoping to empower them. Various features of representative democracy have been raised by O’Neill (2001). In particular, environmental policy brings to the fore the question of representation for the silent voices of non-humans and future generations. Another aspect of representation is how information is delivered through the interaction of science and policy. The role of expert and ‘objective’ scientific information to inform policy on uncertain and complex environmental problems has been brought into question on issues ranging from nuclear power to genetically modified crops. That there is a problem with the traditional approach to environmental policy processes is apparent from the greater attention given in recent years to a range of alternative institutional approaches (Royal Commission on Environmental Pollution 1998). However, while several techniques have been increasingly deployed they also have their own limitations. The use of focus groups in the UK was seen as providing some additional legitimacy for policy, although now such an approach often receives popular derision as a marketing technique which has been overused. Consensus conferences have been employed in several European countries but consensus is itself a questionable goal when facing fundamental differences in perspective and values. There are problems with closing down a debate in an effort to achieve an artificially stable outcome described as consensus. Instead there is a need for mapping out the contours of the debate via an approach such as that suggested by Stirling and Mayer in this volume. The use of ‘stakeholders’ to represent vested interests directly in the decisionprocess has entered into both policy and research with the emphasis on addressing
Exploring alternatives for environmental valuation
17
the needs of ‘end-users’. Both CJs and the appeal to vested interests raise similar questions as to who should be represented and how they should be represented. CJs have become more common as well as being an increasing focus of research. Some environmental economists have also become increasingly concerned to employ such approaches to increase the legitimacy of their valuation work. The rise of DMV is a significant development as an indicator of the desire for greater legitimacy for the values being derived in economic assessments. The reasons driving this move are recognised inadequacies in the economic model of human behaviour. There is now far greater acceptance by economists that psychological motives are important and preferences are often constructed in response to research aiming to discover how people value the environment. The problem which then arises is how far the values being derived from DMV remain valid having been removed from their theoretical basis in economic welfare theory. If values as derived by Blamey and James are to be given weight in decision processes, there is a need for a new theoretical underpinning. At the same time, social psychology raises many insights which economists are only just beginning to discover. The work on CVM has exposed problems with standard approaches and data interpretation. Yet there is still a surprising readiness amongst researchers to throw away large parts of their samples for analytical and/or practical convenience. There are two examples in the current volume where protest bids are excluded, removing 29 per cent of the sample in one case and 56 per cent of the sample in another. Rather than excluding such information, researchers need to investigate and understand what individuals are telling them. Perhaps the technique is failing but with CVM what seems to be happening is that the economic model is being revealed as only a limited frame in which humans are prepared to operate. Some of the people may be homo œconomicus all of the time, all of the people may be homo œconomicus some of the time, but all of the people are never homo œconomicus all of the time. In environmental valuation, rejection of the economic motive for behaviour is common and this cannot be disregarded purely on pragmatic grounds. In the past, the counter to critiques of CBA has often been that ‘there is no option’ and that ‘because society uses a money metric, so must environmentalists’. The treasury departments of government are certainly often the strongest and lie behind many decisions, but there are also other branches of government and civil society, and different forums in which decision processes operate. This volume shows there are alternatives and these can be most rewarding in terms of gaining insight into value conflicts, while aiding and improving policy. There is a need to evaluate the opinions of experts as one perspective on the implications of any project, and set their role within a broader social context where a range of motives operate and different values are expressed with equal legitimacy. However, there is no pretence that alternatives to traditional environmental valuation lack their own problems and can be difficult to implement successfully. Clearly, more work is required which breaks down the boundaries between disciplines and more researchers are required who are prepared to venture out of the comfort zones of mono-disciplinarity into the challenging areas of interdisciplinarity. Ecological
18 Clive L. Spash, Sigrid Stagl and Michael Getzner economics is one such attempt at bringing together diverse perspectives and approaches.
Note 1
Revenue raising is a key objective of taxation rather than designing environmentally efficient and effective taxes. This means taxes which may be described as environmental in fact fail to address the pollution problem to which they are supposedly related. For example, a tax justified as reducing the Greenhouse Effect may select carbon dioxide emissions regardless of the other gases needing control and the tax may then relate only to industrial emissions from point sources, ignoring a whole range of non-point, household and public sources, or select only one specific fuel. Such decisions are taken on the basis of ease of administration and revenue-raising potential rather than their economic efficiency at controlling pollution and matching marginal social costs of environmental use.
References Ajzen, I., T.C. Brown and L.H. Rosenthal (1996) ‘Information bias in contingent valuation: effects of personal relevance, quality of information and motivational orientation’, Journal of Environmental Economics and Management 30(1): 43–57. Brown, T.C., G.L. Peterson et al. (1995) ‘The values jury to aid natural resource decisions’, Land Economics 71(2): 250–60. Burney, J. (2000) ‘Is valuing Nature contributing to policy development’, Environmental Values 9(4): 511–20. Department of the Environment, Transport and the Regions (1999) The Environmental Costs and Benefits of the Supply of Aggregates: Phase 2, London, Department of the Environment Transport and the Regions: 208. European Commission (1998) Aarhus Convention on Access to Information, Public Participation in Decision Making and Access to Justice in Environmental Matters, Brussels. Funtowicz, S.O. and J.R. Ravetz (1990) Uncertainty and Quality in Science for Policy, Dordrecht, Kluwer Academic Publishers. Hanley, N. and C.L. Spash (1993) Cost–Benefit Analysis and the Environment, Aldershot, Edward Elgar. Jacobs, M. (1997) ‘Environmental valuation, deliberative democracy and public decisionmaking institutions’, in J. Foster (ed.) Valuing Nature? Economics, Ethics and Environment, London, Routledge: 211–31. Kahneman, D. and A. Tversky (1979) ‘Prospect theory: an analysis of decision under risk’, Econometrica 47(2): 263–91. Kapp, K.W. (1950) The Social Costs of Private Enterprise, New York, Shocken. Kenyon, W., C. Nevin and N. Hanley (2001) ‘Citizens’ juries: an aid to environmental valuation?’ Environment and Planning C 19: 557–66. Knetsch, J.L. (1994) ‘Environmental valuation: some problems of wrong questions and misleading answers’, Environmental Values 3(4): 351–68. Knetsch, J.L. (1995) ‘Assumptions, behavioral findings, and policy analysis’, Journal of Policy Analysis and Management 14(1): 68–78. Knetsch, J.L. (1999) ‘Environmental valuations and standard theory: behavioural findings, context dependence, and implications’, in T. Tietenberg and H. Folmer (eds) The
Exploring alternatives for environmental valuation
19
International Yearbook of Environmental and Resource Economics 2000/2001, Cheltenham, Edward Elgar: 41. Knetsch, J.L. and J.A. Sinden (1984) ‘Willingness to pay and compensation demanded: experimental evidence of an unexpected disparity in measures of value’, Quarterly Journal of Economics 99(3): 507–21. Martinez-Alier, J., G. Munda and J. O’Neill (1998) ‘Weak comparability of values as a foundation for ecological economics’, Ecological Economics 26(3): 277–86. Niemeyer, S. and C.L. Spash (2001) ‘Environmental valuation analysis, public deliberation and their pragmatic syntheses: a critical appraisal’, Environment & Planning C: Government & Policy 19(4): 567–86. NOAA (1993) ‘Natural resource damage assessment under the Oil Pollution Act of 1990’, Federal Register 58(10): 4601–14. O’Connor, M., S. Funtowicz, F. Agliera-Klink, C.L. Spash and A. Holland (1998) Valuation for Sustainable Environments: The VALSE Project Full Final Report. Ispra, European Commission, Joint Research Centre: 395. O’Neill, J. (1993) Ecology, Policy and Politics: Human Well-Being and the Natural World, London, Routledge. O’Neill, J. (2001) ‘Representation’, Government and Policy 9(4). Ostrom, E. (1998) ‘A behavioral approach to the rational choice theory of collective action’, presidential address, American Political Science Association. American Political Science Review 92(1): 1–22. Royal Commission on Environmental Pollution (1998) Setting Environmental Standards, London, Her Majesty’s Stationery Office: 232. Spash, C.L. (1998) ‘Investigating individual motives for environmental action: lexicographic preferences, beliefs and attitudes’, in J. Lemons, L. Westra and R. Goodland (eds) Ecological Sustainability and Integrity: Concepts and Approaches, Dordrecht, Kluwer Academic. Spash, C.L. (2000a) ‘Ecosystems, contingent valuation and ethics: the case of wetlands recreation’, Ecological Economics 34(2): 195–215. Spash, C.L. (2000b) ‘Ethical motives and charitable contributions in contingent valuation: empirical evidence from social psychology and economics’, Environmental Values 9(4): 453–79. Spash, C.L. (2000c) ‘Multiple value expression in contingent valuation: economics and ethics’, Environmental Science & Technology 34(8): 1433–8. Spash, C.L. (2001) ‘Deliberative Monetary Valuation’. 5th Nordic Environmental Research Conference, University of Aarhus, Denmark. Spash, C.L. (2002a) ‘Empirical signs of ethical concern in economic valuation of the environment’, in D.W. Bromley and J. Paavola (eds) Economics, Ethics, and Environmental Policy: Contested Choices, Oxford, Blackwell: 205–21. Spash, C.L. (2002b) Greenhouse Economics: Value and Ethics, London, Routledge. Spash, C.L. (2002c) ‘Informing and forming preferences in environmental valuation: coral reef biodiversity’, Journal of Economic Psychology 23(5): 665–87. Spash, C.L. and C. Carter (2002) ‘Environmental valuation methods in rural resource management’, in F. Brouwer and J. van der Straaten (eds) Nature and Agricultural Policy in the European Union: New Perspectives on Policies that Shape the European Countryside, Cheltenham, Edward Elgar: 88–114. Spash, C.L. and N. Hanley (1995) ‘Preferences, information and biodiversity preservation’, Ecological Economics 12(3): 191–208. Spash, C.L. and I.A. Simpson (1994) ‘Utilitarian and rights-based alternatives for protecting Sites of Special Scientific Interest’, Journal of Agricultural Economics 45(1): 15–26.
20 Clive L. Spash, Sigrid Stagl and Michael Getzner Spash, C.L., J. van der Werff ten Bosch, S. Westmacott and J. Ruitenbeek (2000) ‘Lexicographic preferences and the contingent valuation of coral reef biodiversity in Curaçao and Jamaica’, in K. Gustavson, R.M. Huber and J. Ruitenbeek (eds) Integrated Coastal Zone Management of Coral Reefs: Decision Support Modeling, Washington, DC, World Bank: 97–118. Ward, H. (1999) ‘Citizens’ juries and valuing the environment: a proposal’, Environmental Politics 8(2): 75–96.
A framework for valuing nature
Part I
Extending the environmental valuation approach
21
22 Michael Getzner
A framework for valuing nature
2
23
A framework for valuing nature Regional biodiversity Michael Getzner
Introduction The contingent valuation method (CVM) seems to divide the economics profession. While there might be higher agreement among a core of (neoclassical) environmental economists on the theoretical foundations of valuing public goods with that specific method, applying the CVM is an important point of scientific discourse. Ecological economists not only doubt empirical applications of the CVM in many respects, but also question its welfare theoretic foundations and more generally those of cost–benefit analysis (CBA). A crucial point in that debate is the understanding of markets as social institutions, and the behaviour of individuals as embedded in a specific social and institutional context. An important element in that debate is also the critique of the concept of Homo oeconomicus who is assumed to have well-behaved, stable preferences, which are ‘home-grown’ and can be elicited by means of CVM surveys – provided several guidelines are followed (e.g. Arrow et al. 1993; Department of the Interior 1994). An aggregation of these individual values is said to point to the (societal) ‘value’ of a natural good, and compared with the costs of protecting the respective natural good, it is assumed that society’s welfare is in principle not much more than the sum of individual welfare aggregated in a social welfare function. Underlying preconditions for this approach are the notion that individuals are in principle willing to trade money for natural goods. That means that even in the case when individuals hold ethical beliefs or commitments often associated with natural goods, they can be ‘bribed’ to give up their beliefs for money. It has to be assumed that individuals who state a willingness to pay (WTP) base their decisions on similar foundations and considerations regardless of the nature and the characteristics of the good in question. Thus, using the CVM in a CBA for societal decisions assumes that individuals do not take different viewpoints when valuing private or public (natural) goods because contingent valuation puts individuals in a situation where they decide on public goods in a market-like setting. Another element in this debate is whether the hypothetical bias can have such an influence that results of CVM studies are fundamentally flawed. The main argument is that people behave differently when actually forced to ‘realize’ their stated WTP.
24 Michael Getzner The ‘standard’ and ‘extended’ model of valuation Figure 2.1 shows the ‘standard’ model of valuation and an ‘extended’ model that includes a number of additionally influential factors regarding an individual’s stated behaviour (stated WTP). Normally, studies explaining WTP in a bid function try to relate a number of individual factors (socio-economic characteristics like age and income, knowledge, perception of environmental quality) to respondents’ WTP bids.1 The explanatory power of such models is often very low (Mitchell and Carson 1989, argue that models should at least explain 15 per cent of the dependent variable’s variation). A low explanatory power might indicate omitted variables bias indicating that those individual variables chosen to explain stated behaviour are not the variables which fully influence a respondent’s stated WTP. Following that argument, other variables might be able to reasonably explain stated behaviour with a higher degree of accuracy. Recently, a number of authors have stressed the importance of the social and institutional context of decisions, especially when these decisions refer to public goods (e.g. Hodgson 1997; Hayden 1993; see with reference to bathing water quality: Georgiou et al. 1998). An extension of the ‘standard’ model of valuation2 to account for additional factors is presented in the lower part of Figure 2.1. First, extending individual factors may play a vital part in a valuation survey when respondents are not willing to trade natural goods for money (lexicographic preferences, see this section). The institutional context of the survey is on the one hand, of course, set up by the payment vehicle and the valuation questions themselves.3 On the other hand, respondents’ beliefs about and perceptions of the CVM instrument may alter their behaviour (next section). The social context comprises both beliefs about the referendum and about assumptions on others’ contributions, and the consumer–citizen dichotomy (third section). Figure 2.1 also shows the labels of variables that operationalize the different additional factors (to be used in the econometric model in the empirical part of this chapter). The central questions for which answers are sought by means of the empirical survey in the third section are: how important is the notion of lexicographic preferences and what is the effect of the respondents’ beliefs about institutional and social context variables in explaining WTP bids for a regional biodiversity programme in Carinthia (Austria)? The structure of this chapter is as follows. First, the elements of the ‘extended’ model of valuation are briefly discussed, presenting a focused literature review of the specific issues which build the basis of the empirical study presented in the second part of the chapter. Second, results of a valuation study designed to include ‘extended’ factors into the bid function are presented.
The extended context of valuation: some important issues Lexicographic preferences and the willingness to exchange When it comes to valuing biodiversity4 in economic (monetary) terms, the issue of substitutability is an important aspect: economic valuation becomes much more
Elements of the institutional context • Individual beliefs on setting and instrument and epistemological issues (e.g. ‘Protest’, ‘Influence’) • Hypothetical referendum (‘Cheap’)
Reasoning and processing • Bounded rationality • Risk aversion
Note: For a detailed description of variables see Table 2.3 on page 38.
Figure 2.1 The ‘standard’ and ‘extended’ model of valuation
Additional individual factors • Lexicographic preferences
Individual factors • Socioeconomic characteristics (e.g. ‘Age’, ‘Gender’, ‘Income’, ‘Children’) • Knowledge, perceptions and behaviour (e.g. ‘Biodiversity’, ‘Member’, ‘Donations’)
Elements of the social context • Individual beliefs about the context of the referendum, personal contribution above/ below average (‘Referendum’, ‘Average’, ‘Majority’) • Attitudes to public goods (‘Public’, ‘Minority’) • Consumer vs citizen (‘Con_cit’)
Extending the model
The ‘standard ’ model of explaining WTP bids
Stated behaviour • WTP in contingent markets • WTA • Other stated preferences
A framework for valuing nature 25
26 Michael Getzner simple if the natural good ‘biodiversity’ can simply be exchanged for man-made capital (private goods, money). Up to now there has been much discussion of the possibilities of substituting environmental goods or single characteristics of these goods such as biodiversity.5 On the one hand, ecologists argue that ‘[b]iodiversity is a resource for which there is absolutely no substitute; its loss is irreversible on any time scale of interest to society’ (Ehrlich and Ehrlich 1992: 225). In contrast to this argument, Solow (1993: 181) finds that history tells us an important fact, namely, that goods and services can be substituted for one another. … There is no specific object that the goal of sustainability, the obligation of sustainability, requires us to leave untouched. … Sustainability doesn’t require … that any particular species of fish or any particular tract of forest be preserved’ (emphasis in the original). While there might be different approaches to the substitutability issue in natural science, it is crucially important to separate findings of natural science from viewpoints (perceptions) and preferences held by respondents in a CVM survey.6 The CVM is part of a social science research method focusing on people’s (as economic subjects) preferences and perceptions in a very specific manner (Luhmann 1986). Elicitation of monetary values in contingent valuation surveys requires welldefined utility functions on the good in question as well as fulfilment of the basic welfare postulates of CBA. According to that foundation there must exist a utility function Ui = U(Xj, E) which includes Xj (a vector of marketed, private goods) and E (a vector of public goods, e.g. biodiversity). The WTP of an individual i to prevent a decrease from Ek1 to Ek2 is given by WTPi = e(pi, Ek1, U1) – e(pi, Ek2, U1) with e(.) being the respective expenditure function (e.g. Spash and Hanley 1995: 193). U1 denotes the utility of i before the quality change of Ek took place. However, in the context of environmental valuation, respondents often exhibit behaviour corresponding to the opinion ‘that some preferences are superior to others’ (Edwards 1992: 121). Such preferences are often formed on the basis of ethical judgements or commitments where a ‘right’ or a ‘wrong’ exists, but where a willingness to trade at the margin does not exist. Such preferences are labelled ‘non-compensatory’ preferences (Lockwood 1997; Gowdy 1996; Hanley et al. 1995). Such preferences can be subsumed under the broader concept of lexicographic preferences, which have quite a long history (e.g. Jeffrey 1974; Sen 1977; Menger 1871; Peukert 1998; de Dios 1995; Encarnación 1990; Stork and Viaene 1992). The neoclassical model of exchange rests on the assumption of the willingness to trade (willingness to exchange, ‘model of pure exchange’; Gowdy and Olson 1994: 162). A conceivable trade-off, which is nothing more than the well-known indifference curve between goods, must exist in the mind of respondents being asked a CVM question.7 While the ‘axiomatic development of economic utility theory does not a priori exclude any motive or any good from an agent’s utility function’ (Carson 1998: 18; Johansson 1992; Aldred 1994), the willingness to trade is at the core of contingent valuation. Lexicographic
A framework for valuing nature
27
preferences imply a hierarchical ordering of wants and therefore unwillingness to trade a specific good at the margin (e.g. environmental goods such as biodiversity) for private consumer goods. There is some empirical evidence from psychological research indicating that decisions are often made by lexicographic ordering (ranking) rather than by matching costs and benefits as usually assumed by economists. While observed preferences are actually constructed in the elicitation process in a context-sensitive way, the prominence hypothesis is often strengthened by empirical research (Tversky et al. 1988). That hypothesis holds that choice is expected to be more lexicographic than trading (matching). Individual decisions on biodiversity or on the survival of single species can be considered as a typical form of choice (no matching or tradeoffs will be made) if preferences are non-compensatory as outlined above (e.g. Spash and Hanley 1995). Interestingly, the critique of CBA made by Boulding (1969) in part comprises what nowadays could be called lexicographic preferences. While Boulding in principle acknowledges the role and importance of CBA in the evaluation of social choice and even social institutions, he admits that two fundamental problems arise out of the concept of CBA: one which he rejects is the inclusion of ‘benevolence’ and ‘malevolence’ in CBA (selfishness, altruism) which can in principle be included in interdependent utility functions. What might be ‘fundamental and harder to repulse’ is the notion of a ‘heroic ethic’ (Boulding 1969: 9), which cannot be included in CBA because a typical trade-off (and willingness to trade) does not exist. While Boulding only sees three main fields of this ‘heroic ethic’ (the military, religion, sports), he admits that the picture of economic man painted by economists is a caricature because humans also hold ethical beliefs. In those cases, trade is thought to be ‘somehow dirty’, and the ‘principle of prostitution’ would apply ‘to virtually all areas of human life’ (Boulding 1969: 10). Ethical commitments and rights-based beliefs towards the environment and other species are often seen as origins of lexicographic preferences (Common et al. 1997; Boyce et al. 1992).8 Sagoff puts such preferences in the light of Kantian ethics (Sagoff 1998). Moreover, Vatn and Bromley write: ‘Commitment and moral judgements are concepts often attached to those domains where life, quality of life, and personal integrity are at stake. These are areas where social norms restrict or reject the commodity fiction’ (Vatn and Bromley 1994: 135). Recently, some empirical evidence has been collected on the extent to which respondents in contingent valuation studies show lexicographic preferences.9 Stevens et al. (1991: 398) concluded that, when valuing wildlife, a significant proportion of respondents gave answers either consistent with lexicographic preference orderings or inconsistent with both neoclassical and lexicographic models of behaviour. Spash and Hanley (1995) find – depending on the base line and definition of lexicographic preferences – a large proportion of respondents (about one quarter) exhibiting rights-based beliefs on which lexicographic preferences may be founded. In contrast, Lockwood (1998: 79), constructing preference maps, shows that only a small percentage of participants had lexicographic preferences.10
28 Michael Getzner The institutional context of CVM surveys Hypothetical bias is considered as a major threat to the reliability and validity of the concept of stated preferences in constructed hypothetical markets in which respondents are invited to express their values for different states of the environment (see recent reviews of Schulze et al. 1996: 97ff.; Jakobsson and Dragun 1996: 84ff.; Getzner 2000). However, there is mixed evidence on the question to what extent people behave differently in hypothetical situations compared with real economic commitments. Critics argue that even if referenda are made as realistic as possible (e.g. by conservative wording of the questionnaire or incentive-compatible auction designs), ‘there does not appear to be a close substitute for The Real Thing’ (Cummings et al. 1995a; see also the discussion in Blackburn et al. 1994, who deliver methods for ‘correcting’ results from surveys to account for hypothetical bias if the distribution of error due to the bias is systematic). Cummings et al. (1995b) and Harrison (1996) report a number of approaches to reduce the hypothetical bias in CVM surveys by directly addressing this problem in the entire survey situation. They designed experiments that compared a real referendum involving subjects making real economic commitments with two hypothetical referenda in which hypothetical bias was taken into account by presenting adequate information on hypothetical bias to respondents. They called their experiments ‘light cheap talk’ and ‘heavy cheap talk’ which presented information to respondents in various densities, e.g. by presenting written and oral text as well as charts (‘cheap talk’ refers to additional information which can be provided to respondents without high cost and effort). The aim of including cheap talk information in the survey situation is to make respondents directly aware of the hypothetical situation in which they act and in which the whole valuation procedures takes place. Cummings et al. (1995b) report that – compared with a ‘normal’ CVM questionnaire without any additional information on the possible hypothetical bias of such surveys – cheap talk can reduce WTP (thus hypothetical bias) by 21 per cent compared with a reduction of WTP of 23 per cent in an actual (real) referendum. They conclude that ‘there exists one set of words which will result in a hypothetical survey generating the same results as a comparable survey eliciting real economic commitments’ (Cummings et al. 1995b: 13). These results indicate that the embedding of CVM questions in a realistic context is crucial to the reliability of CVM results. In the empirical part of this chapter (next section), the possible differences of a ‘cheap talk’ sample compared with an ordinary sample are statistically explored (Harrison 1996). As Boulding (1969: 3) noted a while ago, economics as a science tries to obtain knowledge about a system by changing its inputs and outputs of information, but these inputs and outputs will change the system itself, and under some circumstances they may change it radically. […] [The] generalised Heisenberg principle predominates because knowledge of the social sciences is an essential part of the social sciences itself. […] As science develops it no longer merely investigates the world; it creates the world which it is investigating.
A framework for valuing nature
29
Boulding clearly makes this point when he describes the epistemological problems arising out of a science that examines those facts entirely produced by it. In our context, respondents in a CVM survey might never have thought about valuing natural goods in a monetary way. Economics in this sense extends the fields of research to social relations that have not been considered as economically significant by those being asked a valuation question. CVM questions may produce values which were not formed before or which the respondent was not aware of. Such questions might thus provide incentives for formulating preferences rather than merely eliciting existing and stable preferences that were formed in the past. Applying Boulding’s critical comments on CVM surveys, one may feel curious about the attempts to elicit ‘home-grown values’ in CVM surveys. It is often assumed that respondents a priori have ‘home-grown values’ for biodiversity in a monetary form. It can be doubted whether respondents taking part in a CVM survey have generally ever thought about valuing biodiversity in a monetary form (Bishop et al. 1983). Respondents therefore construct their value statements during the CVM interview (Goodman et al. 1999). The very form of valuation questions – which are worded by the researcher with the theory of Homo oeconomicus in the background – pushes respondents into a certain role (see next section). These theoretical problems have often been addressed. For instance, in the context of strategic behaviour which has been considered as crucial for valuing public goods due to the supposed free-rider behaviour, it turns out that many assumptions made by economists and expressed by valuation questions do not meet respondents’ actual attitudes, perceptions and behaviour (see e.g. Burgess et al. 1998; Marwell and Ames 1981). There are a number of attempts to overcome possibly serious problems with value construction in CVM surveys. Lazo et al. (1992) present a methodology for CVM questionnaires in order to design a shortened version, which can efficiently be administered to a large sample without loss of information. Thus, value construction is not considered as a major problem of the CVM methodology by some researchers but something that needs to be taken into account when designing surveys (McDaniels and Roessler 1998). In the case of CVM surveys eliciting monetary values for natural goods like biodiversity, it is of great interest to determine to what extent respondents are used to, and influenced by, such questions. Besides the fact that some respondents merely dislike valuation questions, the empirical results presented below concentrate on the influence of CVM questions (surveys) on respondents’ perceptions of natural goods as goods to be further protected. If the influence of the valuation question and information presented is overwhelming (e.g. if reality is solely constructed in the valuation process), this can nevertheless be a major problem for CVM studies (see discussions in Hanley et al. 1995; Holm-Müller et al. 1991; Bishop and Welsh 1992). Closely related to the epistemological questions dealt with above, one can be curious if economists or economics students are ‘different’, i.e. if they perceive valuation questions in ways different to those of the general public, or if they express more or less solidarity or altruism than students of other disciplines (see the discussion in the Journal of Economic Perspectives, starting with Carter and Irons
30 Michael Getzner 1991). In the study described in this chapter, economics students and students of other disciplines are compared in order to examine the influence of studying economics (and business administration) on the amount of WTP bids. Eventual differences are taken as an auxiliary measure for the possibly influential effect of learning the economic view. The social context: ‘consumer’ vs. ‘citizen’ As mentioned above, lexicographic preferences can constitute an important argument when it comes to a monetary valuation of natural goods like biodiversity. One reason for such preferences may lie in the individual view that there is only one ‘right’ or ‘wrong’. However, the divergence between two roles individuals play when valuing public goods is more a social than an individualistic argument. The utilitarian viewpoint that – in the case of biodiversity – the result of an action is the adequate measure might be weakened by the social context11 in which such a valuation takes place (see below). The social role individuals play when asked for their WTP refers to their role as ‘consumer’ exchanging consumption for a higher environmental quality. The hypothesis put forward here is that when deciding the appropriate amounts of public goods (like biodiversity), individuals adopt a different social role, namely the one of a ‘citizen’. In other words, it is hypothesized that private households (respondents in a survey) act differently in the realm of the market than in a situation where societal affairs (e.g. public goods) are at stake (one of the first authors trying to clarify this argument was Sagoff 1988).12 That does not mean that economic considerations are completely left out in societal decisions. Such behaviour of respondents as ‘citizens’ might not meet Becker’s claim that ‘all human behaviour can be viewed as involving participants who maximize their utility from a stable set of preferences and accumulate an optimal amount of information’ (Becker 1976: 14). If we take Becker’s statement, it can be considered as being tautological. There is, of course, no human behaviour that is not maximizing the individual’s utility as defined by neoclassical theory. Even suicide can be thought of being the ‘best’ alternative for the person committing suicide. Altruism can also fit perfectly into the model of economic behaviour as one’s utility increases when another individual’s utility increases (for a critical view and arguments for a multiple-utility conception, see Etzioni 1986). Furthermore, rational behaviour can be considered as a constituting element of humans. However, it is the opinion of the author that such argumentation does not contribute very much to the understanding of economic and social behaviour. Economic reasoning and rationality might play an important role in nearly all decisions, be they ‘private’ or ‘public’, ‘individualistic/egoistic’ or ‘altruistic’. Economic arguments are often only implicit in individual decisions. It is the task of economics to clarify those preferences particularly in the context of environmental valuation. It is not put forward here that ‘altruistic citizens’ do not behave rationally (they can be thought of having a societal goal that should be achieved with least cost). It is the author’s understanding that discussing different approaches or explanations can be more fruitful than trying
A framework for valuing nature
31
to find a common theory to all kinds of human behaviour.13 The main argument of the dichotomy is that ‘citizens’ do not maximize their individual utility in a narrow sense (seeking personal advantages) but include broader societal arguments in their decision (see also Frey 1997).14 Besides different roles of respondents, some problems with respondents’ role differences might also be attributable to limited abilities to process information (‘bounded rationality’; Conlisk 1996). Bounded rationality in the sense discussed here could explain why respondents do not answer WTP questions in a way that would maximize their individual utility. Answers or behaviour departing from the individual utility maximization assumption can nevertheless be rational to the extent that citizens answer differently compared with consumers given a different context and different points of view and reference. Schkade and Payne (1994) offer an interesting focus on the social context. They let respondents ‘think aloud’ when making their WTP statement. Respondents primarily considered their ability to pay, referring to the respondent’s economic situation or to the payment vehicle. Some additional considerations can be interpreted as including elements of respondents acting as citizens, but no connection has been made to the sole benefits of society or its future development.15 Some evidence from psychological research stresses the argument that CVM questions mainly focus on the consumer role of the respondent. Surveys might include a tendency ‘to emphasize hedonic (self-interest) rather than social values by asking respondents for their personal opinions’ (Fischhoff 1991: 843). ‘If one wants to predict how people will behave in situations presenting a particular perspective, then one should elicit their values in ways evoking that perspective’ (Fischhoff 1991: 843). Boyce et al. (1992: 1367) make a similar argument, when they write that ‘WTP only measures intrinsic value to the extent that consumers accept moral responsibility in a WTP context’ (emphasis added). A number of publications appearing since Sagoff ’s book have discussed this issue from several viewpoints. Bingham et al. (1995: 76) argue that it is important ‘to understand the various roles that individuals play and how these roles affect the notion of “valuing” something. … [Individuals] value ecosystems differently when expressing personal values and when serving in some advisory or decisionmaking role for a public institution. People also reason differently when acting as decision makers or advisors … than they do as private agents’ (also Blamey 1996; Kosz 1997; Cameron 1997; Stevens et al. 1994; Common et al. 1993; Fischhoff and Furby 1988; Common and Perrings 1992). The central point of the divergence between the social roles individuals take is that contingent valuation is a concept, which derives out of a (normative) political theory founded on a distinct model of private choice in markets. As Jacobs (1997: 213f.) puts it: ‘Using these valuation institutions is essentially to treat choices about public goods as if they were choices about private ones. … The principal difference concerns the frame of reference people employ when making choices about public goods. … They are seen as having value to society over and above the value they give to individuals.’ It is this distinctive framework, which might not be appropriate because in reality decisions about environmental quality are indeed made by
32 Michael Getzner individuals (in elections or referenda within a democratic process; Bingham et al. 1995: 77). This argument additionally involves the idea that not only the mere outcome of a decision regarding societal problems matters but also how these results are achieved. In this sense, environmental valuation via CVM is – besides the utilitarian foundations – based on a teleological view of decision-making. Public goods require public discourse (the discourse ethics approach proposed by O’Hara 1996). It can be very important for those affected by a public project to learn how a decision is being made.16 There is some recent literature on procedural rationality. Citizens might in the first instance not be interested in the result of a public decision as such, but rather concentrate on the process of public decision making (e.g. Toman 1999; Holland 1997; Burgess et al. 1998; O’Hara 1996; Getzner 1999).
Empirical results17 Explaining WTP-bids: incorporating the social context The empirical models presented below are a logit model explaining respondent’s principal WTP, and a regression model on the respondents’ WTP amounts estimated by OLS. Both include several dummy variables to take context variables (see Figure 2.1) into account. Thus the model does not merely consider the usual socio-economic variables, but is extended by additional individual factors and by the institutional and social context. The models estimated can be described in analytical terms as Pr (WTPi > 0 ) = f (Si , Pi , I i , Ci ) and WTPi = f (Si , Pi , I i , Ci ),
(3.1a) (3.1b)
where Pr(WTPi > 0) is the probability that respondent i exhibits some positive WTP. WTPi refers to respondent i’s WTP bid (in Austrian schillings). Four groups of variables are included in the analysis. The first group S describes the socio-economic characteristics of respondent i (e.g. income, gender, age, job); P denotes respondent i ’s preferences and behaviour regarding natural goods in general (e.g. knowledge concerning biodiversity, donations to environmental organizations, memberships). I is a set of institutional context variables, which describe not only ‘objective’ arguments (e.g. whether hypothetical bias has been explicitly mentioned in the survey by ‘cheap talk’) but also respondent’s beliefs about the CVM setting. Finally, C comprises arguments regarding ‘subjective’ perceptions and citizen’s values about the social context of the survey. These variables include the respondent’s opinions about the outcome of the referendum and the consumer–citizen dichotomy. The WTP function increases or decreases in arguments18 (for a complete list of variables, see Table 2.3).
A framework for valuing nature
33
Design of the survey The survey for testing the relevance of extending the ‘standard’ valuation model took place in December 1997 and January 1998 at the University of Klagenfurt (Austria). Subjects were mainly undergraduate students of business administration as well as sociology, English and French language. The surveys were held in the classes; about 20 to 25 minutes were necessary to complete the questionnaire (sample size 189). The ‘design guidelines’ of Arrow et al. (1993) regarding the appropriate design and setting of the CVM survey were followed as closely as possible. After a number of pre-tests (in classes and with focus groups), the survey was finally structured as follows. After filling out a number of questions regarding socio-economic variables (age, gender, studies/courses enrolled, income, children), a set of indicators dealt with the respondent’s attitudes towards the environment (membership of environmental organizations, preferred activities in recreational areas, e.g. sports) and with their knowledge of and familiarity with the concept of biodiversity (using open-ended as well as closed questions). After these introductory ‘warming-up’ questions, the sample was divided into two groups: one group went straight on answering the following valuation questions without any further explanation from the instructor. The other group was asked to take a short break and listen to the instructor’s short presentation of the problems associated with the hypothetical bias of such CVM surveys. The next set of questions asked for the respondent’s WTP for a nature protection programme in Carinthia.19 These open-ended questions were worded as conservatively as possible,20 including references to the respondent’s budget constraint. The payment vehicle was an earmarked contribution to a nature protection fund. The payment motives as well as reasons for protest or zero bids were asked for. Additionally, a question on the respondent’s vote on a public referendum regarding the introduction of an earmarked nature protection tax in Carinthia was included. A final set of questions tried to reveal respondents’ attitudes comprising the ‘extended’ WTP model, asking for lexicographic preferences as well as the abovediscussed distinction between social role and epistemological questions (results for the most important questions of these survey sections, including the wording of the questions, are presented below). Descriptive overview of the empirical results The first questions dealt with respondents’ familiarity with the concept of biodiversity. It is often asserted in the literature that respondents should be familiar with and informed about the good to be valued (Mitchell and Carson 1989). It turned out that the vast majority of respondents had never heard about the concept of biodiversity. Over 75 per cent of respondents had never heard of this concept, only 5 per cent of respondents were at least familiar with it.21 Many interpreted the word ‘biodiversity’ in open-ended questions in the correct sense, meaning
34 Michael Getzner diversity of species and ecosystems.22 However, the meaning of ‘biodiversity’ was explained thereafter in the questionnaire. Quite a large number of respondents were in principle prepared to contribute to a Carinthian nature protection programme securing the survival of certain species and ecosystems (written information on the content of this programme was provided in the survey). Out of 189 respondents, 148 (78.3 per cent) were willing to pay a personal contribution in the form of an earmarked tax that would flow into a fund solely committed to financing the programme (16 per cent of those refusing to pay rejected the entire valuation question). Interestingly, 62.4 per cent of all respondents stated that such a programme should not merely be financed out of public budgets, but also by private contributions. Those who refused to pay mainly thought that inefficiencies in the public sector (e.g. bureaucracy) should first be cut significantly before citizens should contribute with personal donations. Respondents’ mean WTP bids in the survey amounted to 509 ATS (Austrian schillings) per person as an annual contribution to the above-mentioned fund providing financial resources for implementing the Carinthian nature protection programme. The payment is limited to a time span of 5 years (this is the time span when the implementation costs of the whole programme have to be financed). The standard deviation was 964 ATS, and the median was 250 ATS (all numbers were calculated without truncation as some of the high bids proved to be ‘reasonable’ in the sense that respondents bidding so high indeed had the purchasing power to do so). Respondents’ main motive to state a WTP was the existence motive which has been described as comprising a rights-based protection motive in the questionnaire. A share of 47.3 per cent of respondents marked the existence motive as the main motive to state a WTP; 45.9 per cent voted for the bequest motive, the rest thought about the option value of natural resources. A majority of respondents (about 60 per cent) would vote for a (mandatory) Carinthian nature protection tax amounting to 160 ATS, which is lower than the open-ended mean and median WTP elicited in prior questions. Forty per cent of respondents would vote against the proposition. Furthermore, 32.5 per cent of respondents thought that their WTP might lie above the sample mean, while 67 per cent were of the opinion that their WTP was at or below the average (a similar question regarding the median was left out of the questionnaire due to space restrictions). A large proportion, 66 per cent of respondents, showed motives that can be interpreted as being altruistic. They would donate a personal contribution to the Carinthian nature protection programme regardless of the outcome of the referendum; only 7.5 per cent would donate nothing, 5.7 per cent would lower their original donation, and around 20 per cent would await the forming of a new majority. That means that a majority of respondents would be willing to contribute to a nature protection fund even if not forced to do so. Students as subjects seem to be a group where nature and in particular species protection is a high priority. Regarding lexicographic preferences, nearly all
A framework for valuing nature
35
respondents stated that species should be protected regardless of the cost of protection; apparently they hold rights-based beliefs, as the wording of the questions and the appropriate answers could suggest (around 94 per cent of respondents believed that species should be protected regardless of the economic costs associated with it). The wording of this question is consistent with other surveys (e.g. Stevens et al. 1991; Spash and Hanley 1995). Unfortunately, this question could not discriminate between groups of respondents (that circumstance did not show up in the pre-tests). The consumer–citizen differences in respondents’ roles were explored in a simple setting similar to Sagoff ’s (1988) historic example. Respondents were simply asked for their preferences regarding skiing in the Nockberge national park, which is an important regional natural area of about 18,500 hectares. The national park lies 50 kilometres to the north-west of Klagenfurt (the capital city of Carinthia) near the Hohe Tauern national park. While one-quarter of respondents would like to go skiing in the ‘Nockberge’ national park, about 95 per cent of all respondents would vote for the protection of the national park in a referendum (see Table 2.1). Only two respondents would not go skiing there, but would vote for a ski resort in the Nockberge. A number of epistemological questions were raised in the survey. Only 32 per cent of respondents stated that they were not influenced by the survey and the specific CVM questions, while nearly two-thirds of the respondents had not thought about their WTP before answering the survey questions. Fourteen per cent were explicitly influenced by the ‘cost–benefit’ approach, and again about 7 per cent rejected that kind of questioning (see Table 2.2). Explaining WTP bids: the importance of the ‘extended’ valuation framework The estimated equations for explaining respondent’s WTP on the basis of the explanatory variables (for a detailed description, see Table 2.3) are presented for the logit model in Table 2.4 and for the OLS regression model in Table 2.5. Generally, all included variables have the expected signs (e.g. higher income, knowledge or a positive vote in a public referendum all increase the respondent’s acceptance of a personal contribution and the WTP bid). Regarding the different variables, the respondent’s WTP should increase with higher income, education and knowledge about biodiversity. Furthermore, it is expected that members of an environmental organization are, ceteris paribus, willing to pay more while those opposing the question format exhibit a lower WTP. Those with a probably positive vote in a public referendum regarding a nature protection tax should again be willing to pay more – like those willing to pay a voluntary contribution, even if they would be in a minority in a respective referendum. These expectations regarding the direction of the influence of the variables are generally satisfied with the models estimated. The respondent’s principal WTP is estimated with a logit model. The dependent variable is WTP1 (=1, if respondent in principle is willing to contribute some
36 Michael Getzner Table 2.1 Consumer vs. citizen and students’ perspectives Respondent as consumer ‘The Nockberge national park is not (yet) an internationally acknowledged national park, but is protected by provincial law. Let us assume that the Nockberge would be changed into an attractive and also reasonably priced ski and recreation resort with numerous lifts and sports facilities, and the protection of nature would be abandoned. Would you personally like to go skiing in the Nockberge national park or enjoy other attractive sports and recreation facilities?’
No Yes Total
Frequency
Per cent
138 51 189
73.02 26.98 100.00
Respondent as citizen ‘Regardless of your personal possibilities or wishes to use the Nockberge national park for your sports and recreation enjoyment: would you as a citizen taking all arguments into account vote in a referendum for the use of the Nockberge as an attractive ski resort, or do you think that the Nockberge should in all cases remain a national park?’ Frequency I think that the national park should strictly be protected regardless of the economic benefits of a ski resort. I would vote for the ski resort because of the economic benefits. Sum Missing/no answer Sum
Per cent
178
94.18
7 185 4 189
3.70 97.88 2.12 100.00
Note: All questions are translated from the original German questionnaire.
amount). Explaining the respondent’s yes-vote is done stepwise. Four different models are estimated to show the influence of several extensions to the ‘standard’ WTP model, which is shown in the first two columns in Table 2.4. In the ‘standard model’, the explanatory variables ‘Age’, ‘Economics’, ‘Biodiversity’, ‘Member’, ‘Donations’ and ‘Recreation’ have a significant influence on the respondent’s accepting a personal contribution to a nature protection fund. The log-likelihood of the model is –64.18 with a corresponding McFadden’s R² of about 26 per cent (which is well above the usually applied thresholds for CVM survey validity). About 85 per cent of the answers can be correctly predicted. If the model is extended to account for an individual’s beliefs about the CVM setting, the variables ‘Protest’ and ‘Influence’ are additionally significant in explaining the respondent’s WTP. While the coefficients of the other explanatory variables stay roughly in the same order of magnitude, these additional (‘epistemological’) variables improve the explanatory power of the model (McFadden’s R²) to 34 per cent. Extending the model even further to account for the social context, variables ‘Referendum’ and ‘Public’ turn out to be significantly correlated with WTP1,
A framework for valuing nature
37
Table 2.2 Epistemological problems of WTP surveys ‘Surveys asking respondents about their personal willingness to pay for environmental improvements are relatively rare. Answering the questions and this specific kind of questioning is perceived by some as strange or is rejected. Which of the following statements would you agree to?’ [Multiple answers] Frequency This sort of questioning has neither influenced me personally nor my opinions about natural protection. Before this survey, I had not thought about my willingness to pay for a nature protection programme. This kind of question has influenced me. I will in the future pay more attention to whether nature protection yields economic benefits compared to the costs. I reject these kinds of questions on the willingness to pay for nature protection, because species protection cannot be valued in monetary terms. N
Respondents (%)
61
32.28
124
65.61
26
13.76
14 189
7.41
leading to an R² of 42 per cent. Finally, the inclusion of the variable ‘Con_cit’ only marginally improves the quality of the model. About 87 per cent of all answers can correctly be predicted. After having explained the respondent’s principal WTP, the second part of the ‘nested’ model was estimated by OLS. Extending the valuation framework to explain the respondent’s WTP bids by means of a simple OLS regression model improves the model, too. As Table 2.5 shows, the usually significant socio-economic variables in regression models are significant in this estimation as well. Most influential are ‘Income’ and ‘Children’, and (interestingly) ‘Gender’. Some expectedly influential variables regarding environmental behaviour and environmental consciousness also play a significant role in explaining the respondent’s WTP bid. Variables describing the institutional and social context in which decisions on nature protection programmes are made are of crucial importance. A highly significant explanatory power is revealed by those variables that show the respondent’s decisions (behaviour) in public referenda as well as their willingness to donate a personal contribution even if they are in a minority position. That means that the higher the supposedly altruistic motives operationalized with this concept, the higher the respondent’s WTP bid is. Beginning with the usual socio-economic characteristics of respondents, variables ‘Income’, ‘Gender’, ‘Biodiversity’, ‘Member’, and ‘Recreation’ play a significant role. The explanatory power of the OLS model for the respondent’s WTP bids is 19 per cent. If we include institutional context variables (respondent’s perception of the CVM setting), about 24 per cent of the variance of the dependent
38 Michael Getzner Table 2.3 WTP regression model variables Variable name
Dimension
Description
Dependent variable WTP1 WTP2
1=yes
Principal WTP for nature protection programme in Carinthia ATS (natural WTP including zero bids log)
Individual factors: socio-economic variables Income Profession Age Gender Children Economics
in 1,000 ATS (natural log) 1=yes in years 1=female 1=yes 1=yes
Income of respondent At least in part-time employment Age of respondent Gender of respondent Has children or children in the household Studies business admin. and has introductory economics
Individual factors: environmental consciousness and behaviour Biodiversity Member Donations Recreation
1=yes 1=yes ATS (natural log) 1=yes
Partially familiar with the concept of biodiversity Member of an environmental organisation Monthly average donation to environmental groups Likes recreation in the open countryside
Institutional context Protest Influence Thoughts
1=yes 1=yes 1=yes
Cheap
1=yes
Explicitly disliked questions valuing natural goods Admits influence by the kind of valuation question Never thought about economic valuations of biodiversity before The valuation question was framed by verbal explanations regarding hypothetical bias (‘cheap talk’)
Social context Majority
1=yes
Minority 1=yes Referendum 1=yes Average Public
1=yes 1=yes
Con_cit
1=yes
Thinks that the majority of citizens would vote for the introduction of a nature protection tax in Carinthia WTP regardless of the outcome of the referendum Votes yes for the introduction of nature protection tax in Carinthia Believes contribution is above the average Believes that nature protection programmes should be financed by public budget Respondent exhibited different preferences as consumer and citizen (questions as in Table 2.1)
–3.8269 3.3227 0.5326 0.4344 –0.1196 0.5024 –0.0680 0.0401 –0.0929 0.5381 0.0970 0.5960 0.8620 0.5311 1.5275 0.7147 2.3797 1.1823 0.4708 0.2740 1.6453 0.4952
Constant
Influence
Protest
Recreation
Donations
Member
Biodiversity
Economics
Children
Gender
Age
Profession
Income
Coefficient Std. Error
Variables
–2.6699
3.3224***
1.7181*
2.0127**
2.1372**
1.6232(*)
0.1627
–0.1727
–1.6955*
–0.2380
1.2260
–1.1517
z-Statistic
Model 1
–2.0176**
–2.7654
3.1943***
1.4663
2.4445**
2.3544**
1.6698*
–0.1963
–0.4931
–0.5187
–0.7505
0.8216
–0.9982
z-Statistic
Model 2
–3.6743 3.6810 0.3908 0.4756 –0.4198 0.5594 –0.0245 0.0473 –0.2845 0.5770 –0.1270 0.6469 0.9594 0.5746 1.9659 0.8350 3.2502 1.3296 0.3961 0.2702 1.7444 0.5461 –2.0462** 1.3048 –1.3761 0.6820
Coefficient Std. Error
Table 2.4 Logit model estimated coefficients for WTP1
–2.4258 3.8487 0.1895 0.5030 –0.2037 0.5813 –0.0201 0.0495 –0.2645 0.6113 –0.3359 0.6659 1.1795 0.6160 1.7591 0.8788 3.0327 1.2793 0.3627 0.2753 1.8223 0.5632 –1.9380* 1.4270 –1.6679 0.7135
Coefficient Std. Error
–2.3378**
–2.8254
3.2355***
1.3175
2.3707**
2.0018**
1.9147*
–0.5044
–0.4327
–0.4056
–0.3503
0.3767
–0.6303
z–Statistic
Model 3
–1.9814 4.4592 0.1995 0.5919 0.0570 0.6661 –0.0117 0.0557 –0.7243 0.6998 –0.3869 0.7891 1.1912 0.7324 2.2522 1.1446 3.0285 1.3305 0.3136 0.2854 1.9002 0.6899 –1.7794* 1.5878 –2.3451 0.8160
Coefficient Std. Error
continued …
–2.8739***
2.7542***
1.0986
2.2762**
1.9677**
1.6263(*)
–0.4902
–1.0350
–0.2095
0.0856
0.3371
–0.4443
z–Statistic
Model 4
A framework for valuing nature 39
–3.8269
Constant Thoughts
Notes *** p<0.01 ** p<0.05 * p<0.1 (*) 0.1<=p<0.12
n 171 df 10 S.E. of regr. 0.3509 Log likelihood –64.1778 McFadden R² 25.95%
Con_cit
Public
Referendum
Coefficient Std. Error
Variables
Table 2.4 Continued
–1.1517 0.7090
z-Statistic
Model 1
171 13 0.3353 –57.2930 33.89%
–3.6743 1.3168 0.5384
Coefficient Std. Error –0.9982 0.2858
z-Statistic
Model 2
170 14 0.3248 –53.8363 36.71%
–2.4258 0.4951 0.5772 1.0153 0.5215
Coefficient Std. Error
–1.3579
1.9468*
–0.6303 0.6531
z–Statistic
Model 3
–1.9814 0.9298 0.7024 1.0553 0.6120 –2.2454** 0.6047 –0.8726 0.7042 159 16 0.3067 –43.2056 44.92%
Coefficient Std. Error
–1.2391
1.7244*
–0.4443
z–Statistic
Model 4
40 Michael Getzner
–0.8344 2.5186 0.5458 0.3381 0.3333 0.4019 –0.0486 0.0346 –0.7283 0.4102 0.7330 0.4918 0.4958 0.4345 1.5229 0.4472 1.3432 0.5385 0.1766 0.1166 1.1439 0.4287
Constant
Influence
Protest
Recreation
Donations
Member
Biodiversity
Economics
Children
Gender
Age
Profession
Income
Coefficient Std. Error
Variables
–1.4219
2.6680***
1.5144
2.4941**
3.4049***
1.1410
1.4903
–1.7754*
–1.4058
0.8294
1.6144(*)
–0.3313
t-Statistic
Model 1
–2.4679 2.6356 0.6477 0.3410 0.2783 0.3927 –0.0444 0.0351 –0.7238 0.3988 0.7270 0.4852 0.5573 0.4251 1.5703 0.4383 1.6912 0.5349 0.1210 0.1158 1.1361 0.4212 –1.7735* 0.8018 –0.6708 0.5282
Coefficient Std. Error
–1.2698
–1.3956
2.6972***
1.0451
3.1619***
3.5827***
1.3111
1.4984
–1.8147*
–1.2650
0.7088
1.8992*
–0.9364
t-Statistic
Model 2
Table 2.5 Log-normal OLS estimated coefficients for WTP2
1.1999 1.9869 0.2054 0.2637 –0.0667 0.2949 –0.0259 0.0280 –0.5672 0.3025 0.7278 0.3730 0.2805 0.3306 0.4179 0.3379 0.7102 0.4182 0.0520 0.0851 0.8460 0.3212 –1.8278* 0.7635 0.2236 0.4206
Coefficient Std. Error
0.5317
–1.6480
2.6339***
0.6108
1.6980*
1.2368
0.8485
1.9513*
–1.8753*
–0.9231
–0.2264
0.7789
0.6039
t-Statistic
Model 3
2.3239 2.0407 0.0233 0.2655 0.0463 0.2991 0.0023 0.0280 –0.5906 0.2955 0.8268 0.3692 0.4593 0.3363 0.4625 0.3312 0.7955 0.4129 0.0059 0.0821 0.7025 0.3288 –2.2465** 0.7336 0.2430 0.4127
Coefficient Std. Error
continued …
0.5889
2.1364**
0.0721
1.9267*
1.3963
1.3659
2.2392**
–1.9984**
0.0836
0.1548
0.0877
1.1388
t-Statistic
Model 4
A framework for valuing nature 41
172 10 19.11% 2.3338 5.0410***
Coefficient Std. Error
0.5516
t-Statistic
Model 1
172 14 23.79% 2.2654 4.8131***
1.3532 0.4076 0.8214 0.3653
Coefficient Std. Error
0.3139
2.2484**
–0.0500
t-Statistic
Model 2
144 18 39.04% 1.5449 6.0877***
1.1957 0.2989 0.8607 0.3099 1.1955 0.3254
–0.1547 0.3232 0.6309 0.2750 –0.0621
Coefficient Std. Error
–1.1655
3.6743***
2.7771***
–0.1979 0.3150 4.0001***
2.2944**
0.3301
t-Statistic
Model 3
***p<0.01 **p<0.05 *p<0.1 (*)0.1<=p<0.12 a Results of Tobit estimation are available on request; however, the significance of variables stays roughly the same.
Notes
n df Adj. R² S.E. of regr. F–statistic
Con_cit
Public
Average
Referendum
Minority
Majority
Cheap
Thoughts
Variables
Table 2.5 Continued
1.2641 0.3014 0.7873 0.3099 0.9334 0.3355 –3.6273*** 0.3213 –0.5637 0.3137 135 20 45.20% 1.464 6.5255***
0.9629 0.3428 0.5194 0.2786 0.1655
Coefficient Std. Error
–1.7969*
2.7823***
2.5404**
4.1934***
0.5252
1.8641*
t-Statistic
Model 4
42 Michael Getzner
A framework for valuing nature
43
variable can be explained. Variables ‘Protest’ and ‘Cheap’ are additionally significant. Extending the model even further to include individual beliefs about the social context, variables ‘Minority’, ‘Referendum’, and ‘Average’ increase the power of the model to 39 per cent. If we try to account for the consumer–citizen dichotomy (‘Con_cit’) and beliefs about the appropriate financing mechanism of public goods (‘Public’), an explanation of 45 per cent of the variance can be achieved. The high explanatory power of the model shows that not only are the usual individual characteristics of respondents as individuals and consumers crucially influential, but knowing about respondents’ attitudes and behaviour in the institutional and social context improves the quality of the model and adds much explanatory power to give a richer picture of the framework in which respondents decide upon their contribution to a regional biodiversity programme.
Discussion and conclusions Criticism of the CVM is as old as the method itself, and its attackers come from a number of different corners. One major strand of critical work focuses on the assumption of welfare economics and CBA, which holds that individual costs and benefits can be aggregated in a way to judge the economic reasonability of a project or policy on the level of society. However, society, as the term suggests, comprises more than the narrow economic (or market-based) context. Ultimately, the institutional and social context is manifest in individual perceptions, attitudes, beliefs and behaviour. Ethical commitments and rights-based beliefs towards other species constitute problems with the narrow economic notion of efficiency and the concept of contingent valuation and CBA, as individuals might not be willing to make the usual trade-offs between income (or consumer goods) and biodiversity. These so-called lexicographic preferences are more ‘individualistic’ than the much-discussed divergence between social role as consumers and citizens. The latter dichotomy refers to a broader model of human behaviour taking not only individual utility maximization into account, but also referring to social values held by individuals. However, epistemological problems with the assumption of the Homo oeconomicus and the consequent use of methods appropriate to a narrow economic theory rather than methods appropriate to more broader approaches targeted at the perception and behaviour of individuals in an environmental context, threaten the validity of CVM-based valuation. There has been much controversy over these topics in the literature; but numerical or statistical analysis is rather rare compared with more qualitative or explorative approaches. In this chapter, empirical evidence is presented on the extent and importance of some of the most crucial elements of the institutional and social context in which the valuation of regional biodiversity takes place. While the sample chosen for the survey is not representative for the whole population, it can be assumed that the significant variables found in regression models will in principle also prove influential for the whole population.
44 Michael Getzner A large proportion of respondents exhibited lexicographic preferences, and the results of the survey suggest that the consumer–citizen difference in perceiving a context-inherent question is also relevant. Although these two crucial elements of the valuation procedure are of great importance, some of the theoretical considerations regarding the amount of WTP bids cannot be proven empirically. Theory would suggest that – due to the priority of biodiversity in the case of lexicographic preferences – WTP bids could be zero or infinite. The WTP bids in this study are quite reasonable in the sense that the respondents considered their budget constraint when stating a WTP bid. One possible explanation could be that respondents want their preferences to be respected in the decision process. They take those instruments offered to them, and a monetary expression of their preferences is a way to influence the outcome even if they distrust CVM-based research. Respondents answered a number of epistemological questions. A majority of respondents was not used to answering such valuation questions, and thought about it for the first time when dealing with the questionnaire. Variables comprising these elements of instrument-based influence were statistically significant in explaining WTP bids in the regression model. WTP bids are influenced by respondents’ perception about the CVM instrument as well as the framing of the WTP question. Variables encoding the respondents’ perceptions about the outcome of a similar referendum additionally exhibited significant influence (supposedly altruistic motives). It seems that respondents who would vote publicly for a nature protection tax are prepared to state higher private WTP bids. Furthermore, the perception of one stating higher WTP bids as well as the WTP in a situation when a general nature tax is not approved contribute significantly to the respondent’s WTP amount. The social (and political) context of decisions thus turns out to be crucially important in explaining respondents’ WTP bids. Beliefs about the respondent’s behaviour in a minority position as well as supposedly altruistic motives influence the respondent’s answers in a CVM survey.23 The empirical results indicate that, besides socio-economic variables, the institutional and social context in which respondents answer valuation questions can be of crucial importance. The explanatory power of the WTP bid models improves as more context variables are entered into the equation. Respondents of CVM surveys behave more like citizens in a political and societal debate – with their personal beliefs and perceptions of the institutional and social context of the valuation decision – rather than fulfilling the economic model of pure exchange by trading money vs. natural goods. CVM surveys without reference to institutional and social context variables thus may lack important arguments in explaining respondents’ WTP bids because environmental decisions apparently are more political than economic because they include morality (rights-based beliefs) and community rather than individual consumer interests. As a final conclusion regarding the use of contingent valuation results, it seems that contingent valuation exhibits economic values useful for evaluating policy programmes. However, environmental policy decisions cannot be based solely on
A framework for valuing nature
45
such results, as the valuation of environmental goods regularly includes a broad range of different values, perceptions and assumptions. CVM results might be an important element of decision making from an economic perspective (e.g. Carson 2000). Multi-criteria valuation frameworks and participatory decision processes may thus account for such elements and integrate them in a broader and more inclusive decision-making framework.
Acknowledgements The author would like to thank all participating students of the University of Klagenfurt as well as the instructors of those courses who ‘donated’ 25 minutes of their course time. Particular thanks go to G. Haber, W. Hediger, S. Niemeyer, M. Riedel and N. Wohlgemuth for commenting on earlier drafts of this chapter. The author accepts the responsibility for all remaining errors.
Notes 1 Based on the recommendation of Arrow et al. (1993), the concept of respondents’ WTP is used. However, many results of this chapter might also be transferred to respondents’ willingness to accept. 2 The ‘standard’ model referred to in this chapter is a sketch of the usually applied approach to explain the respondent’s WTP (see also the theory of reasoned action by Fishbein and Ajzen 1975). 3 The term ‘institutional’ is taken from the auction literature where the ‘institution’ refers to the elicitation instrument and auction setup. 4 The concept of biodiversity has become ‘popular’, particularly after the United Nations Conference on Environment and Development in Rio de Janeiro (1992) which produced a convention on the protection of biodiversity and biological resources. This convention defines biodiversity in natural science terms as the diversity within species (genetic diversity), species diversity (number of different species living in an ecosystem), as well as ecosystem and landscape diversity (different ecosystems and their protection). 5 Diversity is a characteristic of a species as well as of an ecosystem. Thus, valuation becomes much more complicated. Respondents in CVM studies are usually asked to assess policy-relevant proposals that often involve a number of biodiversity issues. Thus, respondents must consider their sympathy towards single species as well as taking into account their perceptions towards the importance of a (frequently unknown) number of species, their interrelations to secure the stability of ecosystems as a whole, including risk and (fundamental) uncertainties, as well as expected needs of future generations. In the light of some ecological fundamentals including uncertainty, the task of appropriately valuing biodiversity seems particularly difficult. 6 That means that respondents in a CVM survey might consider several elements of an ecosystem to be substitutable or not vital for the function of that ecosystem while ecologists and biologists would strongly oppose such beliefs. 7 For further critique, see Boulding (1994: 8): ‘It is curious that economists, almost without exception, take human valuations for granted and assume that everybody has a set of preferences that can be expressed as an indifference curve.’ Preferences are often found to be imprecise, as there is ‘plentiful evidence of intervals within which respondents found it difficult to be sure whether they would or would not pay or accept compensation for marginal changes in risks of injury’ in road accidents (Dubourg et al. 1994: 127).
46 Michael Getzner 8 It can be argued that preferences based on existence values are more likely to be of lexicographic nature since existence values are often founded on rights-based beliefs (Stevens et al. 1991). Furthermore, a WTP bid based on existence values can additionally be founded on a number of other reasons, e.g. ‘warm glow’ giving (Kahneman and Knetsch 1992), paying for ‘good causes’ (Stevens et al. 1994) or impure altruism (Andreoni 1990). 9 For more examples on the ‘unwillingness to accept’, see O’Connor (1997: 463). 10 Some differences of results of these studies can, of course, be attributed to differences in the institutional settings and contexts of these experiments. 11 Hanemann (1995: 88) argues that statements of respondents and their preferences are ‘inherently context-laden’. Furthermore, he reviews in detail the problems involved with CVM and voting, and offers a number of methods to overcome problems with inadequately defined context variables. 12 Note, the diverging roles assumed here are not generally accepted in mainstream economics ‘since citizens can be expected to contribute, in one way or another, toward the costs of policy, the respondent trades off the prospect of an amenity improvement against the prospect of a reduction in disposable income’ (Hoehn and Randall 1987: 229). However, the strict distinction between ‘citizen’ and ‘consumer’ is made for illustrative purposes. Those two concepts are better thought of as two poles of a continuum from self-interested utility maximization to purely altruistic social behaviour. Individuals can, of course, behave according to both patterns (i.e. they are not substitutes). 13 System theory particularly stresses the importance of differences between systems in order to be able to strictly divide between circumstances that are inside or outside a particular system (e.g. Luhmann 1986). 14 ‘Individual behavior is always mediated by social relations’ (Arrow 1994: 5). 15 In discussing their results, Schkade and Payne (1994: 107) explicitly mention the ‘citizen input (values) into policy decision’. Their study shows that within CVM questionnaires citizen’s considerations are generally excluded rather than acknowledged. 16 Public projects may run into problems not only because of project characteristics but also due to missing or inadequate involvement of citizens. In this sense CVM as a hypothetical market transaction is ‘blind to reasons’ (Keat 1997: 36). Nevertheless, a CVM survey is one way of involving affected citizens. 17 The questionnaire, additional written material as well as all data, are available from the author on request. 18 There are a number of hypotheses regarding the sign of the coefficients in the model. For instance, WTPi / Yi 0 , with Yi being the income of respondent i as part of the group of Sm variables in equations (3.1a) and (3.1b). 19 An actually existing nature programme of the regional government of Carinthia to protect and enhance biodiversity was valued. The programme was described in terms of what was to be valued (¾ page of description of the programme), how it would be provided and at what level (regional government and nature protection fund), who would pay (partly regional government budget and the fund), and how long it would last (5 years). 20 Due to budgetary constraints (small sample) and lack of previous valuation estimates, a question format including dichotomous choice was deemed infeasible. In any case, the significance of institutional and social context variables seems independent of the chosen question format. 21 The figures presented in this section illustrate preferences of the entire sample consisting mainly of students. The values do not provide a basis for aggregation to the whole population. Rather than producing WTP figures for Carinthia, the aim was to show the influence of extended context variables on WTP bids, which can be assumed to exist in the whole population.
A framework for valuing nature
47
22 Respondents’ problems with understanding the concept of biodiversity can also be in part interpreted as being a semantic difficulty: In German, the word Biodiversität is apparently poorly known. The words Artenschutz and Artenvielfalt would be understood more clearly, but these concepts only comprise species protection and diversity, leaving out genetic as well as ecosystems diversity. 23 Of course, the credibility of the proposed programme can also be of crucial importance. Respondents thus might particularly value policy programmes and environmental goods as ‘public’ goods rather than context-free discrete ‘bits’ of environmental goods (Brouwer et al. 1999).
References Aldred, J. (1994) ‘Existence value, welfare and altruism’, Environmental Values, 3(4): 381– 402. Andreoni, J. (1990) ‘Impure altruism and donations to public goods: a theory of warmglow giving’, The Economic Journal, 100 (June): 464–77. Arrow, K.J. (1994) ‘Methodological individualism and social knowledge’, American Economic Review, 84(2): 1–9. Arrow, K.J., Solow, R.M., Leamer, E., Portney, P., Randner, R. and Schuman, H. (1993) ‘Report of the NOAA panel on contingent valuation’, Federal Register, 58(10): 4602–14. Becker, G.S. (1976) The Economic Approach to Human Behavior, Chicago: University of Chicago Press. Bingham, G., Bishop, R., Brody, M, Bromley, D., Clark, E., Cooper, W., Costanza, R., Hale, T., Hayden, G., Kellert, S., Norgaard, R., Norton, B., Payne, J., Russell, C. and Suter, G. (1995) ‘Issues in ecosystem valuation: improving information for decision making’, Ecological Economics, 14(2): 73–90. Bishop, R.C. and Welsh, M.P. (1992) ‘Existence values in benefit-cost analysis and damage assessment’, Land Economics, 68(4): 405–17. Bishop, R.C., Heberlein, T.A. and Kealy, M.J. (1983) ‘Contingent valuation of environmental assets: comparison with a simulated market’, Natural Resources Journal, 23(3): 619–33. Blackburn, M., Harrison, G.W. and Rutström, E.E. (1994) ‘Statistical bias functions and informative hypothetical surveys’, American Journal of Agricultural Economics, 76(5): 1048–88. Blamey, R.K. (1996) ‘Citizens, consumers and contingent valuation: clarification and the expression of citizen values and issue options’, in V. Adamovicz (ed.) Forestry, Economics and the Environment, Wallingford: CAB International. Boulding, K.E. (1969) ‘Economics as a moral science’, American Economic Review, 59(1): 1–12. Boulding, K.E. (1994) ‘Economic ethics and the environmental crisis’, in E. Hickey and L.A. Longmire (eds) The Environment: global problems, local solutions, Westport, CT/London: Greenwood Press. Boyce, R., Brown, T.C., McClelland, G.H., Peterson, G.L. and Schulze, W.D. (1992) ‘An experimental examination of intrinsic values as a source of the WTA-WTP disparity’, American Economic Review, 82(5): 1366–73. Brouwer, R., Powe, N., Turner, R.K., Bateman, I.J. and Langford, I.H. (1999) ‘Public attitudes to contingent valuation and public consultation’, Environmental Values, 8: 325–47.
48 Michael Getzner Burgess, J., Clark, J. and Harrison, C.M. (1998), ‘Respondents’ evaluations of a contingent valuation survey: case study based on an economic valuation of the wildlife enhancement scheme, Pevens Levels in East Sussex’, Area 30(1): 19–27. Cameron, J.I. (1997) ‘Applying socio-ecological economics: a case study of contingent valuation and integrated catchment management’, Ecological Economics, 23(2): 155–65. Carson, R.T. (1998) ‘Valuation of tropical rainforests: philosophical and practical issues in the use of contingent valuation’, Ecological Economics, 24(1): 15–29. Carson, R.T. (2000) ‘Contingent valuation: a user’s guide’, Environmental Science and Technology, 34(8): 1413–18. Carter, J.R. and Irons, M.D. (1991) ‘Are economists different, and if so, why?’, Journal of Economic Perspectives, 5(2): 171–7. Common, M. and Perrings, C. (1992) ‘Towards an ecological economics of sustainability’, Ecological Economics, 6(1): 7–34. Common, M., Blamey, R. and Norton, T. (1993) ‘Sustainability and environmental valuation’, Environmental Values, 2(4): 299–334. Common, M., Reid, I. and Blamey, R. (1997) ‘Do existence values for cost benefit analysis exist?’, Environmental and Resource Economics, 9(2): 225–38. Conlisk, J. (1996) ‘Why bounded rationality?’, Journal of Economic Literature, 34 (2): 669–700. Cummings, R.G., Harrison, G.W. and Osborne, L.L. (1995a) ‘Are realistic referenda real?’, Economics Working Paper B-95-06, Division of Research, College of Business Administration, Columbia: University of South Carolina. Cummings, R.G., Harrison, G.W. and Osborne, L.L. (1995b) ‘Can the bias of contingent valuation surveys be reduced? Evidence from the laboratory’, Economics Working Paper B-95-03, Division of Research, College of Business Administration, Columbia: University of South Carolina. de Dios, E.S. (1995) ‘Re-reading Menger’s table’, European Journal of Political Economy, 11(2): 317–34. Department of the Interior (1994) ‘Natural resource damage assessment’, Federal Register, 59(85): 23098–111. Dubourg, W.R., Jones-Lee, M.W. and Loomes, G. (1994) ‘Imprecise preferences and the WTP-WTA disparity’, Journal of Risk and Uncertainty, 9(2): 115–33. Edwards, S.F. (1992) ‘Rethinking existence values’, Land Economics, 68(1): 120–2. Ehrlich, P. and Ehrlich, A. (1992) ‘The value of biodiversity’, Ambio, 21(3): 219–26. Encarnación, J. (1990) ‘Consumer choice of qualities’, Economica, 57(1): 63–72. Etzioni, A. (1986) ‘The case for a multiple utility conception’, Economics and Philosophy, 2(2): 159–83. Fischhoff, B. (1991) ‘Value elicitation: is there anything in there?’, American Psychologist, 46(8), 835–47. Fischhoff, B. and Furby, L. (1988) ‘Measuring values: a conceptual framework for interpreting transactions with special reference to contingent valuation of visibility’, Journal of Risk and Uncertainty, 1(1), 147–84. Fishbein, M. and Ajzen, I. (1975) Belief, Attitude, Intention and Behavior: An Introduction to Theory and Research, Reading: Addison-Wesley. Frey, B.S. (1997) Not Just for the Money, Cheltenham: Edward Elgar. Georgiou, S., Langford, I.H., Bateman, I.J. and Turner, R.K. (1998) ‘Determinants of individuals’ willingness to pay for perceived reductions on environmental health risks: a case study of bathing water quality’, Environment and Planning A, 30 (3): 577–94. Getzner, M. (1999) ‘Risk, uncertainty and discounting in practical environmental decision making’, in K. Steininger and F. Prettenthaler (eds) Proceedings of the Workshop ‘Risk and
A framework for valuing nature
49
Uncertainty in HD-Research’, Austrian Programme on Human Dimensions of Global Environmental Change, Graz: Department of Economics of the University of Graz. Getzner, M. (2000) ‘Hypothetical and real economic commitments, and social status in valuing a species protection program’, Journal of Environmental Planning and Management, 43(4): 541–59. Goodman, S., Jaffry, S. and Seabrooke, B. (1999) ‘Assessing public preferences for the conservation quality of the British coast, in M. O’Connor and C.L. Spash (eds) Valuation and the Environment: Theory, Method and Practice, Cheltenham: Edward Elgar: 165–82. Gowdy, J. (1996) ‘Discounting, hierarchies, and the social aspects of biodiversity protection’, International Journal of Social Economics, 23(4,5,6): 49–63. Gowdy J. and Olsen P.R. (1994) ‘Further problems with neoclassical environmental economics’, Environmental Ethics, 16(2): 161–71. Hanemann, W.M. (1995) ‘Contingent valuation and economics’, in K.G. Willis and J.T. Corkindale (eds) Environmental Valuation: New Perspectives, Wallingford: CAB International: 79–117. Hanley, N., Spash, C. and Walker, L. (1995) ‘Problems in valuing the benefits of biodiversity protection’, Environmental and Resource Economics, 5(3): 249–72. Harrison, G.W. (1996) ‘Experimental economics and contingent valuation’, Economics Working Paper B-96-10, Division of Research, College of Business Administration, Columbia: University of South Carolina. Hayden, F.G. (1993) ‘Ecosystem valuation: combining economics, philosophy, and ecology’, Journal of Economic Issues, 27(2): 409–20. Hodgson, G.M. (1997) ‘Economics, environmental policy and the transcendence of utilitarianism’, in J. Foster (ed.) Valuing Nature? Ethics, Economics and the Environment, London/New York: Routledge: 48–66. Hoehn, J.P. and Randall, A. (1987) ‘A satisfactory cost indicator from contingent valuation’, Journal of Environmental Economics and Management, 14(3): 226–47. Holland, A. (1997) ‘The foundations of environmental decision-making’, International Journal of Environment and Pollution, 7(4): 483–96. Holm-Müller, K., Hansen, H., Klockmann, M. and Luther, P. (1991) Die Nachfrage nach Umweltqualität in der Bundesrepublik Deutschland, Berlin: Erich Schmidt Verlag. Jacobs, M. (1997) ‘Environmental valuation, deliberative democracy and public decision making’, in J. Foster (ed.) Valuing Nature? Ethics, Economics and the Environment, London/ New York: Routledge: 211–31. Jakobsson, K.M. and Dragun, A.K. (1996) Contingent Valuation and Endangered Species, Cheltenham: Edward Elgar. Jeffrey, R.C. (1974) ‘Preferences among preferences’, Journal of Philosophy, 71(39): 377–91. Johansson, P.-O. (1992) ‘Altruism in cost-benefit analysis’, Environmental and Resource Economics, 2(4): 605–13. Kahneman, D. and Knetsch, J.L. (1992) ‘Valuing public goods: the purchase of moral satisfaction’, Journal of Environmental Economics and Management, 22(1): 57–70. Keat, R. (1997) ‘Values and preferences in neo-classical environmental economics’, in J. Foster (ed.) Valuing Nature? Ethics, Economics and the Environment, London/New York: Routledge: 32–47. Kosz, M. (1997) ‘Probleme der monetären Bewertung von Biodiversität’, Zeitschrift für Umweltpolitik und Umweltrecht, 20(4): 531–50. Lazo, J.K., Schulze, W.D., McClelland, G.H. and Doyle, J.K. (1992) ‘Can contingent valuation measure nonuse values?’, American Journal of Agricultural Economics, 74 (6): 1127– 32.
50 Michael Getzner Lockwood, M. (1997) ‘Integrated value theory for natural areas’, Ecological Economics, 20(1): 83–93. Lockwood, M. (1998) ‘Integrated value assessment using paired comparisons’, Ecological Economics, 25(1): 73–87. Luhmann, N. (1986) Ökologische Kommunikation, Opladen: Westdeutscher Verlag. Marwell, G. and Ames, R.E. (1981) ‘Economists free-ride, does anyone else?’, Journal of Public Economics, 15(3): 295–310. McDaniels, T.L. and Roessler, C. (1998) ‘Multiattribute elicitation of wilderness preservation benefits: a constructive approach’, Ecological Economics, 27 (3): 299–312. Menger, C. (1871) Einführung in die Volkswirthschaftslehre: erster, allgemeiner Theil, Vienna: Braumüller. Mitchell, R.T. and Carson, R.C. (1989) Using Surveys to Value Public Goods: The contingent valuation method, Washington DC: Resources for the Future. O’Connor, M. (1997) ‘The internalization of environmental costs: implementing the polluter pays principle in the European Union’, International Journal of Environment and Pollution, 7(4): 450–82. O’Hara, S. (1996) ‘Discursive ethics in ecosystem valuation and environmental policy’, Ecological Economics, 16(2): 95–107. Peukert, H. (1998) Das Handlungsparadigma in der Nationalökonomie, Marburg: Metropolis Verlag. Sagoff, M. (1988) The Economy of the Earth, New York: Cambridge University Press. Sagoff, M. (1998) ‘Aggregation and deliberation in valuing environmental public goods: a look beyond contingent pricing’, Ecological Economics, 24(2/3): 213–30. Schkade, D.A. and Payne, J.W. (1994) ‘How people respond to contingent valuation questions: a verbal protocol analysis of willingness to pay for an environmental regulation’, Journal of Environmental Economics and Management, 26(1): 88–109. Schulze, W., McClelland, G., Waldman, D. and Lazo, J. (1996) ‘Sources of bias in contingent valuation’, in D.J. Bjornstad and J.R. Kahn (eds) The Contingent Valuation of Environmental Resources, Cheltenham: Edward Elgar: 97–116. Sen, A.K. (1977) ‘Rational fools: a critique of the behavioral foundations of economic theory’, Phil. and Public Affairs, 6(4): 317–44. Solow, R.M. (1993) ‘Sustainability: an economist’s perspective’, in R. Dorfman and N.S. Dorfman (eds) Economics of the Environment, 3rd edn, New York: W.W. Norton. Spash, C. and Hanley, N. (1995) ‘Preferences, information, and biodiversity preservation’, Ecological Economics, 12(3): 191–208. Stevens, T.H., Echeverria, J., Glass, R.J., Hager, T. and More, T.A. (1991) ‘Measuring the existence value of wildlife: what do CVM estimates really show?’, Land Economics, 67(4): 390–400. Stevens, T.H., More, T.A. and Glass, R.J. (1994) ‘Interpretation and temporal stability of CV bids for wildlife existence: a panel study’, Land Economics, 70 (3): 355–63. Stork, P. and Viaene, J.-M. (1992) ‘Policy optimization by lexicographic preference ordering’, Journal of Policy Modeling, 14 (5): 655–73. Toman, M.A. (1999) ‘Sustainable decision making: the state of the art from an economics perspective’, In M. O’Connor and C.L. Spash (eds) Valuation and the Environment: Theory, Method and Practice, Cheltenham: Edward Elgar: 59–72. Tversky, A., Sattath, S. and Slovic, P. (1988) ‘Contingent weighting in judgement and choice’, Psychological Review, 95 (3): 371–84. Vatn, A. and Bromley, D.W. (1994) ‘Choices without prices without apologies’, Journal of Environmental Economics and Management, 26 (2): 129–48.
Non-use values and attitudes
3
51
Non-use values and attitudes Wetlands threatened by climate change Jürgen Meyerhoff
Introduction Climate change is expected to cause extensive damage in coastal areas due to a rise in sea level and increasingly frequent violent storms. Scientists expect that, in particular, the nature and landscape of the German Wadden Sea, a coastal wetland ecosystem, will be negatively influenced. The adjacent salt marshes, a habitat for endangered species, have already been adversely affected in quality and quantity by human interference, such as by dam construction and waste disposal into the North Sea, and are expected to deteriorate even faster. However, it is possible to mitigate the consequences of climate change by taking suitable measures such as removing dams (Reise 1993, 1996). To obtain information about the population’s support for the protection of the Wadden Sea against the consequences of climate change, the willingness to pay (WTP) of German households was determined by using the contingent valuation method (CVM).1 Several studies have shown that non-use values can form an important part of the total economic value of environmental assets (Garrod and Willis 1996; Bateman and Langford 1997), particularly for highly unique resources. Therefore, non-use values were also thought to form an important part of the total economic value of the Wadden Sea. This area is a very valuable natural refuge, ranking third out of thirteen national parks in Germany. However, despite an ongoing debate among environmental economists about the compatibility of non-use values with neoclassical theory, it is far from obvious whether all bids from non-users express economic values and should therefore be used in cost-benefit analyses. As a guideline, some authors require that it should at least be possible to explain stated non-use values partly by predictors suggested by economic theory. ‘If it is argued that non-use values are theoretically included in economic benefits, these values should therefore be suitable for economic models of choice, this would imply, among other things, that the value for a good with non-use value should be partly explained by income, and possibly by other economic determinants of value’ (Wierstra 1996: 118; also Blamey et al. 1995: 272). One aim of this study was to clarify the relationship between attitudes and non-use values, testing the results by Wierstra. In his study On the Domain of Contingent Valuation, he found that the variation of stated non-use values could be explained solely based on attitudes.
52 Jürgen Meyerhoff This chapter is organized in the following way. In the second section, a brief overview of the discussion about the relationship of attitudes and WTP is given. The third section presents the design of the contingent valuation and its results in relation to the use and non-use of the Wadden Sea area. Section four presents findings on the relationship between measured attitudes and WTP, while in section five the results of multiple analyses by SPSS-CHAID and OLS regression are discussed. Further conclusions are drawn in section six.
Attitudes as predictors of WTP within the CVM There is increasing debate about how knowledge of social psychology can and should be incorporated into contingent valuation studies. In general, two different approaches exist. On the one hand, economists are looking for ways to extend their basic models in order to be able to explain stated WTP more comprehensively. For example, Georgiou et al. (1998: 592) conclude: ‘These diverse influences upon WTP suggest that simplistic models of preference formation are inadequate bases for CVM studies. Rather an extended model of how individuals form preferences is required’. This approach is also supported by the NOAA panel, as Kotchen and Reiling (2000) point out. On the other hand, attitudes are used as a determinant competing with economic preferences. Particularly when non-use values form an important part of the economic value, there is some disagreement about whether the stated amounts are due to preferences or whether they are just another measure of attitude on an arbitrary scale (Kahneman et al. 1993; 1999). A general definition of attitudes is given by Eagly and Chaiken (1993: 1). They define attitudes as being ‘a psychological tendency that is expressed by evaluating a particular entity with some degree of favour or disfavour’. In order to be able to use attitudes to clarify the question of whether preferences are at work or not, it is necessary to distinguish between different approaches of attitudes. Kaiser et al. (1999: 2) differentiate between three attitude approaches in connection with ecological behaviour: first, the Fishbein and Ajzen tradition with attitudes towards a specified behaviour; second, attitudes towards the environment commonly referred to as environmental concern; and, third, the New Environmental Paradigm (NEP) as an alternative, single component measure of environmental attitude. The theory of reasoned action by Fishbein and Ajzen (1975) and its successors as well as the NEP by Dunlap and van Liere (1978) are at the centre of attention within the debate. Although sometimes connected in CVM studies, both approaches represent in the end different concepts of environmental attitudes. The theory of reasoned action and its successors try to explain the relationship between attitudes and behaviour. Based on this relationship, it should be possible to predict actual behaviour. The main component of these theories is the behavioural intention, which is influenced by attitudes towards a specified behaviour and a subjective norm towards that behaviour. Under certain circumstances the behavioural intention accurately predicts real behaviour. Among others, one requirement is that the principle of compatibility is fulfilled (Ajzen 1996: 387; also Bamberg 1996). This principle demands that attitudes and behaviour should be
Non-use values and attitudes
53
measured at the same level of generality. In this case the likelihood increases that the same set of salient beliefs will be activated and strong correlations between attitudes and behaviour could be observed. Within the theory of reasoned action the expression of WTP can be seen as an intention to behave in a certain way; that is to actually pay the stated amount of money. Bishop and Heberlein were among the first authors to point out the potential for CVM studies to incorporate attitudes towards WTP. In accordance with their work the theory of reasoned action can help to validate measured values and, therefore, supports the assumption of articulated values. ‘Still, if conditions for high attitude–behaviour correspondence are fulfilled, some grounds would exist for arguing that legitimate economic values are being established at least to a rough approximation’ (Bishop and Heberlein 1986: 146). An example of a combination of both CVM and theory of reasoned action is the work by Moisseinen (1999) which examines the WTP of residents of an area in Finland to prevent the extinction of the Saimaa seal. As the results of a former CVM study on the protection of the Saimaa seal were contradictory in itself, Moisseinen used, among other things, the theory of reasoned action to explain the variations of WTP. Following this theory she asked interviewees not only about their WTP, but also about their attitude towards it and about their subjective norm.2 With an extended behavioural model based on the theory of reasoned action and the standard economic approach, it was possible to explain a significantly larger amount of variation (adjusted R-square at a value of 0.43) than usual. Thus, as these attitudes correspond to the ‘principle of compatibility’ they should be a good predictor of the actual WTP. The NEP is an alternative method used to measure environmental attitudes. In its original version the scale consists of 12 items. According to Kaiser et al. (1999), it represents a shift towards a more evaluative conception of attitude because its proponents regard one’s moral values as the core concept of environmental attitude. Therefore, NEP is an instrument to measure value orientations rather than attitudes. Critics have pointed out that NEP and ecological behaviour correlate only weakly at best. An example of a combination of both CVM and NEP is the study conducted by Kotchen and Reiling (2000). They used an expanded and updated version of the NEP measure to elicit the WTP of 1200 Maine residents for the protection of two endangered species. From this study it could be shown that attitudes, measured with the NEP scale, are a significant explanatory variable both of the general WTP and also of the amount people were willing to pay. But although they are a good explanatory variable of WTP it is necessary to keep in mind that these attitudes are not a good predictor of behaviour. Rather it is the use of the NEP scale that provides the possibility to measure what Fischhoff (1991) called basic values. Using attitudes to explain WTP, Wierstra (1996) explored under what conditions the CVM produces valid and reliable measures of value for environmental goods. Wierstra had hypothesized that a reliable measure of value for environmental goods by the CVM is dependent on the positioning of the good to be valued in the dimensions of ‘use/non-use value’ and ‘concrete/abstract’. More specifically, it was tested whether the quality of CVM results declined with increasing degree of
54 Jürgen Meyerhoff non-use value in the good and with increasing degree of abstraction in the good. To test this hypothesis, he carried out three CVM surveys with different environmental goods. In the case of the protection of the island of Rottumeroog within the Wadden Sea off the Dutch coast, only non-use values could arise because the island is uninhabited and visiting it is not allowed. In addition to economic factors such as household size, net income, and fixed housing expenses, Wierstra also determined attitudes from his interviewees. However, although the theory of reasoned action is referred to, attitudes towards a specific behaviour, such as WTP, were not determined. For example, in the case of Rottumeroog, attitudes recorded were concerned more with subjects such as the government’s areas of responsibility, cultural or historical meaning of the island, and how respondents saw nature as being an important aspect of the island. For this reason, however, the ‘principle of compatibility’ wasn’t fulfilled as these attitudes are not suitable as a basis for statements about the relationship of attitude– behavioural correspondence. Rather the elicited attitudes also represent some kind of general attitudes which measure basic values. In the end, Wierstra found that especially for the non-use good some attitudinal factors were sufficient to explain the variation in WTP, while none of the economic factors were significant in this case. Therefore, conditions for the validity of CVM results were not met because the value of a good should at least be partly explained by income, and possibly by other economic determinants of value.
Study design and main results The aim of the contingent valuation survey was to place a monetary value on measures which would allow the protection of the German Wadden Sea against the consequences of climate change. Therefore, a random sample was drawn from the German population, and 2390 households in Germany were contacted by professional interviewers in April and May 1999. The sample was drawn in a three-stage process: first, 140 sample points representing the size of German towns were determined; second, random walk was used to select the household; and, third, the kish-selection-grid was used to address the target person within the household. In order to test for distance-decay effects, the sample was divided into four groups, defined by concentric zones around the Wadden Sea (each with a radius-increment of 220 km). For various reasons, such as ‘nobody at home’, or ‘household or target person not willing to participate in an interview’, 1412 faceto-face interviews were completed.3 This corresponds to a response rate of 59 per cent. In order to inform interviewees about climate change and its possible consequences for the Wadden Sea, they were presented with various short texts, a map of the Wadden Sea area, and a card with photos of parts of the Wadden Sea. The card also contained a map illustrating the distribution of important biotypes in the area and a short description of the main consequences of climate change for the Wadden Sea. As Bateman and Langford (1997: 572) point out, studies which try to measure non-use values are presented with a definitional problem. Generally, non-use values
Non-use values and attitudes
55
are associated with benefits from the knowledge that an environmental asset exists or is maintained independent of any actual or potential use. Moreover, they are to a large extent expected to be elicited by people who live far away from the site involved and who never have and probably never will visit it (Brouwer and Slangen 1998). However, non-use values can also be attached to environmental assets by people who have visited the site. Despite an ongoing debate, there is no generally accepted approach to decompose values held by users into use and non-use values. On the other hand, those who were, at the time of the survey, non-users may have valued the protection of the Wadden Sea area even if they have no intention or desire to visit it. Therefore, in this study non-use values are defined as the WTP solely of non-users. Accordingly, use values are defined as the WTP of users. This way of proceeding gives at least a well-founded lower bound of the non-use benefits which arise from the environmental good (Silberman and Gerlowski 1992: 226). Moreover, to inform politicians about the economic value of an environmental good such as the Wadden Sea area it is not necessary to know which part of the users’ WTP arises from non-use values. Crucial is, rather, the total economic value which comprises user and non-users’ total WTP. Users were, in contrast to non-users, defined as persons who have visited the North Sea coast or any of the coastal islands at least once. In accordance with this definition 49 per cent (695 people) out of the entire sample were users. Out of this group more than 60 per cent had visited the North Sea coast within the last five years, more than 80 per cent within the last 10 years. In addition, 72 per cent had visited the coast more than once, implying that the North Sea coast is an important area for recreation in Germany.4 On the other hand, 51 per cent (717 people) had never visited this area before and were therefore taken as non-users. The spatial distribution of users and non-users was as expected. The portion of users decreased and that of non-users increased proportionally to the distance from the coast, between 30 per cent non-users in zone 1 (mainly covered by Schleswig-Holstein and Hamburg) to 65 per cent in zone 4 (covered by Bavaria and BadenWürttemberg). As Table 3.1 shows, out of the whole sample 336 people or 25 per cent were in principle willing to pay for the offered programme. This group consisted of 225 users and 111 non-users. Before the data were examined further, protest bids were excluded from the following analysis. Protest responses are often determinated within CVM studies on an ad hoc basis and in almost every CVM study different approaches are used to determine these responses (Jorgensen et al. 1999: 135–7). However, all approaches Table 3.1 Distribution of interviews and in-principle WTP
WTP Zone 1 Zone 2 Zone 3 Zone 4 Total
All
User
300 407 403 302 1412
Yes 61 74 46 44 225
Non-user No 148 153 108 61 470
Yes 3 23 32 53 111
No 88 157 217 144 606
56 Jürgen Meyerhoff share the aim to differentiate real zero bidders from people who stated zero bids but for reasons other than not valuing the good in question. Protest responses are normally excluded from further analysis because they do not represent true economic values. Despite the fact that a lot of problems are associated with this practice of censoring protest responses, a simple follow-up question was also used in this study, asking the respondent to give a reason for not being willing to pay. People who answered ‘current taxes deemed already too high’, ‘I do not trust payments addressee’ or ‘not I but the state is responsible’ were determined as protest voters and excluded from further analysis. Taken together, this results in 56 per cent (797 people) protest responses. This is, compared with other studies, a relatively high amount of protest bids. For example, Spaninks et al. (1996) report 98 protest bids or 11 per cent out of 841 responses. One explanation for the high rate of protest bids could be that shortly before the main survey an eco-tax on fuel was introduced in Germany (1 April 1999). Out of the protest group, 62 per cent (497 people) stated, at least as one of the two most important reasons for not being willing to pay, that current taxes are already too high. Compared with the study by Spaninks et al. it is also necessary to keep in mind that they used a mail survey instead of personal interviews. As the 841 responses represent only a response rate of slightly more then 30 per cent, many people who did not respond were perhaps also ‘protesting’. Within the remaining group of users, 67 per cent (225 people) were in general willing to pay for the protection of the Wadden Sea while within the group of remaining non-users only 37 per cent (108 people) were in general willing to pay. As expected, the portion within non-users is substantially lower than the percentage of users who are willing to pay. Also the mean amounts show significant differences. While users are willing to pay DM 11.51 per household per month, non-users were willing to pay DM 3.57. To test whether there is a statistically significant difference in WTP between the two groups, the non-parametric Mann–Whitney test was used. It requires no assumptions about the distribution and tests whether the observations in two samples have the same distribution. The result (z = –8.38; p < 0.001) allows the rejection of the null hypothesis of equal WTP of users and non-users at a 1 per cent significance level. Table 3.2 summarizes the main results of the WTP statistics for users and non-users pooled and for each group alone. Table 3.2 Summary statistics of the WTP (in DM) for users and non-users
n Protest votes Used sample Willing to pay Mean WTP Standard deviation Median WTP Mode WTP Maximum WTP
All
User
Non-user
1412 790 622 336 7.8 17.2 1.0 0 170
695 362 333 225 11.1 20.8 5 0 170
717 428 289 111 3.6 10.3 0 0 100
Non-use values and attitudes
57
Influence of attitudes on WTP As mentioned in section one, it was the aim of this study to test Wierstra’s results about the relationship between attitudes and WTP. According to him, it was hypothesized that the conditions for economic values were not met if the WTP of non-users could be sufficiently explained by attitudinal factors; in this case the survey would support the results by Wierstra. Also following Wierstra, attitudes towards more general objects such as nature and climate change were gathered. The reason for this is that only people with enough information about the good are in a position to hold site-specific attitudes. Less informed people, such as nonusers, will possibly construct attitudes (Slaby 1996). Similar to the problem of constructed preferences, these attitudes will supply no sound basis for any analysis. Therefore, the gathering of attitudes was designed more generally, allowing for uninformed non-users to be considered. Figure 3.1 shows the conceptual model which is assumed to determine the WTP. Attitudes were mainly elicited towards three attitude objects: climate change, nature, and valuation of nature with money. The following subsections present the items and results regarding attitudes towards climate change and nature, and the results regarding attitudes towards money. Attitude towards nature and climate change The items which were used to measure the attitude towards climate change and nature are presented in Tables 3.3 and 3.4. Attitudes were measured on a fivepoint scale, ranging from ‘completely agree’ to ‘completely disagree’. The scales were developed in accordance with items used in previous studies involving contingent valuation (Goodman et al. 1996; Wierstra 1996), and in psychological studies (Hofer 1987; Preisendörfer 1996). As a measure of reliability, Cronbach’s • Nature Attitudinal factors
• Climate change • Valuation with money
• User/non-user • Planned visits Landscape-related factors
• Knowledge about the Wadden Sea • Interest in conservation • Distance to Wadden Sea
• Household net income Economic factors
• Household size • Age
Figure 3.1 A conceptual model of WTP determinants
WTP for protection of the Wadden Sea against consequences of climate change
58 Jürgen Meyerhoff Table 3.3 Attitudes towards nature 1. A nature conservation area is only valuable for me if it is allowed to be entered. 2. I am very confident that our children and grandchildren will experience nature in a similar way as we do today. 3. We are allowed to form nature regarding only our own needs. 4. Animal and plant species threatened by extinction should be protected without regard of the cost.* 5. Nature and landscape in Germany did not change during the last 20 years. 6. Spending time amidst nature is a repeatedly wonderful experience.* 7. More consumption goods can compensate the losses of nature and landscape. 8. To protect nature for its own sake should not be a goal of environmental politics. 9. For the protection of nature and landscape I would give up a part of my present income.* In the scale all except item 2 were taken into account; Cronbach’s alpha is 0.67. Notes Items marked with an asterisk were transformed before adding scores. The transformation of a negative formulated item with scores 1 to 5 is: ‘new score’ = 6 – ‘original score’.
Table 3.4 Attitudes towards climate change 1. I myself can contribute no further to protect the climate. 2. Within the discussion on environmental problems the meaning of climate change is exaggerated. 3. For us today the consequences of climate change are without meaning because they occur only in a few decades. 4. Unless causes of climate change are not completely known, measures to protect the climate should not be carried out. 5. For me it doesn’t matter if the following generations will suffer any disadvantages from climate change. 6. I am very confident that policy will be successful in protecting our climate. All items were taken into account; Cronbach’s alpha is 0.75.
alpha was calculated. It is the current standard statistic assessing the reliability of a scale composed of multiple items and it is the most appropriate reliability measure to use for Likert scales. Because it considers the degree to which items on a scale intercorrelate, it is often referred to as a measure of internal consistency (Eagly and Chaiken 1993: 67). The associated alpha is 0.67 for the scale ‘nature’, and 0.75 for the scale ‘climate change’. These low values of Cronbach’s alpha illustrate the problem arising from the use of a scale which is not as well developed as the NEP scale, for example. Compared with a generally accepted value of 0.80 (Schumann 1999: 42), the study fails to gather reliable attitudes. However, if compared with the range of values for Cronbach’s alpha obtained in other studies, our data prove reasonably reliable. Wierstra reported values of Cronbach’s alpha of between 0.45 and 0.85. Diekmann (1997: 220–3) mentioned that in the General Population Survey (Allgemeine Bevölkerungsumfrage (ZUMA-Allbus)) short scales of four to six items were used and Cronbach’s alpha values of approximately 0.60 were obtained. Therefore, the gathered attitudes towards nature and towards climate changes were subjected to further evaluation. Tables 3.5 and 3.6 illustrate the relationship between the attitude score and the mean WTP.
Non-use values and attitudes
59
Table 3.5 Attitudes towards nature and WTP Score
User and non-user
User
Non-user
Mean
Std.
N
Mean
Std.
N
Mean
Std.
N
up to 24 25 to 30 31 to 36 37 to 42
1.54 4.92 11.22 19.54
3.98 11.25 19.36 37.09
92 231 260 31
2.63 8.13 14.12 22.22
4.73 15.15 21.52 40.94
34 104 170 20
0.87 2.29 5.74 14.68
3.30 5.29 12.86 30.04
58 127 90 11
X2
93.85
p < 0.001
32.15
p < 0.001
42.25
p < 0.001
Table 3.6 Attitudes towards climate change and WTP Score
User and non-user
User
Non-user
Mean
Std.
N
Mean
Std.
N
up to 12 13 to 18 19 to 24 25 to 30
0.03 3.24 7.86 13.10
0.12 7.98 16.09 23.83
18 156 288 160
0.08 6.17 11.30 15.27
0.20 11.73 19.86 25.10
7 56 160 110
X2
61.9
p < 0.001
18.62
p < 0.001
Mean 0.00 1.60 3.56 8.31 27.59
Std.
N
0.00 3.96 7.59 20.16
11 100 128 50
p < 0.001
The relationship between attitudes towards nature and mean WTP is positive for all three groups. The results confirm the expectation that people with stronger attitudes towards nature are on average willing to pay more than people with weaker attitudes. The mean WTP of users is for each level of the attitude score higher than for the corresponding group of non-users. The last line of Table 3.5 presents results of the Kruskal–Wallis H-test. This non-parametric test is a test for the null-hypothesis that the distribution of some variable (WTP in this case) is the same in all samples defined by a certain grouping variable (attitudes towards nature in this case). The results allow the rejection of the null-hypothesis of equal distribution within each attitude class at a high significance level. The figures in Table 3.6 show the same pattern as in Table 3.5. The mean WTP is at each level of attitude score higher than at the preceding level. Again, mean WTP of users is in each case higher than WTP of non-users. The results of the Kruskal–Wallis H-test allow once more the rejection of the null-hypothesis of equal distribution of WTP at a high significance level. Thus, attitudes towards climate change also have, like attitudes towards nature, a significant influence on WTP. In both cases, a higher score on attitude corresponds to a higher WTP on average. In addition, the analysis of the relationship between attitudes and WTP confirms two more findings presented by Kotchen and Reiling (2000). First, respondents with stronger pro-environmental attitudes are more likely to participate in the valuation procedure. People who were in-principle willing to pay show a higher mean score on both the nature and the climate change scale. Second, people who were defined as protest voters attain lower mean scores on both scales. The Mann– Whitney test indicates that both differences are significant at the 0.1 per cent level.
60 Jürgen Meyerhoff Attitude towards money as a measure of value As another expression of attitude, the degree of agreement with the statement ‘I reject money as a measure to express the value of nature and landscape’ was used. Nearly 45 per cent out of the whole sample ‘completely agree’ or ‘agree’ with this statement, while 25 per cent ‘disagree’ or ‘completely disagree’. Slightly more than 29 per cent are in between both of these groups. The relationship between the stated answers and mean WTP for both users and non-users is given in Table 3.7. In both cases the results of the Kruskal–Wallis H-test allow the rejection of the null-hypothesis of equal distribution of WTP within the defined samples. At a first glance, it is interesting that also people who ‘completely agree’ or ‘agree’ with the presented statement were in-principle willing to pay, and, especially in the group of users, that their mean WTP is relatively high compared with the amounts stated by people belonging to the other categories. However, a post hoc test (Sachs 1997: 395–7) qualifies these findings. In the group of users, their mean WTP does not differ between those who ‘completely agree or agree’ or ‘neither agree nor disagree’ at a 5 per cent significant level. In the group of non-users, only the mean WTP of those who ‘disagree or completely disagree’ differs from the means of the other samples. Nevertheless, the expected result remains unchanged. People who accept money as a measure to express the value of nature and landscape show a significantly higher mean WTP than those who reject money as such a measure.
Multiple analysis of WTP determinants For a multiple analysis of factors influencing WTP, two tools were used. First, CHAID (Chi-squared Automated Interaction Detector) was used to analyse the relationship between the in-principle WTP and the potential explanatory variables. Second, multiple regression equations were estimated using ordinary least squares. This allows us to examine the relationship between the stated amount of WTP and the independent variables. In both cases, equations for users and non-users are presented separately. The aim of this analysis was to test whether use and nonuse values were influenced by totally different factors as Wierstra (1996) found. Table 3.7 Relationship between attitudes towards money and mean WTP ‘I reject money as a measure to express the value of nature and landscape’
User
Non-user
n
% per column
Mean
n
% per column
Mean
Completely agree Agree Neither agree nor disagree Disagree or Completely disagree
41 91 101 100
12.3 27.2 30.2 30.2
11.7 6.6 9.7 17.8
59 95 81 54
20.4 32.9 28.0 18.7
3.4 2.2 3.7 6.0
χ2 (p)
30.4; df. 3; p < 0.001
24.77; df. 3; p < 0.001
Note: Because of a low number of observations in the group ‘completely disagree’, it was merged with the group of people who ‘disagree’.
Non-use values and attitudes
61
Unlike his study, both groups of users have to be analysed separately because only one good with both characteristics was the subject of the CVM. Table 3.8 reports the potential explanatory variables. SPSS-CHAID analysis SPSS-CHAID is a market segmentation technique that is designed for use with categorical variables and continuous variables which have been divided into categories (e.g. income, age). Based on the chi-square test, CHAID divides a population at each level of the analysis into two or more segments that differ with respect to a designated criterion. The analysis stops if the program is not able to detect further significant relationships between the dependent and any remaining variable or if a defined minimum of observations is reached. The first line of each box in the tree diagrams in Figures 3.2 and 3.3 shows the corresponding variable. The second line, except for the root of the tree, shows the category of the respective variable by which CHAID divides the population. In the third line, the percentage of people who are willing to pay within the subgroup is presented, while the fourth line shows the total number of people within that group. In both the user and the non-user group only a very limited number of variables have influence on the dependent variable ‘in-principle WTP’. Among these variables none is predicted by economic theory as an important predictor of WTP. Rather, the attitudinal factors dominate. The attitude towards nature is in both cases the most influential variable. People who show stronger attitudes towards Table 3.8 Potential explanatory variables Variable
Description
AGE INCOME
Respondent’s age (continuous variable) Total monthly net household income (income was set equal to the midpoint of the interval of the 9 intervals in which the respondent’s income fell) Household size including children (continuous variable) Distance to Wadden Sea in kilometres (continuous variable) Previous knowledge of the Wadden Sea area as a natural landscape; ranges from 1 (no knowledge) to 4 (thorough knowledge) 1 if respondent had previous knowledge of negative impact of climate change on Wadden Sea Intention of respondents to travel to North Sea coast: 1 if respondent will travel, 2 if respondent wishes to travel, 3 if respondent neither plans nor wishes to travel 1 if respondent has participated in a hike through the Wadden Sea area Degree to which respondent is interested in conservation of Wadden Sea area; ranges from 1 if respondent is very interested to 5 if respondent is not interested at all Attitudes towards nature (continuous variable) Attitudes towards climate change (continuous variable) Attitudes towards valuation with money (reject money as a measure); ranges from 1 if respondent completely agrees to 5 if respondent completely disagrees
HSIZE DISWAD KNOWS KNOWCL TRAVEL WADHI CONSER ATTNAT ATTCLI ATTMON
62 Jürgen Meyerhoff In-Principle Willingness to Pay yes: 37% n = 289
ATTNA
ATTNA
ATTNA
up to 24 yes: 10%
24–30 yes: 31%
31–42 yes: 61%
n = 58
n = 127
n = 104
TRAVEL 1–2 yes: 48% n = 41
TRAVEL 3
TRAVEL 1–2 yes: 10% n = 50
yes: 22% n = 86
TRAVEL 3 yes: 50% n = 88
Figure 3.2 Tree diagram for non-users
In-Principle Willingness to Pay yes: 67% n = 334
ATTNAT 19–24 yes: 38% n = 34
ATTNAT 25–30 yes: 58% n = 104
ATTNAT 31–54 yes: 77% n = 196
ATTMON tot-agree – nn yes: 71% n = 123
ATTMON disag – tot-disag yes: 88% n = 73
Figure 3.3 Tree diagram for users
nature are more likely to pay for the proposed programme to protect the Wadden Sea. On the second level, variables differ between both models. In the case of the non-user group, TRAVEL is the second most important variable. People who plan a visit or wish to travel to the North Sea are more likely to pay than people who neither plan nor wish to travel. In the case of the user group, the variable ATTMON appears with two new segments merged out of the five-point scale. The more non-users agree to value nature in monetary terms, the higher is the percentage of people who are in-principle willing to pay.
Non-use values and attitudes
63
However, the third attitudinal variable, that is attitudes towards climate change, did not show any influence on the in-principal WTP answer either in the group of users or in the group of non-users. One explanation could be that people perceived the offered programme to protect the Wadden Sea as not being something that is connected to climate change. It is supported by the finding that most of the people who were willing to pay stated that they understood that their payment was not solely to finance measures against consequences of climate change but to finance a programme which would also protect the landscape against other negative interventions. Multiple regression analysis Tables 3.9 and 3.10 report the results from the regression analysis for both the group of non-users and the group of users. Various functional forms were fitted to the data. The preferred models, which show the best statistical fit, are based on the double-log form. As the Breusch–Pagan test indicated presence of heteroscedasticity, the White estimator was used and instead of the F-test the non-parametric Wald test was conducted (Greene 2000: 506–11). At a value of 44.61, the Wald test shows that all variables together have a significant influence on the dependent variable. Furthermore, all variables in the non-user model show the expected sign. INCOME is highly significant and shows a positive influence on the dependent variable. AGE is also significant, but only at the 10 per cent level. In contrast, the third variable out of the group of economic variables, HSIZE, is even at the 10 per cent level not significant and was therefore removed from the equation. Out of the group of landscape-related variables CONSERV (‘very interested’ and ‘interested’) shows a positive influence and proves significant at the 1 per cent level. Finally, ATTMON has a negative influence on the stated WTP. Non-users who completely agree with the statement that they reject money as a term to value nature are less willing to pay than people who agree with this statement at a lower degree or disagree with it. Also ATTNA is highly significant and shows the expected positive influence. In view of these results, the requirements set up by Wierstra seem to be fulfilled. Stated bids by non-users are influenced at a significant level by explanatory variables which economic theory predicts. As shown in Table 3.10, this is not the case when bids stated by users are analysed. As opposed to the non-user group, none of the predictors out of the group of economic variables was significant, either at the 5 per cent level or at the 10 per cent level. Therefore, these variables were removed completely from the equation. Slightly more influential were some of the landscape-related variables. WADHI and DISWAD prove significant at the 1 per cent level and at the 5 per cent level and show the expected sign. Users who had participated in a hike through the Wadden Sea area have a higher mean WTP than users who had not. Distance to the Wadden Sea, in contrast to the non-user model, proves significant and supports findings by Sutherland and Walsh (1993) or Pate and Loomis (1997) about the negative relationship between distance and WTP. Similar to the non-user group, both ATTNA and ATTMON show a high and significant influence. As the
64 Jürgen Meyerhoff Table 3.9 OLS results of non-user group
CONSTANT AGE INCOME CONSERV (very interested) CONSERV (interested) ATTNAT ATTMON (completely agree)
Coefficient
Std. error
t-ratio
–7.62 –0.24 0.26 0.48 0.39 2.12 –0.55
1.62 0.14 0.09 0.16 0.12 0.41 0.15
–4.71 –1.75 3.41 3.08 3.21 5.21 –3.81
**** * **** *** *** **** ****
Notes: **** significant at p < 0.001, *** significant at p < 0.01, ** significant at p < 0.05, * significant at p < 0.1
Table 3.10 OLS results of user group
CONSTANT WADHI DISWAD ATTNAT ATTMON (completely agree) ATTMON (agree)
Coefficient
Std. error
t-ratio
–6.5261 0.4127 –0.1717 2.6759 –0.6208 –0.4727
1.4299 0.1511 0.0871 0.4268 0.2343 0.1508
–4.5638 2.7311 –1.9721 6.2694 –2.6492 –3.1339
**** *** ** **** *** ***
Notes **** significant at p < 0.001, *** significant at p < 0.01, ** significant at p < 0.05, * significant at p < 0.1 R2 = 0.17; adjusted R2 = 0.16; Breusch–Pagan = 7.04 (5df); Wald = 17.29 (p <0.001)
Breusch–Pagan test indicated presence of heteroscedasticity also for this model, the White estimator was used once again. The Wald test with a value of 17.29 allowed the rejection of the global hypothesis of no joint influence of the explanatory variables. Compared with the non-user group, the coefficient of determination is 9 per cent lower than in the coefficient in the user group. In conclusion, one can say that the presented results support findings of other studies such as those by Green et al. (1993), Spash (1998), or Kotchen and Reiling (2000) about the influence of attitudes on WTP. The measured attitudes are significant predictors of both in-principle WTP and mean WTP. Moreover, they are also predictors in both user and non-user groups.
Discussion The aim of this chapter was to test whether non-use values are only influenced by attitudes or whether economic predictors influence these values as well. According to Wierstra (1996), goods which have a high degree of non-use value could be out of the domain of contingent valuation because stated WTP for these goods may only be influenced by attitudes. His results about the WTP for the protection of
Non-use values and attitudes
65
the uninhabited island of Rottumeroog support that hypothesis. As opposed to this, contingent valuation of public benefits of protecting the Wadden Sea against consequences of climate change shows exactly the opposite results. While the bids stated by non-users were influenced among other things by economic variables, in the case of users they show no influence at all. However, attitudes and selected landscape-related variables are significant in both models. Confronted only with the results of the presented analysis, one could draw the conclusion that non-use values are influenced more by economic variables than use values. According to Randall (1998) there is no crucial experiment which allows us to answer the question ‘Does the CVM work?’ Therefore, the findings presented in this chapter do not allow us to draw the conclusion that non-use values are always influenced by economic variables. They are just another contribution to the map of performance characteristics of contingent valuation. In this sense, they support findings of several other studies that environmental attitudes are significant explanatory variables of both principle WTP and of the amount of WTP. Furthermore, they also raise the question of whether it is meaningful to draw a dividing line between use and non-use values. Instead, it seems to be more promising to study further which factors determine people’s WTP for collective goods (Green 1997; Green and Tunstall 1999). It is also important to examine in which case the Fishbein–Ajzen attitude-model could be applied to non-users. Fishbein and Ajzen’s work has recently been recognized as being important in environmental valuation studies. One example of a study which was similar to their approach is that conducted by Moisseinen (1999). However, her study aimed at measuring use values. To be in a position to apply this approach to non-users, several aspects have to be considered. For example, the theory of reasoned action demands that some requirements such as the ‘principle of compatibility’ be fulfilled. But in the case of non-users it is not possible to completely fulfil this requirement. As already mentioned, non-users are, normally, people with less information about the good in question. Therefore, it could be misleading to use the theory of reasoned action when non-users’ WTP is of interest. The predictability of actual behaviour which follows the behavioural intention is much weaker if instead of the theory of reasoned action a more general attitude approach such as the NEP, the one used by Wierstra or in this study is applied. Therefore, Bishop and Heberlein’s (1986) point that some grounds exist for arguing that legitimate economic values are being established, at least to a rough approximation, is perhaps no longer valid. A next step in the analysis of the relationship between attitudes and WTP could be to develop and test models which incorporate general attitudes (e.g. environmental concern) and attitudes towards specific behaviour. Within these models, general attitudes are seen as determinants of attitudes towards specific behaviour (Bamberg et al. 1999; Fransson and Gärling 1999). Furthermore, the amount of knowledge people have about the environmental goods should be incorporated in more detail.
66 Jürgen Meyerhoff
Notes 1 This was part of a study on costs of climate change under the project ‘Effects of climatic change on the Island of Sylt: An interdisciplinary research project’. Further information is available at http://soel.geographie.uni-kiel.de/sylt/english/Welcome.htm. 2 At this stage of the sampling procedure interviewees were not informed about the type of survey. Therefore, people who refused to participate did not refuse because they don’t agree with monetary valuation of environmental resources. They refused generally to participate in any kind of survey. 3 The wording of the corresponding items is: ‘What is your attitude towards the extra payment to protect the Saimaa seal? The idea of collecting money from the households is: good ´ bad’ (attitude towards the behaviour) and ‘Majority of people I know would participate by paying the asked amount: likely ´ unlikely’ (subjective norm), (Moisseinen 1999: 202). 4 The definition of ‘users’ as people who have visited the North Sea coast at least once before the interview is very broad because people who have ‘actually used’ the coast many years ago are also defined as users. Further analysis showed no significant difference between people who used the coast many years ago and people who have visited it only one or two years ago. Therefore, this broader definition was applied during the whole evaluation.
References Ajzen, I. (1996) ‘The directive influence of attitudes on behavior’ in P.M. Gollwitzer and J.A. Bargh (eds) The Psychology of Action. Linking Cognition and Motivation to Behavior, New York/London: The Guilford Press: 385–403. Bamberg, S. (1996) ‘Allgemeine oder spezifische Einstellungen bei der Erklärung umweltschonenden Verhaltens? Eine Erweiterung der Theorie des geplanten Handelns um Einstellungen gegenüber Objekten?’, Zeitschrift für Sozialpsychologie, 27(1): 47–60. Bamberg, S., Kühnel, S.M. and Schmidt, P. (1999) ‘The impact of general attitudes on decisions’, Rationality and Society, 11(1): 5–25. Bateman, I.J. and Langford, I.H. (1997) ‘Non-users’ willingness to pay for a national park: an application and critique of the contingent valuation method’, Regional Studies, 31(6): 571–82. Bishop, R.C. and Heberlein, T.A. (1986) ‘Does contingent valuation work?’, in R.C. Cummings, D.S. Brookshire and W.D. Schulze (eds) Valuing Environmental Goods. An Assessment of the Contingent Valuation Method, Totowa, NJ: Rowman & Allanheld: 123–47. Blamey, R, Common, M. and Quiggin, J. (1995) ‘Respondents to contingent valuation surveys: consumers or citizens?’, Australian Journal of Agricultural Economics, 39(3): 262–88. Brouwer, R. and Slangen, L.H.G. (1998) ‘Contingent valuation of the public benefits of agricultural wildlife management: the case of the Dutch peat meadow land’, European Review of Agricultural Economics, 25(1): 53–72. Diekmann, A. (1997) Empirische Sozialforschung, Reinbeck: Rowohlt. Dunlap, R.E. and van Liere, K.D. (1978) ‘The “New Environmental Paradigm” ’, Journal of Environmental Education, 9(4): 10–19. Eagly, A.H. and Chaiken, S. (1993) The Psychology of Attitudes, Orlando: Harcourt Brace Jovanovich College Publishers. Fischhoff, B. (1991) ‘Value elicitation: is there anything there?’ American Psychologist, 46(8): 835–47. Fishbein, M. and Ajzen, I. (1975) Belief, Attitude and Behaviour. An Introduction to Theory and Research, Reading, MA: Addison-Wesley.
Non-use values and attitudes
67
Fransson, N. and Gärling, T. (1999) ‘Environmental concern: conceptual definitions, measurement methods, and research findings’, Journal of Environmental Psychology, 19(4): 369–82. Garrod, G.D. and Willis, K.G. (1996) ‘Estimating the benefits of environmental enhancement: a case study of the River Darent’, Journal of Environmental Planning and Management, 39(2): 189–203. Georgiou, S., Langford, I.H., Bateman, I.J. and Turner, R.K. (1998) ‘Determinants of individuals’ willingness to pay for perceived reductions in environmental health risks: a case study of bathing water quality’, Environment and Planning A, 30: 577–94. Goodman, S., Seabrooke, W., Daniel, H. Jaffry, S. and James, H. (1996) Determination of Non-use Values in Respect of Environmental and Other Benefits in the Coastal Zone, Portsmouth: Centre for Coastal Zone Management. Green, C. (1997) ‘Are blue whales really simply very large cups of coffee?’, paper presented at Stockholm Water Symposium, Stockholm, January 1997. Green, C. and Tunstall, S. (1999) ‘A psychological perspective’, in I.J. Bateman and K.G. Willis (eds) Valuing Environmental Preferences. Theory and Practice of the Contingent Valuation Method in the US, EU, and Developing Countries, Oxford: Oxford University Press: 207–58. Green, C., Tunstall, S., Herring, M. and Sawyer, J. (1993) ‘Customer preferences and willingness to pay for selected water and sewerage services’, a summary report to the Office of Water Services, Middlesex: Flood Hazard Research Centre. Greene, W.H. (2000) Econometric Analysis, New York: Macmillan. Hofer, D. (1987) ‘Naturschutz als Wertobjekt. Eine exemplarische Studie über Einstellungen zu Schutzgebieten’, PhD dissertation, Munich: Ludwig-Maximilians-Universität. Jorgensen, B.S., Syme, G.J., Bishop, B.J. and Nancarrow, B.E. (1999) ‘Protest responses in contingent valuation’, Environmental and Resource Economics, 14(1): 131–50. Kahneman, D., Ritov, I., Jacowitz, K.E. and Grant, P. (1993) ‘Stated willingness to pay for public goods. A psychological perspective’, Psychological Science, 4: 310–15. Kahneman, D., Ritov, I. and Schkade, D. (1999) ‘Economic preferences or attitude expressions? An analysis of dollar responses to public issues’, Journal of Risk and Uncertainty, 19(1–3): 203–35. Kaiser, G.F., Wölfing, S. and Fuhrer, U. (1999) ‘Environmental attitude and ecological behaviour’, Journal of Environmental Psychology, 19(1): 1–19. Kotchen, M.J. and Reiling, S.D. (2000) ‘Environmental attitudes, motivations, and contingent valuation of nonuse values: a case study involving endangered species’, Ecological Economics, 32(2): 93–107. Moisseinen, E.M. (1999) ‘On behavioural intentions in the case of the Saimaa Seal. Comparing the contingent valuation approach and the attitude-behaviour research’, in M. O’Connor and C.L. Spash (eds) Valuation and the Environment. Theory, Method and Practice, Cheltenham: Edward Elgar: 183–204. Pate, J. and Loomis, J.B. (1997) ‘The effect of distance on willingness to pay values: a case study of wetlands and salmon in California’, Ecological Economics, 20(3): 199–207. Preisendörfer, P. (1996) ‘Umweltbewußtsein in Deutschland. Ergebnisse einer repräsentativen Bevölkerungsbefragung 1996’, report for the Umweltbundesamt, Berlin: Umweltbundesamt. Randall, A. (1998) ‘Beyond the crucial experiment: mapping the performance characteristics of contingent valuation’, Resource and Energy Economics, 20(2): 197–206. Reise, K. (1993) ‘Verschwommene Zukunft der Nordseewatten’, in H.-J. Schellnhuber and H. Sterr (eds) Klimaänderung und Küste, Berlin: Springer: 223–9.
68 Jürgen Meyerhoff Reise, K. (1996) ‘Wattökologische Folgen bei Änderung von Klima und Küste’, Schriftenreihe der Schutzgemeinschaft Deutsche Nordseeküste, 1: 31–45. Sachs, L. (1997) Angewandte Statistik, Berlin: Springer. Schumann, S. (1999) Repräsentative Umfrage, München: Oldenbourg. Silberman, J. and Gerlowski, D.A. (1992) ‘Estimating existence value for users and nonusers of New Jersey beaches’, Land Economics, 68(2): 225–36. Slaby, M. (1996) ‘Einstellungsmessung oder Einstellungsgenerierung? Die Bedeutung der informationellen Basis bei Befragten für die empirische Rekonstruktion von Einstellungen zu gentechnischen Anwendungen (Measuring or generating attitudes? Subjective knowledge base for the empirical reconstruction of attitudes towards new applications of genetic engineering)’, Stuttgart: Institut für Sozialforschung of the Universität Stuttgart. Spaninks, F.A., Kuik, O. and Hoogeven, J.G.M. (1996) ‘Willingness to pay of Dutch households for a natural Wadden Sea. An application of the contingent valuation method’, Amsterdam: Instituut voor Milieuvraagstukken of the Free University Amsterdam. Spash, C.L. (1998) ‘Environmental values and wetland ecosystems: CVM, ethics and attitudes’, in M. O’Connor (ed.) The VALSE Project: Social Processes for Environmental Valuation, full final report to the DG-XII, European Commission, chapter 6–B. Sutherland, R.J. and Walsh, R.G. (1993) ‘Effect of distance on the preservation value of water quality’, Land Economics, 61: 281–91. Wierstra, E. (1996) On the Domain of Contingent Valuation, Enschede: Twente University Press.
Modelling environmental behaviour
4
69
Modelling environmental behaviour Socio-psychological simulation Hans-Joachim Mosler
Introduction In economically highly developed countries, many of the conditions that would allow people to behave in environmentally responsible ways are already in place. We have a lot of knowledge; for years surveys have shown that people give top priority to the need to act on environmental issues. The necessary technical and economic resources are also available. But there is little sign of a real about-face except in limited areas. We believe that the much-cited discrepancy between cognition and behaviour, between lip service and a person’s own contribution to conserving the environment, can be better understood if we also take people’s perceptions of the social surround into consideration. To view human beings as Homo economicus (Diekmann 1996; Harsanyi 1977) falls short, because the economic man approach does not explicitly take into account inner psychological factors, such as motives, values, norms and attitudes, nor does it consider social influences on a person’s behaviour. The present research conceives a model of behaviour that in addition to economic factors includes personal and social factors. Environmental consciousness is determined to a significant degree by social systems, or in other words, by people’s corresponding social representations. An individual’s personal contribution seems insignificant in the face of massive destruction of the environment caused by many. This perception – that there is nothing we can do personally, that each one of us is powerless – as well as a reluctance to be the ‘sucker’, are important causal factors in behaviours that overuse environmental resources. It does not seem rational to exercise personal restraint (for example, by not driving), because not only will we suffer from the harm caused by the general public’s overuse (the consequences of air pollution), but from a reduction in our own direct return as well (time saved, comfort). However, as this state of affairs applies equally to all individuals in a society overusing environmental resources, people mutually trap each other in patterns of actions that harm the environment. It is for this reason that we are particularly interested in examining the psychological conditions that would form the basis of a collective reorientation towards environmentally sustainable behaviour. Starting out from new environmentally friendly behaviours of some ‘pioneer’ individuals, we wish to discover the social psychological conditions that would ensure that the number of persons joining ranks with
70 Hans-Joachim Mosler such pioneers would continue to automatically increase and result in a true, largescale ‘turn-around’ of previous, environmentally harmful patterns of behaviour. The focal question of our research can thus be framed as follows: what are the conditions that foster widespread, effective inner dynamics that change collective environmental overuse (in thinking and action) to collective, environmentally responsible thinking and acting? Findings generated by this research will lend scientific support to the planning of environmental protection campaigns.
Method Procedure If we start from the assumption that environmental problems originate in the overusing behaviours of very many individuals, we need to consider how new solutions might be tested in a problem area of this magnitude. The instrument of the questionnaire, based upon imaginary situations or conditions, seems ill-suited (‘How expensive would gasoline have to be for you to change to public transportation?’). Massive field experiments that translate the issue directly in real social systems can also be eliminated; given our present state of scientific knowledge, such experiments would be both financially and ethically irresponsible. Laboratory experiments, which would not require intervention in existing social systems, cannot be carried out with large groups of persons (1000 and more). Computer simulation provides a possible solution. Simulation aims to reconstruct the relevant causeand-effect relationships in a problem area in the form of a model. With the aid of empirical data, the relationships can then be validated. In this way, we can test ‘experimentally’ the most various and unconventional ideas of ways to spread environmentally responsible thinking without incurring the risk of intervention in real social systems. It is important to remember, however, that simulation results are not empirical results found in reality. Simulation results can generate fruitful hypotheses that must be tested in reality. The great advantage is that the hypotheses are derived consistently using a clearly defined procedure. Another important point to consider is that findings gained through the simulation method allow only comparative conclusions, for example that under the conditions studied, Measure A produced better results than Measure B. The simulation method, furthermore, does not allow statements to be made about the means to be used to implement certain measures. For this, real experiments with real people are required. Despite all of this, computer simulation is the only method available for investigating, by means of many systematically varied experiments, how environmentally responsible changes can be instituted in large populations. Our simulation procedure consisted of the following steps. Following a preliminary selection of the most important, empirically wellfounded theoretical approaches within the field of social influence, the theories’ most relevant and significant core statements were formulated, according to content criteria.
Modelling environmental behaviour
71
In the ensuing modelling, the core statements of the theories were described with the existing variables and set in relation to one another according to certain systems-theoretical rules (see ‘Modelling of the theoretical concepts’). For lack of space, we here dispense with discussion of programming/technical implementations of the theories (see Mosler 2000 for details). The design of the simulation model was validated through the aid of experts’ evaluations and replication of findings from the fields of environmental and social psychology. Again, due to space considerations, the validation process is not reported here (cf. Mosler 2000 for details). Experiments with various strategies for the spread of behaviours were then conducted. The following will report on the most important and meaningful experiments. From the findings of the simulations, conclusions were drawn pertinent to both environmental practice and basic research. The most well known forms of intervention stemming from environmental research receive a new interpretation; wellfounded recommendations for the field of intervention are derived. The reader may wish to note that the following deals intensively with psychological concepts (to psychologists, however, the descriptions will appear somewhat superficial). The author sees the present chapter as a contribution that takes a perspective that differs greatly from the economic viewpoint, although some relevant connections are pointed out. This becomes clear in the next section. The reader interested in the exact derivation of the findings should read all sections, while others may wish to read only the sections on the simulation experiments and results. The most important findings are reported and discussed in the concluding section. The framework model Very many simulations work from the start at the level of collective variables (macroanalytic simulation, such as of, for example, the influence of prices on total consumption by the public). However, if we aim to tap into the dynamics of the spread, or dissemination (or non-spread, respectively), of environmentally responsible thinking and behaving, we must first of all begin at the level of individuals. We then allow them to interact with each other within the framework of the simulation (microanalytic aggregative simulation). The difficult task is to model the relevant processes of the individual in this area of conflict on the computer. We assume that humans are in principle beings that are free to make their own decisions, and we hold that a computer program cannot represent the richness of human individuality, as it is primitive in comparison. On the other hand, in many areas we do find empirically well-founded uniformities in human behaviour. Our way of proceeding springs from the will and hope to develop, from such empirically proven knowledge, a useful working model of the processes taking place within the individual. Usefulness is measured according to whether or not the model can be validated by the behaviour of real collectives and in terms of whether the simulation based upon the model widens and furthers our understanding of the dynamics of these processes.
72 Hans-Joachim Mosler For this reason, we first design a basic model of an individual (see Mosler 2000 for details), which – as 10,000 identically structured copies (equipped, however, with individually different characteristics) – serves as the basis for the simulated influencing and resource-use processes (see Figure 4.1). These persons differ individually only in their values of the variables. They all function according to the same social psychological principles. These principles are based on a few central and well-founded essences of theories, which are presented in detail in the presentation of results. The framework model first specifies the input and output variables of the theory-based processing by individuals. The model yields information about the inner psychological processes that take place when people use environmental resources (for example, use the resource of air when they drive or heat their homes) or influence each other mutually, whether deliberately or unintentionally. Inner psychological processes are triggered as people communicate with one another in daily life and observe themselves and others. These processes change, depending on further internal and external conditions, the ways in which we feel, think, and argue about the environment and the way we act toward the environment. These simulated individuals have at their disposal differently structured social contact nets (number of friends, acquaintances, neighbours, and strangers they observe). In its basic form, the model is an expanded ‘rational choice’ model (Diekmann 1996; Becker 1976). This means that the variables and processes specified in the model more or less consciously enter into the decisions relevant to using behaviour. The variables and processes thus having an effect are not, however, purely cognitive components. They include, in addition, motives, emotions, norms, and so on. This is what distinguishes our model from the usual ‘rational choice’ models: with a psychological conception, it is not a problem that non-rational components enter into rational decision-making. The following explains the external and internal input variables as well as the inner processes. External input variables are the influences that exert upon a person from the outside and are perceived by the person in some form. Possible distorted perceptions of individuals on the basis of biases are not taken into consideration in this model. For each individual some output variables are calculated that function as external input variables for other persons with whom the individual has contact (‘contacts’) (see Figure 4.1). Also, some non-social input variables, such as specific situational parameters and the momentary state of the resource being used, enter into the calculation of an individual’s output value.
•
•
Use, contact: A summary of all environmentally related behaviours shown by a contact. These behaviours towards the environment are conceived as resourceusing behaviours. For resource utilization, a particular resource can be entered into the model with its characteristic parameters (resources: for example, water, air, wild game populations; parameters: for example, rate of regeneration). Attitude, contact: A summary of a contact’s opinions and evaluations of environmental issues, objects, and behaviour towards the environment, as expressed in various ways.
Figure 4.1 Framework model of behaviour toward the environment
Modelling environmental behaviour 73
74 Hans-Joachim Mosler
• •
•
•
• • • •
Status, contact: This variable represents the sum of a contact’s relevant personal resources (for example, social competence, trust, prestige, knowledge, power, possessions). Persuasiveness, contact: The intensity and quality with which a contact makes a case for various attitudes related to the environment. While direction is given for an attitude (in other words, ‘for’ or ‘against’ environmental protection), persuasiveness indicates the intensity and quality with which these attitudes are presented. Situational factors and incentives: These include all influences from the societalinstitutional surround (rules and prohibitions, positive and negative incentives, and the like). From the entire spectrum of the possible scope of a behaviour, situational factors ‘filter’ a partial spectrum. State of resource: State of the environment, or state of a particular environmental resource, as ‘noticed’ and established by the person. Internal input variables have an effect on the simulated psychological processes ‘from the inside’. The emergence of individual degrees of markedness of these variables due to the individual’s learning history is not examined in this simulation. Values: Stable orientation with regard to environmental facts, objects, and behaviours. Knowledge: Extent and quality of information about the environment, such as knowledge of the regeneration parameters of a specific resource. Self-responsibility: Describes the extent to which people attribute responsibility to themselves for the state of the environment (as opposed to holding other persons, organizations, or institutions responsible). Motives: A summary of various motives, such as curiosity and laziness, that enter into a person’s readiness to act in certain ways. Dependent upon these motives, an act – independently of its consequences – will be viewed as ‘easy’ or ‘desirable’.
External and internal input variables are processed in different sub-models (see Figure 4.1) according to the theory being applied. The output values of the submodels have either a direct outward effect on other persons with whom the individual has contact (attitude, persuasiveness), or they affect variables leading up to behaviour and decisions on the use of resources and consequent actions. The following sub-models were designed and simulated (simulations of sub-models II and IV will not be presented):
• • • •
Processing of group influences upon attitudes towards the environment: application of the theory of Social Comparison Processes. Processing of discrepancy between environmental behaviour and environmental attitude: application of Dissonance Theory. Processing of observation of others’ behaviour towards the environment: application of Bandura’s Social Learning Theory. Processing of resistance to environmental protection measures: application of Reactance Theory.
Modelling environmental behaviour
• •
75
Processing of information on the collective utilization of an environmental resource: application of Commons Dilemma research. Processing of communicative influence in view of the individual’s feeling of concerned consternation about the state of the environment, knowledge about the environment, and biases: application of the Elaboration Likelihood Model.
Variables leading to behaviour are direct preliminary steps towards behaviour. On the basis of the Theory of Resource Mobilization (Klandermans 1984) and of the Theory of Planned Behaviour (Frey et al. 1993b; Ajzen 1991; Ajzen and Madden 1986), we start from the assumption that the following five components play an important role in behavioural intentions relating to the environment:
• • • • •
Attitude towards the environment as the attitude towards environmental protection and individual behaviour regarding the environment. The subjective norm as ‘perception of pressure from the social surround’. This expresses the expectations that confront the individual and the degree to which the individual is willing to fulfil these expectations. Behavioural control as the subjective conviction that one can in fact carry out the behaviour. Cost–benefit analysis as a motive for behaviours, which results from weighing the direct costs and direct benefits of actions. Sustainability as a motive, leading to willingness to restrict personal use of a resource. This readiness depends upon both the absolute value that a person places on an environmental resource and on the current discrepancy between collective, sustainable use and the actual pattern of use shown by others. The higher the subjective value of the environment, or the smaller the discrepancy between actual and sustainable patterns of utilization, the greater the effect of the sustainability motive in the sense of a person’s own restraint in use.
All of these variables are conceptualized at a general level, thus according with Ajzen and Fishbein’s (1977) correspondence hypothesis, which states that there is a high correlation between behaviour and the determining variables if they have been operationalized with the same degree of specification. As a fundamental extension of the theory of planned behaviour (Ajzen and Madden 1986), we have added an additional factor, cost–benefit analysis. Although the theory of planned behaviour does contain a kind of cost–benefit analysis, in that its components are broken down in ‘expectation evaluation products,’ we conceptualise this factor as an independent component. With this approach we expect to achieve better explanations of the variance for intention, but more importantly, to gain added starting points for intervention measures. Cost–benefit analysis models people’s ‘economic’ considerations, as they determine whether it will pay for them to act in certain ways. The model is thus the first to integrate economic and social psychological components (see also Mosler and Tobias 2000). These components all influence behavioural intention, which leads to behaviour towards the environment. Additionally, we introduce volition into the model, which
76 Hans-Joachim Mosler is closer to actual behaviour than behavioural intention. Psychological research (Frey et al. 1993b) has shown that the act of making a decision commits a person to the when and where of attempting to realise an intention. As the entire model is deterministic in construction, random variables play absolutely no role in determining individual behaviour towards the environment.
Simulation experiments and results In the following, after a brief introduction of the various sub-models, the conditions for the simulation experiments are described, and the results are presented. In describing the sub-models, we focus on experiments with populations of 10,000 persons and omit the experiments that we conducted with 1 to 10 persons in order to reach a basic understanding of the processes and to validate the model with the aid of existing empirical findings. All variables range within a dimension from 0 to 100. For variables related to environmental orientation, 100 signifies maximum environmental friendliness, and 0 indicates maximum lack of environmental friendliness; 50 represents a point of ‘neither/nor’. In the graphics presenting the findings, average behaviour towards the environment, or the average attitude towards the environment by the population, is always put on the ordinate. If a curve rises, this means that the population is changing in the direction of environmental responsibility. On the abscissa, we find the simulation steps. A step means that in a complete calculation cycle, all individuals exert influence on their contacts, or are themselves influenced, and – together with the others – utilize a resource. The linear progression of the steps should be interpreted as the progression of time; more exact temporal pinpointing is not possible. For the experiments, we usually started out from populations that were rather eco-unfriendly (mean of 40), whereby the values of this variable in the individuals strew around this mean. In order to avoid the reproach that we may have built the results of the simulation right into the program, all experiments were conducted according to a baseline/control group design. The results of a number of steps with and without manipulations (control group) form the basis for comparison; populations were always identical (which is only possible in computer simulation). In the following, we will examine important issues in social intervention and campaign planning with the aid of the simulation model. The following sections on the sub-models are each organized according to a research issue and a reference to the theories being applied. Group influences upon attitudes How must a minority of people behaving in an environmentally friendly way be distributed and networked within a population in order that the environmentally unfriendly majority comes to change its attitudes and behaviour? Frey (1993a: 114ff.) developed an integrative concept of social comparison processes that encompassed Festinger’s theory (1954), Tajfel’s theory of social identity (1978, 1979; cited in Frey et al. 1993a), and Thibaut and Kelley’s theoretical concept of
Modelling environmental behaviour
77
the comparison level for alternatives (1959; cited in Frey et al. 1993a). According to the integrative conception of social comparison processes (Frey et al. 1993a: 111ff.), persons in groups change their own attitudes in dependency upon the existing pressure to conform, the attractiveness of the group, and the perception of threat to their self-concepts should they change their own positions. A change in attitude results from the pressure to conform, weighted by the attractiveness of the group. If, however, the person’s self-concept is threatened, there will be no attitude change (see Figure 4.2). Threats to a person’s self-concept arise when their values differ to a certain degree (a value of 30) from the average attitudes of other members of the group. Individuals experience too great a discrepancy between their own values and the group’s attitude as threatening: in order to adapt to the group, the individual would have to deviate too far from personal values. Under social pressure, they would become untrue to themselves. People are more likely to change an attitude the more attractive the group is and the greater the group pressure to conform. Attitudes will change in the direction of the average attitude of the group. Attitude change occurs, however, only if the change does not pose a threat to self-concept. The attractiveness of the group results from two components: (1) the difference between a person’s own attitude and the average attitude of the group – a group is all the more attractive the more that person and group are on the same ‘wavelength’; and (2) the average status of the group – its prestige, power, social resources. The conformity pressure of a group can be calculated according to Tanford and Penrod’s (1984) well-documented formula. Pressure to conform, in reference to the total INDIVIDUAL
PROCESSING OF GROUP INFLUENCES
Values
Attitudes of contact persons
Status of contact persons
Threat to self-concept
Pressure to conform
Attitude
Attractiveness of group
Figure 4.2 Representation of the ‘processing of group influences’
78 Hans-Joachim Mosler number of group members, is calculated in an exponential function separately for a minority and a majority member. The population is structured in groups of 10 persons, whereby each group is in contact with two other groups. For the intervention, we assume that an environmental campaign can bring 10 per cent of the population to clearly embrace more friendly attitudes towards the environment than previously (an increase in 25 points), for a limited period of time (for example, for one week = one step in the simulation). These persons we call (attitude) ‘pioneers’. Pioneers are, moreover, more strongly sensitized to deviations from their own positions. Through this, they acquire a particularly low susceptibility to social influencing attempts: even minor deviations of other individuals in the group from their own, pioneer positions imply a potential endangering of their sense of self-esteem. This leads to the less susceptible attitudes of the ‘pioneers’ in the group. As a further form of intervention, pioneers are variously organized. Following the short environmental campaign, they remain in their groups (isolated) with no wider contacts to the outside; they remain in their groups, but are in contact with other pioneers (networks); or they form their own groups (core groups), which may have few or many contacts with other groups. Figure 4.3 documents the experiments conducted as well as their results. The intervention applies only to the fifth step; the dynamics described below result from this short-term intervention. It is extraordinarily important for pioneers’ susceptibility to be low, in order that they do not too soon once again adapt to the surrounding majority. If susceptibility is not low, there will be no positive effects from being networked with other pioneers and from no longer experiencing themselves as a minority. Core groups (also with low susceptibility) only then have an effect on the surrounding majorities in the group if they have numerous outer contacts (curve F in Figure 4.3 continues to climb in further steps, but not as high as curves B and D). In sum, pioneers have to hold to a stronger environmentally responsible attitude, and their susceptibility to influence must be low. In effect, they feel quickly threatened in their self-concept. Pioneers should be activated or active in as many groups as possible and should not be concentrated in just a few groups. Further measures are not required. Social learning of behaviour What factors lead the population to follow the examples of behavioural role models? According to Bandura (1979) people learn not only through direct personal experience, but also through observing others. In this way they gain an idea of how an action is performed. The probability that a learned behaviour will be carried out depends on motivation, and motivation is dependent upon expectations of efficacy (self-efficacy, Bandura 1986: 390ff.) and determined by anticipated selfregulation (Stalder 1985; Bandura 1986: 390ff.). Bandura does not, however, give any indication of the effect on a person’s behavioural repertory when there are several, contradictory behavioural models. For example, someone might observe environmentally responsible behaviour in some people and behaviour that is
Modelling environmental behaviour
79
Environmentally responsible 55
Attitude of the population
50
45
40
35 0 Not environmentally responsible
5
10
15
20
25
Runs Intervention
Control population A: Isolated ‘pioneers’, normal susceptibility B: Isolated ‘pioneers’, low susceptibility
D: Networked ‘pioneers’ with low susceptibility E: ‘Pioneer’ core groups, few contacts, low susceptibility F: ‘Pioneer’ core groups, many contacts, low susceptibility
C: Networked ‘pioneers’ with normal susceptibility
A. B. C. D. E. F.
In each group there is an isolated pioneer whose susceptibility to influence is the same in degree as that of the members of the group majority. In each group there is an isolated pioneer whose susceptibility is lower than that of the members of the group majority. In each group there is a pioneer whose susceptibiity is the same in degree to that of group members; in addition, the pioneer is in contact with 10 other ‘pioneers’ in other groups. They thus no longer experience themselves to be a minority. As in C above, but here pioneers have a lower degree of susceptibility. All pioneers are concentrated within their own core groups and show low susceptibility to influence. Each pioneer is in contact with another person outside of the core group. As in E above, whereby here pioneer group members are in contact with 10 other persons outside of the core group.
Figure 4.3 Change in average attitude due to various group-oriented interventions
damaging to the environment in others. In order to model contradictory influences on self-efficacy, we turned to Latané’s theory of social impact (Latané 1981; Latané and Wolf 1981; further specified in Nowak and Latané 1994; Nowak et al. 1990). In this theory, social influence bases upon the following factors, which stand in a multiplicative relationship: •
Strength: power, importance, intensity, unusual quality or features of the source person for the target person.
80 Hans-Joachim Mosler • •
Immediacy: directness, immediacy in space and time, absence of barriers or filters. Number of sources: number of group members, number of persons present.
The multiplicative relation of the three variables expresses the fact that the effect of one of the variables is greater, the greater the value of the other variables. There is no effect at all if one of the variables equals zero. According to the theory, moreover, the effect of the variable N (number) is not linear, but rather is an exponential function: I = sN t, where I is impact, s is a constant, and the exponent t is a value less than 1. The parameters s and t are different for each situation and have to be determined empirically. The factor ‘number’ thus has the effect that the first person has the greatest impact and each person thereafter has ever less of an impact. With an increasing number of influencing source persons, the social impact on a person rapidly decreases. Expectations of efficacy can be approximately represented by behaviour control, because we define behaviour control as a person’s conviction that they can indeed execute the behaviour. In accordance with the framework model, behaviour control affects the behavioural intention and, thus, behaviour. According to Latané’s theory of social impact (1981), behaviour arises from observation of the using behaviour of contact persons: social influence is a multiplicative function of the number, strength, and immediacy of contact persons. In our model (see Figure 4.4), the variable ‘number’ is determined by the number of the individual’s contact persons who behave in an environmentally responsible, or irresponsible, way. The ‘strength’ of contact persons’ influence is defined by their status. ‘Immediacy’ is conceived INDIVIDUAL
PROCESSING OF OBSERVED BEHAVIOUR Using behaviour
Using behaviour of contact persons
Behaviour control Intention
Status of contact persons
Values
Self-confirmation Volition
Self-responsibility Using behaviour
Figure 4.4 Representation of ‘processing of observed behaviour towards the environment’
Modelling environmental behaviour
81
of as psychological immediacy. This means that friends have an influence of 1/1, acquaintances of 3/4, neighbours of 1/2, and strangers of 1/4. The social influence of contact persons who behave in an environmentally responsible way and that of those who behave irresponsibly are calculated separately according to the formula and then compared by dividing them (compare Nowak and Latané 1994; Nowak et al. 1990). This ratio determines the behaviour control of the person: a person’s belief that they are able to perform the concrete environmentally friendly behaviour (that is, their behaviour control is the more environmentally responsible) is stronger the greater the number and the higher the status of individuals behaving environmentally responsibly in comparison to persons not behaving environmentally responsibly that they observe. Self-confirmation arises when people compare their own values (personal choice of behavioural standards) with their behaviour. Self-confirmation is also weighted by self-responsibility in order to incorporate the element of feeling responsible. A particular behaviour is all the more important to a person the better it accords with the person’s values and the more strongly the person feels responsible. Selfconfirmation affects volition and, thus, behaviour. Within the framework of an intervention, incentives can move persons having above-average status, for example, to change their behaviours during the intervention phase in the direction of environmental acceptability. On the other hand, it is possible to raise the status of persons displaying above-average environmentally friendly patterns of behaviour through various measures (for example, by means of public commendation awards, coverage in the media, and so on). The experiments were conducted with 300 or 500 persons and included varying degrees of ‘visibility’, in that selected persons displayed their environmental behaviours to many (15) or a few (5) contacts (Figure 4.5). Visibility of resource use has already been demonstrated to be an effective factor (Jorgenson and Papciak 1981; Mosler 1993). A. In this experiment, 500 persons of high status are selected and caused to adopt a behaviour that is more eco-friendly for the duration of the intervention. They are to demonstrate this behaviour to many others (15 contacts). B. As in A above, but here only 300 persons are selected. C. As in A, but here the behaviour is demonstrated to only 5 contacts. D. Here we selected 500 persons showing very environmentally sustaining behaviour patterns, which they demonstrate to 15 contacts. In this case, their status is raised for the duration of the intervention. It appears to be more effective to select persons having high status as role models and to induce them to behave in a more environmentally friendly way for the duration of the intervention than to temporarily raise the status of persons already showing such behaviours. An increase in the number of contacts, that is, an increase in the visibility of the environmentally sound behaviour, achieves relevant effects. An increased number of role models also has a strong effect. The dissemination process continues to progress automatically for some time after the intervention, as the entire social system must again adapt to the changes.
10
Intervention
5
15
20
25
30
35
40
45
50 Runs
Control situation
D. 15 CP/500 P/Status +
C. 5 CP/500 P/Act +
B. 15 CP/300 P/Act +
A. 15 CP/500 P/Act+
Figure 4.5 Average population behaviour with differing interventions based on observational learning
CP: contacts per person; P: number of persons addressed (role models); Act+, Status+: more responsible acts towards environment.
40 Not environmentally 0 responsible
42
44
46
48
50
52
54
Environmentally responsible 56
82 Hans-Joachim Mosler
Modelling environmental behaviour
83
In developing concepts for environmental campaigns, simulation can contribute support with regards to the implementation of efficient means. It can answer questions as to the number of role models required, how these should be selected, and what changes they should show, the degree of visibility necessary, and so on. In particular, simulation allows us to estimate the effects that can be achieved by means of compensation in other values, which makes a direct contribution to an increase in efficiency (for example, an increase in visibility at the expense of the number of role models). However, the simulation method does not allow conclusions about implementing the results, such as statements on how realistic it is to induce high status people to behave in a more environmentally friendly way. As discussed above, the simulation method permits only comparative conclusions. The commons dilemma How must the social surround be perceived for people to become willing to use an environmental resource sustainably? When people act upon the environment, they usually utilize an environmental resource. Utilization of a resource available to all (common property) involves an interpersonal conflict (Liebrand et al. 1992; Mosler 1995). It is in each individual’s interest to keep the personal benefit as large as possible, while the harm incurred or the depletion of the resource must be borne by all (Spada and Ernst 1990; Mosler and Gutscher 1996). The sustainability motive expresses the degree to which a person is willing to protect common property, the environment. This motive depends upon both the absolute value a person places on environmental goods and the current discrepancy between collective, sustainable use and actual patterns of use shown by others (expectation of sustainable utilization). The more highly the individual values the common property of the environment have, or the smaller the discrepancy between actual and sustainable patterns of use is, the greater the effect a sustainability motive will have. Under these conditions, the individual is motivated to make a personal contribution to sustainable patterns of use. The sub-model of the commons dilemma (compare Figure 4.6) yields a sustainability motive as an expression of the desire to use the common property in a resource-sustaining way through cautious use. The sustainability motive results from a multiplicative connection between the expectation of sustainable utilization and valuation of the resource ‘environment’. Valuation of the resource ‘environment’ is determined by the perceived state of the resource (state of the environment) and a person’s environmental values. Environmentally-oriented people use exhausted resources less. Various studies have shown the legitimacy of equating social orientation with environmentally-oriented values (Liebrand 1986; Liebrand et al. 1986; Mosler 1990). The degree of exhaustion of a resource is calculated on the basis of its state, as a percentage of deviation from its maximal state, and weighted by values. This means that a person’s evaluation of the resource is higher the more depleted a resource is and the more the person is environmentally oriented. Our model presently conceives of the perception of the state of the resource as direct perception, that is, without any distortion
84 Hans-Joachim Mosler INDIVIDUAL PROCESSING OF INFORMATION ON COLLECTIVE USE
Estimated proportion Expectation of of resourcesustainable sustaining users utilization
Using behaviour of contact persons
Knowledge about the resource
Perception of the state of the resource Values
Required proportion of resourcesustaining users
Sustainability motive
Intention
Volition Valuation of the resource Using behaviour
Figure 4.6 Representation of ‘processing of information on the collective use of a resource’
stemming from inner psychological biases or from external influences in the form of media reports, disputes among the experts, and so on. The expectation that one’s own way of using the environmental resource will make a contribution towards preservation of the resource is formed from the required proportion of resource-sustaining users in the population and the estimated proportion of resource-sustaining users. In other words, the greater the proportion of nonsustaining users of a resource as estimated by the person in comparison with the required proportion of resource-sustaining users in the population, the lower the person’s expectation that they themselves can make an effective contribution to sustainable use of the resource. This expectation rises in the degree to which the proportion of estimated resource-sustaining contact persons exceeds the required proportion of resource-sustaining users. Knowledge of the number of resourcesustaining users required in order to preserve the resource comes from resource knowledge and the perception of the state of the resource. The model presumes people to have – at present not a very realistic assumption! – knowledge about the regeneration rate, regeneration characteristics, and the maximal state of the resource. From the discrepancy between the maximal state of the resource and the current state of the resource, the person derives the required proportion of resource-sustaining users. The greater the discrepancy between the current and the maximal state of the resource, the greater must be the proportion of the population that uses the resource in a sustainable way if the resource is to recover. The estimated proportion of resource-sustaining users in the population is based upon the behaviour of contact persons perceived by the person. According to the ratio of contact persons using the resource sustainably and non-sustainably, the person makes an estimate for the entire population. This factor is at present not
Modelling environmental behaviour
85
yet weighted by socially, or environmentally, oriented values. For example, people who are not environmentally oriented overestimate the number of like-minded people. If no particular interventions take place, there is a danger that dynamics such as those shown in Figure 4.7a will develop. With an optimistic starting value with regard to average use within a population (over 50 = environmentally friendly use), the state of the resource briefly improves. Due to its improved state, its value declines (only goods in short supply are valuable), and the individuals in our simulated population resume stronger exploitation of the resource. From the fifth step onwards, this tendency results in clear over-utilization (average use under 50). This increases the discrepancy between actual and sustainable use patterns, and expectations of sustainable use correspondingly decrease. The individual is less motivated to make a personal contribution to sustainable patterns of utilization (‘… personal restraint on my own part would not make any difference; no one else is showing restraint, so it is better for me to help myself to the resource so long as it is still up for grabs …’). And so the state of the resource deteriorates. Its value rises, which does not, however, lead to a marked reduction in utilization. There is no stopping the course of these ‘downhill’ dynamics, and the resource is destroyed. To counteract these negative dynamics, we ran a campaign in the fifth step that aimed to (a) lead persons, in their own use behaviour, to orient themselves less to other persons’ patterns of use and (b) lend heavier weighting to the importance of the resource. Through this, the common property becomes more highly esteemed. We found that the utilization behaviour of the population becomes more environmentally sustaining (see Figure 4.7b). If individuals use a resource in an environmentally sustainable fashion on the average, the state of the resource improves. This line of development continues until the resource has regenerated. At this time, the resource is available in over-abundance so to speak, whereby its value again declines, and intensity of use increases. Downhill dynamics develop until that crucial point where they are again brought under control. A dynamicstable balance has emerged, in which utilization continually adapts to the state of the resource by means of inner personal factors. The resource will never be completely destroyed. Through the interaction of the social system and the resource, the resulting system behaviour shows large fluctuations. Following Forrester (1972: 48ff.), fluctuating system behaviour results when in a system of interlocking feedback loops two or more temporal delays occur. In the present simulation model, there are delays both within the social system and between the social system and the resource system. The average expectation that one can make an effective contribution to sustainable use of a resource reacts with a delay to the average resource use behaviour of the social surround. The rise and fall of the average value of the common good reacts with a delay to the development of the resource. Our simulation approach is not directly comparable to Forrester’s approach. Forrester observes only one macrosystem, while the simulation builds upon numerous microsystems (individual persons) which join to form the macrosystem (the social system).
86 Hans-Joachim Mosler Environmentally responsible/ resource maximal 100 `´ ýP
80
60
40
20
0 Not 0 environmentally responsible/ resource minimal
5
10
15
20
25
30 Runs
resource in % average use average value of common resource average expectation of sustainable use
Environmentally responsible/ resource maximal 100
80
60
40
20
0 0 Not environmentally responsible/ resource minimal
5
10
15
20
25
Resource in % Intervention
30 Runs
Average use Average value of common resource Average expectation of sustainable use
Note: Also illustrates the course of significant inner variables. 4.7a (top): control situation; 4.7b (bottom): with intervention.
Figure 4.7 Use of an environmental resource
Modelling environmental behaviour
87
Processing communicative influence What is the effect of convincing attempts (persuasion) in populations, according to individuals’ concern about the environment, knowledge of the environment, and biases? Here we apply the Elaboration Likelihood Model (Petty and Cacioppo 1986a, b). People process information in varying degrees of thoroughness: the ‘intensity’ of processing is a function of processing motivation and processing ability; areas of great personal relevance (concerned consternation about the environment) and self-responsibility (here responsibility for the state of the environment) increase motivation to process (Stahlberg and Frey 1993). Where deep, complete processing of information occurs, the results of campaigns to convince are primarily dependent upon the quality of the arguments presented; in the case of superficial processing, superficial cues, such as status, credibility and attractiveness of the communicator, gain more weight. Furthermore, individuals holding extreme attitudes or values tend to process information in a biased way: when a person’s own position is too removed from the attitude advocated by the influencing campaign, there can be a ‘boomerang’ effect, whereby the person changes his attitude in a direction opposing the persuasive arguments. If the advocated position is close to one’s own, there will be convergence between them. The principle mechanisms involved may be illustrated by a simplified, information-campaign experiment. The four external input variables in Figure 4.8 represent external influences on the processing individual. The three variables in the upper section of the figure represent influences stemming from the source of persuasion, a communicator. The communicator communicates a particular attitude position with a certain degree of persuasiveness. This persuasiveness variable corresponds to ‘argument quality’ in Petty and Cacioppo (1986b). The fourth external input variable that is processed is the (perceived) current state of the environment. From the left, two free, internal input variables – self-responsibility and values – flow into the submodel. These are not affected by the processes within the model, just as the status of the individual also remains constant. Concern arises from the processing of information on the state of the environment and from a person’s own values as related to the environment. Concern corresponds to the discrepancy between these personal values (the desired value for the state of the environment) and the actual value for the state of the environment (the effective, perceived state of the environment). This means that the greater the deviation between the individual’s values in relation to the environment and the perceived state of the environment, the greater the feeling of concerned consternation. This is true only as long as those values are at a higher level than the state of the environment, or in other words, only as long as the environment is in worse shape than the individual would like to see. The knowledge that the individual has, changes depending upon the degrees of persuasiveness the individual has. Environmental knowledge, in its function as an ability variable, determines the degree of processing depth, or intensity. Processing motivation results from the sum (addition) of the feeling of selfresponsibility – how strongly the person experiences a sense of responsibility for
88 Hans-Joachim Mosler INDIVIDUAL PROCESSING OF SOCIAL INFLUENCING UNDER CONDITIONS OF KNOWLEDGE AND CONCERN Communicator's persuasiveness Communicator's attitude Communicator's status Self-responsibility Processing motivation Perception of state of environment
Status
Concern Processing intensity
Attitude
Values Persuasiveness Knowledge
Processing ability
Figure 4.8 Representation of ‘processing of communicative influencing under conditions of knowledge and environmental concern’
the state of the environment – and environmental concern. Processing intensity (elaboration likelihood) arises in multiplicative fashion from processing motivation and processing ability. In addition, it is influenced by the status of the communicator (but only where neither processing motivation nor processing ability equals zero). A communicator’s positive status raises processing intensity, while negative status lowers processing intensity. The difference between the attitude position of the individual and the position advocated by the communicator determines attitude change – in dependency, however, upon the influences from central and peripheral processing. If processing intensity is low, peripheral processing will predominate, and the status of the communicator is the main determinant of attitude change. High status of the communicator leads to a change of attitude to the position advocated. Low status of the communicator leads to change of attitude opposing the advocated position. If processing intensity is high, however, central processing takes over, and the persuasiveness of the communicator is the main determinant of attitude change. High persuasiveness is expressed by high quality arguments, and the result is a change of attitude in the desired direction. Low persuasiveness, accompanied by weak arguments, leads to a change of attitude in the opposing direction. Biased processing, moreover, modifies central processing. The individual’s environmental values are compared with the attitude position advocated by the communicator. If they agree (when both are environmentally friendly or environmentally unfriendly), the communicator will successfully influence the person in the desired direction. If there is a discrepancy, however, the person’s attitude will change in the opposing direction.
Modelling environmental behaviour
89
To the individual’s initial persuasiveness, a central and peripheral proportion is added in proportion to processing intensity. Where processing intensity is high, the individual’s persuasiveness increases if the persuasiveness of the communicator is higher than the individual’s persuasiveness. Here the individual integrates the communicator’s good arguments into his own argumentation scheme. Where processing intensity is low, the individual’s persuasiveness increases if the persuasiveness of the communicator is higher than the individual’s persuasiveness. In contrast, if the persuasiveness of the communicator is low, the individual’s persuasiveness decreases. This assumption is our own and is not based upon the Elaboration Likelihood Model. The idea is that unreflected acceptance of weak arguments tends to weaken a person’s persuasiveness. We assume to this purpose (as an exception) that the individuals do not influence each other mutually, but that in five steps they are presented with information campaigns (the individuals do not discuss the subject of the environment among themselves, but they all stand under the influence of the campaigns). The campaigns promote strong environmentally friendly attitudes by means of good arguments (leading to high persuasiveness) and/or by means of strong peripheral cues (communicators having high status, as for example esteemed public figures). In addition, we assume that in the campaigns, an appeal can cause persons’ selfresponsibility to increase. Average attitudes and values are low on environmental friendliness, the extent of the sense of self-responsibility is low, and the state of the environment is poor. Persons having environmentally friendly values experience high concern under the condition of poor state of the environment. The following interventions were ‘tested’ (see Mosler et al. 1998):
• • • •
an information campaign presenting good arguments (Figure 4.9, persuasiveness+); an information campaign which works with well-known and esteemed personalities (Figure 4.9, status+); an information campaign based on good arguments and well-known, esteemed personalities (Figure 4.9, persuasiveness+/status+); an information campaign with well-known, esteemed personalities that moreover appeals to people’s sense of environmentally-oriented selfresponsibility (Figure 4.9, self-responsibility+/status+).
Let us first examine results with individual persons selected from the population. Person A – who is biased due to an environmentally non-friendly attitude – places a negative value even on good arguments (curve Persuasiveness+). In contrast, good arguments effect even more environmental consciousness in Person B, who has environmentally responsible attitudes from the start, as information processing is hardly biased here. A campaign using esteemed public figures is effective with both Persons A and B, if they tend to process information relatively superficially (curve Status+). But if Person A’s depth of processing is increased by means of self-responsibility (curve Self-responsibility+/Status+), they then show biased processing, and their attitude
90 Hans-Joachim Mosler Environmentally responsible 70 Person A 60
70
50
50
40
40 B C
B C
30 20 10
Person B
C B D A
60
C B D A
BC DA
30 20
BC DA
Not 0 environmentally responsible
A D
A D
0
5 Intervention 10
10 0
15 Runs
0
Environmentally responsible 90 Population
5 Intervention 10
15 Runs
Campaigns with: A
80 70
C B
60
A D
50
C B A D
Campaign: persuasiveness+
B
Campaign status+
C
Campaign: persuasiveness+/status+
D
Campaign: self-responsibility+/status+ Control situation
40 30 Not environmentally responsible
5
Intervention 10
15 Runs
Note: Above, individuals in the population. Below, population values. The three graphs present courses of attitude change in four different experiments and the control situation.
Figure 4.9 Information campaign experiments without environmentally relevant communication
changes in the direction away from that advocated by the campaign. In the total population, the average attitude becomes most changed by means of good arguments and presentation by esteemed personalities. But as shown in Figure 4.9, we must always take into consideration that these average values within the population hide very contrasting effects upon differing individuals. For various experiments with multiplicators (persons who advocate their environmental attitudes to others), a communication net was constructed. Each simulated person ‘talks’ about environmental issues with five friends and one stranger during each step, whereby the friendships are assumed to be lasting and reciprocal. The relationships to strangers are different in each step. Initial values in the population are the same as those described above for the information campaign experiment. Within the framework of an environmental action campaign, 500 persons (multiplicators) having very environmentally friendly attitudes receive training in argumentation (increasing persuasiveness, see Gonzales et al. 1988). In a variant of this experiment, the status of these persons is raised (as for example through public commendation awards). The following interventions were carried out:
Modelling environmental behaviour
• • • •
91
Multiplicators with good arguments (Figure 4.10, Persuasiveness+). Multiplicators with raised status (Figure 4.10, Status+). Multiplicators with good arguments and raised status (Figure 4.10, Persuasiveness+/Status+). Control situation.
In Figure 4.10 we again see two selected individuals A and B, who both stand under the influence of the action campaign. A glance shows that these graphs are not as smooth as the information campaign graphs. The curves would be smoother if, after a short levelling-off phase, there were only the influence of a constant group of friends. But the ever-changing influence of strangers, upon Persons A and B and also upon their friends, causes fluctuations in attitudes. We can still recognize certain tendencies in the development of attitudes, however. Person A’s attitude becomes less environmentally friendly relatively quickly. Later changes to a more responsible attitude are only brief and are repeatedly destroyed by the environmentally unfriendly social surround. Person B shows initial swings in attitude, but then develops a tendency towards a less environmentally friendly attitude. Both Persons A and B react to the introduction of multiplicators having high status (curve Status+) with an increasingly environmentally friendly attitude. This effect is even stronger if multiplicators in addition show high persuasiveness (curve Persuasiveness+/Status+). But if multiplicators demonstrate only persuasiveness (curve Persuasiveness+), they have no effect upon Person A. There is an initial effect in this case upon Person B, but the influence of the rather environmentally unfriendly social surround cannot be cancelled out. The same effects find expression at the level of the population: multiplicators with raised status are just as convincing as those who also show increased persuasiveness. Multiplicators who have only high persuasiveness at their disposal have a counterproductive effect.
Summary and conclusions The concept of the model, which is based on an elaborated rational choice theory, has proved to be – in connection with the simulation method – a valuable tool for the drafting, designing, and testing of environmental psychological interventions. The most important results of the experiments with the various sub-models can be summarized as follows: People who as a minority wish to effect more environmentally responsible attitudes in their group absolutely must have a low susceptibility to influence. If this condition is fulfilled, no measures for networking the minority persons need be undertaken. Role models can be implemented more successfully if these models have high status and their behaviour is highly visible. To achieve appropriate use of a resource, it is important to lead persons to orient themselves less to other people’s patterns of use and to pay more attention to the state of the resource.
92 Hans-Joachim Mosler Environmentally responsible 100
Person A
100
80 60
C
40
B
20 0 Not environmentally responsible
C D A B
C B
B C
C A D B
B 60 C
C B
B DA D A
40
C B
D A
C
C B
B
D A
A D
20 A D
A D
10
D A
D A
20
A D 30 Runs
0 10
60 C B A D
C B D A
30 Runs
A
Multiplicators: persuasiveness+
B
Multiplicators: status+
C B D
C B D
C B
C
Multiplicators: persuasiveness+/status+
D
D
A
Control situation
A
A
30 20 Not 10 environmentally responsible Intervention
20
Intervention
Environmentally responsible 80 Population 70
40
C
80
CB
Intervention
50
Person B
20
30 Runs
Note: Individual graphs show the course of attitude change in three different experiments and the control situation.
Figure 4.10 Experiments with multiplicators
Both in general campaigns and in specific action campaigns using multiplicators, it is advantageous to work with people who have high status. If attitudes within a population tend to be environmentally irresponsible, there is little to be gained through the use of persuasiveness or appealing to people’s sense of environmentallyoriented self-responsibility. A common denominator of the results can be formulated as follows, in analogy to Latané’s (1981) theory of social impact, according to which social influence is a function of the strength, immediacy, and number of sources. • • •
Strength: the concepts of susceptibility to influence, suggestibility, status, and persuasiveness belong here. Immediacy: this involves the way that persons are networked, the visibility of behaviour, and the degree to which a person may be influenced by others. Number: while it is always advantageous to work with a large number of pioneers, role models, and multiplicators, the financial means to do so may be limited. In this case, the same effect can be achieved with a smaller number of persons if measures are implemented to improve strength or immediacy.
Modelling environmental behaviour
93
In summary, we find a certain number of people equipped with a certain degree of strength and immediacy who will stimulate others to behave in more environmentally responsible ways. Success in getting 3, 5, or even 10 pe rcent of a community to become active does, however, seem an impossible task. According to the Swiss Environmental Survey (Diekmann and Franzen 1995), 16 per cent of people in Switzerland are members of environmental groups. This existing potential, dispersely distributed, needs to be seen as a resource that can be activated and brought together within the framework of concerted environmental efforts. If efforts are insufficiently co-ordinated, they fall flat. The conception of our experiments, with various forms of intervention as well as the application of terms suited to translation into action, might create the impression that we underestimate the problems of implementing in reality the forms of intervention studied. We are conscious of the fact that there are great gaps (theoretical as well as empirical) and incalculabilities between ‘population simulations’ based upon simulated individual behaviour and possible processes in real populations. Many sub-components are missing in our model or are as yet not adequately designed or validated. In spite of this, from our perspective there is no other alternative: we should indeed take on the challenge of examining the complexity of such dynamic processes by means of appropriate instruments. Simulation is an instrument well suited to enlarging our understanding of basic, underlying social processes and developing further those explanatory approaches that are almost exclusively static and based on one-person models. Against the possible view that our approach is far from reality, it must be objected that the forms of intervention we propose are founded upon empirical knowledge gained in small-scale field experiments (see Dwyer et al. 1993; Mosler and Gutscher 1998). In the next phase of research, we plan stage-by-stage testing of the simulation model as a logical step in its proof in practice. Here, in the sense of a practiceoriented ‘acid test’, we will strive towards process-oriented validation. To gain well-founded understanding of the dynamics of existing social systems, data must be collected in real social systems that are relevant to indications resulting from the simulation and that can then be entered into the simulation itself. The focus of the investigation would be the topic of dissemination of new behaviour patterns throughout existing social networks. The optimal method would consist in largescale field experiments. It is not our intention to propagate as ‘tested’ or ‘problem-free’ the application of those forms of intervention judged effective on the basis of our simulations. Rather, our purpose is to show environmentally conscious people, responsible persons, and politicians how worthwhile it can be to expand the old triad of ‘traditional information campaigns’, ‘legal measures’, and ‘economic measures’ by means of additional, novel strategies and to test these potentially successful forms of intervention in practice.
94 Hans-Joachim Mosler
References Ajzen, I. (1991) ‘The theory of planned behaviour. Some unresolved issues’, Organizational Behaviour and Human Decision Processes, 50: 179–211. Ajzen, I. and Fishbein, M. (1977) ‘Attitude–behaviour relations: a theoretical analysis and review of empirical research’, Psychological Bulletin, 84: 888–918. Ajzen, I. and Madden, T.J. (1986) ‘Prediction of goal directed behaviour: attitudes, intentions, and perceived behavioural control’, Journal of Experimental Social Psychology, 22: 453–74. Bandura, A. (1979) Sozial-kognitive Lerntheorie, Stuttgart: Klett-Cotta. Bandura, A. (1986) Social Foundations of Thought and Action, New York: Prentice-Hall. Becker, G.S. (1976) The Economic Approach to Human Behavior, Chicago: University of Chicago Press. Diekmann, A. (1996) ‘Homo ökonomicus. Anwendungen und Probleme der Theorie rationalen Handelns im Umweltbereich’, in A. Diekmann and C. Jaeger (eds) Umweltsoziologie, Sonderband 36 der Kölner Zeitschrift für Soziologie und Sozialpsychologie: 89– 118. Diekmann, A. and Franzen, A. (1995) Der Schweizer Umweltsurvey 1994, Codebuch, Bern: Institut für Soziologie. Dwyer, W.O., Leeming, F.C., Cobern, M.K., Porter, B.E. and Jackson, J.M. (1993) ‘Critical review of behavioural interventions to preserve the environment. Research since 1980’, Environment and Behaviour, 25(3): 275–321. Festinger, L. (1954) ‘A theory of social comparison processes’, Human Relations, 7: 117–40. Forrester, J.W. (1972) Grundzüge einer Systemtheorie, Wiesbaden: Gabler. Frey, D., Dauenheimer, D., Parge, O. and Haisch, J. (1993a) ‘Die Theorie sozialer Vergleichsprozesse’, in D. Frey and M. Irle (eds) Kognitive Theorien 2, Bern: Hans Huber: 81–122. Frey, D., Stahlberg, D. and Gollwitzer, P.M. (1993b) ‘Einstellung und Verhalten: Die Theorie des überlegten Handelns und die Theorie des geplanten Handelns’, in D. Frey and M. Irle (eds) Kognitive Theorien 2, Bern: Hans Huber: 361–98. Gonzales, M.H., Aronson, E. and Costanzo, M.A. (1988) ‘Using social cognition and persuasion to promote energy conservation: a quasi-experiment’, Journal of Applied Social Psychology, 18: 1049–66. Harsanyi, J.C. (1977) Rational Behaviour and Bargaining Equilibrium in Games and Social Situations, Cambridge: Cambridge University Press. Jorgenson, D.O. and Papciak, A.S. (1981) ‘The effects of communication, resource feedback, and identifiability on behaviour in a simulated commons’, Journal of Experimental Social Psychology, 17: 373–85. Klandermans, B. (1984) ‘Mobilization and participation: social-psychological expansions of resource mobilization theory’, American Sociological Review, 49: 583–600. Latané, B. (1981) ‘The psychology of social impact’, American Psychologist, 86: 343–56. Latané, B. and Wolf, S. (1981) ‘The social impact of majorities and minorities’, Psychological Review, 88: 438–53. Liebrand, W.B.G. (1986) ‘The ubiquity of social values in social dilemmas’, in H.A.M. Wilke, D.M. Messick and C.G. Rutte (eds) Experimental Social Dilemmas, Frankfurt am Main: Verlag Peter Lang: 113–33. Liebrand, W.B.G., Wilke, H.A.M. and Wolters, F.J.M. (1986) ‘Value orientation and conformity. A study using three types of social dilemma games’, Journal of Conflict Resolution, 30: 77–97.
Modelling environmental behaviour
95
Mosler, H.-J. (1990) Selbstorganization von umweltgerechtem Handeln: der Einfluss von Vertrauensbildung auf die Ressourcennutzung in einem Umweltspiel, Zürich: Zentralstelle der Studentenschaft. Mosler, H.-J. (1993) ‘Self-dissemination of environmentally-responsible behaviour: the influence of trust in a commons dilemma game’, Journal of Environmental Psychology, 13: 111–23. Mosler, H.-J. (1995) ‘Umweltprobleme: eine sozialwissenschaftliche Perspektive mit naturwissenschaftlichem Bezug’, in U. Fuhrer (ed.) Oekologisches Handeln als sozialer Prozess, Basel: Birkhauser: 77–86. Mosler, H.-J. (2000) Computersimulation sozialpsychologischer Theorien. Studien zur Veränderung von Umwelteinstellung und Umweltverhalten, Weinheim: Psychologie Verlags Union. Mosler, H.-J. and Gutscher, H. (1996) ‘Kooperation durch Selbstverpflichtung im AllmendeDilemma’, Kölner Zeitschrift für Soziologie und Sozialpsychologie, Sonderheft Umweltsoziologie, 36: 308–23. Mosler, H.-J. and Gutscher, H. (1998) ‘Umweltpsychologische Interventionsformen für die Praxis’, Umweltpsychologie, 2: 64–79. Mosler, H.-J. and Tobias, R. (2000) ‘Die Organisation kollektiver Aktionen durch Beeinflussung der individuellen Teilnahmeentscheidung. Eine Simulationsstudie’, Kölner Zeitschrift für Soziologie und Sozialpsychologie, 52(2): 264–90. Mosler, H.-J., Ammann, F. and Gutscher, H. (1998) ‘Simulation des Elaboration Likelihood Model (ELM) als Mittel zur Entwicklung und Analyse von Umweltinterventionen’, Zeitschrift für Sozialpsychologie, 29: 20–37. Nowak, A. and Latané, B. (1994) ‘Simulating the emergence of social order from individual behaviour’, in N. Gilbert and J. Doran (eds) Simulating Societies, London: UCL Press: 63–84. Nowak, A., Szamrej, J. and Latané, B. (1990) ‘From private attitude to public opinion: a dynamic theory of social impact’, Psychological Review, 97(3): 362–76. Petty, R.E. and Cacioppo, J.T. (1986a) ‘The elaboration likelihood model of persuasion’, in L. Berkowitz (ed.) Advances in Experimental Social Psychology, New York: Academic Press: 123–205. Petty, R.E. and Cacioppo, J.T. (1986b) Communication and Persuasion: Central and Peripheral Routes to Attitude Change, New York: Academic Press. Spada, H. and Ernst, A.M. (1990) ‘Wissen, Ziele und Verhalten in einem ökologischsozialen Dilemma’, Forschungsbericht, Freiburg: Psychologisches Institut der Universität Freiburg. Stahlberg, D. and Frey, D. (1993) ‘Das Elaboration–Likelihood–Modell von Petty und Cacioppo’, in D. Frey and M. Irle (eds) Kognitive Theorien 2, Bern: Verlag Hans Huber: 327–60. Stalder, J. (1985) ‘Die soziale Lerntheorie von Bandura’, in D. Frey and M. Irle (eds) Gruppenund Lerntheorien, Bern: Verlag Hans Huber: 241–72. Tajfel, H. (1978) Differentiation Between Social Groups, London: Academic Press. Tajfel, H. (1979) ‘Individual and groups in social psychology’, British Journal of Social and Clinical Psychology, 18: 183–90. Tanford, S. and Penrod, S. (1984) ‘Social influence model: a formal integration of research on majority and minority influence process’, Psychological Bulletin, 95(2): 189–225. Thibaut, J.W. and Kelley, H.H. (1959) The Social Psychology of Groups, New York: John Wiley.
96 Hans-Joachim Mosler
Assessing the quality of different MCDA methods
Part II
Taking multiple criteria into account
97
98 Andrea De Montis et al.
Assessing the quality of different MCDA methods
5
99
Assessing the quality of different MCDA methods Andrea de Montis, Pasquale De Toro, Bert Droste-Franke, Ines Omann and Sigrid Stagl
Introduction The awareness that environmental problems have a large impact on the economy and society has risen continuously during the last three decades. Lifting sustainable development to the political agenda nurtured the understanding that policy analysis is characterized by conflicting economic, environmental, societal, technical, and aesthetic objectives. Consequently numerous research projects on the interdependence of environment, society and economy were conducted in this period. With increasing knowledge about the economic and environmental processes and their interlinkages, it became clear that decisions about environmental problems are especially characterized by uncertainties, high stakes, urgency, and dispute (Funtowicz and Ravetz 1990). These characteristics influence the choice of the method. In the field of environmental decision-making, multiple criteria decision aid (MCDA) has become a widespread tool, able to tackle problems involving more than one objective. A number of theoretical (e.g. Munda 1995; Castells and Munda 1999) and empirical studies (e.g. Janssen 1992; Andreoli and Tellarini 2000; Mendoza and Prabhu 2000) showed that MCDA provides a useful tool for decision aid in this field, as it allows for multiple objectives, for the use of different types of data and the involvement of different stakeholders. The choice of a particular method depends on the specific problem as well as on users’ demands. As the number of existing methods is already quite large and is still increasing, the choice of the ‘right’ method is not an easy task. The aim of this chapter is the comparison of frequently used multiple criteria methods. A number of authors have compared multiple criteria algorithms with regard to their consistency, relative accuracy, speed, etc. (e.g. Olson 2001; Salminen et al. 1997). Since the underlying philosophies and assumptions, and hence the applicability, vary significantly, we chose, in contrast, a less mechanistic approach and describe the methods with reference to a comprehensive list of quality criteria. The criteria were chosen with particular emphasis on their relevance for questions of sustainable development. For the evaluation we apply the list of quality criteria to seven different MCDA methods (MAUT, AHP, Evamix, ELECTRE III, Regime, NAIADE, and programming methods) and compare them. This list of criteria
100 Andrea De Montis et al. provides a basis to assess the usefulness of each method for a specific application. The chapter is structured as follows. First, the list of quality criteria are presented, then the seven methods are characterized and evaluated by use of the list of criteria. This evaluation leads to a comparison of the methods, which is followed by conclusions.
Criteria for the quality assessment of methods To ease the selection of the appropriate method for a specific decision-making situation, a list of quality criteria was developed in De Montis et al. (2000) which can be used to reveal strengths and weaknesses of MCDA methods with respect to their application for environmental problems. These MCDA quality criteria can be grouped according to three main aspects: • • •
operational components of MCDA methods, applicability of MCDA methods in the user context, and applicability of MCDA methods considering the problem structure.
Table 5.1 lists the criteria together with a summarizing description. The first group of criteria includes operational components and characteristics of the method resulting from its theoretical underpinnings. The first important point for the method characterization is whether interdependence of criteria can be addressed in its application (e.g. equity or efficiency). Different technical procedures used to derive the decision may lead to different results in the same decision-making situation, because they use different mechanisms to compare criteria and alternatives. Some methods, for instance, do allow for incomparability, while others do not. Additionally, with regard to the characteristics of the criteria and the weighing process, the methods are distinct from each other. Beside the theoretical foundations and the technical components of the method, specific properties concerning the applicability of the method with respect to the user’s/users’ situation are important factors for the method characterization. Quality criteria concerning the project constraints (costs and time), and on the other hand criteria addressing the envisaged type of the structure of the problemsolving process (e.g. the possibility of stakeholder participation) are relevant factors. Another type of criteria which are relevant in the context of the methods’ applicability, are criteria which give information about the possibility for consideration of important issues within the specific problem-solving situation. One group of these criteria aims to examine the possibilities for the inclusion of indicators with specific characteristics necessary for the problem solution, e.g. the evaluation of indicators over different geographical scales. Another group of criteria characterizes the data situation, i.e. the type of data which can be used and the situation of data availability for the indicators’ evaluation in the context of the specific problem tackled. According to the milestone textbooks by Roy (1985; 1996), multiple criteria methods can be divided into three operative approaches (Roy 1996: 241):
Assessing the quality of different MCDA methods 101 Table 5.1 List of quality criteria for MCDA methods (De Montis et al. 2000) Quality criterion
Description
Operational components Criteria Interdependence Completeness Non-linear preferences Weights Transparency of process, type of weights Meaning Solution finding procedure Issues addressed by results
Allowance for the interdependence of different criteria Need for the completeness of the criteria Possibility to express non-linear valuation patterns Type of the procedure of deriving values for the weights Interpretation and role of weights in the evaluation process Type of procedure used for the comparison of alternatives Interpretation of the results generated by the use of method
Applicability of MCDA methods – user context Project constraints Costs Time
Implementation costs in the specific user situation Implementation time in the specific user situation
Structure of problem solving process Stakeholder participation Possibility to include more than one person as decision maker Problems structuring Existence of mechanisms supporting the structuring of the problem Tool for learning Support of learning processes Transparency Promotion of transparency in the decision making process Actor communication Support of the communication between opposing parties Applicability of MCDA methods – problem structure Indicator characteristics Geographical scale Applicability of different geographical scales for one case Micro–macro link Applicability of different institutional scales for one case Societal/technical issues Possibility for the consideration of both societal and technical issues Methods combination Possibility of methods combination Data situation Type of data Risk/uncertainties Data processing amount Non-substitutability
Type of data supported as values for the indicators Possibilities for the consideration of evaluation risk and/or uncertainties Processing amount needed to compile the data required for the method Possibility to consider sustainability standards and nonsubstitutability
102 Andrea De Montis et al. (a) methods based on the use of a single synthesizing criterion without incomparability, (b) methods based on the synthesis by outranking with incomparability, and (c) methods based on interactive local judgements with trialand-error iteration. While the first two groups embody a clear mathematical structure, the third does not use any formalized or automatic procedure. Another possible division refers to two groups of multiple criteria methods, namely discrete and continuous, depending on whether the set of alternatives is finite or infinite (Voogd 1983). In the following we will first present two types of discrete methods (single synthesizing criterion and outranking methods) and then the continuous methods (programming methods).
Single synthesizing criteria MCDA methods using a single-criterion approach are trying to convert the impacts concerning the different criteria into one criterion or attribute, which build the base for the comparison of the alternatives. A well-known example for a singlecriterion approach is cost–benefit analysis (CBA), which is using the Net Present Value as the criterion. This requires the conversion of impacts into monetary values. Hence, CBA cannot be seen as an MCDA method, because by use of MCDA methods each impact is evaluated on its most suitable scale; this scale can be both cardinal (monetary and non-monetary) and ordinal (qualitative). In the case of MCDA methods using a single synthesizing criterion, the aggregation of all the evaluations concerning each criterion results in a single index associated with each alternative and it expresses the performance of the alternative with respect to all the evaluation criteria simultaneously. Usually the higher this index the better ranks the alternative. Three such methods are described in the following: multiple attribute value theory (MAUT), analytic hierarchy process (AHP) and evaluation matrix (Evamix). Multiple attribute value theory (MAUT) The foundations of MAUT were laid by Churchman, Ackoff and Arnoff (1957) who first treated a multiple criteria decision problem using a simple additive weighting method. It was developed further using a quasi-additive and multi-linear utility function. Its early development is attributed to Fishburn (1968, 1970, 1978). Keeney and Raiffa devoted much of their work (e.g. 1976) to MAUT. Since then the method has been developed further in the way that methodologies to elicit the utility functions are elaborated, and the single-criterion utility functions are aggregated to a multi-attribute utility function, and it now provides a formal mechanism for handling multiple objectives, intangible factors, risk, qualitative data and timesequence effects in ex-ante appraisals based on the decision-maker’s preferences (Dillon and Perry 1977). A frequently used software package, which is based upon the estimation of an additive utility function and on interaction with the decisionmaker, is the so-called PREFCALC software. Other software packages are ‘Decide Right’, which is based on SMART, the Simple Multi-Attribute Rating Technique
Assessing the quality of different MCDA methods 103 (for explanation, see later in this chapter) and ‘HIPRE 3+’, which integrates two methods of decision analysis and problem solving, namely AHP – Analytic Hierarchy Process (see also p. 106 and Chapter 6 in this volume) and SMART. As the name MAUT indicates, it is based on utility theory (von Neumann and Morgenstern 1947). It relies upon the basic von Neumann–Morgenstern axioms of preference and thus upon a utility function, which allows the comparison of risky outcomes through the computation of expected utility. Risk can arise in two ways, as risk about the outcomes or about the attribute values. MAUT can be used for the first form of risk1 (Hwang and Yoon 1981). MAUT uses directly assessed preferences with general aggregation; this involves direct questioning of the decision-makers and the choice on the basis of an aggregate measure for each alternative (Dillon and Perry 1977). It can be seen as an extension of ordinary subjective expected utility procedures to the case of choice between multi-attributed alternatives. To prepare a multiple attribute decision by use of MAUT requires the following steps (Dillon and Perry 1977): 1 2 3 4 5
specify the project alternatives (including combinations) as discrete entities, elicit the decision-maker’s set of probability distributions for outcomes associated with each project alternative in each attribute if there is risk, elicit the decision-maker’s utility function ui(xi ) for the range of outcomes on each attribute, use the appropriate global multi-attribute utility function U(x) to find the expected utility of each project alternative, and choose the project or project combination with the highest expected utility; thus the function U should be maximized.
Step 1 comprises the definition of the alternatives, the objectives, the attributes and the outcomes of each alternative in each attribute. This step is not specific to MAUT; this is how all discrete MCDA methods start. Step 2 concerns risk viewed as subjective probability. If there is risk concerning the impact of the actions on the attributes, probabilities pij can be assigned to the outcomes x1, y1, … of the alternatives on the attributes. It is assumed that the value of the action a for attribute i is given by the expected value of ui (Vincke 1985; Dillon and Perry 1977): ui(xi ) = Sj pijui(xij ), where pij is the probability of the j-th outcome in the i-th attribute or in continuous terms, and ui(xi ) = Úui(xi ) ¶i (xi )dxi, where ¶i (xi ) is the probability distribution of outcomes in the i-th attribute. There are different methods to elicit subjective probability distributions, which are not without difficulties. Which method to choose depends on the problem and often on the experience of the researchers. Direct involvement methods are simple and allow the decision-makers to see and understand the built distribution. For these reasons, such direct involvement methods seem preferable to others, such as lottery-type questions (Dillon and Perry 1977). The simplest method of all is asking for a few point estimates of probability; this is crude but practicable. Step 3 requires the elicitation of the decision-maker’s utility function ui(xi ) for the range of outcomes on each attribute. A utility function ui(xi) is developed for
104 Andrea De Montis et al. each attribute. This is done by asking the decision-maker a series of questions based on the axiom of continuity (e.g. Dillon and Perry 1977: 10). Additionally weights wi are assigned to the attributes, presenting trade-offs, by asking for the decision-maker’s order of importance of the attributes. The assessment of the appropriate utility function is complex and intricate. Thus it is a crucial step in MAUT. This procedure leads us to three difficult questions (Vincke 1985): (a) what must the properties of the decision-maker’s preferences be so that U is a certain function of ui ?; (b) how to test the properties?; and (c) how to construct the function U? The method can only be applied if these questions can be answered in a meaningful way. After the establishment of the utility functions ui (xi ) and the elicitation of the attribute probability distributions, the expected utility is aggregated across the attribute distributions for each alternative in Step 4. For each attribute a function ui is built and the functions ui are aggregated in a global criterion U, such that U(u1(x1),…, un(xn )) > U(u1(y1),…,un (yn )), if the action represented by (x1,x2,…,xn ) is better than the action represented by (y1,y2,…,yn ), when considering all the attributes simultaneously (Vincke 1985; Roy and Bouyssou 1985). In that way the multiple criteria problem is reduced to a single-criterion decision problem. In order to build the aggregate function, different aggregation procedures are possible; the additive or the multiplicative aggregations are most widely applied. The additive form is the simplest form of a utility function, requiring two strong assumptions, namely utility independence and preferential independence of any subset of criteria (Fandel and Spronk 1985).2 SMART (simple multiple attribute rating technique) presents one method to build the additive form (Martel, in Climaco 1997). If the additive model holds, the expected utility of each attribute for each alternative is added. Each attribute must get a weighting factor wi to get a common scale for the utility functions ui(xi ). The utility for a certain alternative with uncertain consequences is then given by: U(x) = Swi [ui (xi )], where in discrete terms, ui (xi ) = Sj pij ui (xij ), pij being the probability of the j-th outcome in the i-th attribute; or in continuous terms, ui(xi ) = Úui (xi ) ¶i (xi )dxi , where ¶i (xi ) is the probability distribution of outcomes in the i-th attribute. If independence does not hold, the multiplicative form is usually applied. Another form is the quasi-additive model, which can be used if neither marginality nor preferential independence prevails; but this procedure becomes very complicated for cases with more than three attributes (Martel, in Climaco 1997). Step 5 consists of summarizing the results and interpreting them. A sub-methodology of MAUT is multiple attribute value theory (MAVT). The two approaches deal with risk in different ways. While MAUT relies upon a utility function, which allows the comparison of risky outcomes through the computation of an expected utility, MAVT is unable to take risk into account. It uses a value function to represent the outcome of the alternatives, not allowing for risky outcomes. Value functions preserve deterministic ordering, whereas utility functions preserve the expected utility hypothesis. MAUT is suited for ex-ante evaluations for multiple objective problems with risky outcomes as long as one accepts the assumptions on which the method is based.
Assessing the quality of different MCDA methods 105 MAVT is based on four key assumptions. First, a complete system of preferences pre-exists, which is an adequate representation of the axioms in the decision-maker’s mind. Second, the preference relation of the decision-maker is of a weak order, which means that all states are comparable, transitivity of the preferences and transitivity of the indifferences is given. The last one is particularly difficult as it can lead to a lack of realism; see for example the famous cup-of-coffee example by Luce (1956).3 Third, the main hypothesis underlying MAUT is that in any decision problem there exists a real-valued function U defined on A (the set of actions), which will be examined. This function aggregates the utility functions ui (Vincke 1985). Fourth, in the probabilistic case, when there is risk concerning the outcomes, the axioms of von Neumann and Morgenstern (1947) must hold. In this way the concept of rationality comes into MAUT, which can be seen as an extension of ordinary subjective expected utility procedures to the case of choice between multiattribute alternatives. They imply an existing unique set of subjective probability distributions for outcomes in each of the attribute dimensions associated with any risk multi-objective alternative and a utility function whose expected value gives a measure of the decision-maker’s preference for each of the risky multi-attribute alternatives. If the axioms are accepted, there exists both ‘a unique set of subjective probability distributions for outcomes in each of the attribute dimensions associated with any risky alternative and a utility function U(xi) whose expected value (in terms of the decision-maker’s subjective probabilities) gives a measure of the decision-maker’s preference for each of the risky multi-attribute alternatives’ (Dillon and Perry 1977: 9). The development of MAUT was a milestone in the history of MCDA, as it presented a formal and transparent method able to deal with multiple objectives, risk concerning the consequences, the pattern of consequences over time, and a start for using qualitative data. But MAUT’s main problem is the strong assumptions that need to be fulfilled to derive the global utility function. Basically, MAUT relies upon the same axioms as social welfare theory like CBA does and therefore a number of the points of critique towards CBA also apply to MAUT. A review of case studies shows that MAUT is used frequently for decision-making for economic, financial, and actuarial problems as well as problems concerning environmental issues, like water management, energy management, agriculture, but not really in the field of sustainable development, which is much broader as it covers an economic, a social, and an environmental dimension. Concerning the first group of quality criteria (see Tables 5.1 and 5.2) for MCDA methods, interdependence of attributes is not allowed; on the contrary the independence assumption is crucial. This assumption makes it hard for use in complex decision problems, as is the case for problems in the field of sustainable development. The meaning and assignment of weights is clear (weights stand for the importance of criteria) and transparent. The solution is a complete ranking of actions, according to their expected global utility. This again is clear and transparent, but the derivation of the utility function is the most delicate part of the whole procedure. Different issues concerning the type of decision problem are allowed, as long as they can be addressed by attributes; a utility function is designed for
106 Andrea De Montis et al. each attribute. Thus the more issues are addressed the more time it takes and the higher are the costs. For most decisions relating to sustainable development, stakeholder participation is necessary and desirable. Stakeholders are normally involved twice during the decision-making process. First, when developing the utility function for each attribute, they have to elicit their preferences. Second, when calculating the expected utility, they have to assign probabilities to certain outcomes. Preferences are directly assessed and generally aggregated and used to develop the utility functions. Under MAUT, transparency of the process is given as well as a clear structuring. Concerning the indicators and data, MAUT can handle different scales as well as different issues. The required amount of data is usually quite high. The data itself can be of qualitative or quantitative character. Risks of the outcome are accounted for by the expected utility theory, while uncertainties are checked with a sensitivity analysis. In sum, MAUT ensures in risky project choices the correspondence of the choice with the decision-maker’s preferences and it is a mechanism for doing this in cases too complex to be handled satisfactorily by intuition, as long as the underlying assumptions are known and always in the minds of the researchers. Information about the preferences of the decision-maker are necessary to build the singular utility functions for each attribute and to build the aggregate utility function based on all attributes. These individual utility functions build the base for the decision. No social function is constructed, thus no interlinkages or interdependencies between the individuals are taken into account. The building of a group or social function is a problem in every MCDA method, as we learned from Arrow’s impossibility theorem that it is not possible to aggregate individual preference functions to group opinions. The individual preferences and estimations can be seen as an acceptable proxy as long as the decision projects stay on a rather microrelated level, like the decision about an irrigation system. But if we have to deal with macro-related problems, additional influences gain importance, such as the impact on the different dimensions of sustainable development or on different groups in society. This decreases the appropriateness of individual utility functions as an acceptable proxy for social preferences. Thus MAUT, like CBA, is hardly suited for macro-related problems. Analytic hierarchy process (AHP) AHP was developed by Thomas Lorie Saaty. The method is implemented by the software package ‘Expert Choice’. From a procedural point of view this approach consists of three steps: (1) construct suitable hierarchies; (2) establish priorities between elements of the hierarchies by means of pairwise comparisons; (3) check logical consistency of pairwise comparisons (Saaty 1980, 1988; Saaty and Alexander 1989; Saaty and Forman 1993). Step 1 concerns the construction of suitable hierarchies. This step is based on findings indicating that when elaborating information, the human mind recognizes objects and concepts, and identifies relations existing between them. Because the human mind is not able to perceive simultaneously all factors affected by an action
Assessing the quality of different MCDA methods 107 and their connections, it helps to break down complex systems into simple structures: this simplification is possible by means of a logical process which aims at the construction of suitable hierarchies. Hence, hierarchies can be useful in helping the human mind to make decisions by constructing a framework of different and numerous elements, which are separated and connected at the same time. ‘A hierarchy is a particular type of system, which is based on the assumption that the entities, which we have identified, can be grouped into disjoint sets, with the entities of one group influencing the entities of only one other group, and being influenced by the entities of only one group’ (Saaty 1980: 7). The simplest model of hierarchy consists of three steps. The first step coincides with the main objective (called the ‘goal’) of the decision-making problem; the second and third steps include criteria and alternatives. It is, however, possible to develop more complex hierarchies (i.e. with more levels), which include a certain number of sub-criteria. This means that factors affecting the decision are organized in gradual steps from the general, in the upper level of the hierarchy, to the particular, in the lower levels. With reference to the same decision-making problem it is also possible to construct more than one hierarchy, for example, a first hierarchy for benefits, a second one for costs, and a third one for risks. In order to obtain satisfactory results, hierarchies should be large enough to capture all the major factors relevant for the decision-making problem but small enough to remain sensitive to changing crucial factors. Step 2 establishes priorities between elements of the hierarchies by means of pairwise comparisons (i.e. comparing elements in pairs with respect to a given criterion). In the AHP approach, pairwise comparisons are used for establishing priorities or weights among elements of the same hierarchical level. They are compared in pairs with respect to the corresponding elements in the next higher level, obtaining a matrix of pairwise comparisons. For representing the relative importance of one element over another, a suitable evaluation scale is introduced (Saaty 1988, 1992), called ‘Saaty’s scale’. It defines and explains the values 1 to 9 assigned to judgements in comparing pairs of elements in each level with respect to a criterion in the next higher level. In particular, the following values which express the intensity of importance between the elements are used: 1 3 5 7 9
if if if if if
two elements have ‘equal’ importance; ‘moderate’ importance of one element over another is recognized; ‘strong’ importance of one element over another is recognized; ‘very strong’ importance of one element over another is recognized; ‘extreme’ importance of one element over another is recognized.
The numbers 2, 4, 6, 8 can be used to express intermediate values between two adjacent judgements (i.e. between ‘equal’ and ‘moderate’, ‘moderate’ and ‘strong’, etc.). When element i compared with j is assigned one of the above numbers, then element j compared with i is assigned its reciprocal. In particular, for each criterion C, an n-by-n matrix A of pairwise comparisons is constructed. The components aij (i, j = 1, 2, …, n) of the matrix A are numerical
108 Andrea De Montis et al. entries, which express (by means of Saaty’s scale) the relative importance of the element i over the element j with respect to the corresponding element in the next higher level. Thus the matrix A has the form: a11 a21 A ≡ an1
a12 a1n a22 a2n an 2 ann
where: aii = 1 a ji =
1 aij
aij ≠ 0
In order to calculate relative priorities among the n elements considered, the ‘principal eigenvector’ (v, with usually Svi ¹ 1) of the matrix A is computed. Then this eigenvector is normalized obtaining the ‘priority vector’ (x, with Sxi = 1), which expresses the priorities among the elements belonging to the same node of the hierarchy. Each component of the vector x represents the ‘local priority’ of an element (i.e. a node of the hierarchy) of the pairwise comparisons; the ‘global priority’ of that element is the product of its local priority with the global priority of the upper node. Of course, the local priority of nodes at the first and second levels is equal to their global priority. Thus each element at each level has a weight (i.e. its global priority) assigned to it. The composite weight for each of the elements at the final level (i.e. the alternatives) is obtained by multiplying the weights along each path of the hierarchy from the apex to the final element and adding the resultant weights from all paths to the element. The result is a vector of final weights for the alternatives under consideration: the higher its weight the better the alternative is. Moreover, it can be examined how priorities among alternatives change according to the variation of weights assigned to the upper elements (‘sensitivity analysis’). Step 3 checks for the logical consistency of pairwise comparisons. In comparing elements, inconsistency of a certain degree can arise: in the AHP approach the ‘maximum or principal eigenvalue’ (called lmax) of each matrix of pairwise comparisons is computed for checking the degree of inconsistency.4 In particular the consistency of a positive reciprocal n-by-n matrix is equivalent to the requirement that its maximum eigenvalue lmax should be equal to n; if the matrix is inconsistent lmax > n it is possible to estimate the departure from consistency by the difference (lmax –1) divided by (n –1) (Saaty and Vargas 1991; Saaty 1994). The closer lmax is to n the more consistent is the matrix A and its deviation from consistency may be represented by the so-called ‘consistency index’ (C.I.): C .I . =
l max - n n -1
Assessing the quality of different MCDA methods 109 Saaty proposes a ‘consistency ratio’ in order to appreciate the consistency of the matrix which is calculated by dividing the C.I. by R.I., where R.I. represents an experimental ‘random index’ which increases as the order of the matrix increases.5 A consistency ratio of 0.10 or less is considered acceptable; if this ratio is more than 0.10, it is necessary to reformulate the judgements by means of new pairwise comparisons. The AHP method can take into account the interdependence existing between the different evaluation criteria by means of the construction of suitable hierarchies. The elements of a hierarchy (i.e. criteria, sub-criteria, alternatives) are not linked to all the others but they are grouped into disjoint sets; if an interdependence between some elements is recognized, they are placed in the same hierarchical level and connected to the same element of the next higher level. If the choice of the number of hierarchical levels is unlimited (although, as a rule of thumb, no more than seven elements for each node is best) it is possible to insert in the hierarchy all the criteria required by the decision-making problem. In this sense a complete (i.e. exhaustive) list of evaluation criteria is allowed. A cardinal weight is assigned to each element of the hierarchy by means of opportune pairwise comparisons. The weights express the importance of one element over the others. The final result offered by the method is a complete ranking of the evaluated alternatives. AHP facilitates stakeholder participation. Different social actors are invited to assign weights to the evaluation criteria; in this perspective weights reflect different social views. In comparing elements in pairs the different judgements given by each actor can also be integrated – by means of an arithmetic average – obtaining only one weight for each criterion, which expresses synthetically the points of view of all the involved stakeholders. It is also possible to promote the actors’ communication if they are called to work together in order to identify general objectives, evaluation criteria, sub-criteria, and deduce their importance. The method has been applied to a range of decision-making problems (e.g. corporate policy and strategy, public policy, political strategy, environmental planning) at different geographical scales. Sometimes also using other evaluation methods it is possible to make a pairwise comparison between criteria as carried out in the AHP if normalized weights are required. In this sense a combination with other methods is possible. AHP allows use of qualitative and quantitative data; in particular, cardinal numbers can be directly normalized without the need of pairwise comparisons. Evaluation matrix (Evamix) Evamix was developed by Henk Voogd. The evaluation matrix in this method may include mixed (i.e. qualitative and quantitative) data. Both the method and the software are called Evamix. From a procedural point of view this approach consists of five steps: (1) make a distinction between ordinal and cardinal criteria; (2) calculate dominance scores for all ordinal and cardinal criteria; (3) calculate standardized dominance scores for all ordinal and cardinal criteria; (4) calculate overall dominance scores; and (5) calculate appraisal scores (Voogd 1981, 1983).
110 Andrea De Montis et al. Step 1 makes a distinction between ordinal and cardinal criteria. The first step is the construction of an evaluation matrix E, which is an m-by-n matrix characterized by m evaluation criteria and n alternatives. Its components are qualitative or quantitative entries, which express by rows the performance of each alternative with respect to a certain criterion. Given a set of evaluation criteria j (j = 1, 2, …, m) and a finite set of alternatives i (i = 1, 2, …, n), the evaluation matrix E will be characterized by its qualitative and quantitative6 components ej i : Ê e11 e12 Áe e22 21 E∫Á Á Áe Ë m1 em2
e1n ˆ e2n ˜ ˜ ˜ emn ˜¯
The set of criteria j is divided into two subsets, denoted O and C, where O is the set of the ordinal (qualitative) criteria and C the set of cardinal (quantitative) criteria, obtaining two distinct evaluation matrices: EO (ordinal criteria/alternatives) and EC (cardinal criteria/alternatives). This way the differences among alternatives can be expressed by means of two dominance measures: the first one based on ordinal criteria and the second one based on cardinal criteria. In particular, in order to construct the cardinal dominance score, the components ej i Œ EC are standardized to a common unit; in this way all the quantitative criteria are expressed on a scale from 0 to 1. All the standardized scores should have the same direction; this means that a higher score implies a better score. In the case of criteria for which lower scores mean better scores, the standardized scores are transformed by subtracting them from 1. Additionally, a qualitative weight can be assigned to each criterion constructing a vector wj (j = 1, 2, …, m); the associated cardinal weights wj can be obtained – in an approximated way – by looking at the ‘extreme weights sets’ which delimit the values that metric weights may have.7 Because weights represent subjective value judgements, it is preferable to use different priority sets so that the consequences of different political viewpoints can be illustrated. Step 2 requires calculation of dominance scores for all ordinal and cardinal criteria. Differences between alternatives are expressed by means of two dominance scores, for ordinal (aii¢) and cardinal (aii¢) criteria:
(
a ii ¢ = f e ji , e ji ¢ , w j
(
aii ¢ = g e ji , e ji ¢ , w j
)
)
("j ŒO ) (" j ŒC )
where eji ( e ) represents the score of criterion j and alternative i ( i ¢ ), w j represents ji ¢ the weight attached to criterion j, and f and g are two different functions. If each extreme (quantitative) weight set w is substituted into the above formulae instead of the qualitative weights w j , it is possible to define – for each extreme weight set – a new dominance measure for ordinal ( a ii ¢ ) and cardinal ( aii ¢ ) criteria:
Assessing the quality of different MCDA methods 111
(
a ii ¢ = f e ji , e ji ¢ , w j
(
aii ¢ = g e ji , e ji ¢ , w j
)
)
("j ŒO ) ( "j Œ C )
The scores expressed by means of these formulae reflect the degree in which alternative i dominates alternative i¢ for ordinal and cardinal criteria respectively. Step 3 calculates standardized dominance scores for all ordinal and cardinal criteria. The dominance scores a ii ¢ and aii ¢ are standardized into the same measurement unit in order to make them comparable. If h is a standardization function, the ‘standardized dominance measures’ for all ordinal (d ii ¢ ) and cardinal (dii ¢ ) criteria have the following expression: δ ii ' = h(α ii ' ) d ii ' = h(a ii ' )
The standardization of the scores α ii ' and a ii ' can be done in three different ways: the ‘subtractive summation technique’, the ‘subtracted shifted interval technique’ and the ‘additive interval technique’.8 Also in this case the standardized dominance scores reflect the degree in which alternative i dominates alternative i ¢ for ordinal and cardinal criteria. Different scores are obtained using the three different techniques. Step 4 calculates overall dominance scores. The ‘overall dominance measure’ mii ' for each pair of alternatives (i, i ¢ ) – giving the degree in which solution i dominates solution i ¢ – is calculated by means of the following formula: mii ' = wO δ ii ' + wC d ii '
where wO represents the weight of the qualitative criterion set O and represents the weight of the quantitative criterion set C: wO = ∑ w j j∈O
wC
wC = ∑ w j j∈C
Step 5 calculates appraisal scores. The overall dominance measure mii ¢ may also be considered as a function k of the ‘appraisal scores’ si and si ¢ of the alternatives i and i ¢ : mii ' = k (s i , s i ' )
At the same time this means that the appraisal score of an alternative depends on the overall dominance score such as calculated in Step 4. In particular, three different appraisal scores of alternative i with respect to the n other alternatives are obtained if the subtractive summation technique, subtracted shifted interval technique or additive interval technique has been used in Step 3. Additionally, by
112 Andrea De Montis et al. using the appraisal score of all three techniques (t = 1, 2, 3) a standardized ‘average appraisal score’ ai for alternative i can be calculated: 3 Ê s - s- ˆ ail = Â Á ti+ ti- ˜ t =1 Ë sti - sti ¯
where sti is the appraisal score of alternative i for each technique; sti- is the lowest sti -score for each technique; sti+ is the highest sti -score for each technique. The result is a complete ranking of alternatives and the higher ai is the better is alternative i. Since the starting point of Evamix is the construction of an evaluation matrix, it is not possible with this method to take the interdependence between the different evaluation criteria into account. But if we consider that there is no limit on the choice of the number of criteria to insert in the evaluation matrix, it is possible to consider all the criteria required by the decision-making problem. In this sense a complete (i.e. exhaustive) list of evaluation criteria is allowed. An ordinal weight, expressing the importance between all the relevant criteria, is assigned to each criterion, and subsequently ordinal weights are transformed into cardinal ones. Evamix supports stakeholder participation. Different social actors are invited to assign weights to the evaluation criteria, and in this case the different weights given by each actor are not integrated but they are used to show the different points of view (i.e. economic, social, environmental) expressed by the stakeholders involved. It is also possible to promote actor communication if they are called to work together in order to identify general objectives, points of view, evaluation criteria, sub-criteria, and deduce their importance. The method is used to cope with different decision-making problems at different geographical scales. Most frequently, Evamix been used in urban and regional planning. Evamix allows using qualitative and quantitative data. In particular, quantitative data are expressed in an ordinal scale and the cardinal numbers are standardized in a scale from 0 to 1.
Outranking methods The development of the outranking methods was started in France in the late 1960s by Bernard Roy and his team (Roy 1985). Outranking methods were developed in an attempt to manage with less strong assumptions (about existence of utility function, additivity, etc.) and to require less information from decisionmakers (especially preference intensities, rates of substitution) than the methods described above. The outranking methods also differ fundamentally from other methods in the approach insofar as they start from the assumption that the decision is a process during which decision-makers may change their preferences in response to the information provided and/or after thorough reflection about the problem. For this reason the developers of outranking methods considered it important to allow at the beginning for incomparability of options. This avoids a complete ranking being identified too early; the aim is therefore to stimulate thorough consideration
Assessing the quality of different MCDA methods 113 and to avoid premature conclusions. Hence, more than other methods, these methods encourage interaction between the model and the decision-maker/s. The outranking algorithms also account for the political reality that options which perform very poorly in one dimension are likely to face severe opposition from at least some of the stakeholders and are therefore deemed unacceptable. ELECTRE III Roy (1985) developed one of the most widely used software packages, known as ELECTRE, which is available in four different main releases (ELECTRE I, II, III, IV) and two other versions (ELECTRE IS and TRI). A comprehensive comparison of these approaches can be found in Vincke (1992). The underlying principle of the ELECTRE approach stems from processing a system of pairwise comparisons of the alternatives. This procedure involves the use of a system of preference relations (SPR) that is based on the exploitation of the outranking relation. According to this relation, an action will outrank another if the first action is considered to be at least as good as the second. Roy (1996) defines a system like this as a basic system of outranking relations. Recently it was pointed out that release III is remarkable among the ELECTRE family, since it takes explicitly into account the indifference and preference thresholds but, furthermore, it is based on an outranking relation, which is less sensible to the variation of the input data (Scarelli 1997). The system solves ranking problems, dividing the set A of actions into equivalence classes (Scarelli 1997). In release III of ELECTRE, the process can be sketched in the following steps. A cardinal impact table is prepared for a finite number of alternatives ‘measured’ according to a set of pseudo-criteria. The latter are criteria whose SPR includes weak preference, indifference and incomparability, and a coherent system of thresholds has to be set. The program allows to set thresholds for indifference and for preference, according to a linear relationship: T = a • g(a) + b, where T is the total threshold, a is a proportional threshold, b is a fixed threshold and g is the criterion function. This threshold is defined directly when it refers to the maximum value of the criterion, and indirectly when it refers to the minimum value. This allows constructing fuzzy outranking relations, given that it is possible to state how much an action outranks another. In fact, a series of pairwise comparisons is set through the use of concordance and discordance indices. A general framework for the calculation of these indices is given by Voogd (1983). The degree of dominance of alternative i over alternative i¢ is measured by the following index: 1/ a
È Â j Œy w aj ˘ ˙ cii ¢ = Í a Í Â jwj ˙ Î ˚
where y is the set of criteria that indicate a preference for the alternative a with respect to the alternative a ¢ , w the criterion weight and a a scaling parameter greater than one. ‘By means of this scaling parameter the analyst is able to vary
114 Andrea De Montis et al. the importance of the small weights and small divergences between the scores’ (Voogd 1983: 123). In many applications, the value of the scaling parameter is set equal to one, which allows managing a simpler functional algorithm. The degree of discordance between the same alternatives is measured by the following index: 1/ a
a È Â j Œj w j | e ji - e ji ¢ | ˘˙ Í dii ¢ = a Í w | e - e ji ¢ | ˙˙ ÎÍ Â j j ji ˚
(
(
)
)
where j is the set of criteria that do not agree about the preference of alternative i with respect to alternative i ¢ . In the generalized concordance analysis the values of the concordance and discordance indices individuate the subsets of alternatives that are accepted and rejected. The first group consists of the alternatives that fulfil the following requirements: Tc e ≥ h and Td ≥ m. In ELECTRE III, the concordance and discordance indices are calculated with reference to the values that the criterion functions take for the consequences of each alternative. In the same pattern, the threshold functions are modified as linear functions of the pseudo-criterion g. For each criterion gj a preference threshold pj and an indifference threshold qj are set. Given the pseudo-criterion g and two variable alternatives i and i ¢ , two concordance indices cj (i, i ¢ ) are calculated for the comparison of alternative i with respect to alternative i ¢ and vice versa. For the comparison between alternative i and i ¢ , the concordance index takes values between 0 and 1. The discordance index dj (i, i ¢ ) is calculated with reference also to the veto threshold vj . This threshold is associated with a criterion that is considered to be very important. If the difference between two functional values in the case of a number of alternatives is greater than the threshold value, this condition acts as a veto to the affirmation: the first alternative outranks the second. The construction of the outranking relation is based upon the credibility index s(i, i ¢ ), which ranges between 0 and 1 and is a function of the concordance and of the discordance index. The credibility index takes the following values: s (i, i ¢ ) = c (i, i ¢ ), if dj (i, i ¢ ) < cj (i,); otherwise s (i , i ¢ ) = c j (i , i ¢ ) ’
1 - d j (i , i ¢ ) 1 - c j (i , i ¢ )
The credibility index provides the information for the construction of two rankings of the alternatives. First, from the worst to the best (ascending distillation) and the second from the best to the worst (descending distillation). The intersection of these distillations yields the final distillation graph. This graph indicates the
Assessing the quality of different MCDA methods 115 system of relationships among the alternatives, according to a system of oriented arcs. This framework reveals incomparability relationships that the other distillations do not consider. Roy (1996) considers that in every decision-making situation, four problem areas have to be taken into account: Pa (choice), Pb (sort), Pg(rank), and Pd (description). With reference to this outline set forth by Roy (1996), the procedure does not deliver a complete ranking of the evaluated alternatives, leaving partially unresolved what Roy (1985) calls the problematic P. g (ranking of the alternatives). On the other hand, the procedure of ELECTRE III is helpful for what Roy (1985) names the problematic P.a (choosing the best action). ELECTRE III is based on the assumption that the system of individual preference relationships can be described by means of a sophisticated analysis of the relationships of preference, indifference, and incomparability. As a matter of fact, incomparability enters as a puzzling factor into the final ranking. On the one hand, this complex analysis involving the assessment of several thresholds and preference relationships may be judged as a remarkable effort. On the other hand, it may lead to cumbersome work. Often it is very difficult to express credible threshold values and the resulting ranking can end up being hardly understandable. As a matter of fact, in the situation where results were described by a Kernel graph, the complex meaning stemming from this kind of representation would be hardly transmissible to a group of laypersons, such as a local community. Local communities seem to be more keen to comprehend and discuss definitive positions and comparable outcomes than to give personal interpretation of relationships. The nature of the ranking, which is presented by a graph, allows also for incomparability among the alternatives: such an incompletely resolved outcome challenges the analyst who should then, if equipped with abilities of good advocacy, foster communication among stakeholders to discuss the incomparability. Politicians often are unwilling to accept the complexity of this structure and show a tendency to sceptical behaviour against an incomprehensible technique for decision-making. Often it is the expectation that the multiple criteria analysis will solve the problem at hand completely. The obvious fact that MCDA supports decision-making, but must not substitute decision-makers, becomes even more apparent with outranking methods than with the previously discussed ones. These are the main reasons why in Table 5.2 ELECTRE is rated highly with respect to criteria completeness and interdependence, while it is rated poorly with respect to transparency and to applicability of criteria. It should be noted that in terms of ‘meaning of weights’ ELECTRE received a zero score, since it does imply weighting procedures connected to the combinatory use of the so-called ‘weights’ and to the thresholds. Regime The Regime method was developed by Hinloopen et al. (1983), assessed and refined by Hinloopen (1985) and Hinloopen and Nijkamp (1990). This method can be considered as a multiple criteria evaluation framework belonging to the family of dominance analysis approaches, even though it has been pointed out that it is
116 Andrea De Montis et al. quite distinct from other methods in this family, mainly because Regime is a qualitative multiple criteria method. Qualitative multiple criteria methods aim to provide a tool for analysing complex situations, which are impossible to model by means of quantitative information. Hinloopen and Nijkamp (1990) list several such methods: (a) the extreme expected value method (Kmietowicz and Pearman 1981; Rietveld 1980), (b) the permutation method (Mastenbroek and Paelink 1977), (c) the frequency method (Van Delft and Nijkamp 1977), (d) the multidimensional scaling method (Nijkamp and Voodg 1981), and (e) the mixed data method (Voogd 1983). Since human activities and settlements are characterized by phenomena that are often impossible to measure, studies have focused on the construction of techniques able to deal with ‘soft’ and ‘fuzzy’ information. ‘These qualitative multidimensional judgement methods are essentially more in agreement with “satisfying behaviour” (based on Simon’s bounded rationality principle) than with conventional “optimising decision making”, as imprecise information precludes the attainment of an unambiguous maximum solution’ (Hinloopen and Nijkamp 1990: 38). As in Evamix, an evaluation table is given and composed by eij scores of a number n of alternative scenarios with respect to m criteria. In the case of ordinal information the weight can be represented by means of rank orders wj in a weight vector w: w = (w1, w2, …, wj )T. The higher the value of the weight, the better the correspondent criterion. The following main features characterize the multiple choice regime method. First, with this method the evaluation table can contain cardinal as well as ordinal data; this is accomplished by treating cardinal information as ordinal, with reference to the ranking position of each alternative. Second, the basis of the method is the regime vector. For two alternative choice options i and i ¢ , the difference of the criterion scores is assessed by considering the difference between the scores sii¢j = eij – eei¢j. In case of ordinal information, the order of magnitude of sii¢j is not relevant, but only its sign qii¢j . Therefore, if qii¢j = +, then alternative i is better than alternative i¢ for criterion j; i¢ qii¢j = –, then alternative i is better than alternative i¢ for criterion j. The extension of the pairwise comparison to all the alternatives leads to the following J ¥ 1 regime vector: rii¢ = (sii¢1,sii¢2, sii¢3, … sii¢n ). The regime vector, composed of + or – signs, or eventually zeros, ‘reflects a certain degree of (pairwise) dominance of choice option i with respect to i¢ for the unweighted effects for all J judgement criteria’ (Hinloopen and Nijkamp 1990: 41). For all the combinations of comparisons, a regime matrix is drawn with the following pattern: R = ÍÎ r12r13 … rI 1r21 ……… rIa ……… rI ( I -1) ˙˚
Usually, the regime vector is not completely composed of a series of ‘+’ or a series of ‘–’. Instead it is often composed of both signs. Additional information has to be processed by means of ordinal values, on the assumption that the ordinal nature of the weights reflects the inability to measure human preferences, which in principle are cardinal. This weight vector can be considered as a ‘rank order
Assessing the quality of different MCDA methods 117 representation of an (unknown) underlying cardinal stochastic weight vector’ (Hinloopen and Nijkamp 1990: 41): w* = (w1* , w2* ,…w*j )T with 0 < w*j < 1. The consistency hypothesis implies that the weight vector is supposed to be consistent with the quantitative information incorporated in an unknown cardinal vector. The next assumption is that the dominance relation can be represented by means of the following stochastic expression referred to the following weighted linear summation of cardinal entities: n ii ¢ = Â s ii ¢j w*j j
If nii¢ is positive, then a dominance relation of choice option i over i¢ is revealed. Since the information about w*j is unknown, the probability of the dominance of i with respect to i¢ is introduced in the following pattern: pii¢ = prob (nii¢ > 0). Thus an aggregate probability measure (or success score) can be defined as follows: pi =
1 Â pii ¢ I - 1 i ¢π i
Hence, pi is the average probability that alternative i is given a higher value than another alternative. The rank order of choice options is then determined by the rank orders (or the order of magnitude) of the pi. Since the assessment of the pii¢ is the crucial point, a uniform density function is usually assumed for the probability distribution of the w*j . This argument is referred to as the ‘principle of insufficient reason’, and as the Laplace criterion, in case of decision-making under uncertainties (Taha 1976). According to this statement, without any prior information, ‘there is no reason to assume that a certain numerical value of w* has a higher probability than any other value’ (Hinloopen and Nijkamp 1990: 42). Regime differs in some important points from other outranking methods. While ELECTRE requires cardinal rankings, the Regime method allows using mixed data so that cardinal and ordinal criterion can be included. The Regime method adopts ordinal weights; this feature seems to depict real systems of preferences better than other multiple criteria weighting methods. This approach is based on more prudent assumptions as it refrains from the difficult task of attaching cardinal values to measure the intensity of criteria importance. In terms of shortcomings and benefits, this method is in a sense the counterpart of ELECTRE III. It has a simpler structure of the preference model than ELECTRE III, since qualitative and quantitative data can be used for criteria. In other words, this method is able to process mixed data, because it can adopt an ordinal scale for the impact scores. The result of the evaluation process with Regime is a complete ranking of the alternatives. The outcome of the analysis is quite easy to communicate and discuss. However, the Regime method does not admit incomparabilities among alternatives. Also it is not clear how the system operates in the stochastic domain. This is mainly due to the random generation system adopted for the weight vector: the analyst, involved in the task of explaining the
118 Andrea De Montis et al. exact mechanism of the algorithm, may meet significant difficulties in communicating the meaning and in involving the decisional community. Hence, despite the apparently simple structure of the Regime method, the resulting decision-making process may not be so transparent. Novel approach to imprecise assessment and decision environments (NAIADE) This was a MCDA method developed by Giuseppe Munda (1995), whose impact or evaluation matrix may include quantitative and qualitative data; in particular the values assigned to the criteria for each alternative may be expressed in the form of either crisp, stochastic, fuzzy numbers or linguistic expressions. Hence it allows the use of information affected by different types of uncertainty. It is a discrete method, and no weighting of criteria is used explicitly. The method is implemented by a software application called NAIADE. From a procedural point of view the method consists of three steps: (1) the pairwise comparison of alternatives; (2) the aggregation of all criteria; (3) the evaluation of alternatives (Munda et al. 1994; Munda 1995; Menegolo and Guimarães Pereira 1996). Step 1 requires pairwise comparison of alternatives. The first step is the construction of an evaluation matrix E, which is an m-by-n matrix characterized by m evaluation criteria and n alternatives. Its components are qualitative or quantitative entries, which express by rows the performance of each alternative with respect to a certain criterion. Given a set of evaluation criteria j ( j = 1, 2, …, m) and a finite set of alternatives i (i = 1, 2, …, n), the evaluation matrix E will be characterized by its qualitative and quantitative9 components eji: Ê e11 e12 Áe e22 21 E∫Á Á Áe Ë m1 em2
e1n ˆ e2n ˜ ˜ ˜ emn ˜¯
and the pairwise comparison of alternatives (with reference to each criterion) is carried out by means of the concept of ‘semantic distance’ between two fuzzy sets.10 More precisely, if S1 and S2 are two fuzzy sets and m1 and m2 are respectively their membership functions, it is possible to define two new functions: f ( x ) = k1m1 ( x )
and
g ( y ) = k2m 2 ( y )
obtained by rescaling the ordinates of m1 and m2 through k1 and k2 such as: +•
Ú
-•
+•
f ( x )dx =
Ú g ( y )dy = 1
-•
Assessing the quality of different MCDA methods 119 The semantic distance between the two fuzzy sets (that is the distance between all points of their membership functions) is computed. The same distance concept can be extended to stochastic measures considering f(x) and g (y) probability density functions of the measures. The pairwise comparison of alternatives is based on six ‘preference relations’: 1 2 3 4 5 6
much greater than (>>) greater than (>) approximately equal to (@) very equal to (=) less than (<) much less than (<<)
expressed for each criterion starting from the distance between alternatives. The above preference relations are analytically defined by means of six functions that express (for each criterion on the base of the distance between alternatives) an index of credibility of the statements that an alternative is ‘much greater’, ‘greater’, approximately equal’, ‘very equal’, ‘less’ or ‘much less’ than another.11 This credibility index goes, increasing monotonically, from 0 (definitely non-credible) to 1 (definitely credible). In particular, given a criterion j and a pair of alternatives i and i¢, it is possible to define six membership functions (different for crisp criterion scores and for stochastic or the fuzzy criterion scores), which are respectively named12: m >> (i , i ¢ ) j
(for much greater than)
m > (i , i ¢) j
(for greater than)
m @ (i , i ¢ ) j m = (i , i ¢ ) j
(for approximately equal to) (for very equal to)
m < (i , i ¢) j m << (i , i ¢ ) j
(for less than) (for much less than)
Membership functions related to stochastic or fuzzy numbers consider in their mathematical expressions the semantic distance defined above. Step 2 requires aggregation of all criteria. In order to take into account all criteria simultaneously, it is necessary to aggregate the evaluations related to the pairwise performance of alternatives according to each single criterion. A ‘preference intensity index’ of one alternative with respect to another is calculated:13 m * (i , i ¢ ) = f ( m * (i , i ¢ ) j )
and then an aggregate ‘fuzzy preference relation’ can be obtained:
120 Andrea De Montis et al. Ïm >> (i , i ¢ ) Ôm ( i , i ¢ ) Ô > Ôm @ (i , i ¢ ) Ì Ôm = ( i , i ¢ ) Ôm < ( i , i ¢ ) Ô Óm << (i , i ¢ )
H ( >> ) H (>) H (@) H (=) H (<) H ( << )
where m* (i , i ¢ ) is the overall evaluation of a given fuzzy relation for each pair of actions and H(*) is the associated entropy level.14 Step 3 is the evaluation of alternatives. The evaluation of the alternatives derives from the information provided by the above aggregate fuzzy preference relation, defining for each action two functions f+ and f –, from which separate rankings of alternatives are obtained. The function f+(i) is based on the ‘better’ and ‘much better’ preference relations and with a value from 0 to 1 indicates how the alternative i is ‘better’ then all other alternatives. The second function f–(i) is based on the ‘worse’ and ‘much worse’ preference relations, its value going from 0 to 1 which indicates how the alternative i is ‘worse’ than all other alternatives (Menegolo and Guimarães Pereira 1996). The final ranking of alternatives comes from the intersection of two separate rankings obtained by means of the functions f+(i) and f-(i), taking into account that it can also be an incomplete ranking because nondominated alternatives are calculated. Additionally, NAIADE allows for another type of evaluation. It analyses conflicts between different interest groups and the possible formation of coalitions according to the proposed alternative options. Besides the ‘impact matrix’, also an ‘equity matrix’ is constructed, which contains linguistic evaluations of different social groups for each alternative. In particular, ‘equity analysis is performed by the completion of an equity matrix from which a similarity matrix is calculated. Through a mathematical reduction algorithm, it is possible to build a dendrogram of coalitions which shows possible coalition formation, and a level of conflict among the interest groups’ (Menegolo and Guimarães Pereira 1996: 1). Because the starting point of NAIADE is the construction of an evaluation matrix, it is not possible with this method to take into account the interdependence existing between some different evaluation criteria. Since there is no limit on the choice of the number of criteria that can be inserted in the evaluation matrix, it is possible to consider all the criteria required by the decision-making problem. In this sense a complete (i.e. exhaustive) list of evaluation criteria is allowed. No weights are assigned to the criteria. The final result offered by the method can be an incomplete ranking of the evaluated alternatives. In fact, the final ranking comes from the intersection of two separate rankings which allows to calculate also the non-dominated alternatives. Stakeholder participation is explicitly supported in NAIADE. The method allows the construction of an equity matrix which reflects how the different social groups
Assessing the quality of different MCDA methods 121 involved evaluate the alternatives. It is also possible to promote the actors communication if stakeholders are called to discuss their evaluation of the alternatives. The method is used to cope with different decision-making problems (it has especially been used for problems of unsustainability) at different geographical scales. NAIADE allows using qualitative and quantitative data. In particular, qualitative data are expressed by means of linguistic evaluations and quantitative data may be expressed either in crisp, stochastic or fuzzy numbers.
Programming methods Multi-Objective-Programming (MOP), Goal Programming (GP) and their variants are continuous multiple criteria methods. In contrast to the methods presented so far, the programming methods do not rank or sort a finite number of alternatives, but the alternatives are generated during the solution process on the basis of a mathematical model formulation. The roots of GP lie in the work by Charnes and Cooper in the early 1960s. After the mid-1970s it became widely applied due to seminal work by Lee (1972) and Ignizio (1976). A number of specific software tools were developed, but often general programming software packages (like GAMS) are applied. From a procedural point of view MOP consists of two steps: (1) identification of Pareto efficient (also called non-dominated) solutions; and (2) identification of the most preferred alternative together with the decision-maker(s). Step 1 consists of finding the non-dominated solutions. The problem is formulated as a task of simultaneous maximization/minimization of several objectives subject to a set of constraints. In a so-called criterion space it can be defined as follows: ‘max’ q, s.t. q Œ Q, where Q is a so-called feasible region in the criterion space. The set Q is not specified directly, but by means of decision variables as usually done in single optimization problems: ‘max’ q = f(x) = (f1(x),...,fk(x)), s.t. x Œ X, where X is a feasible set. The functions fi, i = 1, 2, ..., k are objective functions (Korhonen 2001). Since the simultaneous optimization of all objectives is impossible – given a certain level of conflict between them in most real problems – MOP tries to find the set of Pareto efficient solutions. According to Ballestero and Romero (1998) a necessary condition to guarantee the rationality of any solution to an MCDA problem is the following: ‘A set of solutions in a MCDA problem is Pareto efficient (also called non-dominated), if their elements are feasible solutions such that no other feasible solution can achieve the same or better performance for all the criteria being strictly better for at least one criterion’ (p.7). The goal is therefore to divide the feasible set into a subset of Pareto efficient solutions and a subset of Pareto inefficient solutions. Step 2 requires choosing the most preferred solution. Any choice from among the set of efficient solutions is an acceptable and ‘reasonable’ solution, as long as we have no additional information about the decision-maker’s preference structure. The final solution q Œ Q is called ‘the most preferred solution’. It is a solution preferred by the decision-maker to all other solutions (Korhonen 2001). Since we
122 Andrea De Montis et al. are always facing a number of conflicting goals in multiple criteria problems, a solution can never be optimal (or ideal), but rather a compromise. With regards to how much we need to know about the decision-maker’s preferences, different approaches were developed: there are (a) those that need knowledge of the decisionmaker’s preference; (b) those in which the decision-maker’s preferences are preemptively determined; and (c) those that progressively reveal the preferences of the decision-maker through man–machine interactions (interactive methods) (Kalu 1999). The latter typically operate with decision-maker’s aspiration levels regarding the objectives on the feasible region. The aspiration levels are projected via minimizing so-called achievement scalarizing functions (Wierzbicki 1980). No specific behavioural assumptions, e.g. transitivity, are necessary (Korhonen 2000). At the beginning of MOP research the expectation was that computing all efficient extreme points and then asking the decision-makers to examine their criterion vectors to select the best one could solve the problem. However, after vector-maximum codes became available, it was found that the number of efficient extreme points generated by MOP was far larger than had been expected. As a result, interactive procedures for exploring efficient sets for the best solution became popular. Many different non-reference point and reference point procedures are available now (Steuer and Gardiner 1990). In GP the impossibility of building a reliable mathematical representation of the decision-maker’s preferences due to conflicts of interest and incompleteness of available information is addressed in a different way. Two major subsets of GP models can be distinguished. In the first type it is assumed that within this kind of decision environment, decision-makers attempt to achieve a set of relevant goals as close as possible to the set of targets established (Ballestero and Romero 1998). The algebraic representation is given as the sum of unwanted deviations from target values if a number of objectives are minimized (weighted GP): k
min z (ui ni + vi pi ), s.t. f i (x ) + ni - pi = bi , i = 1…Q , x ŒC5 , i =1
where fi(x) is a linear function (objective) of x, and bi is the target value for that objective. ni and pi represent the negative and positive deviations from this target value. ui and vi are the respective positive weights attached to these deviations in the achievement function z (Tamiz et al. 1998: 570). The methodology rests on the following basic scheme (Vincke 1992): (1) set the values one wishes to attain on each criterion (the objectives); (2) assign priorities (weights) to these objectives; (3) define (positive or negative) deviations with respect to these objectives; (4) minimize the weighted sum of these deviations; and (5) perform a sensitivity analysis. In practice, the weights are sometimes derived from the pairwise comparison of AHP. Alternatively, they can be derived by use of an interactive MCDA method, like the one by Zionts and Wallenius (1976) (as shown in Lara and Romero 1992). Variants of this type (Chebyshev GP, MinMax GP, and Fuzzy GP) seek to minimize the maximum deviation from amongst the set of unmet objectives instead. In the case of another variant (Preference-Function GP) the relationship between the deviation
Assessing the quality of different MCDA methods 123 from the target and the penalty imposed is not the standard linear one, but it may be piecewise linear, discontinuous, or even non-linear (Tamiz and Jones 1997). In the other major subset of GP the deviational variables are assigned to a number of priority levels and minimized in a lexicographic sense (lexicographic GP). The assumption of non-continuity of preferences implies the impossibility of ordering the decision-maker’s preferences by a monotonic utility function. Lexicographic minimization means a sequential minimization of each priority while maintaining the minimal values achieved by all higher priority level minimizations. In mathematical terms this means: Lex min a = (g1(n,p), g2(n,p), ... , gL(n,p)), s.t. fi(x) + ni – pi = bi, i = 1, ..., Q The given model has L priority levels, and Q objectives. a is an ordered vector of these L priority levels. ni and pi are deviational variables which represent the underand over-achievement of the i-th goal, respectively. x is the set of decision variables to be determined. The g (within a priority level) function is given by
gl ( n, p) = ulq n1 + … + ulq nq + vl1 p1 + … + vlQ PQ , where u and v represent inter-priority level weights (Tamiz et al. 1998: 570). The possibility to account for the non-continuity of preferences seems particularly important for the analysis of environment problems (Spash 2000). Instead of optimization, GP is based on Simon’s satisficing philosophy (Simon 1955, 1979). In comparison, GP is a more pragmatic approach. It was used in the 1960s and 1970s by practitioners in corporations and agencies and only slowly received more attention by theorists. While the techniques applied are the same programming techniques as for the other MOP methods, the underlying philosophy differs. The idea of optimizing is abandoned; only satisficing is attempted. MOP and particularly GP are the most frequently applied multiple criteria methods. They have been applied to a wide array of problems, namely production planning, oil refinery scheduling, health care, portfolio selection, distribution system design, energy planning, water reservoir management, timber harvest scheduling, problems of wildlife management, etc. Their popularity can be explained by their flexibility; they can account for diverse variable types (continuous, integer, Boolean, etc.) as well as constraints and objective functions (linearity, convexity, differentiability, etc.) (Vincke 1992). More recent additions to the field of MOP have been: (a) to allow for fuzziness in the data which is a way to address the problem of uncertainty to some degree; and (b) evolutionary algorithms which were developed as a response to the fact that for models which are large in size and/or significantly non-linear, traditional solution methods are often unable to identify the global optimum. In such cases, typically linear approximations are applied instead in order to find a solution. Genetic algorithms were shown to be applicable to examples of large non-linear models (e.g. Mardle et al. 2000). They follow the concept of solution evolution, by stochastically developing generations of solutions populations using a given fitness statistic.
124 Andrea De Montis et al. MOP and GP allow addressing a completely different type of multiple criteria problems. Instead of choosing from a limited number of alternatives, a continuous set of alternatives is generated during the solution process. They are the only methodologies in the field which provide the model-based calculation of nondominated alternatives. Another advantage is that restrictions which are for instance needed for the analysis of decision-making processes based on criteria of strong sustainability (non-substitutability) can be accounted for. These special features turn out to be important for the decision-making situation in many case studies. GP has often been criticized for being analytically inferior to MOP. While GP is indeed a somewhat simpler method, it is often preferred because of its capacity to handle large-scale operational real-life problems. MOP works efficiently for moderate-size problems defined within a well-structured environment. When programming methods are applied, interdependence between criteria is not allowed. The completeness description of the phenomenon is desired, but not explicitly mentioned. While MOP is very appealing theoretically it has the disadvantage that for large problems and particularly if non-linear functions are included, often no optimal solution can be found. The development of interactive methods, which allow to reveal progressively the preferences of the decision-maker through man–machine interactions, made MOP much less reliant on doubtful assumptions about the decision-maker’s preferences. A number of useful software tools were developed which allow the decision-makers an important role in the decision process and enable ‘psychological convergence’ (i.e. termination of the interactive process as and when the decisionmaker chooses). Hence, interactive MOP leads not only to a good structuring of the problem but can also provide a useful tool for learning for the decision-maker involved. A problem remaining with interactive methods is that mostly they do not provide an objective basis upon which the decision-maker could base the decision to terminate the interactive process (Schniederjans 1995). The next step in perceiving the role of the decision-makers in a more democratic way, namely accounting for stakeholder participation (group decision-making), is however explicitly accounted for in few tools and is therefore exhausted in only some case studies. Another drawback in the user context is that the process cannot be made very transparent. MOP allows for the use of different geographical scales, as well as different general scales (micro–macro levels). Concerning data, MOP is able to deal with quantitative as well as qualitative and fuzzy data and allows for risk via fuzziness. An interesting aspect with regard to the fuzzy version of MOP was reported in several case studies: the large amount of time which is normally needed to select all the data required can on some occasions be bypassed by using the fuzzy relations (e.g. Chang et al. 1997). GP is not able to deal with a lot of different data, such as qualitative and fuzzy sets.
Comparison of the methods The evaluations of the individual methods using the quality criteria pointed to the differences in the characteristics of the presented methods. The summary in Table
Assessing the quality of different MCDA methods 125 5.2 gives a detailed characterization of each method. Without reference to the specific characteristics of the case study for which the methods should be applied, the selection of the best applicable method is not possible. Consequently the aim here is a characterization of the methods based on the quality criteria with regard to sustainability issues in general. From the list of quality criteria we find the following four groups most crucial:
• • • •
possibility to deal with complex situations (criteria, consideration of different scales and aspects, i.e. geographical scales, micro–macro link, societal/technical issues, type of data, uncertainties), possibility to consider non-substitutability (strong sustainability) issues, possibility to involve more than one decision-maker (stakeholder participation, actors’ communication, and transparency), and information of stakeholders in order to increase their knowledge and change their opinion and behaviour (problem structuring, tool for learning, transparency, type of weights).
Some criteria could not be considered here because they are especially designed to help the user in the selection of a method for a concrete decision-making situation, and not designed for a theoretical method comparison with regard to general sustainability issues without any specific application in mind, as done in this comparison. Reference to the whole list of criteria was made in De Montis et al. (2000), which included selected case studies of sustainable water use. In handling complex situations, all methods show similar performance with respect to the aspects and scales which can be considered; but they all show weaknesses in the characteristics of the evaluation criteria which should be used as operational components. Only AHP allows the interdependence of evaluation criteria, while only MAUT and NAIADE allow non-linear preferences. Evamix, ELECTRE III, and Regime do not permit either of these characteristics. In each case quantitative and qualitative data can be addressed. In ELECTRE III qualitative data have to be transformed first to a quantitative scale. Additionally, some of the methods like GP/MOP and NAIADE include features to deal with fuzzy data and stochastic numbers, which is the way that these methods deal with risk and uncertainties. MAUT allows to deal with risk (not uncertainty) concerning the outcomes of the alternatives by assigning probabilities to the utility functions (von Neumann–Morgenstern utility functions). All of the methods provide the possibility to carry out a sensitivity analysis or to apply qualitative data to consider uncertainties and risks. Non-substitutability aspects can only be properly analysed by using GP/MOP and ELECTRE III which allow to set constraints and thresholds explicitly. Within the remaining methods non-substitutability can only be considered by assigning flags to the respective criteria. The performance to inform stakeholders in order to increase their knowledge and change their opinion and behaviour is satisfactory for all methods. In MOP/ GP, ELECTRE III and Regime the transparency is only of medium quality, which hinders stakeholder participation. The other methods are satisfactory or even good
Interdependence Completeness Non-linear preferences Transparency of process, type of weights Meaning
Structure of problemsolving process
Project constraints
Stakeholder participation
Time
Costs
Applicability in user context
Solution-finding procedure
Weights
Criteria
Operational components of methods
Necessary
Depending on the numbers of attributes, stakeholders involved, etc. Depending on the numbers of attributed, stakeholders involved, etc.
Necessary
Depending on the numbers of attributes, stakeholders involved, etc. Depending on the numbers of attributed, stakeholders involved, etc.
Cardinal weights assigned Importance Complete ranking
Allowed Not allowed
Allowed Allowed
Cardinal weights assigned Trade-offs Complete ranking
Allowed
AHP
Not allowed
MAUT
Table 5.2 Summary of method comparison
Supported
Depending on the numbers of attributes, stakeholders involved, etc. Depending on the numbers of attributed, stakeholders involved, etc.
Ordinal weights assigned Importance Complete ranking
Allowed Not allowed
Not allowed
Evamix
Supported
Depending on the numbers of attributes, stakeholders involved, etc. Depending on the numbers of attributed, stakeholders involved, etc.
Cardinal weights assigned Importance Nondominated option/s calculated
Allowed Not allowed
Not allowed
ELECTRE III
Supported
Depending on the numbers of attributes, stakeholders involved, etc. Depending on the numbers of attributed, stakeholders involved, etc.
Ordinal weights assigned Importance Complete ranking
Allowed Not allowed
Not allowed
Regime
Necessary
Depending on the numbers of attributes, stakeholders involved, etc. Depending on the numbers of attributed, stakeholders involved, etc.
No weights Nondominated option/s calculated
No weights assigned
Allowed Allowed
Not allowed
NAIADE
Supported by a few variants
Depending on the numbers of attributes, stakeholders involved, etc. Depending on the numbers of attributes, stakeholders involved, etc.
Required Allowed for some variants Weights assigned for some variants Importance Nondominated option/s calculated
Not allowed
MOP/GP
126 Andrea De Montis et al.
Data situation
Indicator characteristics
Low Not possible
Not possible
Possible
Medium
Possible
Different issues are possible
Qualitative and quantitative possible Via ordinal criteria only
Possible
Different issues are possible
Possible
Different can be treated
Via the construction of the evaluation matrix Not appropriate High Via the integration of stakeholders
Qualitative and quantitative possible Via sensitivity analysis
Different issues are possible
Possible
Micro–macro link Societal/ technical issues Methods’ combinations Type of data
Qualitative and quantitative possible Risk/ Risky uncertainties outcomes: probabilities, sensitivity analysis Data processing High amount NonNot possible substitutability
Possible
Different can be treated
Geographical scale
Different can be treated
High Via the integration of stakeholders
High Via the integration of stakeholders
Applicability for problem structure
Appropriate
Appropriate
Tool for learning Transparency Actors’ communication
Via the construction of suitable hierarchies
Via the construction of utility functions
Problems structuring
Not possible
High
Via thresholds
Mainly quantitative
Possible
Different issues are possible
Possible
Different can be treated
Medium Via the integration of stakeholders
Sometimes cumbersome
Not possible
Medium
Qualitative and quantitative possible
Possible
Different issues are possible
Possible
Different can be treated
Medium Via the integration of stakeholders
Via the construction of the Regime vector
Not possible
Medium
Qualitative and quantitative possible Via stochastic and fuzzy numbers
Not possible
Different issues are possible
Possible
Different can be treated
High Via the integration of stakeholders
Via the construction of the evaluation matrix Appropriate
ordering
Possible via constraints or lexicographic
Medium
Via sensitivity analysis and fuzzy numbers
Mainly quantitative
Possible
Different issues are possible
Possible
Different can be treated
Interactive MOP Low Via the integration of stakeholders
Model based
Assessing the quality of different MCDA methods 127
128 Andrea De Montis et al. concerning these aspects. NAIADE does not allow to assign weights explicitly but is the only methodology which provides an explicit structure for stakeholder participation. For MOP/GP there are tools for interactive decision-making as well as for group decision-making. While the majority of applications did not make use of them, these tools have been available for more than ten years now. The methods also differ in the possibilities for involving more than one decisionmaker in the process. For MOP/GP there are tools for group decision-making, but again, they have not been applied much. Some methods like AHP support consideration of preferences of several decision-makers. In practice, each decisionmaker is asked individually. Concerning MAUT, if n persons are involved in the decision, each person is asked for their single utility function for any attribute; these single functions are then aggregated to n multi-attribute utility functions. If they are in conflict with each other, group solutions are possible, but are not really treated by the literature about MAUT. Concerning stakeholder information, MAUT, AHP, and Evamix provide good transparency and allow to give weights explicitly. In a very interesting way, NAIADE supports the involvement of more than one person in the decision-making process, namely with reference to their different interests, which allows for an explicit conflict analysis. In general, an important aspect in group decision-making is how transparent the decision-making process is, as this enhances possibilities of participation and goal-oriented discussion within the group. Evamix, ELECTRE III, and Regime provide only satisfactory performance for complex situations, so they are not the first choice in this context. Nonsubstitutability aspects could best be considered by using ELECTRE III or GP/ MOP. MAUT and Evamix have acceptable performance in general, however they lack important aspects crucial for sustainability issues, namely the ability to account for uncertainty and non-substitutability. NAIADE is a specialist to involve more than one decision-maker, while GP/MOP and ELECTRE III allow dealing with non-substitutability. Thus, a clear ranking of the methods concerning the different quality criteria is impossible, as each has its own strengths and weaknesses in different areas. The characterization shows, however, that depending on the specific problem the choice of one method over the other is advisable and that some methods are largely useless for the analysis of sustainability problems.
Conclusions The goal of more sustainable development has been studied extensively during the last fifteen years; but a key criticism has been vagueness as to recommendations and lack of operational applications. The acknowledgement of the limits of growth has led to the revision of the paradigms of economics regarding human and natural resources. Besides well-known goals like efficiency and income distribution, many other conflicting factors need to be included in the analysis. MCDA constitutes a powerful tool that is able to take into account several concerns and to foster participation and learning processes among politicians and citizens. Thus, MCDA is useful for decision-making processes in the context of sustainability (e.g. Munda
Assessing the quality of different MCDA methods 129 1995; Castells and Munda 1999; Janssen 1992; Andreoli and Tellarini 2000; Mendoza and Prabhu 2000). In contrast to earlier studies which compare MCDA methods with respect to more technical issues (e.g. Olson 2001, Salminen et al. 1998) we used a list of quality criteria for the assessment (De Montis et al. 2000) and concentrated on the most relevant criteria for sustainability issues. The comparison shows that the indicator list is applicable and useful for the evaluation of MCDA methods in the context of sustainability. As a result, a detailed characterization of the different MCDA methods was derived. Some rough guidelines which can be given from the comparison of the methods are: 1
2 3 4
5 6
If the respective decision problem is such that relying upon social welfare theory and its assumptions is possible and if the data to build utility functions is available (risk and qualitative data are possible) then MAUT is a good choice. If working with different conflicting interest groups is important for the case, NAIADE and AHP provide the best performance. If the decision-makers involved should primarily learn from the application of the MCDA tool, it is advisable to use MAUT or AHP. If thresholds and constraints are central for the problem under investigation, which means that there is non-substitutability of some criteria, ELECTRE III or GP/MOP should be chosen. If the problem is a continuous one, i.e. there is no discrete number of alternatives which comes out of the specific situation, GP or MOP should be chosen. If a complete ranking of the given alternatives as a result of the analysis is indispensable, MAUT, AHP, Evamix, or Regime should be applied.
Further research should focus on the relationship between sustainability and evaluation with respect to the process of consensus building. Recent research studies (Hewitt 1995) state that general principles of environmental action consist of environment management and of democratic decision-making. In this sense, the technical features are not that important; rather, each multiple criteria technique should be evaluated according to its capacity to foster learning processes and to allow users to become aware of society’s strategic role in environmental decision-making.
Notes 1 Risk and uncertainty are sometimes used interchangeably in the MCDA literature. As the difference is very important in the context of sustainability, we would like to distinguish between them. We use ‘risk’ if a probability distribution can be assigned for the unknown effects; otherwise the term ‘uncertainy’ is used (Knight 1921). 2 Utility independence: ui(xi ) is unaffected by levels of the other attributes. Preferential independence: the preferences for some attribute levels do not depend on the levels fixed for the other attributes. 3 No individual will sense a difference between a cup of coffee with i milligrams of sugar and one with i + 1 milligrams of sugar, but will express a preference between a cup with a lot of sugar and one with none; this contradicts transitivity of indifferences (Vincke 1992).
130 Andrea De Montis et al. 4 ‘In general what we mean by being consistent is that when we have a basic amount of raw data, all other data can be logically deduced from it. In doing pairwise comparison to relate n activities so that each one is represented in the data at least once, we need n - 1 pairwise comparison judgements. From them all other judgements can be deduced simply by using the following kind of relation: if activity A1 is 3 times more dominant than activity A2 and activity A1 is 6 times more dominant than activity A3 then A1 = 3A2 and A1 = 6A3. It should follow that 3A2 = 6A3 or A2 = 2A3 or A3 = ½A2. If the numerical value of the judgement in the (2, 3) position is different from 2 then the matrix would be inconsistent. This happens frequently and it is not a disaster’ (Saaty 1980: 18). 5 For more information on the ‘random index’ (R.I.) see Saaty (1980: § 1.4). In the following table values of the random index for matrix of orders from 1 to 11 are reported: Order of the Random index
1 2 3 4 5 0.00 0.00 0.58 0.90 1.12
6 7 1.24 1.32
8 9 1.41 1.45
10 1.49
11 1.51
6 Elements of qualitative criteria are expressed by means of symbols like +++, ++, +, 0, –, – –, – – –, where . … + + + ++ + 0 - -- - - - …. 7 An example of the way to obtain cardinal weights from ordinal preferences is given by Voogd (1983: 175): ‘if w1 £ w2 £ w3 and it is assumed that weights meet the condition
m  wj =1 j =1 then the following extreme sets w , = 1, 2, 3 can be distinguished:
w1 = (1, 0, 0), w2 =
Ê1 1 ˆ ÁË , , 0˜¯ , 2 2
w3 =
Ê 1 1 1ˆ ÁË , , ˜¯ ' . 3 3 3
8 For more information about the mathematical aspects of Evamix, see Voogd (1983: ch. 10). 9 Using stochastic numbers, it is necessary to choose the probability density functions; using fuzzy numbers, it is necessary to define their membership functions; using qualitative evaluations, some pre-defined linguistic variables such as ‘good’, ‘moderate’, ‘bad’ and so on are considerate. It is not possible to assign different types of measurements (i.e. crisp, stochastic, fuzzy, linguistic) to the same criterion for different alternatives. 10 In the case of two crisp numbers, the ‘distance’ is defined as their difference; in the case of stochastic and fuzzy numbers, the ‘semantic distance’ measures the distance between two functions (i.e. probability density functions or fuzzy membership functions); the linguistic variables are treated as fuzzy sets. 11 For linguistic variables,the equivalent preference relations may be used : much better than (>>); better than (>), approximately equal to (@); very equal to (=); worse than (<); much worse than (<<). In this case the six associated analytic functions express an index of credibility of the statements that an alternative is ‘much better’, ‘better’, ‘approximately equal’, ‘very equal’, ‘worse’ and ‘much worse’ than another. 12 For more information about the mathematical aspects of NAIADE, see Munda (1995: ch. 7). 13 The symbol * stands for >>, >, @, < or <<. 14 ‘Entropy measure of fuzziness can be easily extended to the notion of fuzzy relations … to have information on the diversity among the assessments of the single fuzzy relations, according to each criterion’ (Munda 1995: 138). More precisely ‘entropy is calculated as an index varying from 0 to 1 that gives an indication of the variance of the credibility indexes that are above the threshold, and around the crossover value 0.5 (maximum fuzziness). An entropy value of 0 means that all criteria give an exact indication (either definitely credible or definitely non-credible), whereas an entropy value of 1 means that all criteria give an indication biased by the maximum fuzziness (0.5)’ (Menegolo and Pereira 1996: 5). For the mathematic formulae of entropy, see Munda (1995: § 7.4.)
Assessing the quality of different MCDA methods 131
References Andreoli, M. and Tellarini, V. (2000) ‘Farm sustainability evaluation: methodology and practice’, Agriculture, Ecosystems and Environment, 77 (1–2): 43–52. Ballestero, E. and Romero, C. (1998) Multiple Criteria Decision Making and its Applications to Economic Problems, Boston/Dordrecht/London: Kluwer Academic. Castells, N. and Munda, G. (1999) ‘International environmental issues: towards a new integrated assessment approach’, in M. O’Connor and C.L. Spash (eds) Valuation and the Environment – Theory, Method and Practice, Cheltenham, UK/Northampton, MA: Edward Elgar: 309–27. Chang, N.-B., Wen, C.G. and Chen, Y.L. (1997) ‘A fuzzy multi-objective programming approach for optimal management of the reservoir watershed’, European Journal of Operational Research, 99: 289–302. Churchman, C.W., Ackoff, R.L. and Arnoff, E.L. (1957) Introduction to Operations Research, New York: Wiley. Climaco, J. (1997) Multiple Criteria Analysis. Proceedings of the XIth International Conference on MCDM, Coimbra, Portugal: Springer, 1–6 August 1994. De Montis, A., De Toro, P., Droste-Franke, B., Omann, I. and Stagl, S. (2000) ‘Criteria for quality assessment of MCDA methods’, paper presented at the 3rd Biennial Conference of the European Society for Ecological Economics, Vienna, 3–6 May 2000. Dillon, J.L. and Perry, C. (1977) ‘Multiattribute utility theory, multiple objectives and uncertainty in ex ante project evaluation’, Review of Marketing and Agricultural Economics, 45 (1, 2): 3–27. Fandel, G. and Spronk, J. (eds) (1985) Multiple Criteria Decision Methods and Applications. Selected readings of the First International Summer School Acireale, Sicily, September 1983, Berlin: Springer. Fishburn, P.C. (1968) ‘Utility theory’, Management Science, 14: 335–78. Fishburn, P.C. (1970) Utility Theory for Decision Making, New York: John Wiley and Sons. Fishburn, P.C. (1978) ‘A survey of multiattribute/multiple criteria evaluation theories’, in S. Zionts (ed.) Multiple Criteria Problem Solving, Berlin: Springer: 181–224. Funtowicz, S.O. and Ravetz, J. (1990) Uncertainty and Quality in Science for Policy, Dordrecht: Kluwer. Hewitt, N. (1995) European Local Agenda 21 Planning Guide – How to Engage in Long Term Environmental Action Planning Towards Sustainability? Brussels: European Sustainable Cities and Towns Campaign. Hinloopen, E. (1985) De Regime Methode, MA thesis, Interfaculty Actuariat and Econometrics, Free University Amsterdam. Hinloopen, E. and Nijkamp, P. (1990) ‘Qualitative multiple criteria choice analysis’, Quality and Quantity, 24: 37–56. Hinloopen, E., Nijkamp, P. and Rietveld, P. (1983) ‘Qualitative discrete multiple criteria choice models in regional planning’, Regional Science and Urban Economics, 13: 73–102. Hwang, C. and Yoon, K. (1981) Multiple Attribute Decision Making, Berlin: Springer. Ignizio, J.P. (1976) Goal Programming and Extensions, Lexington: Lexington Books. Janssen, R. (1992) Multiobjective Decision Support for Environmental Management, Dortrecht: Kluwer Academic. Kalu, T.C.U. (1999) ‘An algorithm for systems welfare interactive goal programming modelling’, European Journal of Operational Research, 116: 508–29. Keeney, R.L. and Raiffa, H. (1976) Decisions with Multiple Objectives, New York: John Wiley. Kmietowicz, Z.W. and Pearman, A.D. (1981) Decision Theory and Incomplete Knowledge, Aldershot: Gower.
132 Andrea De Montis et al. Knight, F. (1921) Risk, Uncertainty, and Profit, Boston: Houghton Mifflin. Korhonen, P. (2001) ‘Multiple objective programming support’, in C.A. Floudas and P.M. Pardalos (eds) Encyclopedia of Optimization, Dortrecht: Kluwer: Vol. 3, 566–74. Lara, P. and Romero, C. (1992) ‘An interactive multigoal programming for determining livestock rations: an application to dairy cows in Andalusia (Spain)’, Journal of the Operational Research Society, 43: 945–53. Lee, S.M. (1972) Goal Programming for Decision Analysis, Philadelphia, PA: Auerbach. Luce, R.D. (1956) ‘Semiorders and a theory of utility discrimination’, Econometrica, 24: 178–91. Mardle, S.J., Pascoe, S. and Tamiz, M. (2000) ‘An investigation of genetic algorithms for the optimization of multi-objective fisheries bioeconomic models’, International Transactions in Operational Research, 7(1): 33–49. Mastenbroek, P. and Paelinck, J.H.P. (1977) ‘Qualitative multiple criteria analysis – applications to airport location’, Environment and Planning A, 9 (8): 883–95. Mendoza, G.A. and Prabhu, R. (2000) ‘Multiple criteria decision making approaches to assessing forest sustainability using criteria and indicators: a case study’, Forest Ecology and Management, 131(1–3): 107–26. Menegolo, L. and Guimarães Pereira, A. (1996) NAIADE Manual, Joint Research Centre – EC, Institute for System, Informatics and safety, Ispra (VA), Italy. Munda, G. (1995) Multiple Criteria Evaluation in a Fuzzy Environment – Theory and Applications in Ecological Economics, Heidelberg: Physika Verlag. Munda, G., Nijkamp, P. and Rietveld, P. (1994) ‘Fuzzy multigroup conflict resolution for environmental management”, in J. Weiss (ed.) The Economics of Project Appraisal and the Environment, Aldershot: Edward Elgar. Nijkamp, P. and Voogd, H. (1981) ‘New multiple criteria methods for physical planning by means of multidimensional scaling techniques’, in Y. Haimes and J. Kindler (eds) Water and Related Land Resource System, Oxford: Pergamon Press: 19–30. Olson, D.L. (2001) ‘Comparison of three multicriteria methods to predict known outcomes’, European Journal of Operational Research, 130(3): 576–87. Rietveld, P. (1980) Multi Objective Decision Making and Regional Planning, Amsterdam: North Holland. Roy, B. (1985) Méthodologie Multicritère d’Aide à la Décision, Parigi: Economica. Roy, B. (1996) Multiple Criteria Methodology for Decision Aiding, Dordrecht: Kluwer Academic. Roy, B. and Bouyssou, D. (1985) ‘An example of comparison of two decision-aid models’, in G. Fandel and J. Spronk (eds) Multiple Criteria-Decision Methods and Applications. Selected readings of the First International Summer School Acireale, Sicily, September 1983, Berlin: Springer: 361–81. Saaty, T.L. (1980) The Analytic Hierarchy Process for Decision in a Complex World, Pittsburgh, PA: RWS Publications. Saaty, T.L (1988) Decision Making for Leaders, Pittsburgh, PA: RWS Publications. Saaty, T.L. (1992) Multicriteria Decision Making – The Analytic Hierarchy Process, Pittsburgh, PA: RWS Publications. Saaty, T.L. (1994) Fundamentals of Decision Making and Priority Theory with the Analytical Hierarchy Process, Pittsburgh, PA: RWS Publications. Saaty, T.L. and Alexander, J.M. (1989) Conflict Resolution – The Analytic Hierarchy Process, New York: Praeger. Saaty, T.L. and Forman, E. (1993) The Hierarchon, Pittsburgh, PA: RWS Publications. Saaty, T.L. and Vargas, L.G. (1991) Prediction, Projection and Forecasting, Dordrecht: Kluwer Academic Publishers.
Assessing the quality of different MCDA methods 133 Salminen, P., Hokkanen, J. and Lahdelma, R. (1998) ‘Comparing multicriteria methods in the context of environmental problems’, European Journal of Operational Research, 104: 485–96. Scarelli, A. (1997) Modelli matematici nell’analisi multicriterio, Viterbo: Edizioni Sette Città. Schniederjans, M.J. (1995) Goal Programming Methodology and Applications, Boston: Kluwer Publishers. Simon, H. (1955) ‘A behavioral model of rational choice’, Quarterly Journal of Economics, 69: 99–118. Simon, H. (1979) ‘Rational decision making in business organizations’, American Economic Review, 69: 493–513. Spash, C.L. (2000) ‘Multiple value expression in contingent valuation: economics and ethics’, Environmental Science & Technology, 34(8): 1433–8. Steuer, R.E. and Gardiner, L.R. (1990) ‘Interactive multiple objective programming: concepts, current status, and future directions’, in C.A. Bana e Costa (ed.) Readings in Multiple Criteria Decision Aid, Berlin/Heidelberg/New York: Springer: 413–44. Taha, H.A. (1976) Operations Research, New York: Macmillan. Tamiz, M. and Jones, D. (1997) ‘An example of good modelling practice in goal programming: means to overcoming incommensurability’, in R. Caballero, F. Ruiz and R.E. Steuer (eds) Advances in Multiple Objective and Goal Programming, Berlin/Heidelberg/ New York: Springer: 29–37. Tamiz, M., Jones, D. and Romero, C. (1998) ‘Goal programming for decision making: an overview of the current state-of-the-art’, European Journal of Operational Research, 111: 569–81. Van Delft, A. and Nijkamp, P. (1977) Multiple Criteria Analysis and Regional Decision Making, The Hague/Boston: Martinus Nijhoff. Vincke, P. (1985) ‘Multiattribute utility theory as a basic approach’, in G. Fandel and J. Spronk (eds) Multiple Criteria-Decision Methods and Applications. Selected readings of the First International Summer School Acireale, Sicily, September 1983, Berlin: Springer: 27–40. Vincke, P. (1992) Multiple Criteria Decision-Aid, New York: John Wiley & Sons. von Neumann, J. and Morgenstern, O. (1947) Theory of Games and Economic Behaviour, Princeton, NJ: Princeton University Press. Voogd, H. (1981) ‘Multicriteria analysis with mixed qualitative–quantitative data’, Delft University of Technology, Department of Urban and Regional Planning, Planologisch Memorandom, 81: 86. Voogd, H. (1983) Multiple Criteria Evaluation for Urban and Regional Planning, London: Pion. Wierzbicki, A. (1980) ‘The use of reference objectives in multiobjective optimization’, in G. Fandel and T. Gal (eds) Multiple Objective Decision Making, Theory and Applications, New York: Springer. Zionts, S. and Wallenius, J. (1976) ‘An interactive programming method for solving the multiple objective programming problem’, Management Science, 22(6): 652–63.
134 Wendy Proctor
6
MCDA and stakeholder participation Valuing forest resources Wendy Proctor
Introduction Making decisions about the environment often involves balancing the conflicting, incommensurate and incompatible values of many users and uses of a resource. One of the most fundamental and difficult tasks involved, therefore, is the effective integration or synthesis of all values related to the resource issue in question. Consideration and effective integration of all resource values, whether they are environmental, economic or social, is a necessary first step to achieving and maintaining ecologically sustainable development. In 1992, after decades of environmentalist versus forest industry conflict, the Australian Government embarked on the largest and most expensive environmental planning exercise ever undertaken in Australia. This was the programme of Comprehensive Regional Assessment (CRA) of Australia’s forests that would result in the signing of Regional Forest Agreements (RFAs) and ensure the sustainable management of forests in Australia. One of the most difficult tasks for planners and policy-makers was that of integrating all of the different forest values upon which a final decision about reserved and unreserved areas could be made. Without an adequate process of integration, important information could have been lost in the volumes of complex assessment documents that were created. As well, the process would fail to adequately take account of the preferences of all stakeholders. This chapter outlines a practical application of the decision support tool multiple criteria decision analysis, also known as multiple criteria analysis (MCA). The case study concerns the integration of various forest values using a stakeholder forum from one of the assessment regions. In recent years, MCA, as a structured procedure for aiding complex environmental, social and economic decisions, has gained popularity. However, MCA has had limited application in environmental policy-making in Australia. Although not part of the official process, the case study utilizes the results of the assessments themselves and the decision-making criteria and priorities placed on different forest values provided by a representative stakeholder group that was part of the CRA of the Southern Forest Region of New South Wales, Australia.1
MCDA and stakeholder participation
135
Forest assessments Of the 157 million hectares of forest in Australia, around 20 per cent comprise rainforest and open eucalypt forests, both of which are major sources of timber production. Around 30 per cent of this rainforest and open eucalypt forest area is situated on state government timber reserves and 18 per cent in nature conservation reserves. The day-to-day management of these areas rests largely under the control of the relevant state government with the Commonwealth (i.e. Federal) Government having jurisdiction over the issuing of woodchip export licences and obligations to ensure sustainable forest management under various international protocols, such as the Montreal Process (Commonwealth of Australia 1997a). However, the management of Australia’s public forests has been characterized by intense public debate for several decades. As a result, Australian governments are now carrying out a series of CRAs of these forests that are designed to determine the major part of Australian forest policy for the next twenty years. In 1992, the Commonwealth and State Governments of Australia reached agreement on a National Forest Policy that would provide a long-term management strategy for Australia’s forests (Commonwealth of Australia 1992). As part of this strategy, the governments agreed to: undertake CRAs of environment, heritage, social and economic values of forests; set up Comprehensive, Adequate and Representative (CAR) conservation reserves on the basis of these assessments; sign RFAs about the long-term management and use of Australia’s forests (Commonwealth of Australia 1997b). The CAR Reserve system is intended to safeguard biodiversity, old growth, wilderness and other natural and cultural values of forests. Forests outside of these reserves will be available for timber or other types of suitable ecologically sustainable production. The conservation criteria set by a panel of experts and agreed to nationally by governments (known as the JANIS criteria) aim to reserve where practicable: 15 per cent of the estimated extent of each forest ecosystem prior to European arrival; at least 60 per cent of old growth forest; and 90 per cent or more of high quality wilderness. Other criteria have been similarly set for flora and fauna species. The assessments have resulted in volumes of documents for each region that need to be considered carefully before the drawing up of a Regional Forest Agreement between the Commonwealth government and the relevant state government. The RFAs will be maintained for a period of twenty years. The long-term nature of these agreements has been designed to provide a secure basis on which the timber industry can make long-term investment decisions. These agreements are, however, to be reviewed every five years, to ensure that the objectives are being satisfactorily met. In 1994, the first of a series of extensive assessments of 25 million hectares of forests began. Relevant government departments provide assessments on environment and heritage, National Estate, wilderness, resource, economic and social values of the forests under review. An important part of the decision-making phase leading up to these agreements is the ‘integration’ process. That is, the process that ideally allows a rigorous comparison of economic, environmental and other forest values in order to allow decisions to be made about the extent, in these forests, of reserved, logged and other use areas. The following is an overview of
136 Wendy Proctor how these assessments are undertaken and how the official ‘integration’ process takes place (for more information, see for example Commonwealth of Australia 1998). The ‘Environment and Heritage Assessment’ covers biodiversity, endangered species, old growth and world heritage areas. The biodiversity assessment includes a compilation of flora and fauna data and an assessment of species and forest ecosystems. The species assessment includes information on distribution, habitat, life history and risk of extinction factors for each terrestrial and aquatic forest flora and fauna species including those that are rare and threatened. The forest ecosystems assessment attempts to estimate pre-1750 distributions of forest ecosystems and compare these with present-day distributions. The old growth assessment lists and maps areas classified under certain scientific principles. The framework for the world heritage assessment is provided by a panel of experts who identify themes of outstanding universal value relevant to Australia and related to Australian forests. The analysis then seeks to identify if any of these themes can be applied to the forest region under review. National Estate Assessment uses a broad range of technical expertise and public input. The assessment seeks to identify areas of both natural and cultural value. It draws on a wide range of existing flora, fauna, historical, ecological and pre-logging surveys as well as some new data collected especially for the purpose. Also included is an ongoing assessment of indigenous cultural and heritage values. The Wilderness Assessment has been largely based on either existing surveys or on expert opinion or a combination of the two. Data on variation in wilderness quality across the natural areas of the region are mapped. Measures of wilderness quality are essentially ‘remoteness’ from access, structures and settlement and ‘naturalness’, both apparent and biophysical. Other guidelines agreed upon by technical experts (such as a minimum of 25,000 hectares for an area to be classified as wilderness) are also used. Resource and Economic Assessment details the existing resources and uses of the forest area and provides an economic evaluation of the current situation and potential future scenarios. The usage categories may include: native forests (production of sawlogs and residual logs), plantations (native and exotic species), other forest produce (e.g. honey), recreation and tourism, water and mineral resources. The net value of the timber industry is assessed using survey data collected from existing sawmillers. Future returns to the industry are estimated using a GAMS (Generalized Algebraic Modelling System) based constrained linear optimization model of the local timber industry called FORUM (Forest Resource Use Model), and estimates of sustainable yields provided by independent forecasting models of forest growth and production for the particular region (Dann and Clarke 1996). Data on the values of other forest uses is provided by existing surveys. The potential for mineral resources is assessed using standard geological survey techniques that place varying probabilities on the existence of valuable minerals according to the basic geological structure of the area. Social Assessment uses survey, interview and workshop techniques. This analysis provides a socio-demographic profile of the CRA area as well as details of the
MCDA and stakeholder participation
137
current community infrastructure and an outline of community attitudes and perceptions with regard to the use of forest resources. Various techniques based on social assessment theory are employed in the analyses of the data. ‘Integration’ describes the official government process by which the various assessments are brought together and used as the basis of coming to a decision about reserved and other use areas in a region. The integration process differs from state to state depending on the preferences of particular governments and on the availability of data. In general, though, after the assessments have been completed, a series of meetings are held and attended by representatives from the Commonwealth, the state government and various stakeholder groups. The objective of these meetings is to develop feasible scenarios, and subsequently, a negotiated resource use option or set of options for the region, that is released for public comment before a final decision is made. A Geographic Information System (GIS) based reserve selection model called ‘C Plan’ is used extensively as the basis for developing resource use options. The C Plan model maps areas of differing ‘irreplaceability’ and finds the smallest possible set of areas that will achieve a nominated conservation goal (Pressy et al. 1995). Irreplaceability measures the extent to which a particular conservation goal can be achieved. Two definitions of irreplaceability are used: one is the likelihood that any of the areas in a region will be needed to achieve an explicit conservation goal, the other is the extent to which the options for achieving an explicit conservation goal are narrowed if any of the areas in a region are destroyed or made unavailable for conservation. The conservation goals can include the preservation of such items as particular forest ecosystems, species of flora and fauna, etc. Also, the model maps areas of old growth forest, wilderness, potential mineral deposits, high cultural and heritage values, timber production, private and leasehold land as well as areas which have been targeted for indigenous land rights claims. A process of visual interpretation of the irreplaceability indices is followed and each area of high irreplaceability investigated and selected for inclusion as a possible reserve area. A visual interpretation of the areas of wilderness, old growth forests and areas where threatened species are reported to exist is also carried out and these areas included in the possible reserve system. Reserve design is also taken into account. Despite the technical and structured nature of the C Plan method, the reserve design process also includes a degree of ad hoc selection. For example, this is displayed in the choice of criteria being used to select areas for reserve and sometimes on the heavy reliance on local knowledge of conservation groups and government officers. After several selections have been made, the extent to which conservation targets have been met for the various flora and fauna species and ecosystems are assessed. Next, a timber volumes model, the Forest and Resource Management System (FRAMES) is run. This model estimates the amount of commercial timber that could be harvested from the remaining area of forest that had been excluded from reserve and is not on private or leasehold land (Turner 1998). The results are then fed into the FORUM economic model to determine the likely consequences on the value of timber production in the area as a result of
138 Wendy Proctor the change in reserve selection. FORUM provides an analysis of the commercial value of the forest harvest, net returns to the regional industry and employment. Further details on ‘integration’ can be found in Proctor (1999), however, it should be noted that at no stage in the official integration process for the Southern New South Wales (NSW) Region were the data obtained from the various assessments comprehensively and methodically incorporated into the decision-making. It should also be noted that, unlike previous assessments, for this region, the stakeholder groups were largely excluded from the integration process. The Southern Region integration process resulted in the description of five options, along with various data associated with each option, being released for public comment. These options represented five different reserve designs (with implications for economic indicators as well as various levels of conservation and National Estate targets being met) associated with levels of high quality sawlog production of 32,000 cubic metres per annum, 35,000 cubic metres per annum, 45,000 cubic metres per annum, 55,000 cubic metres per annum and 65,000 cubic metres per annum. Details of each option are provided later. The following sections outline an MCA to utilize the forest assessment data in a much more rigorous, objective and structured way than has been undertaken in the RFA integration process. This method also allows for a much more participatory approach to the decision-making process than has so far occurred for any of the forest agreements. The guidelines for the CRA process (see Commonwealth of Australia 1992) outline the need for public participation covering all interest groups and accounting for all stakeholder preferences in a comprehensive and transparent manner. In addition, stakeholder participation is a fundamental principle of ecologically sustainable development. It will be shown that MCA is an effective means of achieving stakeholder participation in the decision-making process.
Method MCA is a means of simplifying complex decision-making tasks which may involve many decision-makers, a diversity of possible outcomes and many and sometimes intangible criteria by which to assess the outcomes. It is one means by which structure and transparency can be imposed upon the decision-making process. Its origins lie in the fields of public policy, mathematics and operations research and it has had a great deal of practical usage by public planners in such areas as the location of health facilities, motorways and nuclear reactors (Massam 1988). An MCA seeks to make explicit the logical thought process that is implicitly carried out by an individual when coming to a decision. In complex decisionmaking tasks, which sometimes involve many objectives and many decisionmakers, this structured process of logic may be lost in the complexity of the issues. In general, an MCA seeks to identify the alternatives or options that are to be investigated and decided upon, a set of criteria by which to rank these alternatives and the method by which the alternatives are to be ranked and preferences aggregated. Finally, a sensitivity analysis is carried out on the results. The final outcome is a preferred option or set of options that is based upon a
MCDA and stakeholder participation
139
rigorous definition of priorities and preferences decided upon by the decisionmaker.2 In general, the following steps are carried out by the decision-maker in an MCA. The first is to identify the feasible options or preferred outcomes involved in the decision problem. Next the decision-maker seeks to identify the overall objective that is to be achieved in the process and then identifies the criteria by which to judge the selected options. An important part of the process is then to apply appropriate weights on each of the criteria that reflect the particular preferences of the decision-maker in how important each criterion is in relation to the overall objective of the decision process. The analyst then chooses an appropriate form of MCA with which to analyse the data and aggregate the preferences. An assessment is then made of each of the options based on the chosen MCA technique, the data on how each option performs under the different criteria and the weightings supplied by the decision-maker. An investigation of the sensitivity of the ranking of options with respect to, for example, the chosen weights and method of aggregation can then be carried out. An iterative process can then be pursued where the analyst reports back to the decision-maker the findings of the MCA and the sensitivities. ‘Fine tuning’ can then be performed by the decision-maker to aspects related to their input. Also this part allows for the identification of trade-offs, if any, that may be made in the overall decision process. A common criticism in public policy-making and in particular environmental policy-making in Australia is the high incidence of ad hoc decision-making that takes place under such processes (see, for example, Dovers 2000; Doyle and Kellow 1995; Ferguson 1996). The need for a more structured and objective approach has been addressed in the public policy literature, with some authors providing general guidelines on how this might be achieved (for example Hogwood and Gunn 1984). Similarly, an important aspect of environmental policy-making in recent years has been the increasing practice of incorporating public and stakeholder participation into some part of the policy-making process. The need for a participatory framework to environmental policy formulation in Australia has grown out of increasing interest by individuals in environmental issues (Lothian 1994), a grass-roots desire by the public to become involved in the policy-making process over regional issues that affect them, and a recognition by governments that involving the general community early in the process can bring benefits by avoiding disagreements and conflicts in later stages and increasing the knowledge base that is available to policymakers (Howlett and Ramesh 1995). Whereas much of the literature on both prescriptive methods for policy-making and the incorporation of public participation into such policy-making has addressed general requirements and concerns, the steps outlined under the MCA process are more explicit and provide a more detailed and structured guide for practical approaches to participatory and other public policy-making. In the context of this research, MCA is primarily regarded as an aid in the process of decision-making and not necessarily as a means of coming to a singular optimal solution. As such, the MCA process is valued for the enlightenment and unravelling of issues that it can provide in the decision-making problem. The process adds to the knowledge
140 Wendy Proctor of the decision-maker and is greatly aided by including the decision-maker in each step of the analysis. This is one reason why some forms of MCA are regarded as superior to other decision-aiding techniques (for example, mathematical programming methods) which are usually carried out in isolation from the decisionmaker. Such techniques are designed to provide an ‘optimal’ result at the end of the analysis, and so they do not necessarily increase the understanding of the important elements of the process, especially to those who are not familiar with such models. Other techniques are not designed to unravel the decision-making problem or provide a process that will enlighten the decision-maker and add to the decision-maker’s knowledge of the problem. The non-transparency and singular solution nature of such techniques may only result in increased mistrust in the process by those who are not familiar with how a decision was finally derived (Prato 1999). The approach followed in this current research is therefore similar to the ‘heuristic’ approach emphasized by Stirling and Mayer (1999) in their ‘multicriteria mapping’ technique (see also Chapter 7 in this volume). The analytic hierarchy process There are many different types of MCA, as discussed in the previous chapter. Saaty (1980) developed a method of analysing decisions based upon a hierarchy of components of the decision, known as the Analytic Hierarchy Process (AHP). His method is essentially an interactive one where a decision-maker or group of decision-makers relay their preferences to the analyst and can debate or discuss opinions and outcomes. His method largely stems from theories of human behaviour including thought process, logic, intuition, experience and learning theories (Saaty 1982). The hierarchies are made up of: a top level, which comprises the overall objective of the decision process; intermediate levels, which comprise the criteria and sub-criteria to analyse the decision; and the lowest level, which lists the alternative plans that are to be analysed and decided between. The AHP is based upon the construction of a series of ‘pairwise comparison’ matrices which compare criteria with one another. This is done to estimate a ranking or weighting of each of the criteria that describes the importance of each of these criteria in contributing to the overall objective. If the criteria are broken down into a number of sub-criteria, the pairwise comparisons are repeated for each level of the hierarchy. A pairwise comparison of J criteria (G1,…, GJ ) to reflect the importance or weighting of each criteria in influencing the overall objective, involves constructing a J by J matrix (G) which shows the dominance of the criteria in the left-hand column with respect to each criteria in the top row.
A1 Aj
A1 AI
A1
a j1
a j1 a j1
a jI
a jI a jI
a j1
a jI
MCDA and stakeholder participation
141
Each cell entry of G reflects a ratio scale of the underlying priority weights assigned to each of the criteria, i.e. gjj ¢ = wj/wj ¢. The weightings (wj ) have to be derived from the cell entries gjj ¢. A nine-point Intensity of Importance scale was developed by Saaty to determine these. It is claimed that the scale is based on psychological experiments and is designed to allow for an accurate reflection of priorities in comparisons between two items whilst minimizing the difficulties involved in doing so (Table 6.1). In G, each cell entry is positive and the diagonal elements (gjj ) are a series of 1’s. If it is assumed that transitivity of preferences prevails (i.e. that if G1 is preferred by a scale of 5 to G2, then G2 is preferred by a scale of 1/5 to G1) then the reciprocal property gjj’ = 1/gjj’ is satisfied and estimates need only be provided for those cells which lie above the diagonal. Saaty proves that, if G displays ‘cardinal’ consistency, in that gjj’ gj’j” = gjj”, then by normalizing the positive reciprocal matrix G so that the columns sum to unity, a solution to w, the vector of overall priority weights, can be obtained by reading any column of the matrix, as each column in this normalized matrix will be identical (Saaty and Vargas 1982). Imposing cardinal consistency on the matrix also means that only one row of the matrix needs to be entered and all other values can be derived. If cardinal consistency were not imposed, then each column vector may be different and it would then be necessary to average across the rows to determine the overall priority weights. In this instance, Saaty provides a ‘check for consistency’ measure and acceptable bounds within which this measure should fall. To gauge how each plan performs with respect to each of the criteria, another series of pairwise comparisons are carried out. For j = 1, …, J criteria and i = 1, …, I different plans, a pairwise comparison matrix comparing all plans under the criteria Gj would be:
A1 Aj
A1 AI
A1
a j1
a j1 a j1
a jI
a jI a jI
a j1
a jI
Table 6.1 The AHP intensity of importance scale Intensity of importance
Definition
1 3 5 7 9
Equal importance of both elements Weak importance of one element over another Essential or strong importance of one element over another Demonstrated importance of one element over another Absolute importance of one element over another
Source: Saaty (1982)
142 Wendy Proctor Again, the nine-point scale is used to provide estimates of the ratio weights, the aji, and the positive reciprocal properties are assumed to hold. The overall priority (OPi) of each of the plans with respect to all of the criteria can be estimated as follows: OP1 = a11(w1) + a21(w2) +…+ aJ1(wJ) OP2 = a12(w1) + a22(w2) +…+ aJ2(wJ) . . . OPI = a1I(w1) + a2I(w2) +…+ aJI(wJ) The following example clearly describes the above process. Assume that three broad criteria groups (which could incorporate a number of sub-criteria) are thought to be important in a decision concerning various natural resource use options. These represent environmental, social and economic considerations. A pairwise comparison for each of the criteria could then be as shown in Table 6.2. The comparison of a criterion with itself always denotes equal importance and so the central diagonal is a series of 1’s. In this example, environmental criteria are considered to be weakly dominant over social criteria and so a 3 is recorded in the appropriate cell. Its reciprocal is placed in the corresponding cell that compares social with environment. This matrix is consistent in the sense that a value of 5/3 is recorded in the space when social criteria are compared with economic criteria. This value was derived by imposing the cardinal consistency property on the matrix although cases could exist where a different value is recorded here thus denoting an inconsistency of preferences of the decision-maker. The matrix is then normalized (by dividing each cell entry by its column total) so that meaningful comparisons can be made amongst the components, and the relative priorities for each criteria with respect to the overall objective can be read from any column. This is shown in Table 6.3. The vector of overall priorities shows the relative importance of each of the three criteria in affecting the final decision. In this example, environmental criteria are considered most important by a large margin and then follow social and finally economic considerations. Using the criteria and priority weights estimated above, three different plans are compared by first carrying out a pairwise comparison of the performance of each plan against the others under the three different criteria identified. An example is shown for the environmental criteria (Table 6.4). The same procedure would then be carried out for the social and the economic criteria. In this example, Plan a performs quite strongly under the social and economic criteria and less so under the environmental criteria. Plan b performs poorly under the environmental criteria and slightly better for the other two, whilst Plan c is strongly preferred under environmental criteria but is least preferred when only social or economic criteria are considered. To obtain an overall ranking of the plans under all of the criteria combined, a weighted sum of each plan’s performance under each of the criteria is calculated as in Table 6.5. From this analysis, Plan c is
MCDA and stakeholder participation
143
Table 6.2 Example pairwise comparison
Environment Social Economic
Environment
Social
Economic
1 1/3 1/5
3 1 3/5
5 5/3 1
Table 6.3 Normalized matrix example
Environment Social Economic
Environment
Social
Economic
Priority weighting
0.65 0.22 0.13
0.65 0.22 0.13
0.65 0.22 0.13
0.65 0.22 0.13
Table 6.4 Pairwise comparison of environmental criteria Environmental
Plan a
Plan b
Plan c
Ranking (priority after matrix is normalized)
Plan a Plan b Plan c
1 1/3 2
3 1 6
1/2 1/6 1
0.30 0.10 0.60
Table 6.5 Overall ranking example
Plan a Plan b Plan c
Environment 0.65
Social 0.22
Economic 0.13
Overall priority
0.30 (0.65) 0.10 (0.65) 0.60 (0.65)
+0.50 (0.22) +0.25 (0.22) +0.25 (0.22)
+0.48 (0.13) +0.32 (0.13) +0.19 (0.13)
= 0.37 = 0.16 = 0.47
chosen as the preferred plan because of its strong performance with respect to the environmental criteria and the associated weighting placed on the environmental issues in achieving the overall objective.
Case study The NSW Southern Region Regional Forest Forum was formed as part of the RFA process. The Forum provided an opportunity for regional stakeholders to share information with governments and offer recommendations. The Forum was not, however, given an official decision-making role in the process. Members of the Forum represented local state and federal government natural resource management organizations, forest industry, conservation, farmer and beekeeping lobby groups as well as academics and indigenous tribal elders and representatives. The Forum met regularly for almost three years – around once every six weeks between 1997 and 2000. Although not part of the ‘official’ process, the Forum members agreed to take part in the MCA as part of their regular meetings, with
144 Wendy Proctor the time allocated to collect information for the analysis forming only a small part. Much of the information was gathered outside the meetings by mail survey, phone or face-to-face interview. The first stage of the analysis was to identify a complete set of criteria by which to assess each alternative plan. The criteria identified relate directly to the stated objectives of the RFAs as outlined in the National Forest Policy Statement. Figure 6.1 shows a hierarchy of objectives and criteria, developed in association with the Forum members. Forum members were then asked to carry out the AHP pairwise comparison exercise outlined previously. A survey asking for details about their pairwise preferences of the identified criteria and sub-criteria was distributed and information on how to fill out the survey was provided at one of the meetings. Follow-up phone calls also provided assistance. A pairwise comparison of the three broad criteria (Environment, Economic and Social) was requested as well as three sub-criteria group comparisons. Information was only required on the matrix cell entries which were above the diagonal. Reciprocals were later entered below the matrix diagonals. Cardinal consistency was not imposed on the comparisons. Time and other constraints did not allow for further interaction with the Forum for ‘fine-tuning’ of their preference data.
Results Southern Forest Forum preferences The purpose of this part of the analysis is to identify issues of relative disagreement and agreement on priorities between the 22 members of the Forum. The potential of the analysis is that those issues of substantial disagreement can be isolated and be subject to more in-depth analysis and discussion so that possible compromises or trade-offs may be found. Criteria consistently given low priority can be excluded from the analysis as they are regarded as being of relatively less importance in meeting the overall objectives. The raw data from each respondent’s matrices were normalized (by dividing the cell entry by the column total) to allow comparisons between criteria and to show the range of relative priorities. The Cardinal Consistency Index was within the acceptable limits for all responses. Figure 6.2 shows the differences in priorities amongst Forum members with respect to the three major criteria. The largest differences lie in the Environment and Economic categories with a distinct polarization of priorities evident amongst many Forum members. In general, those members classed as having forest industry affiliation and conservation group affiliation, show very divergent priorities as expected. There is some consistency with regards to the Social criteria, with most respondents considering this to be of least (or equal) importance out of the three. As an indication of the overall group response, following the method recommended by Saaty (1982), the geometric mean over all respondents of each matrix cell entry was calculated and the resulting geometric mean matrix normalized. For the three broad criteria categories, the Environmental category was given greatest priority in influencing
Conservation of environmental values
Maintenance of long-term economic benefits
Protect conservation values Ensure the long-term ecologically sustainable management of forests Develop an internationally competitive forest products industry Effectively use other regional economic and social resources
Employment and community needs
Other product values
Apiary
Minerals
Timber
Productive capacity, health and vitality of forest
Forest contribution to global carbon cycles
Adequate hazard reduction
Water and soil resources
Wilderness
Old growth
Biodiversity
Figure 6.1 Hierarchy of objectives, criteria and indicators for the Southern Region RFA
Maintenance of social and cultural values
Indicators
Criteria
Objectives
MCDA and stakeholder participation 145
Indigenous
Cultural and heritage
National estate
Recreation and tourism
146 Wendy Proctor
0.90
Normalized priorities
0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 Environment
Economic
Social
Figure 6.2 Preferences of criteria
the overall RFA objective with a figure of 0.41. Next followed the Economic criteria (0.34) and finally the Social criteria (0.25). The normalized priorities of the sub-criteria were then calculated. Within the Environmental sub-criterion, the most important criteria were identified as Water and Soil Resources, the Productive Capacity of the Forest and Biodiversity (Figure 6.3). The normalized priorities of the Economic criteria show less dispersion than those of the Environmental criteria. Disparities in preferences exist in the Productive Capacity, Timber Values and Minerals Values categories although the range of priorities for Productive Capacity is not spread evenly with one substantial outlier influencing the results (Figure 6.4). Disregarding this outlier, the highest priorities were given to Timber Values and Employment and Community Needs. In the category of Social Criteria, the highest priorities were given to Employment and Community Needs and Indigenous Values. The largest range of priorities for the social criteria (ignoring outliers) appeared to be in the Employment and Community Needs category (Figure 6.5). More detailed analysis of the preference structures of the Forum can be found in Proctor (1999) and Proctor (2001). Performance of options against the criteria The next part of the analysis was to assess each of the suggested options devised for the Southern Region against the decision criteria identified by the Forum members. Table 6.6 shows an Impact Table (measuring the performance and implications of each option) of the only available data released on each option at the time of writing. The available data fell well short of those that were identified by the Forum as necessary to the decision-making process. An indication of the relationship between the available data and the decision criteria is provided in Table 6.6. It is important to note that impact data on some of the criteria consistently regarded as important by the Forum members (for example, Water and Soil
MCDA and stakeholder participation
147
0.60
Normalized priorities
0.50 0.40 0.30 0.20 0.10
Prod cap
Carbon
Hazard
Water and soil
Wilderness
Old growth
Biodiversity
0.00
Figure 6.3 Preferences of environmental criteria
0.60
Normalized priorities
0.50
0.40
0.30
0.20
0.10
0.00
Productive capacity
Timber values
Minerals values
Apiary values
Other
Employment and community
Recreation and tourism
Figure 6.4 Preferences of economic criteria
Resources and Indigenous Values) were not provided for the different proposed options. The individual Forum members’ preference matrices of the sub-criteria were re-normalized, after excluding those sub-criteria for which data were not available. Table 6.7 shows the estimated pairwise comparisons of options under the
148 Wendy Proctor 0.70
0.60
Normalized priorities
0.50
0.40
0.30
0.20
0.10
0.00
Employment and community
Recreation and tourism
National estate
Cultural and heritage
Indigenous
Figure 6.5 Preferences of social criteria
environmental criteria based on the performance of each option outlined in the Impact Table. Using the priorities placed on the environmental criteria by the Forum members (i.e. Biodiversity 0.47, Old Growth 0.29 and Wilderness 0.23), Option 1 (with an overall priority of 0.43 = (0.442)0.47 + (0.490)0.29 + (0.333)0.23) ranks highest in terms of the environmental criteria. Then follow Option 2 (0.20), Option 3 (0.14), Option 4 (0.12) and Option 5 (0.11) (Table 6.8). Table 6.9 shows the pairwise comparison of options under the economic criteria. Using the priorities placed on the economic criteria as shown in Table 6.10 (i.e. Timber Values, 0.39; Employment and Community Needs, 0.36; and Other Values, 0.24), the ranking of the options according to their performance under the economic criteria was: Option 5 (0.40), Option 4 (0.26), Option 3 (0.16), Option 2 (0.09), Option 1 (0.08). Using the pairwise comparisons of options under the social and cultural criteria (Table 6.11) and the social and cultural criteria priorities (i.e. Employment and Community Needs, 0.56; National Estate Values, 0.44), the ranking of the options under the social criteria was: Option 5 (0.26), Option 1 (0.22), Option 4 (0.19), Option 3 (0.17), Option 2 (0.16) (Table 6.12). The overall ranking of the options under all criteria was: Option 1 (0.26), Option 5 (0.24), Option 4 (0.19), Option 3 (0.16), Option 2 (0.15). These rankings are depicted in Figure 6.6 with the priorities of the broad criteria and the individual Forum member’s overall priorities are presented in Figure 6.7 and Figure 6.8. An option representing an ‘optimal group decision’ has been provided here for illustrative purposes in Figure 6.7, following the method of averaging over the group’s preferences suggested in the AHP. However, a more valuable and informative part of the MCA is provided by a closer scrutiny of the results than is revealed in a simple geometric averaging procedure. For example, the range of individual overall priorities of options (Figure 6.8) suggests that the greatest disparity of rankings lies in the two extreme options with much greater agreement by Forum members in placing the centre options on a lower priority. The following sensitivity
High 144–151 38 16.4–17.6 –33 3,329.1
High
140–145 36
15.5–16.5 –48 3,327.6
19.3–22.3 +16 3,333.8
172–180 43
High
62 52 72 71 88.1 All
45,000 (for 20 years and at least 43,000 per year thereafter)
Option 3
22.2–25.4 +69 3,338.6
195–200 50
Medium
61 52 68 67 87.7 Some
55,000 (for 20 years and at least 45,000 per year thereafter)
Option 4
3
2
* 1
25.0–27.6 +115 3,343.2
213 55
Low
60 52 66 63 87.2 Some
65,000 (for 20 years and at least 45,000 per year thereafter)
Option 5
Achievable targets refers to the extent to which the JANIS criteria are met on formal or dedicated reserves such as national parks, nature reserves and flora reserves. These data address the broad criteria under Conservation of Environmental Values – Biodiversity (achievable targets met for Flora, Fauna and Forest Ecosystems); Old Growth Forests (achievable targets met); Wilderness (percentage reserved). These data address the broad criteria under Maintenance of Social and Cultural Values – Employment and Community Needs (direct mill employment, harvesting and haulage employment, change in other employment); National Estate (National Estate areas reserved, National Estate values). These data address the broad criteria under Maintenance of Long-Term Economic Benefits – Timber (gross value of timber output); Other Products (total gross value of output) Employment and Community Needs (direct mill employment, harvesting and haulage employment, change in other employment).
Source: Forest Taskforce and Resource and Conservation Division 2000, ‘A Proposal for a Regional Forest Agreement for Southern NSW’.
67 55 73 74 88.5 All
35,000 (for 20 years and at least 35,000 per year thereafter)
Option 2
80 73 79 84 90 All
32,000 (for 20 years and at least 22,000 per year thereafter)
Volume of sawlogs/year (cubic metres)
Achievable targets* met in dedicated reserves (%): 1 Forest ecosystems Old Growth Forest Fauna Flora Wilderness reserved (%) 1 National Estate Areas Reserved (areas with National Estate sites identified within them and placed into reserve) 2 National Estate Values (additions to reserves because of high potential National Estate values identified)2 Total Direct Mill Employment (no.)2,3 Total Harv. and Haul. Employment (no.) 2,3 Gross Value Output ($m)3 Change in other employment (no.) 2,3 Total Gross Value Output ($m)3
Option 1
Indicator
Table 6.6 Impact table for forest use options
MCDA and stakeholder participation 149
150 Wendy Proctor Table 6.7 Pairwise comparison of options under environmental criteria Biodiversity*
Option 1
Option 2
Option 3
Option 4
Option 5
Option 1 Option 2 Option 3 Option 4 Option 5
1 0.33 0.33 0.25 0.25
3 1 0.5 0.5 0.33
3 2 1 0.5 0.5
4 2 2 1 0.5
4 3 2 2 1
Wilderness Option 1 Option 2 Option 3 Option 4 Option 5
Option 1 1 0.5 0.5 0.5 0.5
Option 2 2 1 1 1 1
Option 3 2 1 1 1 1
Option 4 2 1 1 1 1
Option 5 2 1 1 1 1
Old growth Option 1 Option 2 Option 3 Option 4 Option 5
Option 1 1 0.25 0.25 0.25 0.25
Option 2 4 1 0.5 0.5 0.5
Option 3 4 2 1 1 1
Option 4 4 2 1 1 1
Option 5 4 2 1 1 1
* Biodiversity figures were estimated using an average over achievable targets met for forest ecosystems, flora and fauna from Table 6.7.
Table 6.8 Priority of options in terms of environmental criteria Ranking of options for each sub-criteria**
Option 1 Option 2 Option 3 Option 4 Option 5
Biodiversity 0.47* 0.442 0.217 0.153 0.111 0.077
Old growth 0.29* 0.490 0.189 0.107 0.107 0.107
Wilderness 0.23* 0.333 0.167 0.167 0.167 0.167
Priority 0.426 0.195 0.141 0.121 0.106
* Contribution of sub-criteria to overall priority. ** After normalizing the pairwise comparison matrices.
analyses are also important parts of the MCA to help to unravel the complexities of the decision-making problem.
Sensitivity analysis A sensitivity analysis was carried out to test the outcome of the overall ranking of options when the priorities placed on the broad Environmental and Economic criteria were both placed at their maximum values in turn.3 The outcomes are depicted in Figure 6.9 and Figure 6.10, with an environmental preference leading to an even higher ranking of Option 1 and an economic preference leading to Option 5 being ranked first.
MCDA and stakeholder participation
151
Table 6.9 Pairwise comparison of options under economic criteria Timber employment*
Option 1
Option 2
Option 3
Option 4
Option 5
Option 1 Option 2 Option 3 Option 4 Option 5
1 2 3 4 5
0.5 1 3 4 5
0.33 0.33 1 2 3
0.25 0.25 0.5 1 2
0.2 0.2 0.33 0.5 1
*Includes direct mill and harvesting and haulage employment.
Timber values* Option 1 Option 2 Option 3 Option 4 Option 5
Option 1 1 1 4 4 6
Option 2 1 1 3 4 5
Option 3 0.25 0.33 1 2 3
Option 4 0.25 0.25 0.5 1 2
Option 5 0.17 0.2 0.33 0.5 1
Option 2 1 1 1 2 2
Option 3 1 1 1 2 2
Option 4 0.5 0.5 0.5 1 2
Option 5 0.5 0.5 0.5 0.5 1
*Gross value of timber output.
Other values* Option 1 Option 2 Option 3 Option 4 Option 5
Option 1 1 1 1 2 2
*Total gross value of output.
Table 6.10 Priority of options in terms of economic criteria Ranking of options for each sub-criteria**
Option 1 Option 2 Option 3 Option 4 Option 5
Timber 0.39* 0.062 0.067 0.178 0.263 0.429
Value emp. & comm. 0.36* 0.061 0.082 0.169 0.267 0.421
Other values 0.24* 0.141 0.141 0.141 0.249 0.327
Priority 0.080 0.090 0.164 0.259 0.397
* Contribution of sub-criteria to overall priority. ** After normalizing the pairwise comparison matrices.
Sensitivity analysis was also carried out to assess the order of rankings under different assumptions about the sub-criteria. As the Employment and Community Needs criterion displayed the greatest difference in the priorities placed on it by Forum members, two scenarios were tested. The first used the maximum value obtained for this criterion from the individual pairwise comparisons of both the Economic and the Social sub-criteria (with other relative priorities maintained). The effect was to change the ranking to Option 5 (0.29), Option 1 (0.22), Option 4 (0.21), Option 3 (0.16) and Option 2 (0.13). The second scenario used the minimum value obtained from the individual pairwise comparisons for the Employment and Community Needs criterion. Minimizing this criterion’s
152 Wendy Proctor Table 6.11 Pairwise comparison of options under social criteria National Estate*
Option 1
Option 2
Option 3
Option 4
Option 5
Option 1 1 2 3 4 5 Option 2 0.5 1 2 3 4 Option 3 0.33 0.5 1 2 3 Option 4 0.25 0.33 0.5 1 2 Option 5 0.2 0.25 0.33 0.5 1 *The National Estate criterion has been given an arbitrary weighting because only qualitative data were available. Emp. & comm. Option 1 Option 2 Option 3 Option 4 Option 5
Op ion 1 1 1 4 5 6
Option 2 1 1 4 5 6
Option 3 0.25 0.25 1 4 4
Option 4 0.2 0.2 0.25 1 2
Option 5 0.17 0.17 0.25 0.5 1
Table 6.12 Priority of options in terms of social criteria Ranking of options for each sub-criteria**
Option 1 Option 2 Option 3 Option 4 Option 5
Emp. & comm. 0.56* 0.061 0.082 0.169 0.267 0.421
National estate 0.44* 0.416 0.262 0.161 0.098 0.062
Priority 0.217 0.161 0.165 0.193 0.263
* Contribution of sub-criteria to overall priority. ** After normalizing the pairwise comparison matrices.
preference weighting only strengthened the lead of Option 1 further and only slightly changed the lower rankings with an ordering of Option 1 (0.30), Option 5 (0.21), Option 2 (0.17), Option 4 (0.17), Option 3 (0.15). A final sensitivity analysis was carried out on the relatively arbitrary values of the National Estate impacts for the different options. Changing these values from very small differences in impacts to very large differences in impacts had the effect on the final rankings of swapping the placements of Options 1 and 5 between first and second in each case. For small differences in National Estate impacts, the final ranking was Option 5 (0.25), Option 1 (0.24), Option 4 (0.19), Option 3 (0.17), Option 2 (0.15). For large differences in National Estate impacts, the final ranking was Option 1 (0.27), Option 5 (0.24), Option 4 (0.18), Option 3 (0.15), Option 2 (0.15).
Conclusion The outcome of the MCA based on the data released for the Southern Region reveals that the preferred option is Option 1. Contrary to what may be thought
MCDA and stakeholder participation
153
0.45 0.40
Priority weight
0.35 0.30
Environmental
0.25 Economic 0.20 Social
0.15 0.10 0.05 0.00 1
2
3 Option
4
5
Figure 6.6 Overall ranking of options
0.30
Priority weight
0.25 0.20 0.15 0.10 0.05 0.00 1
2
3
4
5
Option
Figure 6.7 Priorities of broad criteria
from initial observations of the data, however, is that the next preferred option is Option 5. On closer inspection of the data, this outcome is seen to be the result of the high performance of Option 1 under the environmental criteria when compared with all the other options and, similarly, the much better performance of Option 5 under the employment criteria when compared with all the other options. This result accurately depicts the polarized nature of the priorities of the Forum members and shows the nature of the trade-off in the forest management debate as being one between conservation and employment. The outcome of the analysis shows
154 Wendy Proctor 0.5
Priority weight
Max 0.25
Min Av
0 1
2
3 Option
4
5
Figure 6.8 Range of individual priorities
0.40
Priority
0.30
0.20
0.10
0.00 1
2
3
4
5
Option Figure 6.9 Environmental preference
that a ‘middle ground’ choice, such as Option 2, 3 or 4 may not be accepted favourably by any stakeholders because none of these options perform strongly enough under either the conservation or the employment criteria. This result is also supported by the outcomes of the individual Forum member’s overall priorities. The sensitivity analysis also supports this conclusion and indicates that the issue of reserving or not reserving areas of land from timber production in the southern New South Wales region is really reduced to a decision between the two extreme forest use options as they currently stand. Maximizing the weighting of the
MCDA and stakeholder participation
155
0.40
Priority
0.30
0.20
0.10
0.00
1
2
3 Option
4
5
Figure 6.10 Economic preference
Employment and Community needs sub-criterion did change the top ranked option to Option 5. However, Option 1 was still ranked second under this variation. Minimizing the weighting for Employment and Community Needs far increased the priority of Option 1 even though Option 5 was still ranked second. The results were sensitive to the arbitrarily assigned weightings used to portray the qualitative differences in the National Estate data but this also only affected to any large extent the rankings of Options 1 and 5. Further analysis of National Estate values in the official process could have resulted in more precise measurement of the impacts of this sub-criterion. The effects of changing the two broad criteria weightings to an economic preference or an environmental preference is to be expected due to the nature of the forest debate. Changes to the social group weightings had little impact on the final results. However, none of the sensitivity analyses performed had any impact on raising the middle three options to the status of the chosen scenario. On the basis of the impact data provided and the preferences of the Forum, the decision appears to be reduced to one between the two extreme forest use scenarios and a choice between conservation and jobs. In carrying out this MCA it seems that there could have been great scope for revising the options (noting that the middle options did not appear to have any impact in the analysis) to accommodate this apparent trade-off between conservation and jobs. For example, one avenue which could have been explored under this analysis is that of following the forest use scenario of Option 1 but with increased emphasis on the encouragement of jobs growth in industries that would not impact on forest ecosystems. At the time of writing, no RFA has been signed by the New South Wales and Commonwealth governments for the New South Wales Southern Region forests. However, the New South Wales government has declared its own ‘Forest Agreement’ choosing, as a result of its own integration process, Option 3 as the preferred
156 Wendy Proctor framework for its forest policy. It is hardly surprising that since announcing this policy, intense public disagreement has resulted, with both conservation groups and timber industry organizations criticizing the decision. Using an approach such as MCA could have led policy-makers to foresee the response to such a proposal and encouraged them to identify possible trade-offs in order to revise and finetune the options. The results therefore show that overcoming the complex problems involved in achieving ecologically sustainable development, such as those of comparing multiple values and incorporating stakeholder participation into the decision process, can be aided by the use of MCA. Some specific merits and flaws deserve mention as a result of carrying out the MCA. MCA imposed a rigorous structure on the decision-making problem that allowed for a methodical approach to the breakdown of the problem which was lacking in the approach to the official decision-making process. This structure and methodical disentangling of the important issues also led to the identification of several key areas of information that should have been collected, but were not, in the official assessments. These included, for example, the issues of water and soil quality and indigenous values under different options. Both of these criteria were given high weightings of importance by most members of the Forum. As this research shows, under the MCA process, data requirements would be identified very early in the process, thus reducing the costs of collecting irrelevant or unnecessary information. Another advantage made evident from this analysis, is that the objectives, criteria and indicators that are made explicit in the MCA technique, could aid the process of future monitoring of environmental policies. Some problems were also identified in using the MCA technique. For example, there seems to be a lack of guidance in the MCA literature in carrying out interactive group analyses. Although the technique does facilitate a rigorous and objective method of incorporating public participation into a policy-making process, some parts of this process require skills that are not addressed in the MCA literature. For example, in carrying out the first part of this analysis to elicit criteria, indicators and preferences, a large amount of time was spent in interacting with the Forum as a group and as part of their regular meetings. Conducting such a facilitating role in issues that are highly contentious in nature requires certain skills, which are not addressed in most textbooks on MCA. The Analytic Hierarchy Process appears to be the only method which addresses (albeit briefly) the problem of conducting interactive group decision-making. To be effective in this part of the process, it is necessary for the analyst to borrow from the literature on negotiation, facilitation and meeting techniques. There exists great scope for incorporating some other techniques, for example, the deliberative approach of the Citizens’ Jury, into the MCA. This approach would also aid the iterative phase that would take place at the end of the MCA.
MCDA and stakeholder participation
157
Notes 1 Under the RFA process, the Southern Region was divided into three smaller zones: the northern, western and coastal sub-regions. Data in this research refer only to the coastal sub-region. 2 In this research, the term ‘decision-maker’ can also apply to multiple decision-makers that may, for example, be comprised of stakeholders, bureaucrats, politicians or a combination of these. 3 The maximum weighting for the environmental and economic criteria were both around 0.75 and so, for the first test for example (Environmental Preference), the environmental weighting was set at 0.75 and the relativities of the other two weightings were maintained as in the original analysis at 0.15 for economic criteria and 0.10 for social criteria. The same was done for the Economic Preference.
References Commonwealth of Australia (1992) National Forest Policy Statement: A New Focus for Australia’s Forests, Online. http://www.rfa.gov.au/nfps/contents.html (accessed April 2000). Commonwealth of Australia (1997a) Australia’s First Approximation Report for the Montreal Process, Canberra: Department of Primary Industries and Energy, Montreal Process Implementation Group. Commonwealth of Australia (1997b) East Gippsland Resource and Economics Report, Joint Commonwealth and Victorian Regional Forest Agreement (RFA) Steering Committee, Canberra and Melbourne. Commonwealth of Australia (1998) Central Highlands: Comprehensive Regional Assessment Report, Canberra. Dann, T. and Clarke, J. (1996) ‘An economic model for comprehensive regional forest assessments, a case study – some issues and considerations’, paper presented at 40th Annual Conference of the Australian Agricultural and Resource Economics Society, Melbourne, 13–15 February 1996. Dovers, S. (ed.) (2000) Environmental History and Policy: Still Settling Australia, Melbourne: Oxford University Press. Doyle, T. and Kellow, A. (1995) Environmental Politics and Policy Making in Australia, South Melbourne: Macmillan. Ferguson, I. (1996) Sustainable Forest Management, Melbourne: Oxford University Press. Hogwood, B. and Gunn, L. (1984) Policy Making for the Real World, Oxford: Oxford University Press. Howlett, M. and Ramesh, M. (1995) Studying Public Policy: policy cycles and policy subsystems, Oxford: Oxford University Press. Lothian, J. (1994) ‘Attitudes of Australians towards the environment: 1975–1994’, Australian Journal of Environmental Management, 1: 78–99. Massam, B. (1988) Multi-Criteria Decision Making (MCDM) Techniques in Planning, Oxford: Pergamon Press. Prato, T. (1999) ‘Multiple attribute decision analysis for ecosystem management’, Ecological Economics, 30: 207–22. Pressy, R., Ferrier, S., Hutchinson, C., Sivertsen, D. and Manion, G. (1995) ‘Planning for negotiation: using an interactive geographic information system to explore alternative protected area networks’, in D. Saunders and J. Craig (eds) Nature Conservation 4: The Role of Network, Sydney: Surry Beaty and Sons. Proctor, W. (1999) ‘A practical application of MCA to forest planning in Australia’, paper presented at IUFRO International Symposium: From Theory to Practice – Gaps and
158 Wendy Proctor Solutions in Managerial Economics and Accounting in Forestry, Czech University of Agriculture, Prague, 13–15 May 1999. Proctor, W. (2000) ‘Towards sustainable forest management – an application of MCA to Australian forest policy’, paper presented at the 3rd International Conference of the European Society for Ecological Economics, Vienna, Austria, 3–6 May 2000. Proctor, W. (2001) ‘Multi-criteria analysis and environmental decision-making: a case study of Australia’s forests’, unpublished PhD thesis, Australian National University, Canberra. Saaty, T. (1980) The Analytic Hierarchy Process, New York: McGraw-Hill. Saaty, T. (1982) Decision-making for Leaders, California: Wadsworth. Saaty, T. and Vargas, L. (1982) The Logic of Priorities, Boston: Kluwer-Nijhoff. Stirling, A. and Mayer, S. (1999) ‘Rethinking risk: a pilot multi-criteria mapping of a genetically modified crop in agricultural systems in the UK’, report for the UK Roundtable on Genetic Modification, SPRU, University of Sussex, August 1999. Turner, B. (1998) ‘Review of FRAMES data for the upper north east and lower north east RFA regions of NSW’, unpublished consultant’s report for the State Forests of NSW and the Bureau of Resource Sciences, Commonwealth DPIE.
Confronting risk with precaution 159
7
Confronting risk with precaution A multi-criteria mapping of genetically modified crops Andy Stirling and Sue Mayer
Introduction The ‘precautionary principle’ is becoming an increasingly prominent theme in the debate over technological risk. Many questions are raised over the implications for policy making. In particular, concerns have been expressed over the relationship between ‘precautionary’ and more traditional ‘science-based’ approaches to decision making such as cost–benefit and risk analysis. Fears are sometimes raised that – unlike risk assessment – a ‘precautionary approach’ is too ambiguous and impractical to serve as a basis for real decision making, that it is somehow antagonistic to science and even that it threatens to stifle technological innovation and economic growth. The first part of this chapter takes a close look at some of the key practical and theoretical issues bearing on the relationship between ‘science’ and ‘precaution’ in the management of technological risk. It is found that – far from being in tension – these two concepts might actually be seen as entirely consistent and even mutually reinforcing. The real distinction is found to lie between a narrow ‘risk-based’ concept of regulatory appraisal and broader precautionary approaches. A series of eight criteria are developed against which appraisal techniques might be evaluated in terms of both their scientific and precautionary validity. The second part of the chapter then reports on a recent pilot exercise that seeks to apply these criteria to the specific case of the regulatory appraisal of a genetically modified crop. The main features of a novel ‘multi-criteria mapping’ methodology are outlined. Key findings are reported, with particular attention paid to the extent to which the method conforms to the scientific and precautionary quality criteria. Conclusions are drawn concerning the degree to which the quality criteria can be made operational in the practical business of regulatory appraisal.
Risk, regulation and precaution Risk is a complex concept. Even under the most narrowly defined of quantitative approaches, it is recognized that risk is a function of at least two variables – the likelihood of an impact and its magnitude. However, it is only very rarely the case that a series of technology, policy or investment options are seen to present only
160 Andy Stirling and Sue Mayer one form of hazard. Normally, the characterization of risks associated with any individual option requires the consideration of a wide variety of disparate risks. In the energy sector, for example, risks can take forms including greenhouse gas emissions, radioactive wastes, heavy metals, persistent organic pollutants, soil erosion, thermal discharges, ambient noise, ecological disturbance or aesthetic intrusion in the landscape. Each of these risks is manifest in a different way, with different physical, biological, social, cultural and economic connotations. The conventional analytical response to this breadth and diversity of issues in regulatory appraisal is to identify a single major yardstick of performance and seek to measure all the various aspects of risk using this as a metric. The chosen unit of measurement in conventional risk assessment is usually human mortality rates, although more complex appraisals sometimes also employ a variety of measures of human morbidity effects. In some areas, the techniques of cost-benefit analysis (CBA) are employed in seeking to subject a wider range of impacts to measurement under a common monetary metric and so allow comparison with the associated benefits. In this way, it is hoped that the multiplicity of magnitudes typically confronted in the regulation of environmental and health risk may usefully be reduced to a single key factor, thus apparently simplifying the process of appraisal. This process of reduction is an essential element in what is sometimes described as a ‘science-based’ approach to the regulatory appraisal of risk. Of course, one crucial consequence of this artificial narrowing and conflation of the full diversity of technological risks is effectively to exclude from consideration many classes of effect. For instance, it is clear that only a minority of the types of energy risks mentioned above is meaningfully addressed by a mortality, morbidity or even a monetary metric. Moreover, even with respect to the single issue of human health, risk is an inherently multi-dimensional concept. For instance, are exposures voluntary or controllable? Are they manifest as disease, injuries or deaths? How familiar are the risks? How immediately are they realized and how reversible once identified? To what extent are they concentrated in large events or dispersed in small routine incidents? How are they distributed across space, time and society? Mortality, and even morbidity, indices fail to capture these important contextual features. Beyond this, further scope for divergent approaches to regulatory appraisal lies in the characteriztics of the assessment process itself. Should appraisal take account of social, economic, cultural and ethical issues, as well as environmental and health factors? With respect to the more narrowly defined physical factors, to what extent should appraisal seek to address the potential additive, cumulative, synergistic and indirect effects associated with particular environmental and health risks? With how wide an array of potential alternatives should each individual technological or policy option be compared in appraisal? Should attention be confined simply to the operation of the options concerned, or should it extend to the manufacture, decommissioning and disposal of plant, as well as to the various inputs (such as energy and materials) and associated risks at each stage? To what extent should the relative benefits of different options be taken into account in appraisal so that they can be offset against the associated risks?
Confronting risk with precaution 161 In an ideal world, the appropriate response to factors such as these is easy to determine. All else being equal, the regulatory appraisal of risk should be as complete and as comprehensive as possible. However, aspirations to completeness concerning the different classes and dimensions of risk and benefit and comprehensiveness concerning the different types of option provide only a rather loose operational guidance in the practical regulation of risk. Moreover, even were appraisal to be fully complete and comprehensive, in some hypothetical sense, then there would still remain the problem of how the different aspects of risk should be framed and prioritized in analysis. For instance, what assumptions should be made about adherence to best practice in the various activities under appraisal? What relative priority should be attached to different effects such as toxicity, carcinogenicity, allergenicity, occupational safety, biodiversity or ecological integrity? What weight should properly be placed on impacts to different groups, such as workers, children, pregnant and breastfeeding mothers, future generations, disadvantaged communities, foreigners, those who do not benefit from the technology in question or even to animals and plants as beings in their own right? Even if they were practically feasible, objectives such as completeness or comprehensiveness do not assist in addressing issues of framing and prioritization of this kind. No one set of assumptions or priorities may be claimed to be uniquely rational, complete or comprehensive. It is here that we come to a classic and well-explored dilemma in the field of social choice theory, but one that is frequently forgotten in risk assessment and regulatory appraisal. The disciplines of risk assessment, economics and decision analysis have developed no single definitive way of addressing the problems of comparing ‘apples and oranges’. Even the most optimistic of proponents of rational choice acknowledge that there is no effective way to compare the intensities of preferences displayed by different individuals or social groups (Bezembinder 1989). Indeed, even where social choices are addressed simply in relative terms, the economist Kenneth Arrow went a long way towards earning his Nobel Prize by demonstrating formally that it is impossible definitively to combine relative preference orderings in a plural society (Arrow 1963). Put simply, the point is that ‘it takes all sorts to make a world’. Different cultural communities, political constituencies or economic interests typically attach different degrees of importance to the different aspects of environmental risk and look at them differently. Within the bounds defined by the domain of plural social discourse, no one set of values or framings can definitively be ruled more ‘rational’ or ‘well informed’ than can any other. Even were there to be complete certainty in the quantification of all the various classes and dimensions of risk, it is entirely reasonable that fundamentally different conclusions over environmental risk might be drawn under different – but equally legitimate – perspectives. It is a matter of the science of risk assessment itself, then, that there can be no analytical fix for the scope, complexity and intrinsic subjectivity of environmental and health risks. The notion that there can be a single unambiguous ‘science-based’ prescription in the regulatory appraisal of risk is not only naïve and misleading; it is a fundamental contradiction in terms.
162 Andy Stirling and Sue Mayer The depths of incertitude This problem may seem serious enough. Unfortunately, the difficulties encountered in the regulatory appraisal of risk are even more intractable than this. Thus far we have considered only the issues associated with the characterization of the ‘magnitude’ aspects of risk. What of the likelihoods? Here we come upon some profound limitations to the applicability and robustness of probabilistic approaches that are as seriously neglected in regulatory appraisal as are the difficulties discussed above concerning the comparison of magnitudes. In economics and decision analysis, the well-established formal definition of risk is as a condition under which both a comprehensive set of all possible outcomes can be defined and a discrete set of probabilities (or a density function) can be defined across this array of outcomes. This is illustrated in the top left-hand corner of the diagram in Figure 7.1. This is the domain under which the various probabilistic techniques of risk assessment are applicable, permitting (in theory) the full characterization and ordering of the different options under appraisal. There is a host of details relating to this picture (such as those hinging on the distinction between ‘frequentist’ and ‘Bayesian’ understandings of probability), but none of these alter the formal scientific definition of the concept of risk. The strict sense of the term uncertainty, by contrast, applies to a condition under which there is confidence in the completeness of the defined set of outcomes, but where there is acknowledged to exist no valid theoretical or empirical basis confidently to assign probabilities to these outcomes. This is found in the lower left-hand corner of Figure 7.1. Here, the analytical armoury is less well developed, with the various sorts of scenario analysis being the best that can usually be managed (Funtowicz and Ravetz 1990). Whilst the different options under appraisal may still be broadly characterized, they cannot be ranked even in relative terms without some knowledge of the relative likelihoods of the different outcomes. Both risk and uncertainty, in the strict senses of these terms, require that the different possible outcomes can be clearly characterized and subject to measurement. The discussion in the previous section has already made it clear that this is often not the case – the complexity and scope of the different forms of environmental risk and the different ways of framing and prioritizing these, can all-tooeasily render ambiguous the definitive characterization of outcomes. This may be so, even where there is relatively high confidence in understandings of the likelihood that at least some form of impact will take place (top right corner of Figure 7.1). An illustrative example here might be the prospects for regional climatic, ecological and socio-economic impacts arising from the human-enhanced greenhouse effect. Where these problems are combined with the difficulties in applying the concept of probability, we face a condition which is formally defined as ignorance (bottom right corner of Figure 7.1) (Loasby 1976; Smithson 1989; Wynne 1992). This applies in circumstances where there not only exists no basis for the assigning of probabilities (as under uncertainty), but where the definition of a complete set of outcomes is also problematic. In short, recognition of the condition of ignorance is an acknowledgement of the possibility of surprises. Under such circumstances,
Confronting risk with precaution 163 KNOWLEDGE
KNOWLEDGE ABOUT OUTCOMES
ABOUT LIKELIHOODS
continuum of outcomes
set of discrete outcomes
outcomes poorly defined
INCERTITUDE RISK Apply: firm basis for probabilities
frequentist distribution functions
discrete frequentist probabilities
AMBIGUITY
Apply: shaky basis for probabilities
no basis for probabilities
Bayesian distribution functions
discrete Bayesian probabilities
fuzzy logic sensitivity analysis precaution
UNCERTAINTY
IGNORANCE
Apply: scenario analysis sensitivity analysis precaution
Apply: diversity flexibility precaution
Figure 7.1 Risk, uncertainty, ambiguity and ignorance defined
not only is it impossible definitively to rank the different options, but even their full characterization is difficult. Under a state of ignorance (in this strict sense), it is always possible that there are effects (outcomes) which have been entirely excluded from consideration. Figure 7.1 provides a schematic summary of the relationships between these formal definitions for the concepts of risk, uncertainty, ambiguity and ignorance. It is quite normal, even in specialist discussion, for the full breadth and depth of these issues to be rolled into the simple concept of ‘risk’ (and sometimes ‘uncertainty’), thus seriously understating the difficulties involved. In order to avoid confusion between the strict definitions of the terms ‘risk’ and ‘uncertainty’ as used here, and the looser colloquial usages, the term ‘incertitude’ can be used in a broad overarching sense to subsume all four subordinate conditions. Either way, it is not difficult to see that it is the formal concepts of ignorance, ambiguity and uncertainty – rather than mere risk – which best describe the salient features of regulatory decision making in areas such as energy technologies, toxic chemicals
164 Andy Stirling and Sue Mayer and genetically modified organisms. Indeed, many of the most high profile technologically-induced ‘risks’ of recent years – such as stratospheric ozone depletion, endocrine disrupting chemicals and BSE, for instance – are all cases where the problem lay not so much in the determination of likelihoods, but in the anticipation of the very possibilities. They were surprises! The crucial point is that intractable uncertainties, ambiguities and ignorance are routinely treated in the regulatory appraisal of technology simply by using the probabilistic techniques of risk assessment. This treatment of uncertainty and ignorance as if they were mere risk effectively amounts to what the economist Hayek dubbed (in his Nobel acceptance speech) ‘pretence at knowledge’ (Hayek 1978). Far from displaying a respect for science in regulatory appraisal, the effect of such scientistic oversimplification is actually to ignore and undermine the scientific principles on which risk assessment itself purports to be based. Given the manifest inapplicability – in their own terms – of probabilistic techniques under uncertainty and ignorance, this is a serious and remarkable error. The selfcontradictions in aspirations to a ‘science-based’ approach reliant solely on quantitative risk assessment, already noted in the last section, are thus further underscored and reinforced. Why is it that pursuit of (and claims to) the definitive authority of ‘sciencebased’ approaches continues to be so prominent in regulatory appraisal? It seems that the elegance and facility of probabilistic calculus has had a seductive effect on many risk analysts and their sponsors. This may be understandable, yet it is also curious. Despite the intractability of the condition of ignorance, there is no shortage of operational tactical and strategic ‘precautionary’ responses. Some specific features of these approaches will be reviewed in some detail later. For the moment, in the specific context of ignorance, the point is simply that there do exist practical alternatives to the use of probabilistic methods. For instance, there exists a variety of institutional procedures for including in the regulatory appraisal process a range of different scientific disciplines and people with pertinent professional and local knowledge and relevant socio-economic perspectives. By providing for the identification of a wider range of possibilities, this effectively helps to convert some part of the domain of ignorance into the more tractable condition of uncertainty. Here, techniques such as scenario and sensitivity analysis can also help systematically to characterize these neglected possibilities and explore their implications under different perspectives. Beyond this, there are a series of broader strategies that may be employed and which will be returned to later. In particular, rather than focusing entirely on efforts to characterize the ‘problem’ (ignorance) attention can also be devoted directly at aspects of the ‘solution’. Although the manifestations of ignorance are, by definition, not characterizable in advance, certain dynamic properties of the different options themselves can offer valuable ways of hedging against ignorance. Here properties such as flexibility, reversibility, resilience, robustness and adaptability are all potentially valuable (Stirling 1999; Salo 1999). Perhaps even more important, are the possible merits of deliberate diversification across a range of options. After all, it is a well-established matter of common sense that, when we don’t know what we
Confronting risk with precaution 165 don’t know, we don’t put all our eggs in one basket! It may be that a persistent preoccupation with probabilistic methods has left these kinds of strategies unduly neglected in the regulatory appraisal of technological risk. The practical consequences for risk assessment The problems discussed so far – the multi-dimensionality of environmental and health risks and the conditions of uncertainty, ambiguity and ignorance – may all seem a little abstract and theoretical. It is perhaps also partly for this reason that they remain relatively neglected in the business of regulatory appraisal. Unfortunately, however, they have some important practical consequences that, though often concealed, hold profound implications for the interpretation of orthodox risk assessment results in all fields, extending from the regulation of energy options, through chemicals and industrial hazards to genetic modification technologies. In all these areas, the typical response to these difficulties in regulatory appraisal is to reduce and simplify – focusing on those aspects that are either the most tractable or the most ‘reasonable’ under certain dominant perspectives. In this way, individual studies can construct a picture of environmental risks, which appears to be quite unambiguous and precise. The scale of the discrepancies only becomes evident on occasions when attention is extended to a series of different appraisal studies, each applying subtly different – but equally ‘reasonable’ and ‘legitimate’ – framing assumptions concerning the different dimensions of appraisal discussed here. When this takes place, it becomes clear that the apparent relative riskiness of different options can vary quite radically, depending on the framings and priorities attached to the ‘hidden variables’ during the process of appraisal. Figure 7.2 illustrates this by showing the results obtained in 32 large-scale risk assessments of eight different energy technologies conducted in industrialized countries over the past two decades. Here, environmental and health effects are characterized using the techniques of CBA as monetary ‘external costs’ expressed in standardized form per unit of electricity production (Stirling 1997). This case is taken as an example because both the techniques employed, and this particular field of application, might arguably be seen as being among the most mature and intensively explored areas of application of comparative risk assessment. The picture is not specific to these techniques or this field. A similar pattern may be found in a variety of other regulatory fields, including transport, toxic chemicals and food safety. The same pattern is also evident in the underlying physical and mortality indices on which these monetary results are based. A number of salient features can be seen. First, individual studies present their results with great precision – often as a single value rather than a range and sometimes expressed with as many as four significant figures (one part in ten thousand). Yet, the variability in the results obtained in the literature as a whole for any one option is radically larger. For instance, the uppermost values of the highest range assessing the risks associated with coal power amount to the equivalent of some twenty dollars per kilowatt-
166 Andy Stirling and Sue Mayer Logarithmic scale 0.001 0.01
0.1
1
0
1
100
1000
0
1
100
1000
ONSHORE WIND PHOTOVOLTAICS BIOMASS HYDROELECTRICITY NUCLEAR FISSION GAS OIL COAL 0.001
0.01
0.1
1
Figure 7.2 Ambiguity of ordering in risk assessments
hour of electricity production. The lowest values of the bottom range in Figure 7.2 are less than four hundredths of a cent per kilowatt-hour. The difference is more than four orders of magnitude – a factor of more than fifty thousand! Detailed analysis of the reasons for these discrepancies show that they do not arise as a result of any single factor. It is not a simple matter of some studies being more ‘accurate’ or ‘reasonable’ than others in any definitive sense. Instead, the variability is the cumulative consequence of the adoption of divergent assumptions and priorities concerning the whole range of the different ‘dimensions of appraisal’ identified in the preceding sections (Stirling 1997). The second crucial feature that is illustrated in Figure 7.2 concerns the ambiguities in the ordering of the different options under appraisal. The lowest values obtained for the worst ranking option (coal) are lower than the highest values obtained for the apparently best ranking options (wind). Since the effect of the particular assumptions adopted in individual studies is to produce results at the high end of the overall range for some options but lower in the distributions for others, the overall picture yielded by the literature as a whole would accommodate virtually any conceivable ranking order for these eight options! By the judicious choice of framing assumptions, then, radically different conclusions can be justified for regulation. This evident disjuncture between precision and accuracy in supposedly ‘sciencebased’ risk assessment paints a rather negative picture. One of the first and most basic tasks in the management of risk is to construct some robust overall notion of the relative merits of the different options under consideration from the point of view of society as a whole. This then serves as a basis for regulatory intervention,
Confronting risk with precaution 167 market-based measures or investment initiatives. Where this cannot be achieved in any absolute (or even relatively robust) sense, then the value of appraisal lies in systematic exploration of the relationships between different assumptions in analysis and the associated pictures of the relative importance of different options. Where aspirations to the ‘science-based’ appraisal of risk lead to the assertion of the intrinsic authority of narrow risk assessment procedures, then these crucial exogenous factors typically remain unacknowledged and unexplored. In this event, the problem is not simply one of a lack of rigour concerning the theoretical contradictions noted in the previous sections. The difficulties are also very concrete and pragmatic. For, without a robust appreciation of the assumptions under which appraisal yields differing pictures of performance, serious questions must be raised over whether the associated results – no matter how confidently and precisely expressed – are of any practical policy use at all. ‘Science’ and ‘precaution’ in the regulatory appraisal of environmental risk It is with increasing realization of these practical and theoretical limitations to the value of orthodox risk assessment in regulatory appraisal, that interest is growing in complementary and alternative approaches. In particular, the ‘precautionary principle’ is becoming an ever more prominent feature of the regulatory debate on environmental risks and of national and international legislation (O’Riordan and Cameron 1994; Fisher and Harding 1999; Raffensberger and Tickner 1999; O’Riordan and Jordan 2001). Although subject to a variety of different definitions, in the broadest of terms, a ‘precautionary’ approach acknowledges the difficulties in risk assessment by granting greater benefit of the doubt to the environment and to public health than to the activities which may be held to threaten these things. A host of different practical instruments and measures are variously proposed in different contexts as embodiments of a ‘precautionary approach’ or as means to implement a ‘precautionary principle’. For present purposes, attention will concentrate on the way in which a precautionary approach offers a direct response to the practical and theoretical problems in regulatory appraisal which have been discussed so far. One key theme in the current lively debate on these matters surrounds the frequent assertion (and sometimes assumption) that – whatever form it takes – a ‘precautionary’ approach to the management of environmental risk is somehow in tension with (or even antithetical to) the generally uncontroversial aspiration that regulatory decision making should be based on ‘sound science’. Of course, this does not address the extent to which orthodox ‘scientific’ approaches such as comparative risk assessment may themselves be claimed to yield ‘sound’ results. The thrust of the discussion thus far has been to raise serious doubts over this. Nevertheless, the important question remains as to what exactly is the relationship between so-called ‘science-based’ and ‘precautionary’ approaches to the regulation of environmental risk? A necessary starting point for this analysis is a clear characterization of exactly
168 Andy Stirling and Sue Mayer what is meant by ‘science’ and ‘precaution’ in the context of decision making on environmental risk. Drawing on a wide literature, Figure 7.3 displays some idealized attributes of scientific approaches to regulatory appraisal (Stirling 1999). In short, a scientific approach to the management of risk should, ideally and at minimum, be transparent in its argumentation and substantiation, systematic in its analytical methods, sceptical in its treatment of knowledge claims, subject to peer review, independent from special interests, professionally accountable and continually open to learning in the face of new knowledge. These aspirations may not always be realized, but they represent fundamental, and relatively uncontroversial, principles guiding any ‘science-based’ approach to regulatory appraisal. Likewise, drawing on an equally extensive parallel literature, it is possible broadly to characterize the essential features of a ‘precautionary’ approach to the management of risk. In short, a precautionary approach involves the application of principles that ‘prevention is better than cure’, that ‘the polluter should pay’, that options offering simultaneously better economic and environmental performance should always be preferred (‘no regrets’), that options should be appraised at the level of production systems taken as a whole and that attention should be extended to the intrinsic value of non-human life in its own right (a ‘biocentric ethic’). In effect, this is variously taken to mean a certain humility about scientific knowledge and an acknowledgement of the complexity and variability of the real world. It implies recognition of the vulnerability of the natural environment and living organisms and the prioritizing of the rights of those who stand to be adversely
SCIENTIFIC DISCIPLINE
Precautionary approach
Risk-based approach
systematic sceptical peer-reviewed independent accountable learning Cornucopian
Apocalyptic BREADTH OF SCOPE
consider all effects
acknowledge possibility of surprise
compare alternatives account for pros as well as cons reverse onus of persuasion
include many perspectives take broad social view extend research and monitoring
Figure 7.3 A model of the relationships between risk, science and precaution
Confronting risk with precaution 169 affected. It requires scrutiny of claims to benefits and justifications as well as risks and costs, with full account given to the available alternatives. Finally, a precautionary approach involves the adoption of long-term, holistic and inclusive perspectives in regulatory appraisal (Stirling 1999). In many ways, these attributes of a ‘precautionary’ approach can be seen to concern different aspects of the breadth of the regulatory appraisal process. A ‘broad’ regime is one that takes account of a wide range of different types of impact, including qualitative as well as quantitative issues and including indirect as well as direct effects. Likewise, a ‘broad’ framework accommodates a diverse array of different points of view (including, importantly, those of potential ‘victims’) and anticipates a wide range of possibilities in the face of uncertainty and ignorance. It extends consideration to the benefits and justifications associated with the introduction of the technology in question and examines a variety of alternative ways in which the benefits of a regulated technology might be realized at lower levels of risk. Taken together, these features constitute a more ‘precautionary’ approach because they increase the number and intensity of the constraints that any technological option must satisfy in order to be approved by the regulatory process, thus making it more difficult for certain innovations to pass through the regulatory ‘filter’. At the same time, however, such measures might equally serve to encourage other technological innovations that might otherwise remain neglected. What is interesting about this characterization of ‘precaution’ in terms of the ‘breadth’ of the associated regulatory regime, is that it reveals an inherently consistent – and in many respects complementary – relationship between ‘precaution’ and ‘science’ in the management of technological risk. Accordingly, Figure 7.3 distinguishes between different approaches to risk management based on the degree to which each embodies the respective characteristics of ‘scientific discipline’ and ‘breadth of framing’ identified here. Of course, both the ‘broad’/‘narrow’ and the ‘scientific’/‘unscientific’ dichotomies drawn here are highly stylized and simplified. However, the general picture revealed in Figure 7.3 is at least richer and more realistic than the prevailing one-dimensional dichotomy between ‘science’ and ‘precaution’. Taken together, the combination of these two dichotomies generates a fourfold array of idealized permutations. The adoption of a ‘narrow’ regime without reference to scientific understandings or disciplines in appraisal might be described as a permissive position. Taken to an extreme, this would amount to an entirely uncritical ‘anything goes’ approach to the regulation of technology of the kind associated with caricature ‘cornucopian’ visions of technological progress. Likewise, a broad-based regime might be similarly unscientific. The resulting restrictive position might be associated with a caricature ‘apocalyptic’ vision of technology. In the extreme, it would lead to a situation of paralysis under which no new technological innovation that offends in the slightest respect would ever be approved for deployment. The crucial point is that neither the ‘permissive’ (cornucopian) nor the ‘restrictive’ (apocalyptic) positions as defined here would be subject to challenge or reversal by the disciplines of scientific discourse associated with the vertical axis.
170 Andy Stirling and Sue Mayer Clearly, neither the established procedures of risk regulation (based on relatively narrowly framed risk assessment methods) nor the emerging precautionary approach (based on broader perspectives and considerations) actually resemble these stylized ‘permissive’ or ‘restrictive’ caricatures. Existing risk-assessment-based regulation includes a host of effective checks and balances. It certainly does not necessarily provide for the uncritical approval of any new technology that may be developed. Likewise, even the most progressive formulations of a ‘precautionary principle’ are circumscribed in their scope, admit an incremental series of instruments and allow for regulatory approval under a host of favourable conditions. Both approaches are compatible – at least in principle – with the requirements of systematic methodology, scepticism, transparency, accountability, quality control by peer-review, professional independence and an emphasis on learning which are held for the purposes of this discussion to be among the key aspirations of a ‘sciencebased’ approach. Now let us return to the earlier discussion of the profound importance of the conditions of uncertainty, ignorance and multidimensionality in risk assessment. Questions over the scope of appraisal, the plurality of different value positions and framing assumptions, the diversity of different anticipated possibilities and the degree of confidence placed in the available knowledge are all matters that are central to the ‘scientific’ status of the appraisal process. That probabilistic approaches are inapplicable under strict uncertainty and ignorance flows directly from the theoretical foundations of risk assessment, and CBA (and, indeed, all ‘rational choice’ approaches to decision-making on risk). Following equally directly from these fundamental theoretical principles, different priorities, framing assumptions and value systems cannot be definitively aggregated across different groups. Thus, for both these reasons, there can be no analytical fix for the definitive ranking of different technology or policy options in the social appraisal of risk. All that can be done to maximize scientific rigour in appraisal is to ensure that the process is as broadly based as possible in terms of the value systems and framing assumptions that are included and the options and possibilities that are addressed. Seen in this way, then, key elements of the ‘breadth’ of the regulatory regime themselves become issues of ‘sound science’ in the management of environmental risk, as well as institutional features of the wider regulatory regime. Precaution, in this sense, is not just entirely consistent with science – it is a necessary pre-requisite for a truly scientific approach to the regulatory appraisal of risk. Implementing a precautionary approach to appraisal In the discussion so far, environmental and health risks have been treated at a high level of generalization. Of course, it is obvious that at a greater level of detail, different technological and policy options will vary radically in the magnitude and character of the risks that they present. Likewise, the range of different possible regulatory interventions do not fall neatly into ‘permissive’ or ‘restrictive’ categories, but lie along a series of continua, and range between being more and less precautionary in their effect. With respect both to precautionary and other approaches
Confronting risk with precaution 171 to the regulation of risk, then, different measures will be appropriate in different contexts. This general picture is the subject of a vast literature (e.g. O’Riordan and Cameron 1994; Fisher and Harding 1999; Raffensberger and Tickner 1999; Stirling 1999; Renn and Klinke 1999; O’Riordan and Jordan 2001). Here, however, the purpose is to focus specifically on some key implications for regulatory appraisal. As a result of the foregoing analysis, we construct a series of eight evaluative criteria against which the regulatory appraisal of risk can be assessed both in terms of its scientific rigour and its precautionary qualities (after Stirling 1999). In short, these are as follows: 1
2
3
4
5
6
7
8
Humility: maintain a culture of humility in the face of the many sources of uncertainty, ignorance and subjectivity in appraisal. Avoid claims to complete or otherwise definitive knowledge. Completeness: broaden the scope of the regulatory appraisal of technological risk to address cumulative, additive, complex, synergistic and indirect effects as well as more direct causal processes. Benefits and Justifications: include systematic consideration of the claimed benefits and justifications as well as adverse effects, in order to allow determination of ‘net’ benefits under different contexts. Comparison: conduct appraisal on a comparative rather than a case-by-case basis, including account of a variety of technological and policy options and the cumulative effects across different cases. Participation: ensure full engagement by all interested and affected parties, both to elicit all relevant knowledge and to include consideration of all pertinent priorities and framing assumptions. Mapping: express appraisal results not as discrete numerical values, but using sensitivity analysis systematically to ‘map’ the consequences of different value judgements and framing assumptions. Transparency: use the most straightforward of methods. Minimize the number of hidden variables. Provide for detailed auditing of how particular results derive from particular inputs. Diversity: extend appraisal to address the way that diverse mixes of different options may help to hedge against uncertainty and ignorance and help accommodate divergent social perspectives.
These are the considerations that have informed the development of the ‘multicriteria mapping’ technique employed in the present case study (Stirling 1997; Stirling and Mayer 1999). In short, the motivation behind this approach is to seek to combine the openness and qualitative flexibility of participatory deliberation with the clarity and focus of quantitative assessment. The specific case study with which this method has been piloted concerns the hotly contested debate over the use of genetically modified crops in UK agriculture.
172 Andy Stirling and Sue Mayer
Applying the multi-criteria mapping technique The UK debate over GM crops The agricultural use of GM technologies is held in some quarters to promise great benefits. On the other hand, there is general agreement that there exists at least the potential for serious, irreversible harm. In the UK, formal regulatory appraisal of GM crops has centred on the question of whether or not they are ‘safe’ for the environment and for human consumption. However, there is considerable scientific incertitude over the form and magnitudes of the possible effects and, as yet (by contrast with chemical or nuclear risks), little accumulated practical experience to draw upon. This has led to the evolution of a set of controls which are intended to be precautionary in nature – where it is accepted that action to avoid harm may be taken in the absence of scientific proof – with the conduct of risk assessment being required before experimental or commercial use of a particular genetically modified organism is allowed. Despite this somewhat precautionary approach to risk regulation enshrined in the European Commission’s Deliberate Release Directive (90/220/EC) and the Novel Foods Regulation, the regulatory appraisal process has failed to gain confidence, either of Non-Governmental Organizations (NGOs), private industry (Mayer et al. 1996) or the general public (EPCAG 1997; Grove-White et al. 1997). This lack of confidence arises among other reasons: because the scope of the regulatory appraisal is still held in many quarters to be too narrow; because there is a general lack of trust in official reassurances of safety (particularly in the wake of BSE); and because justifications and benefits are not explicitly included in the evaluation process. Industry and regulators have expressed frustration, believing that the precautionary approach is being invoked in too burdensome a fashion, with unrealistic demands being made concerning absolute proof of safety. It has also been almost impossible to gain agreement between European Union Member States over whether particular commercial releases of GM crops are environmentally ‘safe’, despite a supposedly common approach to their risk assessment (Von Schomberg 1998; Wynne and Mayer 1999). Disputes routinely emerge over the appropriate scope of regulatory appraisal. Even where there is agreement over the possibility that effects will occur, notions of what constitute adverse effects remain strongly contested. These sorts of problems with the current regulatory appraisal of GM crops are typical of those that beset the use of conventional risk assessment and cost-benefit analysis in other areas. Taken together, these characteristics of the debate over the use of GM technologies in agriculture make it a challenging candidate for a case study concerning the application of the general principles of scientific and precautionary appraisal enunciated above. The pilot study The details of the multi-criteria mapping methodology and this particular application in the GM field in the UK are described at length elsewhere (Stirling 1997; Stirling and Mayer 1999, 2000). Only those features bearing on the
Confronting risk with precaution 173 examination of the quality criteria given above will be outlined here. The pilot study took place between April 1998 and September 1999. It was funded by the transnational food firm Unilever, but was entirely independent in its conception, design, implementation and reporting. The two authors come from an academic and NGO background and the project was overseen by a steering group comprising a wide range of environmental, consumer, farming and food industry representatives. The specific topic chosen for the study was the production of oilseed rape in the UK. Six ‘basic options’ were identified in advance for the purposes of comparison (Table 7.1). In consultation with the steering group, twelve individual high profile protagonists in the GM food debate were selected as participants. These individuals came from a variety of backgrounds, including academics and government advisers, environmental, consumer and religious organizations and representatives from the farming, food and biotechnology industries. Of course, it is impossible to claim any ‘statistically representative’ status for such a small sample. However, care was taken that, between them, the group of participants covered a full ‘envelope’ of specialist and socio-political perspectives – ranging, for instance, from strongly opposed to strongly in favour of the use of GM crops. Given the character of the UK GM debate at the time of the study (1998–9), it was necessary to give undertakings of anonymity in order to secure participants’ involvement. However, the viewpoints of participants are reproduced individually, being identified by a code letter and an affiliation with one of four general groupings of perspectives: ‘academic’, ‘NGO’, ‘industry’ or ‘government’ (Table 7.2). Table 7.1 The definitions of the ‘basic options’ appraised by all participants Option
Definition
Organic agriculture
All farming and food production conducted under presentday organic standards.
Integrated pest management
All farming and food production conducted using systems designed to limit but not exclude chemical inputs and with greater emphasis on biological control systems than conventional systems.
Conventional agriculture
All farming and food production conducted under presentday intensive systems.
GM oilseed rape with segregation and present systems of labelling
Labelling based on the presence of foreign DNA or protein in the final product.
GM oilseed rape with post-release monitoring
Monitoring for effects (mainly environmental) conducted on an on-going basis after commercialization.
GM oilseed rape with voluntary controls on areas of cultivation
Areas of growing of GM oilseed rape restricted on a voluntary basis to avoid unwanted effects such as gene-flow and cross-fertilization of non-GM crops.
Up to six additional options to be specified by participant
Any option of participant’s choice including combinations of the above if desired.
174 Andy Stirling and Sue Mayer Table 7.2 The participants Area
Code
Agriculture and food industry
B, L, H, K
Academic scientists
C, J
Government safety advisors
E, F
Religious and public interest groups
A, D, G, I
During a 2–3 hour individual interview, each participant undertook a fourstage process (Figure 7.4): (i) the identification of additional options; (ii) the defining of appraisal criteria under which the options should be assessed; (iii) the scoring of the performance of each option under each criterion; and (iv) the weighting of each criterion in terms of its relative importance. A straightforward ‘linear additive’ multi-criteria procedure was employed using a simple spreadsheet model mounted on a portable computer in order to display to the participant the results of the scoring and weighting process as it developed. The process then iterated cyclically until the participant was satisfied that the results accurately reflected their own personal perspective and professional judgement on the issues at hand. One crucial feature of the scoring process was the deliberate eliciting of a range of performance scores for each option under each criterion. Rather than providing a single ‘best guess’, this allowed consideration of a range of optimistic and pessimistic assumptions under each perspective. The discussion of the associated issues was then carefully documented and provided an interesting reflection of the importance of technical uncertainties in the final picture of performance derived under each perspective. In the period following the interviews, participants were each contacted and asked to consider one final aspect of appraisal. This concerned the role of deliberate diversification among options as a means to implement two crucial features of the precautionary approach discussed earlier in this chapter: the desire to hedge against intractable uncertainties and ignorance and to accommodate a variety of sociopolitical viewpoints. For this purpose, a simple index of diversity was borrowed from mathematical ecology and information theory in order to model different trade-offs between the value attached to the performance of individual options and the value attached to diversity in the mix of options. The details of this method are described in more detail elsewhere (Stirling 1994, 1997; Stirling and Mayer 1999). In short, participants can choose to assign a ‘zero’, ‘low’, ‘medium’ or ‘high’ weighting to diversity. Where this takes a zero value, then only the best-performing option is included in the favoured mix. Where higher weightings are placed on diversity, progressively greater contributions are included from lower-performing options – drawn on in proportion to their relative performance under the perspective in question. In this way, the study was able to elicit a crude notion of the importance of ignorance and pluralism under the different perspectives. As analysis proceeded, participants were asked to review the results of a thorough sensitivity analysis conducted on their own set of weightings. Along with an
Confronting risk with precaution 175 1 CHOOSE OPTIONS
OPTIONS CRITERIA
WEIGHTS
A
B
C
D
E
environment
2
health
DEFINE CRITERIA
economics
5
SCORES
CALCULATE RANKS
pessimistic / optimistic
other
RANKS
3 ASSESS SCORES
4 ASSIGN WEIGHTS
EXAMPLE
GM 0.14 pessimistic assumptions GM0.572 GM0.566 con0.252 organic IPM0.142 org 0.21
0.14 1
optimistic assumptions
0.5717 1 0.5658 1 0.2521 0 0.1423 1 0.21 1
integrated pest management conventional GM - current GM - monitoring GM - voluntary controls low performance
high performance
Figure 7.4 The multi-criteria mapping process
invitation that participants retain the spreadsheet model and freely experiment with it at their leisure, this provided a further means to verify that the final results represented a robust reflection of the different viewpoints. Finally, all participants were invited to a concluding workshop at which the results were again confirmed and used as a basis for wider-ranging discussion. Findings The procedure summarized above generated a rich body of empirical material, reflecting a very wide range of issues bearing on the regulatory appraisal of GM
176 Andy Stirling and Sue Mayer technologies in agriculture. This is reported on in more detail elsewhere (Stirling and Mayer 1999, 2000) but the key points can be summarized here. In addition to the six ‘basic options’ a further 18 alternative agricultural strategies were identified and evaluated by participants. These included many different labelling and control regimes and the application of a series of different regulatory assessment procedures and criteria. They also involved a variety of different agricultural strategies, including a focus on hypothetical strategies using GM technologies under an organic farming regime. Interestingly these tended to perform relatively well under the perspectives of participants from both sides of the GM debate (Table 7.3). A total of 117 appraisal criteria were defined by different participants. Although some of these are apparently closely related, all embody important differences of emphasis or framing. It is interesting that criteria typically reflect issues not only of consumer choice but also of citizenship and wider questions of participation and agency. Collectively they cover a very wide range of considerations, including environmental, agricultural, health, economic, social and ethical issues. Most of these issues are very remote from the factors included in orthodox risk assessment. Indeed, for no participant was it true that the full range of their concerns is fully addressed by existing regulatory appraisal procedures in the UK (Table 7.4). Table 7.3 Additional options defined by participants Labelling and/or other controls
GM crops with segregation, full labelling and postrelease monitoring and legally binding growing contracts GM crops with segregation, current labelling and postrelease monitoring GM crops with segregation, full labelling and postrelease monitoring GM crops with segregation and labelling according to means of production and source of gene, plus postrelease monitoring GM crops with segregation, comprehensive labelling based on process and generic restrictions on some classes, e.g. in centre of origin GM crops within controlled sectors (compulsory control) GM crops with legally binding threshold for gene transfer to non-GM stream
A F H G I A A
Agricultural system
GM crops, IPM system GM crops, IPM system No GM crops – conventional and organic as now GM crops in conventional and organic systems GM crops, organic agricultural system, plus segregation, labelling and other regulations as required
G J K K J
Assessment criteria
GM crops with quality GM crops with assessment of indirect agricultural impact and assessment of need
B I
Other
GM crops only in USA No GM commodity crops Complete public control over choice
A A C
Confronting risk with precaution 177 Table 7.4 Broad groupings of criteria defined by participants Environment Biodiversity
Health
Agriculture
Economy
Society
Ethics
e.g. field boundary ecology, other environmental risks Chemical use e.g. reduction in use of existing herbicide sprays, benefits of contact herbicides versus soil acting residuals, longer term pollution of air and water Genetic pollution e.g. gene flow to other crops and native flora Wildlife effects e.g. impact of enhanced weed control efficiency on wildlife, other practices affecting wildlife value of agricultural systems Unexpected effects e.g. potential for effects not foreseen under this scheme Visual e.g. amenity impacts Aesthetics e.g. feelings about environment Allergenicity e.g. from food consumption Toxicity e.g. human or animal health Nutrition e.g. to consumers Unexpected effects e.g. unexpected interactions between ingredients, stability of genetic insert Ability to manage e.g. traceability and ease of recall Weed control e.g. invasive volunteers and weedy relatives Food supply e.g. sustainability, tendency to monocultures, global stability food security Agricultural e.g. farmers’ rights, choice and quality of life, land practice requirements Consumer benefit e.g. retail price Producer benefit e.g. shorter term costs, yield or longer term value added Benefit to processor e.g. profitability Socio-economic e.g. welfare of small farmers, substitutions for impact developing countries Individual impacts e.g. consumer choice, transparency, accessibility, participation, pluralism Institutional e.g. concentration of power, institutional trust, impacts regulatory complexity Social needs e.g. new opportunities, opportunity costs, misuse of science, employment, quality of life Fundamental e.g. animal welfare, taking care of nature principles Knowledge base e.g. hubris about scientific knowledge
The pattern displayed by the performance scores assigned by participants under the various groups of criteria shows that the differences between perspectives are not simply due to variations in the weighting placed on different criteria. Instead, the picture yielded in this exercise is very similar to that illustrated above with respect to risk assessment of energy technologies (Figure 7.3). Participants adopt a variety of different ‘framing assumptions’, resulting in significant differences in the scores assigned by different participants under the same criteria. This has implications for conventional multi-criteria analysis, in which scoring is often conducted by a separate body of experts, with an assumption that different value judgements can be captured simply in the weightings. Despite these differences,
178 Andy Stirling and Sue Mayer however, the patterns evident in the scoring do allow certain general conclusions. For instance, under all perspectives the organic option tends to perform well under environmental criteria. Also, the evident divergence between the patterns of performance under health and environmental criteria challenge conventional assumptions that these aspects of performance will necessarily be well correlated. Notwithstanding the complications raised with respect to scoring above, the patterns evident in the weightings do illuminate some interesting, if rather unsurprising, features. Perspectives drawn from the biotechnology industry and food supply chain are conspicuous in their relative under-emphasis of the social and/ or environmental and safety considerations which are prominent under all other perspectives. The perspectives adopted by government advisers have the distinctive characteristic of being at the same time relatively narrow in scope whilst emphasising environmental and safety considerations. The perspectives expressed by the non-industry participants share a markedly lower emphasis on economic or agricultural considerations. Perhaps the most important outputs of the multi-criteria mapping procedure are the patterns in the final rankings obtained under the various participants’ perspectives. Indeed, it is this picture which constitutes the ‘map’ referred to in the name of the technique, reflecting the way that final results vary with starting assumptions. Figure 7.5 displays the results obtained in the present exercise for the ten participants from whom it was possible to elicit all the necessary quantitative inputs. For reasons discussed more fully in the full report (Stirling and Mayer 1999) the remaining two – G and H – expressed difficulties either in the scoring or weighting process. Each chart represents the rankings elicited from a single participant. The sequence of basic options is the same in each case, running top to bottom from organic farming, through integrated pest management and conventional intensive agriculture to GM crops with (respectively) segregation and labelling, monitoring and voluntary controls. The extension of the horizontal bars to the left reflects the overall performance of the option under that perspective – the further to the left, the better the performance. The scale is in arbitrary subjective linear units of performance. The length of the bars shows the difference between the performance under optimistic and pessimistic assumptions under each perspective, providing a useful indication of the relative importance of uncertainties. The differences of ordering in the left- and right-hand ends of the bars show the practical implications for option performance of adopting pessimistic or optimistic assumptions under each perspective. A number of features can be distinguished in this picture. First, it is clear that there exist profound differences between the practical implications of the perspectives adopted by different participants. In an orthodox regulatory appraisal procedure (such as risk assessment) such variability would typically be concealed by the emphasis on a single aggregated position. In the present analysis, on the other hand, the idiosyncrasies of each individual position are clearly displayed. Second, it is clear that the differences between rankings obtained under optimistic and pessimistic scores are generally rather small compared with the differences between perspectives. This suggests that it is not generally the technical dimensions
Confronting risk with precaution 179 C org ipm
Academic scientists J org ipm
cnv
cnv
gm1
gm1
gm2
gm2
gm3
gm3
E org ipm
Government safety advisers F org ipm
cnv
cnv
gm1
gm1
gm2
gm2
gm3
gm3
A org ipm
Public interest D org ipm
cnv
cnv
gm1
gm1
gm2
gm2
gm3
gm3
I org
B org
ipm
ipm
Industry
cnv
cnv
gm1
gm1
gm2
gm2
gm3
gm3
K org
L org
ipm
ipm
cnv
cnv
gm1
gm1
gm2
gm2
gm3
gm3
Each chart shows one perspective on the relative performance of six agricultural options for producing oilseed rape. Individual perspectives are grouped and colour coded according to their affiliation with four different constituencies. In descending sequence on the vertical axis, the options are organic farming, integrated pest management, conventional intensive agriculture and three different strategies for the implementation of genetically modified crops. The horizontal axis shows option performance and uncertainties on a subjective interval scale (low to left, high to right). Figure 7.5 Final individual rankings for the basic options
180 Andy Stirling and Sue Mayer of uncertainty which are the key issue, but rather more intangible qualitative aspects concerning the divergent interests, values and framing assumptions adopted by different participants. Third, despite the variability in the evaluations of different participants, there emerges a series of consistencies. With only a few exceptions, the organic option tends to perform relatively well. Conventional intensive agriculture tends to perform uniformly badly. The voluntary controls regime for GM crops is regarded rather poorly by industry, academic and NGO participants alike, being viewed favourably only by the government advisers. The final issue concerns the merits of diversification as part of a precautionary response to uncertainty, ignorance and divergent values. Figure 7.6 shows the diverse mixes of agricultural strategies that were favoured by each of the seven participants who were able to respond to this part of the analysis. The partitioning of the pie charts represents some arbitrary measure of importance, such as ‘share of output’ or ‘share of land in production’. For present purposes, the details do not affect the broad-brush conclusions. What is interesting is that six of these participants – drawn from all sides of the debate – favour the deliberate pursuit of some degree of diversity. The remaining participant – a senior UK government safety adviser – displays a uniquely high degree of confidence in the validity and robustness of their own perspective. Together with the idiosyncrasies evident in the rankings displayed in Figure 7.5, this is, in itself, an informative result. Taken as a whole, this part of the multi-criteria mapping analysis might raise questions over the extent to which R&D and regulatory policy making should be geared towards active encouragement of a variety of techniques rather than assuming or emphasising a single particular trajectory. Given the broad support expressed from all sides of the debate for the principle of diversity, questions might also be raised over how to treat options that display characteristics that militate against diversity. In summary, then, this pilot multi-criteria mapping exercise permits a series of quite subtle and specific observations to be made concerning the character of the debate and the positions of the different protagonists. However, it also provides a basis for some general normative conclusions concerning the options under appraisal. These findings are all the more robust for being set firmly in the context of systematic exploration and acknowledgement of the underlying variability and dissent. In particular, in the present case, the picture is quite unequivocal with respect to the relatively negative performance of conventional intensive farming and voluntary controls on the use of GM crops. If equal attention is paid to each perspective, then it is clear that the organic farming and integrated pest management options tend to perform significantly better overall.
Conclusions This chapter has sought to cover quite a bit of ground. It has been shown that there are significant theoretical, methodological and practical grounds for concern over the utility of orthodox ‘narrow risk assessment’ techniques in the regulatory appraisal of risk. The frequently cited dichotomy between ‘science-based’ and ‘precautionary’ approaches to regulatory appraisal has been shown to be unfounded
Confronting risk with precaution 181
Notes: Participants were asked their opinion concerning diversity among options. The results for each participant are shown as a pie chart displaying the proportional importance of each option in their favoured mix of options. The six basic options 1–6 are colour coded. The basic options are: 1 organic, 2 IPM, 3 conventional, 4 GM with segregation and current labelling, 5 GM with monitoring, 6 GM with voluntary controls. The additional options are: I7 GM with segregation, labelling, monitoring and other restrictions; I8 GM with full assessment of impacts and need; B7 GM with quality; K7 conventional and organic without GM; K8 mix of GM, conventional and organic; J7 GM in regulated organic system, J8 GM in IPM system; F7 GM with segregation, labelling and monitoring. Figure 7.6 Diverse options mixes favoured by individual participants
182 Andy Stirling and Sue Mayer and misleading. Indeed, it is argued that there exist strong scientific reasons for holding the broader scope of precautionary approaches to be more consistent with the scientific foundations of rational choice and probability theory than are conventional ‘narrow risk assessment’ techniques. The imperatives both of science and precaution can therefore be seen to pull in the same direction. The regulatory appraisal of risk should become more systematic and broader in scope. In particular, a set of criteria are developed concerning the need for greater humility, completeness, transparency and participation in regulatory appraisal, with specific attention to the comparison of different options (including mixtures of options), the consideration of benefits and justifications and the systematic ‘mapping’ of the way in which different framing assumptions lead to different pictures of performance. In order to explore how these criteria might be practically operational, the second part of the chapter examined as a case study a pilot exercise applying a novel ‘multi-criteria mapping’ method to the regulatory appraisal of a genetically modified (GM) crop. The results of this exercise are more complete than orthodox risk assessment, in that they embody consideration of an unlimited array of issues. They also include consideration of a wide range of different strategic alternatives to the use of GM technologies. Equal attention is given to the justifications and benefits, as to the impacts and costs associated with the various options. The appraisal is inherently participatory, in that the results fully reflect the perspectives of a wide range of different socio-economic and cultural interests. The relative simplicity and auditability of the weighting and scoring approach retains a high degree of transparency. A central feature is the focus on the systematic mapping of the way in which specific results relate to specific assumptions in appraisal. This, together with the explicit attention given to more diverse mixes of options, embodies a greater degree of humility in the face of ignorance and pluralism than is normally displayed in risk assessment. In this way, it may be concluded that there seems no reason in principle why conventional regulatory appraisal might not be adapted to better address the imperatives both of science and precaution. The present pilot exercise was not prohibitive in scale, being conducted by two researchers working part-time over a period of 18 months. Although necessarily provisional, the consistencies displayed in the results show that it is possible – despite the breadth of the process – to draw some quite robust conclusions over the relative performance of options such as organic farming, conventional intensive agriculture and voluntary control regimes for the use of GM crops. Although the techniques of risk assessment continue to offer essential tools in addressing specific aspects, there seem to be no compelling theoretical, methodological or practical reasons for persisting in basing the entire business of regulatory appraisal on these inherently constrained and limited methods.
Confronting risk with precaution 183
Acknowledgement This chapter is based on an article published in the International Journal of Occupational and Environmental Health (Vol. 6, 2000: 296–311), the material being reproduced by permission.
References Arrow, K. (1963) Social Choice and Individual Values, New Haven: Yale University Press. Bezembinder, T. (1989) ‘Social choice theory and practice’, in C. Vlek and G. Cvetkovitch (eds) Social Decision Methodology for Technological Projects, Dordrecht: Kluwer. EPCAG/Biotechnology and the European Public Concerted Action Group (1997) ‘Europe ambivalent on biotechnology’, Nature, 387: 845–7. Fisher, E. and Harding, R. (eds) (1999) Perspectives on the Precautionary Principle, Sydney: Federation Press. Funtowicz, S. and Ravetz, J. (1990) Uncertainty and Quality in Science for Policy, Amsterdam: Kluwer. Grove-White, R., Macnaghton, P., Mayer, S. and Wynne, B. (1997) ‘Uncertain world. Genetically modified organisms, food and public attitudes in Britain’, Lancaster: Lancaster University, Centre for the Study of Environmental Change. Hayek, F. (1978) New Studies in Philosophy, Politics, Economics and the History of Ideas, Chicago: Chicago University Press. Loasby, B. (1976) Choice, Complexity and Ignorance: An Inquiry into Economic Theory and the Practice of Decision Making, Cambridge: Cambridge University Press. Mayer, S., Hill, J., Grove-White, R. and Wynne, B. (1996) ‘Uncertainty, precaution and decision making: the release of genetically modified organisms into the environment’, briefings no 8, Sussex: University of Sussex, Global Environmental Change Programme. O’Riordan, T. and Cameron, J. (eds) (1994) Interpreting the Precautionary Principle, London: Earthscan. O’Riordan, T. and Jordan, A. (2001) Interpreting the Precautionary Principle, 2nd edn, London: Cameron May. Raffensberger, C. and Tickner, J. (eds) (1999) Protecting Public Health and the Environment: Implementing the Precautionary Principle, Washington: Island Press. Renn, O. and Klinke, A. (1999) ‘Risk evaluation and risk management for institutional and regulatory policy’, field study for ESTO project on ‘Technological risk and the management of uncertainty’, Sevilla: Institute for Prospective Technological Studies. Salo, A. (1999) ‘Technological risk and the management of uncertainty: the role of decision analytic modeling’, study for ESTO project on ‘Technological risk and the management of uncertainty’, Sevilla: Institute for Prospective Technological Studies. Smithson, M. (1989) Ignorance and Uncertainty: Emerging Paradigms, New York: Springer. Stirling, A. (1994) ‘Diversity and ignorance in electricity supply investment: addressing the solution rather than the problem’, Energy Policy, 22(3): 195–215. Stirling, A. (1997) ‘Multi-criteria mapping: mitigating the problems of environmental valuation’, in J. Foster (ed.) Valuing Nature, London: Routledge. Stirling, A. (1999) ‘On science and precaution in the management of technological risk – final report’, Sevilla: Institute for Prospective Technological Studies. Stirling, A. and Mayer, S. (1999) ‘Rethinking risk: a pilot multi-criteria mapping of a genetically modified crop in agricultural systems in the UK’, report no 21, Sussex: University of Sussex, SPRU.
184 Andy Stirling and Sue Mayer Stirling, A. and Mayer, S. (2000) ‘Precautionary approaches to the appraisal of risk: a case study of a GM crop’, International Journal of Occupational and Environmental Health, 6(4): 296–311. Von Schomberg, R. (1998) ‘An appraisal of the working in practice of directive 90/220/ EEC on the deliberate release of genetically modified organisms’, Luxembourg: European Parliament, Scientific and Technological Options Assessment Unit (STOA). Wynne, B. (1992) ‘Uncertainty and environmental learning: reconceiving science and policy in the preventive paradigm’, Global Environmental Change, 2(2): 111–27. Wynne, B. and Mayer, S. (1999) ‘The release of genetically modified organisms to the environment’, in EU 98, Copenhagen: European Environment Agency.
Consumer valuation and citizen deliberation 185
Part III
Deliberation, participation and value expression
186 Jonathan Aldred
Consumer valuation and citizen deliberation 187
8
Consumer valuation and citizen deliberation Towards a comparison Jonathan Aldred
Introduction There is now a relatively well-developed critique of the contingent valuation method (CVM) for adding decisions about environmental problems. Theories of deliberative democracy have been invoked which question the individualistic, preference-based calculus of contingent valuation. A particular deliberative institution which has recently received much attention is the citizens’ jury (CJ). This is taken here to involve around 16 ordinary members of the public, selected to represent a cross-section of the local community. The jury comes together over a period of usually four days to consider an important public policy question. The members of the jury are briefed by and question expert ‘witnesses’, and discuss the issues with each other in small and large groups, chaired by an independent moderator. However, while the original impetus for the adoption of deliberative institutions in environmental policy-making was undoubtedly dissatisfaction with the CVM and the search for an alternative, a perhaps unexpected development is that advocates of deliberative institutions no longer see them as direct alternatives to the CVM. For instance, as the CJ methodology has been elaborated in more detail, researchers have increasingly realized that the CJ ‘valuation exercise’ serves different purposes, answers different questions, and makes different assumptions about the individual participants than the CVM. Although much of interest must be set aside in order to pursue any sort of comparison between the CVM and CJ, it is still possible to explore the relative merits of public discussion about the environment, vis-à-vis private reporting of individual judgements. What does the presence of discussion per se add to the environmental decision-making process? A number of putative advantages of discussion are explored in detail, and in particular the critique of these deliberative virtues, coming from both neoclassical environmental economists and the ‘rational choice’ school of political theorists, is assessed. An aim has been to address not just existing criticisms of deliberative institutions from this perspective, but to anticipate and respond to criticisms which are likely to be made in the same spirit. Rather than repudiate this perspective altogether (although I do not share it), I show that the criticisms do not even succeed on their own terms. That is, I hope to show that the arguments against discussion fail, in terms which the critics of deliberative democracy must acknowledge.
188 Jonathan Aldred The chapter begins by exploring the extent to which the CVM and CJ can be meaningfully compared. It is argued that this is possible to a limited extent, by treating both as explicit decision-making processes, with only CJs involving discussions before decision, and restricting the comparison to an assessment of epistemic advantages and disadvantages of that discussion. The assessment in this chapter is made from the perspective of rational choice theory, because that approach has been influential amongst critics of deliberative democracy. A final section examines the desirability of using CJs to generative explicit decisions, including a brief discussion of methodologies which attempt to combine the CVM and CJ.
Comparing the CVM and CJ Theories of deliberative democracy1 are often proposed as an alternative to aggregative models, where given preferences2 are aggregated in order to determine a collective decision. At the level of institutional design these two approaches will be said to generate deliberative and aggregative institutions respectively (‘voting’ will be used as a synonym for aggregation in what follows). In the context of environmental decision-making, the CVM might be regarded as an aggregative institution, while CJs could be seen as one type of deliberative institution.3 But can the CVM and CJ be meaningfully compared? Certainly they should be regarded as competing alternatives only with caution. The CVM and CJ answer to different institutional needs, cultural roles and social contexts (O’Connor 2000). Moreover they are incompatible in an important sense – they assume competing, arguably mutually exclusive, accounts of rationality. The validity of one approach, then, would seem to cast doubt on the theoretical underpinnings of the other. But this tension does not rule out a comparison. Public bodies charged with environmental policy are increasingly required to justify their policy decisions through public consultation, and see the CVM and CJ as alternative methods (among others) for doing so (Jacobs 1997a). So the motivation for pursuing a comparison is clear. Public bodies do compare the CVM and CJ, and it would be inadequate to answer a decision-maker seeking advice on the appropriate consultation process with the reply ‘it depends on the kind of consultation you seek’. A less question-begging response would be based on an assessment of which consultation process is superior. A central claim of this chapter is that the comparison involved in such an assessment is coherent, and that much of what is essential to the comparison can be captured by a relatively straightforward question: What are the relative epistemic merits of preceding decision-making by discussion, compared with mere private reporting of individual judgements? (Q ) It is not immediately obvious that the comparison specified by this question maps neatly and helpfully on to that between the CVM and CJ. A number of features of the comparison must be clarified.
Consumer valuation and citizen deliberation 189 Decision-making processes? (Q) implies the end point of the process is a decision. There are two separate issues at stake here. Roughly speaking, the first may be regarded as the question of whether the input to the process is open or closed; the second issue is whether the output is open or closed. In both cases, there is no dichotomy but a continuum of varying degrees of openness. The process input is said to be closed if the participants’ have no control over the inputs to the process. These inputs are principally the task the participants are set, and the information they are given at the beginning of the process, or are able to obtain during it. The more open the process input, the more control participants have over their role, and in particular the subject matter of their thinking, during the process. The process output is said to be closed if the outcome of the process is narrow and determinate. The more open the output, the more it admits diverse interpretations, and the less structure it will have. While these notions of openness and closedness of outputs and inputs seem intuitive, they are hard to define clearly in isolation. A few examples should help to clarify them. The following process outputs are (arguably) progressively more open: decision, valuation, recommendation, competing recommendations, process report suggesting consensus but making no recommendations, process report suggesting dissent. Clearly the openness of the output will be influenced by the openness of the input – for instance, if the participants are set a relatively closed task in being asked to reach a decision. This example might suggest that a relatively closed input entails a closed output. However this is mistaken for two reasons. First, in the example, the participants may be asked to reach a decision, but the process input may be very open in other respects, implying a relatively open input overall. Second, the participants may simply ignore the exhortation to reach a decision. The openness of inputs and outputs are in general independent of one another, and only the former concerns the degree of control the participants possess. A very strict requirement to reach a decision is likely to be followed by a closed output; but it is the closed input rather than the closed output which is responsible for the participants’ lack of control; in contrast, a decision may emerge from a process in which the participants were given an extremely open, sketchy task, but chose to narrow this down into a particular decision problem. Another example of confusion between open inputs and outputs is captured in the dubious claim that if a process requires only a report of deliberation rather than a decision, then the participants have more control. Certainly a decision is a less open output, but this does not entail a less open input. In other words, there is no a priori reason why the requirement to produce a report on a specific task gives participants any more control than the requirement to produce a decision. It does, however, demand less agreement among them. While the process input is principally classified according to control, the focus in classifying the process output is agreement – a closed output will demonstrate more agreement among the participants.4 In a later section, this feature will be used to distinguish different types of CJ, but the more immediate concern is the
190 Jonathan Aldred comparison between the CVM and CJ. The CVM methodology presumes and necessitates a closed input in which the set of policy alternatives, among which a single choice is to be made, is exogenously determined. CJ methodology makes no such assumption. With an open input, the set of alternatives is not well defined, perhaps because there is no fact of the matter, or established understanding, regarding the nature of the problem. The public authority may present the CJ with an open input because the authority’s understanding of the problem does not permit closure. Part of the resolution of the problem is in framing the decision to be made – ‘asking the right questions’. While the open input approach may be the richer and more realistic characterization, it remains incompatible with the CVM methodology. If a comparison of the CVM and CJ is to be pursued, a closed input must be assumed. Another difficulty for the comparison between the CVM and CJ is that they typically have different outcomes. A CJ provides a report, while the CVM gives a valuation. Since comparing a report and a valuation directly is impossible, it is helpful to construe the comparison between the CVM and CJ as one between different decision-making procedures. So, in the absence of unanimity, deliberation in the CJ is terminated with a voting or aggregation procedure; it may then be compared with ‘CV-in-CBA’, which also yields a decision, rather than the CVM alone.5 Once both the CVM and CJ are defined to stretch as far as a final decision, a direct comparison is possible. However adopting this approach is unnecessary in order to compare the CVM and CJ indirectly. Referring again to (Q), it is still possible to examine whether, and in what respect, discussion affects participants’ views. But the comparison is clearly much sharper if the effect of discussion can be traced through to a decision. To sum up, the CVM and CJ are regarded here as decision-making processes in two respects: participants in both are set the task of deciding between exogenously specified options (closed input), and both processes reach a decision (closed output). These features are clearly exhibited by some, but not all, CVs and CJs in practice. For the purpose of tractable comparison, they will be assumed in what follows, although I shall return to the possibility of open input and output processes at the end of the chapter. This approach also has the virtue of consistency with the more abstract comparison between deliberative and aggregative institutions. Insofar as they have been compared, both are assumed to terminate with a decision (Elster 1998a). The comparison between deliberative and aggregative institutions is then a rather strange one – between aggregation preceded by discussion, and aggregation alone. Most advocates of deliberative democracy would of course reject the claim that the only difference between deliberative and aggregative institutions is that in the latter, voting is not preceded by discussion. But this approach reflects the ‘thin’ view of deliberation, as discussion, adopted here. Deliberation and discussion While CJs are usually said to involve deliberation rather than discussion, (Q) adopts the latter term. This is because deliberation is here understood simply as discussion
Consumer valuation and citizen deliberation 191 (Fearon 1998).6 There are a number of reasons for this approach. First, the meaning of discussion as a concrete activity is intuitively clear. Second, CJs involve a relatively small number of people meeting to make a decision. Discussion is not obviously applicable to society at large, whereas deliberation is often urged on this scale. Third, one may deliberate with oneself, while the concern here is with public discussion. Finally, and most importantly, deliberation carries much theoretical baggage. For example, many deliberative democracy advocates distinguish the ‘diagnostic’ approach of preference discovery in the CVM from the ‘constructive’ approach of judgement creation in CJ (Sagoff 1998: 221). But this distinction is not essential to the comparison between deliberative and aggregative institutions. Aggregation procedures as defined by social choice theory are formal structures alone, with no requirement that their inputs be preferences rather than judgements. The picture of the agent constructing her judgements afresh in response to the information given to her is compatible with an aggregative institution. Conversely, the possibility of a utility maximizer entering into discussion with other utility maximizers, is compatible with participation in a deliberative institution. Epistemic merits The approach taken here might appear to bias the debate heavily in favour of deliberative methods; the comparison between aggregation preceded by discussion, and aggregation alone, is, to repeat, a strange one. But the debate is far from over, since advocates of deliberative democracy must claim: decisions reached after discussion will be better decisions than those reached without it (B) As it stands, the term ‘better decision’ is so vague as to be almost meaningless. Yet the extent to which many deliberative democracy advocates regard (B) as obviously true suggests that there is a strong intuitive notion of ‘better decision’ in circulation. Despite the vagueness, claims such as (B) are common. The list in Table 8.1 is adapted from Elster (1998b: 11). Table 8.1 Possible advantages of discussion Positive features Epistemic
Processual
Distributional
Encourages an appeal to principles Pools private information Lessens or overcomes the impact of bounded rationality Legitimizes the ultimate choice Is desirable for its own sake Makes for a larger consensus Improves the moral or intellectual qualities of the participants Makes for Pareto-superior decisions Makes for better decisions in terms of distributive justice
192 Jonathan Aldred This chapter defends (B) in regard to the content of the decision rather than the process by which it is reached; it appeals to the epistemic virtues of discussion, rather than any processual or distributional merits of the process. There are three reasons for this focus. First, it is difficult to see how the legitimating powers of discussion, its educative or moral value to the discussants, or its desirability per se, could be defended without some prior commitment to the virtues of deliberative democracy. Thus, in the spirit of avoiding such commitments in order to engage with the critics of deliberative democracy, the focus must be on the epistemic merits of deliberation, seen as practical consequences of the concrete activity of discussion. Second, another sense in which assessment of these epistemic virtues engages with the critics of deliberative democracy is that it shifts the debate on to the critics’ favoured methodological ground. The list above interprets the advantages of discussion in a way that facilitates analyses drawing on rational choice theory and game theory.7 While this list represents one perspective on the advantages of discussion, there are at least three others, which will not be discussed further here: 1
2
3
Recent work by deliberative democrats has focused on the feasibility rather than the legitimacy of the deliberative ideal, examining various mechanisms which might generate true ‘deliberative uptake’ (Bohman 1996, 1998; Gutmann and Thompson 1996; Macedo 1999). There is empirical research in psychology, particularly social psychology, which supports many of the theoretical claims about discussion made here. While rational choice theory appeals to homo economicus to justify agents behaving in ways which generate beneficial features of discussion, social psychology explains these same benefits using an altogether different model of the agent. There is evidence of non-self-interested behaviour increasing as a result of discussion (Orbell et al. 1988, 1990), and violations of homo economicus more generally (Lane 1991; Liebrand et al. 1992). For a synopsis, see Sally (1995). There has been tentative discussion of the framework that social choice theory might provide for exploring how discussion might help consensus be reached (Buchanan 1954a, 1954b; Hylland 1986; Miller 1992; Sen 1995b, 1997; Aldred 2000).
An obvious drawback of viewing deliberation through the lens of rational choice theory is that much of the richness of the deliberative account of decision-making is inevitably lost. Most obviously, the processual advantages appear to be overlooked. However, these benefits tend to be by-products of the epistemic virtues (Elster 1986b, 1995; Fearon 1998; Gambetta 1998). Pursued directly, they can be selfdefeating. For instance a main aim of CJs for many advocates is to aid in rebuilding the trust of ordinary citizens in government and political processes (Coote and Lenaghan 1997). Telling the participants in a CJ that it is a trust-building exercise will not achieve that effect. So a third reason for the focus on epistemic merits is that a policy-maker attempting to secure the processual advantages directly is probably doomed to failure.
Consumer valuation and citizen deliberation 193
Appealing to principles8 At the heart of deliberative democracy are two claims: an empirical one, that public discussion will encourage a particular mode of justifying arguments (Goodin 1986), and a theoretical claim, that this justificatory requirement will have a beneficial effect on the quality of the final decisions of a deliberative institution. The justificatory requirement is that arguments invoked during discussion cannot be narrowly concerned with self-interest, but must be acceptable to all. In short, arguments must appeal to widely shared principles. There are two broad reasons why we might expect this justificatory requirement to be satisfied in practice (Elster 1995; Fearon 1998). Strategic reasons A speaker (S) might advance justifications which appeal to the public interest for what have been termed ‘strategic’ reasons (Elster 1995). The basic idea is straightforward: if S invites others to choose proposal P merely because P serves S’s private interest, few will be persuaded. Johnson (1998) is characteristic of the deliberative democracy critics in seeing the presence of self-interested motives behind apparently impartial arguments as crucially undermining the normative weight, which deliberative democracy advocates attach to the argumentative content of deliberation. The epistemic justification of deliberative democracy, which appeals to ‘the force of the better argument’, is subverted by the self-interested motives of participants to the deliberative process. The objection runs that listeners are swayed by arguments which have no objective merit but have been proposed simply because they serve the speaker’s interests. Johnson concedes that ‘psychological pressures’ generated by the process of discussion might over time make speakers sincerely come to embrace a principle which they earlier invoked for self-interested reasons alone – one example of Elster’s ‘civilizing force of hypocrisy’ (Elster 1995). But this possibility does not nullify the objection – Johnson (1998: 172) argues that such a mechanism which ‘first induce[s] actors to engage in self-censorship and eventually cause[s] them to embrace the resultant views … does not so much generate a “reasoned agreement” as induce a conformity that is at once rather shallow and normatively suspect’. The reasons for suspicion are not stated, although in a note Johnson (1998: 172 n. 47) worries about the autonomy of actors subject to such mechanisms and quotes Elster (1986a: 113) approvingly: ‘Dissonance reduction does not tend to generate autonomous preferences’. He summarizes his critique by implying that deliberative democracy advocates must demonstrate both ‘(a) that impartial claims are, even tacitly, quite widespread in public debate and (b) such claims are animated by mechanisms other than … strategic calculation and attendant psychological pressures’ (Johnson 1998: 172). It is argued here that the case for deliberative institutions such as CJs need not satisfy (b) and that there are reasons to expect (a) will be satisfied, at least in the ideal case. Regarding (b), a claim can be impartial, in the sense of appealing to widely shared principles or interests, but still be ‘animated by strategic calculation’.
194 Jonathan Aldred Arguments that appeal to the public interest, to widely shared principles, have normative force not because they purportedly reveal the noble character of the speaker, but because they are much more likely – in the ideal case, the only way – to secure the consent of the listener. The motives of the speaker are not in general relevant to winning this consent. For example, in a debating contest S may be acting as a ‘devil’s advocate’, but this does not in itself undermine S’s arguments, even if the listeners know of S’s rejection of S’s arguments. Of course motives matter: if the listener (L) is aware that S’s self-interest is best served by the action urged by S, then deliberative democracy recommends (and expects) that L will examine S’s claim that the public interest is served by the action all the more carefully. L will also be sceptical of emotional appeals to the public interest, or uncorroborated factual claims, in this context. But L should in general be able to establish, independently of S’s motives, whether the action urged by S serves the public interest – and in particular, whether it does so for the reason(s) advanced in S’s argument. Johnson also misrepresents Elster’s ‘civilizing force of hypocrisy’. While advocates of deliberative democracy are not always clear, the essential point is that agents, through discussion, become aware of new ideas and the positions of others. In particular, as S advances arguments in which S has no conviction in order to persuade others, S may come to realize their merits over opposing arguments S privately held at first – a situation I shall term devil’s advocacy conversion. It occurs directly through the recapitulation of arguments to others in order to persuade them: finding new perspectives which enjoy common support, realizing how the position can be strengthened further, or simply coming to understand it more fully. Returning to Johnson’s claim (a): as already suggested, whenever S seeks to win consent, whether for self-interested reasons or otherwise, S will seek to advance impartial arguments – arguments showing how the public interest is served – because by showing how diverse private interests are simultaneously served, an argument will command widespread support. This motivation for appeals to the public interest would not seem to apply to majority groups: a subset of deliberators, if they form a majority of the whole group, will be free to make blatant appeal to their collective self-interest, that is the collective interest shared by the members of the subset but not shared by all. Yet there is anecdotal evidence that even majorities appeal to a broader public interest than their own constituents. This suggests there must be other reasons besides strategic ones for justifications in terms of the public interest. Avoiding appearing selfish A second reason why S might advance justifications which appeal to the public interest is that S will not wish to appear selfish to other members of the deliberative institution, whatever S’s true motives are. A related reason why a speaker or group, particularly a majority group, might not wish to appear selfish is reciprocity. Although the majority group may be able to command majority support in the current argument by appealing to (its) self-interest, in a later argument (which
Consumer valuation and citizen deliberation 195 refers to the same final decision) the members of the majority group may be on opposing sides. Thus the members of the majority group may wish to adhere to strictly public-interest justifications now in the hope of ensuring only public-interest justifications from other majority groups in the future. The desire not to appear selfish is arguably less significant than the strategic motive for public-interested justifications in discussion, for two reasons. First, the desire not to appear selfish is parasitic upon the existence of some widely accepted account of ‘unselfish justifications’ – it depends on the possibility, or at least the belief in, genuinely impartial or public-interested arguments after all. Second, it may have less relevance in problematic cases of environmental decision-making. For the embarrassment or shame a person feels from appearing ‘selfish’ is shorthand for shame from appearing to ignore a moral norm. However in the environmental case there will often be clashing norms, value conflicts and moral dilemmas where no matter which outcome is chosen, it will be wrong in the sense of a norm being broken (Nagel 1979; Williams 1979; Gowans 1987). More generally, in such cases, the judgements ‘public-interested’ and ‘self-interested’ are at the least ambiguous, and often wholly inappropriate.
Pooling private information The information might take the form of, or refer to, facts, preferences, or beliefs. It is often noted that discussion permits participants to express intensities of preference as well as mere orderings. In response, it may be argued that the CVM is also able to capture intensities of preference, subject as always to the caveat that preferences (and judgements) can be adequately represented through the private monetary format. However, while the CVM can capture preference intensities, it is surely unable to capture the subtle nuances of natural language, or even all the nuances of ‘natural language expressions of preference’. This may appear a misleading comparison, because, except when the participants are unanimous, a deliberative institution will ultimately be limited to what can be expressed through a vote. But the point is that nuanced information – both ‘preference’ and ‘factual’ – can be shared with others before a vote is conducted. This has obvious benefits, since I may have good reasons to change my preference once aware of your information, such as altruism, reciprocity, or trust in your judgement. It is not possible with the CVM. For critics of deliberative democracy, the chief difficulty with these arguments is that they overlook the strategic incentives an agent may have to lie or misrepresent information in an attempt to sway the preferences of others (the so-called ‘cheap talk’ models, e.g. Farrell and Gibbons 1989; Austen-Smith 1990a, 1990b; Krehbiel 1991). Of course most voting procedures, including consumer valuation, are susceptible to strategic manipulation by voters exaggerating their preferences. But this is irrelevant here, since a deliberative institution will typically be terminated by voting. The key question is: ‘Given preference diversity and thus incentives to misrepresent, is a group better off simply voting or discussing things before voting?’ (Fearon 1998: 47).
196 Jonathan Aldred Much of the ‘cheap talk’ literature offers a very nihilistic answer, claiming that because ‘talk is cheap’ – reported private information is unverifiable and/or lies cannot be punished – no credible communication can occur (Riker 1982; AustenSmith and Riker 1987; Austen-Smith 1990a, 1990b). The idea is that, because L knows that ‘talk is cheap’, L will ignore S’s communications, regardless of whether S in fact has an incentive to lie. (More precisely, L will not engage in the costly business of establishing S’s incentives or verifying communications.) There are a large number of problems with the game-theoretic cheap talk models used to justify the nihilistic conclusion (Mackie 1998 offers a good critique), but three main ones are that the models are too static, the results are dramatically weakened once repeated interaction is assumed, and equally weakened once multiple speakers and listeners are modelled. The models are too static because they ignore the agents’ responses to the credibility problems. If S knows that L will ignore S’s private information then S will use different information, information that is public but overlooked or private but verifiable by an independent third party, because the original information has no persuasive power. Moreover L has an incentive to help S in S’s attempt to reveal information credibly. Repeated interaction improves the credibility of communication for obvious reasons: ‘When Aristotle was once asked, what a man could gain by uttering falsehoods; he replied, “not to be credited when he shall tell the truth”’ (Johnson 1963, quoted in Mackie 1998: 85). With multiple speakers and listeners, the formal results are complex, but confirm intuitions such as multiple speakers with conflicting interests having an incentive to correct each other. Due to the focus in the cheap-talk literature on the difficulty of credibly communicating any private information, and the idea that L will ignore all communications, the extent to which L might filter the transmission of misleading information is overlooked. The distinction between communication which is not heeded, and misleading communication which is acted upon, is an important one. Only in the latter case can there be a putative argument against discussion. In the case of cheap talk and other attempted strategic manipulation, how might deliberative institutions prevent misleading communication? Put simply, they should be constructed to allow participants to heed Aristotle’s advice, by paying particular attention to the credibility of a speaker when evaluating the speaker’s claims: That credibility has both epistemological dimensions – the speaker must have good sense and be reliable in the formation of judgements – and ethical dimensions – the speaker must have the moral character that allows us to trust their utterance and there must be grounds for believing that they are not inclined to impart falsehoods to their audience. (O’Neill 1997) The Aristotelian account encourages listeners to recognize their inability to evaluate certain claims in the absence of specialist training, and concentrate on discerning on whose authority they can rely for such an evaluation. The emphasis is on assessing the character of the speaker, rather than the speaker’s claims (O’Neill 1993: Ch. 8).
Consumer valuation and citizen deliberation 197 Little more can be said in general, but in terms of institutional design, the following features of a CJ may be helpful: a neutral facilitator selects witnesses, factual information is conveyed through witnesses, witnesses are able to criticize each other’s statements, more time is usually allocated for jurors to question witnesses than for their statements, and jurors are able to recall witnesses at a later point in the proceedings if necessary. In a CJ that I facilitated (Aldred and Jacobs 2000), jurors were able to pick up on inconsistencies in witness presentations. The jurors were choosing between various land management options for wetland areas in East Anglia, UK. In response to a question about its cost, one advocate of a particular option claimed that the option would be ‘largely self-managed’ (reducing labour costs). But later the same witness maintained that the option would ‘create significant numbers of jobs’. The advocate had probably not actively tried to mislead the jury, but the jury were able to force the advocate to be more specific about the advantages proposed. In sum, although the evidence so far is limited and there are obvious problems, it seems that CJs do at least have the potential to allow listeners to assess the credibility of speakers. As well as overlooking many of the cognitive difficulties associated with elaborate schemes for strategic manipulation, critics of deliberative democracy, like the rational choice theories on which they draw, also tend to assume that more information is always better. But without some constraint on the revelation of private information, perhaps through discussing rules of relevance or time limits, ‘information overload’ will be a danger for many decision problems (Fearon 1998: 65 n. 9). Although an aggregative institution might face overload problems because the voting instrument provides the voters with too much information about the alternatives, a deliberative institution, with (at least in principle) open-ended opportunities for private information revelation, appears more susceptible to these difficulties. For a supporter of deliberative democracy, the obvious response to this challenge, besides acknowledging the necessity of rules of relevance and time limits, is to demonstrate how discussion may help participants to overcome the broad class of ‘bounded rationality’ problems, including information overload.
Bounded rationality and new perspectives A central feature of ‘bounded rationality’ is that when making decisions our reasoning and calculating capacities are both limited and prone to error (Simon 1978). Even if a decision was in the hands of a single individual with dictatorial powers and all relevant information, the decision-maker might begin by discussing the alternatives with another, in order to clarify their nature and perhaps suggest new solutions altogether. This suggests two benefits of discussion, namely the sharing of arguments, ideas, and understandings which were previously familiar to only a subset of participants, and the generation of new ideas, etc. which were previously familiar to none of the participants (Fearon 1998: 49–50). So for complex decisions, will many heads be better than one? The subject matter of the decision is crucial. For decisions where the relevant arguments can be readily evaluated in a way that would command the assent of all,
198 Jonathan Aldred then in general it seems likely that many heads will be better. Examples include a player – or team – aiming to win at chess (Fearon 1998: 50), or an individual talking to a friend when trying to decide upon the purchase of some complex financial product such as a pension or insurance. A chess champion or financial expert would be best, but much can be gained through discussion per se. In these cases, certain arguments can be shown to be objectively superior to others, and discussants will recognize and accept their arguments being overthrown by stronger ones. Excepting pathological cases, there is little possibility of social pressure forcing discussants to abandon stronger arguments in favour of weaker ones. However, this case clearly does not apply to most environmental decisions. Rather than cleanly resolving the complexities of the decision in the manner of the chess example, discussion may become blocked by intractable disagreements arising from value conflicts. As described so far, the sharing of arguments and ideas appears to be simply another form of private information pooling. But this interpretation is mistaken, due to the treatment of uncertainty in the Bayesian framework, which underpins rational choice theory. For the Bayesian, there is no such thing as ‘pure’ uncertainty; rather, all uncertainty is ‘risk’ where agents always know the probabilities (subjective or otherwise) of all possible alternatives. Note that as a prerequisite on this account, agents must know the complete list of possible ‘answers’ to some question of uncertainty. This may be adequate with respect to the weather (tomorrow it could be fine, wet, windy or snowy, etc.) and at least coherent with respect to some forms of scientific uncertainty. But it is incoherent with respect to the arguments, interpretations and perspectives of another. Arguments are either known or unimagined, rather than on a list of possibly existing arguments. Thus if L hears a new (unimagined) argument from S it may have a profound and creative effect, whereas if S reveals private information, then according to rational choice theory, it will already have been considered as a possibility by L. So far in this section S and L have been assumed to have common interests. How can I trust the arguments of another with divergent interests? The considerations mirror those of the private information case, although particularly here, a speaker’s communication may not have the intended effect. S may intend to mislead with a bogus argument, but simply because new arguments are generated by creative leaps of the imagination, S’s communication may trigger as a by-product a new idea or argument which serves L’s case. Or S may not intend to mislead but nevertheless present a partial argument, because bounded rationality makes S simply unaware of the argument’s limitations. This latter case is common with environmental problems where the participants bring a number of differing perspectives and interpretations to the deliberative institution. Indeed the idea that CJ participants are exposed to a much wider range of points of view than they would be through responding privately to a CVM survey has been a key claim of CJ advocates (Jacobs 1997b). In terms of the framework here, this case is one of divergent interests without intent to deceive. The arguments and interpretations offered by any speaker will inevitably be partial and incomplete because they reflect that speaker’s particular perspective and experiences. However the partiality is neither strategic, nor necessarily conscious, and notwithstanding the presence of
Consumer valuation and citizen deliberation 199 divergent interests, a speaker will be willing to modify and amend her perspective if she seeks to understand, communicate with, or persuade others. Thus the distinction between the sharing of arguments and the pooling of private information applies a fortiori when there are divergent interests. When L hears S claim to report S’s private information, L can always use S’s speech to update L’s probability estimates regarding S’s true private information, because L can conceive of what it might be. In contrast, it does not matter whether S has actually entertained, let alone expressed, an argument or idea once L has conceived of it, and if L has not conceived of it, then L cannot attach a probability to its existence. Moreover an argument by S can often have indirect creative benefit for L regardless of (whether L knows) S’s interests. In sum, the sharing of arguments is more than simply the pooling of private information when new perspectives are introduced. Bounded rationality considerations are important for discussion in another way: they suggest that the standard for a better argument should not be overstated. An argument is better just in case it can be recognized as superior to existing arguments by all the deliberative institution participants, that is impartially superior. It need not be flawless or objectively compelling; rather, in the presence of bounded rationality the introduction of a superior but flawed argument by one participant should be seen as beneficial not because it is objectively compelling but because it demands a revised response from those who disagree. This point hints at one possible explanation of why critics of deliberative democracy are unwilling to accept that the quality of collective decision-making can be improved as a result of arguments advanced for purely strategic reasons. For such critics, it may be that there can be no impartially superior arguments, or at any rate it is not possible for listeners to identify them when the arguments emanate from speakers who may be partial. A related but distinct explanation, at least for those critics influenced by rational choice theory, appears to be that the possibility of new ‘altruistic’ arguments is widely denied. Either an argument made by S to L benefits only S, or if it benefits both S and L, or L alone, L would already be aware of it. The critics of deliberative democracy tend to overlook the possibility of bounded rationality preventing L from being aware of information or arguments which serve L’s case. In summary, the last three sections have examined some key claims regarding the epistemic virtues of discussion in the light of criticisms from a rational choice theory perspective. At the level of this abstract comparison, it has been argued that the claimed epistemic merits withstand this critique. One possible virtue of the approach here is that it seeks to persuade neoclassical environmental economists of the merits of deliberative democracy in terms which they must acknowledge. But a drawback has been the necessity of assuming the CJ as well as the CVM yields a closed output. While the assumption of a closed output – roughly, the requirement that the process yield a ‘decision’ – has been useful for comparing the CVM and CJ, it is important to assess the merits of imposing this requirement on the process. This is the subject of the next section, and it turns out to be indirectly relevant to the CVM/CJ comparison after all, because it permits evaluation of some methods which seek to combine both the CVM and CJ.
200 Jonathan Aldred
Agreement and deliberative valuation For many advocates of CJs and other deliberative institutions, the acid test of the ‘success’ of some deliberative institution is whether the preferences of participants changed during the deliberations – crudely, whether the deliberation ‘made a difference’ (see Chapter 12). Probably because the notions of ex ante and ex post preference in this context are controversial and little understood, this test appears to have received no explicit discussion in the deliberative democracy literature, and there is no consensus on how it should be operationalized. However, without ex ante unanimity, CJ participants will not reach agreement on the decision facing them unless at least some participants’ preferences have shifted. Thus the fact of agreement usually implies a change in preferences. While agreement may be a sufficient condition for preference change, it is clearly not necessary. Too much of the abstract deliberative democracy literature has assumed that it is – that genuine deliberation, involving preference change, guarantees agreement on the decision faced (Bohman 1998). No assumption that discussion per se will generate agreement has been made here; the epistemic merits of discussion discussed above all tend to encourage, but not ensure, agreement. Rather, the assumption of a closed output has been made on the basis that discussion will usually be followed by a vote. But should the output of a CJ be closed by the imposition of some aggregation mechanism? In the rest of this chapter, I shall briefly consider two related questions: 1
2
Should some procedure be imposed on the CJ which ensures a closed output, or should the participants be left to reach as much or as little agreement as they feel able to? If a closed output is deemed desirable, what form should it take? In particular, should it be a qualitative or quantitative output?
Beginning with the first question, the problem is not the emergence of a closed output per se, but the extent to which this is forced. The requirement that a closed output be reached, is a form of closed input which reduces the control the participants have over the process. In contrast, theories of deliberative democracy generally encourage as much openness as possible. Deliberative democracy advocates argue that, by demanding a decision on a specific ‘charge’ or question, the body which commissions a CJ may exclude important issues, information and perspectives. At best, the CJ may be biased, while if the exclusion is intentional then the CJ is manipulated. At the other extreme: Experiments with complete juror control of the process have found that participants, in the initial stages, do not have enough of an overview on a subject to deal competently with setting the charge, agenda organisation or witness selection. (Smith and Wales 1998: 12)
Consumer valuation and citizen deliberation 201 Thus there is a trade-off between the openness of inputs and deliberative competence. How this trade-off should be resolved depends on the role attributed to the CJ. One view holds that the role of a CJ is simply to ‘read off ’ the given opinions or tastes of the jurors. On this view, it may be plausible to separate entirely the framing of the question from opinions about its answer. In contrast, supporters of CJs usually interpret their role as centrally involving matters of ethical concern, where the neoclassical economists’ mantra de gustibus non est disputandum is irrelevant. Jurors must debate their judgements of ‘right’ and ‘wrong’ rather than simply express their tastes of ‘like’ and ‘dislike’, and a central part of this debate is to explore the issues at stake. It seem indefensible that this exploration be subject to an exogenous constraint. Part of the process of reaching the ‘right answer’ in environmental problems is framing the question correctly, just as it is for, say, abortion or capital punishment issues. For such problems, it seems that any defence of imposing a specific question on a CJ must rely on roughly the claim that ‘the commissioning body knows best’. Whether this claim is based on the view that the commissioning body is more experienced at making complex problem-framing decisions, or better informed, or more public-interested than the jury, it faces many difficulties. An obvious one is that a commitment by the commissioning body to a closed input reflects a lack of trust in participants. This may trigger a corresponding mistrust among participants – in particular a suspicion that the commissioning body will ‘cherrypick’ recommendations from the report of the CJ as it suits them (Aldred 1998; Aldred and Jacobs 2000). It is difficult to overemphasize the need for jurors to maintain trust in the process if a CJ is to be successful (Armour 1996; Crosby 1996, 1999; Coote and Lenaghan 1997; Smith and Wales 1999; Ward 1999). A less obvious difficulty is an apparent tension between the claim that ‘the commissioning body knows best’ and rational choice theory. For on the broader – and arguably more consistent – account of rational choice theory which embraces public choice theory, it is not just ordinary market agents who act as if homo economicus, but politicians and bureaucrats too. Politicians are neither better informed, nor more public-interested, than anyone else. Within this tradition, it is very difficult to defend any claim that ‘the commissioning body knows best’, yet it is from within this tradition that environmental economists emerge who most strongly argue in favour of requiring closed outputs to the CVM and CJ. In summary, reconciling the demand for a closed output with a view of CJs as calling forth ethical judgements of right and wrong, proves difficult. Suppose however that such a reconciliation can be achieved, or that this view of CJs is rejected, and that in answer to the first question, a closed output is deemed desirable. Moving to the second question, should this output be quantitative or qualitative? A common early view of the role of CJs was that they could offer a useful space for discussion before proceeding to a standard CVM exercise: the CJ would finish with the now well-informed participants each privately filling in a CVM questionnaire, a close variant of the ‘focus group CV’ (Brown et al. 1995; Aldred 1996; Jacobs 1997b; Brouwer et al. 1999). Both Ward (1999) and James and Blamey (2001) are more radical in seeking an average valuation from the jury not as a
202 Jonathan Aldred statistical average of individual CVM responses, but as the direct response to asking the jurors collectively for their valuation. They both hope to avoid some of the problems of imposing forced agreement: they hope that an average valuation will emerge naturally as a kind of quantitative consensus from group discussion, rather than being enforced by means of a vote. Nevertheless, they both limit jurors’ control in the sense discussed above; they require a closed output, to be achieved by some kind of voting procedure, if consensus does not emerge. This new approach, termed ‘deliberative monetary valuation’ (DMV) (see the introduction to this volume), seeks to combine the advantages of discussion with the merits of a valuation output. Since DMV aims to be more readily comparable with the CVM than conventional CJs, with their qualitative output, it will undoubtedly attract much attention. In practice, probably the main advantage of the DMV approach is the appeal of quantitative measures in contemporary policy discourse in market economies; they have overtones of scientific rigour and accuracy. However, at a theoretical level, the DMV approach appears to have two main virtues over a conventional CJ. They mimic two of the arguments made for favouring the CVM over CJ. First, DMV captures preference intensities. If, as seems likely, the deliberation in both CJ and DMV processes does not generate unanimity, then some kind of aggregation mechanism must be employed to generate a decision. If the aggregation procedure used in DMV at this point measures preference intensities, then it is argued to be superior because it captures more information about individual preferences. Second, this additional information enables DMV to avoid standard Arrovian results in social choice theory under which there exists no mechanism to aggregate preferences which also satisfies certain weak and plausible properties (Sen 1970). The threat to post-deliberation preference aggregation posed by Arrovian impossibility is arguably a weak one – indeed, as noted above, the language and concepts of social choice theory may help make clear the extent to which deliberation can generate consensus (Miller 1992; Aldred 2000). But the argument that individual valuations provide extra information seems more robust, and can be generalized to the claim that DMV does more than simply decide amongst a given set of policy options, because it provides a willingness to pay which is applicable in broader resource allocation contexts. For instance, the willingness to pay value could help a public authority make budgetary decisions beyond the options facing the jury – say between health care and environmental preservation expenditure. The fundamental difficulty with this argument, and the DMV method, can be easily stated: ‘is the collective valuation question meaningful?’ Ward (1999: 78) writes: ‘The question posed should be “what is a particular level of provision of environmental quality worth to the average citizen?”’ (1). Later he adds (p.79): ‘Jurors are not asked to express their personal evaluations but their judgements about what environmental quality is worth to society as a whole’ (2). In contrast, James and Blamey’s (2000, emphasis added) stylized valuation question takes the form: ‘What is the maximum amount the citizens of City X would be prepared to pay to ensure the proposed harbour development proceeds, if they knew what you now know about it?’ (3). In the actual DMV exercise they conducted, the format was
Consumer valuation and citizen deliberation 203 different: ‘How high would a park management levy have to be, before the jury would recommend Option 1 rather than Option 4 below? In other words, how high would the levy have to be before the New South Wales public would be no better off under Option 4 than Option 1?’ (5). It seems possible that these five different question formats may elicit five substantively different kinds of response. The variety of question formats may reflect an underlying ambiguity in the variable they seek to measure. Although both Ward and James and Blamey suggest they seek to capture ‘citizen’ rather than ‘consumer’ responses (Sagoff 1988), it is not obvious that all the formats would achieve this, particularly formats (1) and (3). Formats (1) and (3) might both be interpreted by jurors as asking them to make some kind of prediction regarding what others (non-jurors) would be willing to pay. Format (5) also appears to ask for a prediction, but this time relative to welfare rather than willingness to pay. In contrast, formats (2) and (4) seem to ask the jurors for their valuations, although particularly in (2) it remains unclear how they are to interpret the question, and, correspondingly, the meaning to be attached to an answer. One conclusion to draw is simply that questions in DMV exercises need very careful formulation, another is that DMV exercises reflect similar problems of meaning as do the CVM exercises, if not more so. There is no space to go into this very large issue,9 but the kinds of questions which both the CVM and DMV exercises raise are easily stated. In the DMV context, do the jurors assume all members of the affected community have to pay? And an equal amount? Or just those who express a hypothetical willingness to do so? Or are the jurors being asked how much they would voluntarily contribute to a joint effort to secure the specified outcome? It is difficult to see how the DMV format, or any decision procedure which seeks a valuation, will ever be as readily comprehensible to participants as the underlying ‘what is best for society?’ in a CJ.
Conclusions This chapter has discussed how a comparison between the CVM and CJ might proceed, focusing on the epistemic merits of discussion. It was argued that the rational choice theory critique of these merits largely fails on its own terms. For the CJ/CVM comparison, the focus in this chapter on the feasibility of competent deliberation seems more informative than the early emphasis in the deliberative democracy literature on an ideal of deliberation. However this focus is subject to a number of caveats. The social facts which constrain and determine the deliberative setting are not fixed but contingent. There is already some evidence of a dynamic feedback effect from deliberative institutions to the social conditions, which determine their feasibility. The evidence is generally positive – increasing the presence of deliberative institutions may improve the confidence, capacity and willingness of ordinary citizens to make decisions about public policy questions (Smith and Wales 1998, 1999). Thus it is important to set the detailed discussion of feasibility constraints in this chapter in context: they will be subject to – and may be nullified by – the deliberative institutions, which they presently constrain.
204 Jonathan Aldred One important feasibility constraint which has not been mentioned here (it receives little attention from the rational choice theory perspective), but which deserves detailed treatment, is inequalities of voice in any community – what might be termed the problem of ‘willingness to say’ (O’Neill 1997). It is arguable that income and wealth inequalities may generate the same unequal distribution of power in influencing the decision, whether they work through constraining willingness to pay in CVM or constraining willingness to say in CJs. More worrying still: the very requirement that discussion be reasoned and serious to qualify as deliberation may be enough to disenfranchise many members of society (Sanders 1997). The question (Q) invokes a comparison both narrower and more general than that between the CVM and CJ: it captures only one element of any overall comparative assessment (although it is not clear in what other respects the methodologies can be helpfully compared), but it applies to the more general comparison of deliberative versus aggregative institutions. It will by now be apparent that the advantages of discussion canvassed here – and the objections raised – are highly interdependent. Together they generate a complex web of forces (see in particular Bohman 1996: 57–66), which even on the account of rationality advanced by rational choice theory, create some significant benefits of discussion. With these caveats firmly in mind, the discussion here has a number of policy implications. Very broadly, where bounded rationality is likely to be severe, publicinterested principles are called forth, or important private information is present, the epistemic merits of discussion are likely to be more significant. CJ appears more attractive in these contexts. This conclusion aims to stimulate further research rather than general judgements about the superiority of one method in all cases. In a particular policy context, there is no substitute for a careful, empirical, caseby-case comparison. While both the CVM and CJ have drawbacks, the final section suggests that appeal to a mixed method such as DMV may not be the best way to overcome them.
Acknowledgements The author would like to thank Alan Holland, Michael Jacobs, Rosemary James, Jürgen Meyerhoff, Simon Niemeyer, John O’Neill, Graham Smith, Andrew Stirling and Hugh Ward for useful discussions on these issues.
Notes 1 See Bohman (1998) for a survey. The leading early statements of deliberative democracy are all in Bohman and Rehg (1997), which also contains important papers. Another key collection is Elster (1998a). 2 I shall ignore the distinction between judgements and preferences in what follows. Much of the literature to be discussed uses ‘preference’ in contexts where ‘judgement’ seems more appropriate – where the content of the ‘preference’ can be assessed for reasonableness. I shall follow the usage of this literature while acknowledging its shortcomings. 3 There is substantial work discussing CJs in an environmental context (Armour 1996; Crosby 1996; Coote and Lenaghan 1997; Jacobs 1997b; Kuper 1997; Aldred 1998; Crosby 1999; Rippe
Consumer valuation and citizen deliberation 205
4
5
6 7
8 9
and Schaber 1999; Smith and Wales 1999; Ward 1999; Aldred and Jacobs 2000), in addition to the contributions elsewhere in this volume. Of course absence of agreement does not ensure an open output. There may be no agreement among the CVM participants, yet an apparently closed output – at least if the CVM is regarded as having only one output, an aggregate valuation, and data on the distribution of individual valuations or protest bids is ignored. The distinction (Jacobs 1997b) between an institution with decision-making power, and one with power only to make a recommendation to a decision-making authority, is ignored in this chapter. A recommendation is termed a decision regardless of whether those making the recommendation have power to enforce it. (Of course the possession or absence of such power will typically affect the recommendation.) Although the term deliberative institution will be retained instead of discursive institution. In this chapter, rational choice theory is understood in a strong sense as comprising both the formal axioms which make up decision theory (Anand 1993 offers a detailed critical treatment), and a substantive assumption of self-interest – in short, the agent is homo economicus. See Pettit (1998). The arguments in the next three sections are developed in more detail in Aldred (2001). There is a large literature on what the CVM question means in theory (as opposed to methodological difficulties in the practical application of the method), most of it written by non-economists. For three different, and powerful, critiques by economists which cast doubt on the coherence of the CVM question, see Sen (1995a), Sugden (1999) and Vatn and Bromley (1994).
References Aldred, J. (1996) ‘Value conflicts in environmental decision-making’, PhD thesis, Faculty of Economics and Politics, Cambridge: University of Cambridge. Aldred, J. (1998) ‘Land use in the fens: lessons from the Ely citizens’ jury’, Ecos, 19(2): 31–7. Aldred, J. (2000) ‘Deliberation and social choice theory’, paper presented at ISUS Conference, Winston-Salem/North Carolina, March 2000. Aldred, J. (2001) ‘Citizens’ juries: discussion and rationality’, Risk, Decision and Policy, 6(2): 71–90. Aldred, J. and Jacobs, M. (2000) ‘Citizens and wetlands: evaluating the Ely citizens’ jury’, Ecological Economics, 34(2): 217–32. Anand, P. (1993) Foundations of Rational Choice under Risk, Oxford: Clarendon. Armour, A. (1996) ‘The citizens’ jury model of public participation’, in O. Renn, T. Webler and P. Wiedemann (eds) Fairness and Competence in Citizen Participation: Evaluating Models for Environmental Discourse, Dordrecht: Kluwer. Austen-Smith, D. (1990a) ‘Credible debate equilibria’, Social Choice and Welfare, 7: 75–93. Austen-Smith, D. (1990b) ‘Information transmission in debate’, American Journal of Political Science, 34(1): 124–52. Austen-Smith, D. and Riker, W. (1987) ‘Asymmetic information and the coherence of legislation’, American Political Science Review, 81 (3): 897–918. Bohman, J. (1996) Public Deliberation, Cambridge, MA: MIT. Bohman, J. (1998) ‘The coming of age of deliberative democracy’, Journal of Political Philosophy, 6 (4): 400–25. Bohman, J. and Rehg, W. (eds) (1997) Deliberative Democracy, Cambridge, MA: MIT. Brouwer, R., Powe, N., Turner, R.K., Bateman, I.J. and Langford, I.H. (1999) ‘Public attitudes to contingent valuation and public consultation’, Environmental Values, 8(2): 325–47.
206 Jonathan Aldred Brown, T., Peterson, G. and Tonn, B.E. (1995) ‘The values jury to aid natural resource decisions’, Land Economics, 71(2): 250–60. Buchanan, J. (1954a) ‘Individual choice in voting and the market’, Journal of Political Economy, 62: 334–43. Buchanan, J. (1954b) ‘Social choice, democracy and free markets’, Journal of Political Economy, 62: 114–23. Coote, A. and Lenaghan, J. (1997) Citizens’ Juries, London: Institute of Public Policy Research. Crosby, N. (1996) ‘Citizens’ juries: one solution for difficult environmental questions’, in O. Renn, T. Webler and P. Wiedemann (eds) Fairness and Competence in Citizen Participation: Evaluating Models for Environmental Discourse, Dordrecht: Kluwer. Crosby, N. (1999) ‘The citizens jury process and environmental decisions’, in K. Sexton and T. Burkhardt (eds) Better Environmental Decisions: Strategies for Governments, Businesses and Communities, Washington, DC: Island Press. Elster, J. (1986a) ‘The market and the forum: three varieties of political theory’, in J. Elster and A. Hylland (eds) Foundations of Social Choice Theory, Cambridge: Cambridge University Press. Elster, J. (1986b) The Multiple Self, Cambridge: Cambridge University Press. Elster, J. (1995) ‘Strategic uses of argument’, in K. Arrow (ed.) Barriers to Conflict Resolution, New York: Norton. Elster, J. (ed.) (1998a) Deliberative Democracy, Cambridge: Cambridge University Press. Elster, J. (1998b) ‘Introduction’, in J. Elster (ed.) Deliberative Democracy, Cambridge: Cambridge University Press. Farrell, J. and Gibbons, R. (1989) ‘Cheap talk with two audiences’, American Economic Review 79(3): 1214–23. Fearon, J. (1998) ‘Deliberation as discussion’, in J. Elster (ed.) Deliberative Democracy, Cambridge: Cambridge University Press. Gambetta, D. (1998) ‘Claro! An essay on discursive machismo’, in J. Elster (ed.) Deliberative Democracy, Cambridge: Cambridge University Press. Goodin, R. (1986) ‘Laundering preferences’, in J. Elster and A. Hylland (eds) Foundations of Social Choice Theory, Cambridge: Cambridge University Press. Gowans, C. (1987) Moral Dilemmas, Oxford: Oxford University Press. Gutmann, A. and Thompson, D. (1996) Democracy and Disagreement, Cambridge, MA: Harvard University Press. Hylland, A. (1986) ‘The purpose and significance of social choice theory’, in J. Elster and A. Hylland (eds) Foundations of Social Choice Theory, Cambridge: Cambridge University Press. Jacobs, M. (1997a) ‘Alternatives to contingent valuation’, paper presented at CSEC/Green Alliance Conference, London, 23 January 1997. Jacobs, M. (1997b) ‘Environmental valuation, deliberative democracy and public decisionmaking institutions’, in J. Foster (ed.) Valuing Nature?, London: Routledge. James, R. and Blamey, R. (2000) ‘Deliberative non-market valuation’, paper presented at ISEE 2000 Conference: People and Nature: Operationalising Ecological Economics, 5–8 July. Johnson, J. (1998) ‘Arguing for deliberation: some skeptical considerations’, in J. Elster (ed.) Deliberative Democracy, Cambridge: Cambridge University Press. Johnson, S. (1963) ‘The adventurer, Saturday 28th April 1753’, in W. Bate (ed.) ‘The Idler’ and ‘The Adventurer’, New Haven: Yale University Press. Krehbiel, K. (1991) Information and Legislative Organization, Michigan: Ann Arbor.
Consumer valuation and citizen deliberation 207 Kuper, R. (1997) ‘Deliberating waste: the Hertfordshire citizens’ jury’, Local Environment, 2(2): 139–53. Lane, R. (1991) The Market Experience, Cambridge: Cambridge University Press. Liebrand, W., Messick, D. and Wilke, H. (eds) (1992) Social Dilemmas: Theoretical Issues and Research Findings, Oxford: Pergamon Press. Macedo, S., (ed.) (1999) Deliberative Politics, Oxford: Oxford University Press. Mackie, G. (1998) ‘All men are liars’, in J. Elster (ed.) Deliberative Democracy, Cambridge: Cambridge University Press. Miller, D. (1992) ‘Deliberative democracy and social choice’, Political Studies, 40: 54–67. Nagel, T. (1979) ‘The fragmentation of value’, in T. Nagel (ed.) Mortal Questions, Cambridge: Cambridge University Press. O’Connor, M. (2000) ‘Pathways for environmental evaluation: a walk in the (Hanging) Gardens of Babylon’, Ecological Economics, 34(2): 175–93. O’Neill, J. (1993) Ecology, Policy and Politics, London: Routledge. O’Neill, J. (1997) ‘Deliberation and its discontents’, working paper presented at the conference Environmental Justice, Global Ethics for the 21st Century, October 1997, Melbourne: University of Melbourne. Orbell, J., van der Kragt, A. and Dawes, R. (1988) ‘Explaining discussion-induced cooperation’, Journal of Personality and Social Psychology, 54(5): 811–19. Orbell, J., Dawes, R. and van der Kragt, A. (1990) ‘The limits of multilateral promising’, Ethics, 100: 616–27. Pettit, P. (1998) ‘Institutional design and rational choice’, in R. Goodin (ed.) The Theory of Institutional Design, Cambridge: Cambridge University Press. Riker, W. (1982) Liberalism against Populism: A Confrontation between the Theory of Democracy and the Theory of Social Choice, San Francisco: Freeman. Rippe, K. and Schaber, P. (1999) ‘Democracy and environmental decision-making’, Environmental Values, 8(1): 75–88. Sagoff, M. (1988) The Economy of the Earth, Cambridge: Cambridge University Press. Sagoff, M. (1998) ‘Aggregation and deliberation in valuing environmental public goods: a look beyond contingent pricing’, Ecological Economics, 24(2–3): 213–30. Sally, D. (1995) ‘Conversation and cooperation in social dilemmas’, Rationality and Society, 7: 58–92. Sanders, L. (1997) ‘Against deliberation’, Political Theory, 25(3): 347–76. Sen, A. (1970) Collective Choice and Social Welfare, Amsterdam: North Holland. Sen, A. (1995a) ‘Environmental evaluation and social choice: contingent valuation and the market analogy’, Japanese Economic Review, 46(1): 23–37. Sen, A. (1995b) ‘Rationality and social choice’, American Economic Review, 85(1): 1–24. Sen, A. (1997) ‘Individual preference as the basis of social choice’, in K. Arrow, A. Sen and K. Suzumara (eds) Social Choice Re-examined, London: Macmillan. Simon, H. (1978) ‘Rationality as process and as product of thought’, American Economic Review, 68 : 1–16. Smith, G. and Wales, C. (1998) ‘Citizens’ juries and deliberative democracy’, mimeo, Glasgow: University of Strathclyde, Department of Government. Smith, G. and Wales, C. (1999) ‘The theory and practice of citizens’ juries’, Policy and Politics, 27(3): 295–308. Sugden, R. (1999) ‘Alternatives to the neo-classical theory of choice’, in I. Bateman and K. Willis (eds) Valuing Environmental Preferences, Oxford: Oxford University Press. Vatn, A. and Bromley, D. (1994) ‘Choices without prices without apologies’, Journal of Environmental Economics and Management, 26: 129–48.
208 Jonathan Aldred Ward, H. (1999) ‘Citizens’ juries and valuing the environment: a proposal’, Environmental Politics, 8(2): 75–96. Williams, B. (1979) ‘Conflicts of values’, in A. Ryan (ed.) The Idea of Freedom, Oxford: Oxford University Press.
Three approaches to valuing nature 209
9
Three approaches to valuing nature Forest floodplain restoration Wendy Kenyon and Nick Hanley
Introduction Given the extensive literature documenting problems related to economic valuation techniques such as the contingent valuation method (CVM), interest has developed in approaches which provide an alternative source of information on public value judgements for natural resource decision-making (Brown et al. 1995). In particular, some authors have suggested that deliberative methods might have a role to play as alternatives or complements to more traditional project appraisal techniques (Jacobs 1997; Sagoff 1998). A number of researchers have assessed the role of the Citizens’ Jury (CJ) as discussed in the preceding chapter (see also, Stewart et al. 1994; Aldred and Jacobs 1997; James and Blamey 1999). Here we take the CJ to consist of a small group of people, selected to represent the general public rather than any particular interest group or sector, who meet to deliberate upon a policy question. The CJ offers a number of advantages over economic appraisal methods, and in particular over the CVM. First, respondents to CVM surveys must understand exactly what it is they are to value if useful information about preferences is to be elicited (Arrow et al. 1993). However, recent literature highlights the fact that many respondents do not appear to be well informed about the issues or the good to be valued (Brown et al. 1995; Jacobs 1997). As Munro and Hanley (1999) show, changing people’s information sets can be expected to change their willingness to pay (WTP). CJs tackle this ignorance problem by combining information, time, scrutiny and deliberation in the preference elicitation process (Coote and Lenaghan 1997). They allow participants to question witnesses, discuss witnesses’ evidence with other jurors, and thereby gradually learn about and reach a richer understanding of the issue (Sagoff 1998). CJs therefore address the information problem better than the CVM. Second, economists and others have suggested that a CVM questionnaire asks respondents the wrong question, assuming that consumers think about environmental goods in the same way they do about private goods (Blamey 1996; Jacobs 1997; Sagoff 1988). Blamey (1996) suggests that respondents should not be treated as consumers of environmental goods, but rather as citizens who think of the welfare of the community when responding to environmental issues. In other
210 Wendy Kenyon and Nick Hanley words, individuals approach decision-making relating to environmental goods as citizens rather than consumers. Whilst the validity of this as a universal description fitting all cases of environmental management can be disputed, the use of CJs as a method of preference revelation allows consumers to be asked what Sagoff and Jacobs call ‘the right question’, as it allows deliberation on the environmental issue in terms of what is best for society. Indeed, while the question for the jury can be framed in the context of individual consumer values and preferences if necessary, CJs were developed specifically to determine opinions that represent the general public, rather than any individual interest (Coote and Lenaghan 1997; Brown et al. 1995). Third, CJs may also be useful in dealing with the equity and distributional issues which the CVM has attracted criticism over. Economic value is effectively determined by preferences, underpinned by ability to pay. Therefore, in the CVM, any value that a consumer places upon a good is not registered unless she is able to pay for it. CJs however allow participants’ opinions and preferences to be expressed and registered regardless of their ability to pay (Crosby 1995).1 Fourth, the United Nations Conference on Environment and Development has suggested that environmental decisions will not be sustainable unless local communities participate fully in the decision-making process (United Nations 1993). The CVM does not include the participation of the community in a central way, and therefore may not encourage sustainable actions. The use of CJs is, in contrast, a means by which public participation can be more fully incorporated into the environmental decision-making process. Finally, the notion of value construction suggests that respondents do not have well-defined preferences for many complex environmental goods prior to the elicitation process, but that these preferences are constructed during this elicitation process itself (Gregory et al. 1997; Payne and Bettman 1999). The way in which people construct their preferences is important as it is argued that decision-makers should attach more weight to the preferences of someone who knows both sides of the argument, than to someone whose knowledge of the problem is more limited (Elster 1983; Payne and Bettman 1999). According to Elster (1983) however, methods which explicitly try to determine how people construct their preferences (for example by aiding value construction, such as Gregory et al.’s (1997) decision pathways, or by asking respondents to ‘think aloud’ whilst determining their value as Schkade and Payne (1994) did) are likely to be contradictory – akin to telling someone to ‘be spontaneous’! CJs, however, provide information about the process of preference construction as a by-product of the process rather than as a central role. Therefore, using CJs may offer a means of circumventing the contradiction inherent in helping respondents construct their values, and provide information to decision-makers on the weight which should be given to those preferences expressed. Elster (1983) argues that because the CVM is based on a ‘thin theory of rationality’, which requires only consistency in the expression of preferences, poor policy choices may be made on the basis of CVM results. CJs on the other hand appear more consistent with a ‘broad theory of rationality’ which examines not only the consistency of expressed preferences, but also the beliefs and desires behind
Three approaches to valuing nature 211 decisions. He argues that decisions should be made by the ‘public and rational discussion about the common good, not the isolated act of voting to private preferences’ (Elster 1983: 35). CJs facilitate this type of rational discussion more than the use of economic appraisal methods. Despite the advantages that CJs appear to have over the CVM, they do not provide an economic estimate of the value of any particular project, nor whether it constitutes an efficient use of resources. These are important weaknesses. A third approach therefore seems potentially desirable, which allows deliberation, but also the estimation of the economic value of a project. Such a method might aim to combine the strengths of the CVM and CJ. This chapter reports on an application of three approaches to one decision-making context, a proposed floodplain restoration project in the south of Scotland. A CVM, a CJ and a combined approach (the ‘valuation workshop’) were conducted to assess the value of the project to local communities. The valuation workshop was an attempt to combine the quantitative outputs of the CVM with the participatory, deliberative and preference construction attributes of the CJ. The rest of the chapter is organised as follows. We briefly describe the context of the three approaches. Then we report on the design and the results of the CVM. The next section describes the design and results of the CJ, and the following section relates to the workshop approach. Then we assess the methods, and look at the implications of the CJ and valuation workshop for dealing with problems related to the CVM. The chapter concludes with a discussion of the complementarity of the three approaches.
The Ettrick Valley Forest floodplain restoration project The context for the three approaches is a forest floodplain restoration project in the Borders Region of Scotland. Forest floodplains have almost disappeared from temperate areas in Europe and North America. In Britain, most forest floodplains were cleared in prehistoric and early historic times (Peterken and Hughes 1995). Until the eighteenth century some tracts survived on large rivers, but interventions by man in the early nineteenth century have meant that much of the remaining forest floodplains were wiped out. Actions such as ‘river rectification’ meant that flooding regimes were altered so that agriculture could take place on land previously prone to flooding. Ecologically the decline of forest floodplains is significant as they are one of the ‘richest components of the landscape’ (Peterken and Hughes 1995: 191). Dechamps et al. (1987) prove that bird species richness depends on the variety of vegetation types, and the fauna depends on the variety of food sources available. The richness of flora found in forest floodplain areas therefore leads to a richness of wildlife inhabiting such areas. The forest floodplain is therefore important in protecting and restoring biodiversity. Restoration of forest floodplains may therefore go some way to maintaining or restoring biodiversity in suitable areas. In addition, Petts (1990, cited in Peterken and Hughes 1995) suggests that the restoration of forest floodplains might have a number of other benefits including timber production,
212 Wendy Kenyon and Nick Hanley reduction of agricultural surpluses, fishing, water quality and pollution control, and landscape benefits. In the last two decades interest has grown in conserving, restoring and expanding those areas of forest floodplain that have survived. In 1981, forest floodplains were the subject of a special study by the Council of Europe (Yon and Tendron 1981). In 1982, the Council of Ministers passed a resolution that they should be protected (No. R(82)12). The World Wide Fund for Nature (WWF) launched a project to conserve central European floodplains in 1985, and ten years later, in 1995, WWF Scotland responded to concerns about the loss of forest floodplain habitats by commissioning a review of their status in Scotland. The study concluded that such habitats were scarce and under threat in Scotland (McGhee Smith Associates 1995). One of the most ecologically interesting areas of floodplain identified in the review was the Upper Ettrick site, which contains a variety of woodland, wetland and grassland habitats of nature conservation interest. The different habitats are distributed in a mosaic of small patches which results in high biodiversity for the area, including species which are recognised as locally and nationally scarce or endangered. The Upper Ettrick presents great potential for the expansion of this valuable habitat, utilising areas which are at present of limited conservation interest such as conifer plantation and improved grassland. Increasing the areas of valuable habitat would both protect the species which are already present and encourage others which would have been present in these habitats in the past. Specifically the project aimed to create 25 hectares of native woodland; to restore 30 hectares of floodplain habitat, including scrub, fen, haymeadow and wetland; to convert 30 hectares of conifer plantation to native broadleaf; to manage 15 hectares of willow scrub and to create 3 km of footpaths, boardwalks and tracks. A group of nongovernment and government organisations in partnership with the local community and landowners co-ordinated the habitat restoration project. The project was led by a local community organisation, the Borders Forest Trust, who were guided by technical and local community steering groups. The Trust was keen to work in partnership with the local community in all aspects of the project, including its evaluation, and were also keen to illustrate the economic benefit the project provided to current and potential future funders. The project therefore provided an appropriate context for the application of the three methods.
The contingent valuation method survey There is much debate in the literature about the appropriate design of CVM questionnaires, with respect to a number of issues, and in particular the choice of elicitation format and payment vehicle. Boyle and Bergstrom (1999: 197) state that ‘despite early evidence that payment vehicles can influence responses to CVM questions no published research has been conducted to address this concern’. However, recent recommendations suggest that a mechanism where the respondent has no choice but to pay (e.g. taxation), is most appropriate. Carson et al. (1999) suggest that the use of charitable donations as a payment vehicle within a CVM
Three approaches to valuing nature 213 survey may lead to strategic behaviour by the respondent. However, others suggest that a charitable bid vehicle is appropriate in certain circumstances, and indeed recommend that the information and questionnaire be in a campaigning style to imitate real scenarios of this nature (Macmillian et al. 1998). Macmillian et al. (1998) found that a charitable donation achieved a high level of convergence with real payments. One of the other important design issues in using the CVM surrounds the elicitation format used. The National Oceanic and Atmospheric Administration panel favoured the dichotomous choice (DC) format, but this contrasts with other recommendations that the most conservative questionnaire design is most appropriate (Arrow et al. 1993). The DC format consistently produces higher estimates of mean WTP than do open-ended elicitation formats (Ready et al. 1996; Boyle et al. 1996), and is therefore not the most conservative design choice. Notwithstanding the preceding discussion, one of the most important considerations in the design of a CVM questionnaire is to make the scenario believable (Mitchell and Carson 1989; Boyle and Bergstrom 1999). In the Ettrick study a charitable donation payment vehicle and an open-ended elicitation format were used. There were two main reasons for this choice. First, a number of focus groups were carried out during the design of the questionnaire, where different payment vehicles and elicitation formats were considered. Participants of these groups indicated that they were most comfortable with the open-ended format and the charitable donation bid vehicle. Second, this format and vehicle is one which respondents to the survey are familiar with, especially given the local nature of the project. A wildwood recreation project in the Borders Region of Scotland was campaigning for funds at a similar time that the questionnaire was designed, tested and conducted. In this real situation an open bid in conjunction with a payment card type elicitation method was used with a charitable donation as the payment mechanism. This method is often seen in other charitable campaigns in the UK. Respondents to the Ettrick survey could therefore be expected to be familiar with this payment context, making the choice most suitable in the circumstances. The questionnaire consisted of three sections. The first requested general information about the respondent’s residential status, their participation in outdoor activities, and attitudes towards the environment. The second provided information about the forest floodplain project and asked a payment principle question, as well as the WTP question.2 This section also reminded the respondent of their budget constraint, that the money would go to fund the Ettrick project only, and that the project would not go ahead if sufficient funds were not raised from public donation. It also contained a question which allowed protest bids to be identified. The final section of the questionnaire requested the usual socio-economic data from respondents. The questionnaires were completed in the late summer of 1998, by an independent market research company, using a face-to-face interview method. Nine towns in the Borders Region of Scotland were selected as sites for the questionnaires to be carried out, and responses were collected from a stratified sample of the Borders population, and a small number of visitors to the Region. The contingent valuation questionnaire provided information on respondents’ WTP for the project and on their views on the project as a whole. Of the 336
214 Wendy Kenyon and Nick Hanley respondents, 29 per cent were protest bids. This number is relatively high and the reason most often stated was that they thought the government, the lottery or some other body should pay. Those respondents who did not provide a reason for bidding zero were classified as protesters. Protest bids were removed from the data set for further analysis, although this means that average WTP is higher than would be the case if protesters were allocated a zero value and included. Table 9.1 shows the resultant descriptive statistics for the project. The mean WTP was £13.18 per person for the project as a one-off payment. Descriptive statistics are shown in Table 9.1.3 Given the results shown in Table 9.1, the benefit of the project can be estimated at £568,677 (based on a once-only mean WTP aggregated over the population of the Borders Region) and can be compared with the costs of the project at £350,000.4 Costs of the project were considered in the initial years only, as the aim of the project was that it would be self-sustaining, perhaps with some small continued help from the local community. The composition of costs and benefits indicates that the project might pass the cost-benefits test and may therefore be considered value for money. However, this data does not provide any recommendations for management, nor does the method involve the participation of the local community in a central way. An alternative approach which does meet these two criteria was therefore also conducted – a CJ.
The citizens’ jury The CJ participants were selected from the CVM questionnaire respondents. The final question on the CVM questionnaire asked respondents whether they would be willing to attend a group meeting to discuss local issues in more depth. Those who responded positively to this question formed the pool from which jurors were selected.5 The CVM questionnaire provided extensive socio-economic information about respondents, so that it was possible to select jurors to be representative of the Borders population. Eleven jurors finally took part in the CJ. The jury met over three days and one evening in December 1998. The aim was to assess the project site and to provide qualitative information on its value and importance to the local community. The jury was carried out in collaboration with a local community environmental organisation, the Borders Forest Trust, who were keen to encourage consultation and participation of the local community in the project. The jurors were asked to deliberate on the following specific questions: (i) What should individual land use and environmental projects in southern Scotland such as the Ettrick Forest Floodplain Project aim to achieve? and (ii) How might
Table 9.1 Descriptive statistics for WTP for the Ettrick project Mean (£ per person)
Std dev
Median
Range
95% Confidence intervals
N
13.18
69.71
1
0–1,000
4.15–22.1
232
Three approaches to valuing nature 215 the success of such projects be determined? Clearly, ‘economic efficiency’ could be one answer to either of these questions, but may not be a priority for participants. To provide information with which the jury could assess the project, ten witnesses attended sessions to provide ‘evidence’ and to answer questions. The witnesses were selected in consultation with the Borders Forest Trust, and in discussions with stakeholders from all sides of the debate to ensure balanced representation. The witnesses came from a variety of backgrounds including Scottish Natural Heritage, the local council, the Forestry Commission, the Scottish Tourist Board, and environmental consultants. Witnesses made short presentations to the jury of 10 to 15 minutes followed by a discussion session with the jury of about 30 to 40 minutes. In addition to sessions involving witnesses, the process included a number of jury-only sessions, where the jurors discussed particular issues as a whole unit, or in smaller groups using participatory appraisal techniques. Participatory tasks were designed to including listening, hearing and watching. According to Pretty et al. (1995: 24) participants are likely to remember ‘10 per cent of what they read, 20 per cent of what they are told, 30 per cent of what they see and 50 per cent of what they hear and see’. Although participants did not need to remember long term what went on in the session, ensuring they were engaged in the tasks was vital. The use of brainstorming techniques, organising comments written on cards, and using large flip-chart paper served a number of purposes: to provide a focus for attention while the discussion ensued; to stimulate discussion between people with different levels of numeracy and literacy; to re-enforce spoken issues; to provide a means of cross-checking what was said, and to record the proceedings. The final recommendations were achieved entirely by discussion and consensus, and approved by all of the jurors. A report on the process and outcome of the jury was written and sent to the jurors for approval, before being sent to the Borders Forest Trust and other interested parties (Kenyon 1999). The jury made recommendations about how the Ettrick project should be managed and co-ordinated in order to achieve the environmental and social objectives which the jurors felt were desirable in southern Scotland. The jury also noted a number of aspects related to the project they felt were valuable (Table 9.2). Interestingly, not only environmental issues were identified, but also aspects which might be classed as social values, such as community involvement, and the role of the project for education. In addition to the aspects of value identified in Table 9.2, the jury noted two areas of concern, namely access and future management. Jurors felt that visitors should be allowed access to the site, but were concerned about the damage they might inflict. Specific recommendations were made regarding arrangements to promote benign access, and the provision of information to visitors to encourage careful use of the site. The second area of concern related to future management of the site and the availability of funding to secure it. The jurors felt that it may be possible to start a trust fund dedicated to the Ettrick project to make sure that money was still available for future management. However, after speaking with a member of the local community, the jurors became less concerned as they were
216 Wendy Kenyon and Nick Hanley Table 9.2 Positive issues identified with the forest floodplain project Environmental issues
Social issues
Preservation of a natural ecosystem – a world resource
Good demonstration scheme for copying
Flood control
Education of the young and encouraging educational studies
Balance of different habitats and getting back to nature
Decrease the number of sheep and fencing of sensitive areas
Encouraging wildlife and preservation of indigenous life forms
Getting rid of ugly blanket plantation forestry
Monitoring of species
Community involvement
assured that the local community was fully involved in the project and that they would ensure that the site was well managed into the future. The jury were able to look at the Ettrick project in the wider context of the south of Scotland, and made a number of recommendations regarding the management of individual environmental projects in the region. They felt that a variety of projects were needed in the region, which met a number of different needs, but that all of these projects should be co-ordinated in an integrated way. For example, some environmental projects might aim to attract tourism to the area, but such projects must be situated in less environmentally sensitive areas. Others might aim to increase biodiversity, but not aim to attract tourists. Finally, the jury considered how the success of environmental projects such as the Ettrick Forest Floodplain Project, might be assessed. Table 9.3 shows the criteria offered by the jury, who also suggested that these measures could be considered as elements of future project design and planning. It is interesting to note that the jury considered environmental and social elements important in judging the success of the project, but that they did not seem to consider economic criteria important.
The valuation workshop The final approach used to evaluate the Ettrick Floodplain Restoration Project was a hybrid of the CVM and CJ, named the valuation workshop. It aimed to build on the strengths of the CVM and CJ exercises, and contained elements of each. In particular, the intention was to combine the quantitative outputs of the CVM with the participatory, deliberative and preference construction aspects of the CJ. The valuation workshop was conducted in three parts. After an introduction, the participants were each given a CVM questionnaire and asked to complete it individually. Next, participants were divided into groups of between 4 and 7 people for discussion. The first task each group was given was to discuss the good aspects of the Ettrick project. Participatory methods (such as those discussed above) were used to facilitate the discussion, and provide a focus for the task (Pretty et al. 1995).
Three approaches to valuing nature 217 Table 9.3 Jury suggestions measuring environmental project success Environmental issues
Social issues
Has the variety of wildlife improved? Is wildlife being protected? Is the site attractive?
Has it got community and farmer approval? Has community spirit improved? Has the project created any unseen problems?
Each group was asked to provide a ranked list of good points related to the Ettrick project. The second task was to consider the problems of the project, and the concerns they had with it. Each group were asked to discuss problems, rank their relative importance and offer suggestions of how each problem might be mitigated or solved. The final task in this section of the workshop asked the participants to make recommendations on how the success of the project might be judged. This involved creating a list of means by which the project could be assessed, with discussion surrounding each suggestion. In the third part of the valuation workshop participants were asked to complete some further questions individually in a survey. These questions asked whether the participants would change the WTP they stated at the beginning of the workshop, and explain why. Participants had now had a chance to discuss the project, assimilate information more fully and consider the project from different perspectives, given the interaction with other participants. The valuation workshops were conducted in December 1999 and took around three hours each. Two workshops were carried out in each of two towns in the Borders, giving a total of four workshops and 44 participants. In order to select participants, 500 letters were sent out to a random selection of addresses in or near each of the towns. The letter invited the addressee to one of the workshops, explained that the Scottish Agricultural College was carrying out research on the local environment and were interested in their views, and asked them to return an enclosed form if they wished to participate. A pool of possible participants was then drawn up from those who replied, and participants were chosen from this pool to be as representative of the local population as possible. Each participant was paid £20 for their attendance. The workshop provided a range of qualitative and quantitative data on the Ettrick project. Of the 44 participants attending the workshops, 27 (61 per cent) made genuine CVM bids initially, the rest being either protest bids (5 per cent) or ‘don’t know’ responses (34 per cent) to the valuation question. After the discussion, in the second part of the questionnaire, the number of valid bids increased as 2 of the ‘don’t know’ respondents offered a new valid bid. Of the 44 participants, only 6 (14 per cent) changed their bid after the discussion (Table 9.4). Participants were clearly considering their budget constraints carefully as the most common reason given for not changing their bid was that their financial circumstances had not changed and they could not therefore afford a higher amount even if they now wanted to. Interestingly none of the participants revised their bid downwards. Although the preceding part of the workshop comprised discussion of both the good and bad aspects of the project, the discussion of the good points clearly carried more weight.
218 Wendy Kenyon and Nick Hanley Table 9.4 Original and revised bids with reasons for change Original bid
Revised bid
Reason
25
50
Wish it would happen
10
20
It would have a beneficial effect on the local community, Borders community, and is a valuable recreational and educational asset
25
50
Awareness of the problem
20
??
Would be willing to sponsor the project on a monthly basis
DK
10
More knowledge
DK
25
Because of points for the project brought up in the discussion
Table 9.5 and Table 9.6 display the descriptive statistics for pre-discussion and post-discussion bids. Although the mean has increased, a two-tailed paired T-test showed no significant difference between the two workshop mean WTPs at 10 per cent significance (t = –0.57, p = 0.57).6 Although the discussion did have an impact on individual bids, it did not therefore have a statistically significant impact on mean WTP. A Mann–Whitney U-test gives the same results, the hypothesis that the pre-discussion median is equal to the post-discussion median cannot be rejected (W = 747.0, p = 0.6735 adjusted for ties). In addition to this quantitative output, the valuation workshop provided information about what the participants felt was good and bad about the project, and how the project’s success might be measured. Environmental aspects of the project were in all but one of the 8 groups (2 per workshop) considered the most important. The one group who dissented felt that the promotion of education and tourism in the area was most important. Table 9.7 shows the results from one discussion group which is typical of the responses from other groups. One strength of the valuation workshop approach relative to the CVM is that it provides additional information on participants’ views of the project – both positive and negative. Participants were asked not only to discuss issues which they thought were problematic, but also to offer solutions to these problems. Table 9.8 shows the typical concerns and the solutions offered by one of the workshop groups. Every group considered the cost of the project to be a problem and many groups differentiated between start-up costs and running costs. One of the other most discussed problems was that of access. Roads to the site were single track and not suitable for a large number of cars, and there was virtually no public transport available to the site. Despite having been provided with very little information, participants were able to offer solutions to these problems in most cases, clearly some more practical than others (Table 9.8). Finally, the participants were asked to suggest how the success of the project might be measured. A range of ideas was suggested including monitoring flora and
Three approaches to valuing nature 219 Table 9.5 WTP descriptive statistics for initial valuation Mean
Median
Std dev
Range
95% CI
N
11.07
10
12.59
0–50
6.09–16.05
27
Table 9.6 WTP descriptive statistics for final valuation Mean
Median
Std dev
Range
95% CI
N
13.59
10
12.59
0–50
7.64–19.54
29
Table 9.7 Good aspects of the Ettrick project Good points
Examples
Rank
Protected area
Place of scientific interest Protected area
1
Wildlife
Increase wildlife Varied wildlife Protect wildlife
2
Jobs
Create jobs Attract people to the area
3
Local/outdoor interest
Somewhere to visit More interesting countryside to walk in Makes people more aware of the environment locally
4
Reduce chemicals on land
Reduce chemicals on land
5
Less cultivated
Less cultivated, more natural
6
fauna, evaluating visitor levels, jobs, the level of community involvement, and interest in the site by educational establishments. As with the CJ, no direct financial criteria were suggested here, although many of the criteria have an economic element, such as jobs and visitor levels.
Conclusions Both the CVM and CJ provide policy-relevant information, the information provided by each being useful in different ways. The results of the CJ identify the project needs and how it should develop. Table 9.2 lists those issues that the jurors felt to be positive, and that might be used as objectives by the managers of the project. Similarly, by identifying concerns relating to the project, jurors provide direction to managers and policymakers as to the development of the project. The information provided by the jury therefore plays a practical role in directing the management of the project. The results of the CVM provided different, but equally policy-relevant information. The CVM is able to measure both the intensity and direction of
220 Wendy Kenyon and Nick Hanley Table 9.8 Project problems, solutions and the importance of bad aspects Problems
Rank
Solutions
Costs – who pays
1
Start-up cost: central funding, lottery, EU Running cost: 3 for admission fee, 3 against Prevent damage by visitors
Rural depopulation – reduction of number of viable farms
2
Change/expansion of jobs related to the exercise, e.g. catering, maintenance, paths, bridges, etc. Rangers Tourism facilities Farm shop/equipment hire shop
Access roads
3
Straighten roads Bridges Improved public transport, i.e. minibuses
Agreement with landowners
4
No solutions offered
Damage by visitors
5
Designated picnic area and toilets Designated play area
Car parking
6
No solutions offered
preferences, and provides data that is consistent with welfare economic analysis which can feed into cost-benefit analysis. The benefit of the project estimated at £568,677 could be compared with the costs of the project at £350,000. The project passes the cost-benefit test, and thus offers a potential Pareto improvement. The workshop approach is an attempt to amalgamate aspects of the two methods and provide both practical guidelines to the project managers and cost-benefit estimates, whilst maintaining the participation of the local community in the project evaluation. The results of the workshop are comparable on the one hand with the CVM results and on the other with CJ results. A T-test shows that neither the prediscussion workshop mean nor the post-discussion workshop mean WTP are significantly different from the CVM mean bid at 10 per cent level (t = –0.41, p = 0.68 and t = 0.01, p = 0.99 respectively). However, a Mann–Whitney U-test does shows a significant difference between the medians (W = 4228.5, p = 0.0350 and W = 4265.5, p = 0.0268 respectively). The results of the discussion of the workshop correspond quite closely to the recommendations provided by the jurors. Workshop participants were also asked to evaluate the final aggregate CVM result at the end of the session. The calculation was explained to the group, and the final figure presented to them. The comments from all the groups can be split into four categories. First, that it is not possible to put a value on such a project. Second, comments surrounded whether or not this aggregate figure would be obtained in reality, as due to the poor economic situation in the Borders there were more important things to spend money on. Third, the clear feeling that money for such projects should come from government, lottery or EU funds was expressed. Finally, there was discussion that information on costs (which was not made
Three approaches to valuing nature 221 available) was required before any pronouncement was made on its value. This additional feature provides interesting information on the CVM figures, and on the authenticity of the economic estimates. Overall the valuation workshop approach goes some way to developing a method which provides economic estimates of environmental projects whilst adding value to traditional CVM exercises, and offering qualitative recommendations. However, the workshop results suffer in this study from a small sample size relative to the CVM survey. The research also related to problems identified with the CVM in the literature, and the role that more deliberative methods may have in mitigating these concerns. One relates to the provision of information to CVM respondents. The respondents to the CVM survey were given visual and verbal information about the project site, and asked, ‘Do you prefer the site with or without the project?’ Thirteen per cent of respondents did not know whether they preferred it with or without. This may indicate that the information was not sufficient for them to be able to determine their preference. This did not appear to be a problem in the CJ. Jurors were all able to determine their own preferences regarding the Ettrick project, and further were able to identify those aspects they preferred most (Table 9.2). The workshop participants were given the same information as the CVM respondents, but were allowed to discuss it in a social setting. The fact that two participants changed their WTP bid from a ‘don’t know’ response to a positive response implies that the discussion did allow the participants to better understand the project they were asked to pay for. A number of researchers suggest that CVM questionnaires ask the wrong question and assume that respondents act as consumers and not citizens when responding. Our results indicate that this assumption may be valid (Sagoff 1998; Blamey 1996). Analysis of the CVM results showed that the most significant variable in influencing whether a respondent would be willing to pay anything was whether she was likely to visit the site if the project went ahead (Kenyon and Nevin 2001). This may indicate that the respondents acted as consumers and not citizens when responding to the questionnaire.7 If this is the case, the CJ can be seen as a complement to the CVM by evaluating the project from a citizen’s point of view. The valuation workshop incorporates elements of both a citizen and consumer approach. The initial CVM responses were from a consumer standpoint, but the output from the discussion was clearly from a citizen standpoint, since community issues were considered an important aspect of the Ettrick project. However, this result also suggests some relationship between a ‘consumer’ standpoint and the respondent accounting mainly for use and option values (and not other components of total economic value). It is clear that further research is required on the relationship between these two issues. The issue of value construction is also a matter of debate within the environmental valuation literature. The CVM results provide no evidence regarding the detailed construction of final values. However, comments from jury participants in the evaluation discussion and questionnaire at the end of the process suggested that breaking down the decision-making process into tasks made the whole task less daunting, and more manageable. The by-product of these tasks is to provide
222 Wendy Kenyon and Nick Hanley information that may indicate how jurors constructed their final response. The evidence of the valuation workshop is mixed on this point. Whilst the pre-discussion WTP was constructed in the normal CVM way, the post-discussion response had the benefit of discussion in a social setting, and the outcome of the discussion may be an indicator of the reasons for a (limited) changed bid. Finally, in a climate where both national and international agreements exist which seek to enhance public participation in environmental decision-making, it seems clear that the process should rely on more than just economic estimates of value as provided by traditional CVM surveys. Whilst such estimates are still useful, policy-makers are increasingly required to incorporate decision-making, planning and management into one process, and include both expert and lay opinion. New methods which are able to provide a more holistic approach to environmental policy are needed so that these national and international targets can be met. The CJ is able to offer such an integrated approach. Evidence from the CJ suggests that the jurors were able to think holistically, offering comment on benefits and pitfalls of the project and a wide range of management recommendations. One of the recommendations was that the Ettrick project should not be considered in isolation, but as part of a suite of projects, to ensure that the Borders environment developed in an integrated way. The valuation workshop also provided a more integrated approach, in which participants were able to complement economic estimates with wider indicators of values and preferences. In many instances where environmental project evaluation is required, conducting both a CVM and CJ would be prohibitive in terms of time and money. The challenge for the future may therefore lie in developing complementarities. Building on the valuation workshop approach, and increasing sample sizes may go further in offering a more appropriate combined approach, so that decisionmakers can benefit from both a wide range of policy-relevant information when evaluating environmental projects.
Acknowledgements This work was funded by the Economic and Social Research Council under the Global Environmental Change Programme, and the Scottish Executive Rural Affairs Division. The authors would also like to thank the Borders Forest Trust, and the people of the Borders for their input and enthusiasm.
Notes 1 Although self-selection of the participants in a CJ could mean that those with higher education and income, for example, may be more likely to participate and the preferences of the ‘excluded’ may not be registered. 2 How much money (on a once-only basis) would your household be willing to donate to the fund in order to ensure this project went ahead? Please bear in mind that this money would only go towards paying for the management of the natural environment in the Ettrick Valley. You might also like to think about the spending on other items that you might have to give up if you did make this payment.
Three approaches to valuing nature 223 3 More detailed statistical analysis of the CVM results can be found in Kenyon et al. (2000). 4 Figure obtained from the Borders Forest Trust. 5 The Borders Jury was selected in this way to minimise recruitment costs. Despite the final CJ being a reasonable representation of the Borders population (when compared by socio-economic characteristics), this is probably not the best means of jury selection. Selection processes utilising a random draw from the electoral register are considered to provide a better sample of jurors. 6 The two respondents who changed from a ‘don’t know’ to a positive bid were taken out of the sample for the purpose of the test. 7 It could be argued that the most important feature of the project was therefore its use or option use value.
References Aldred J., and Jacobs, M. (1997) ‘Citizens and wetlands. What priority, if any, should be given to the creation of wetlands in the fens?’, report of the Ely citizens’ jury, Centre for the Study of Environmental Change (CSEC), Lancaster: Lancaster University. Arrow K., Solow R., Portney P.R., Leamer E.E., Radner R. and Schumann H. (1993) ‘Report of the NOAA panel on contingent valuation’, Federal Register, 58: 4601–14. Blamey, R.K. (1996) ‘Citizens, consumers and contingent valuation: clarification and the expression of citizen values and issue-options’, in W.L. Adamowicz, P.C. Boxhall, M.K. Luckert, W.E. Phillips and W.A. White (eds) Forestry Economics and the Environment, Oxford: CAB International. Boyle, K.J. and Bergstrom, J.C. (1999) ‘Doubt, doubts and doubters’, in I. Bateman and K.G. Willis (eds) Valuing Environmental Preferences. Theory and Practice of the Contingent Valuation Method in the US, EU and Developing Countries, Oxford: Oxford University Press. Boyle, K.J., Johnson, F.R., McCollum, D.W., Desvouges, W.H., Dunford, R.W. and Hudson, S.P. (1996) ‘Valuing public goods: discrete versus continuous valuation responses’, Land Economics, 72: 381–96. Brown, T.C., Peterson, G.L. and Tonn, B.E. (1995) ‘The values jury to aid natural resource decisions’, Land Economics, 71: 250–60. Carson, R.T., Groves, T. and Machina, M.J. (1999)‘Incentive and informational properties of preference questions’, paper presented at the European Association of Resource and Environmental Economists Conference, Oslo, June 1999. Coote, A. and Lenaghan, J. (1997) Citizens’ Juries: Theory into Practice, London: Institute for Public Policy Research. Crosby, N. (1995) ‘Citizens juries: one solution for difficult environmental questions’, in O. Renn, T. Webler and P. Wiedemann (eds) Fairness and Competence in Citizen Participation, Dordrecht: Kluwer. Dechamps, H., Joachim, J. and Lauga, J. (1987) ‘The importance for birds of the riparian woodlands within the alluvial corridor of the River Garonne, SW France’, Regulatory Rivers Restoration Management, 1: 301–16. Elster, J. (1983) Sour Grapes. Studies in the Subversion of Rationality, Cambridge: Cambridge University Press. Gregory, R., Flynn, J., Johnson, S.M., Sattefield, T.A., Slovic, P. and Wagner, R. (1997) ‘Decision-pathways surveys: a tool for resource managers’, Land Economics, 73: 240–54. Jacobs, M. (1997) ‘Environmental valuation, deliberative democracy and public decision making institutions’, in J. Foster (ed.) Valuing Nature? Economics Ethics and Environment, London: Routledge. James, R.F. and Blamey, R.K. (1999) ‘Public participation in environmental decision-making – rhetoric to reality?’, paper presented to the International Symposium on Society and Resource Management, Brisbane, Australia, July 1999.
224 Wendy Kenyon and Nick Hanley Kenyon, W. (1999) Report of the Galasheils Citizens’ Jury, Edinburgh: Scottish Agricultural College. Kenyon, W. and Nevin, C. (2001) ‘The use of economic and participatory approaches to assess forest development: a case study in the Ettrick Valley’, Forest Policy and Economics, 3(1–2): 69–80. Macmillian, D., Smart, T.S. and Thornburn, A.P. (1998) ‘The importance of realism to experiments comparing cash and CV charitable donations: the case of the isle of Eigg Trust’, paper presented at the Agricultural Economics Society Conference, Reading, March 1998. McGhee Smith Associates (1995) ‘Wild rivers. Aspects of the location and ecology of the floodplain woodlands in Scotland’, report to World Wildlife Fund. Mitchell, R.C. and Carson, R.T. (1989) Using Surveys to Value Public Goods: The Contingent Valuation Method, Washington, DC: Resources for the Future. Munro, A. and Hanley, N. (1999) ‘Information, uncertainty and contingent valuation’, in I. Bateman and K.G. Willis (eds) Valuing Environmental Preferences. Theory and Practice of the Contingent Valuation Method in the US, EU and Developing Countries, Oxford: Oxford University Press. Payne, J.W. and Bettman, J.R. (1999) ‘Measuring constructed preferences: towards a building code’, Journal of Risk and Uncertainty, 19: 243–70. Peterken, G.F. and Hughes, F.M.R. (1995) ‘Restoration of floodplain forests in Britain’, Forestry, 68: 187–202. Petts, G. (1990) ‘Forested river corridors: a lost resource’, in D. Cosgrove and G. Petts, (eds) Water, Engineering and Landscape: Water Control and Landscape Transformation in the Modern Period, London: Belhaven Press: 12–34. Pretty, J., Guijit, I., Thompson, J. and Scoones, I. (1995) Participatory Learning and Action: A Trainers Guide, London: International Institute for Environment and Development. Ready, R., Buzby, J.C. and Hu, D. (1996) ‘Differences between continuous and discrete contingent valuation estimates’, Land Economics, 72: 397–411. Sagoff, M. (1988) The Economy of the Earth: Philosophy, Law, and the Environment, Cambridge: Cambridge University Press. Sagoff, M. (1998) ‘Aggregation and deliberation in valuing environmental public goods: a look beyond contingent pricing’, Ecological Economics, 24: 213–30. Schkade, D.A. and Payne, J.W. (1994) ‘How people respond to contingent valuation questions: a verbal protocol analysis of willingness to pay for an environmental regulation’, Journal of Environmental Economics and Management, 26(1): 88–109. Stewart, J., Kendall, E. and Coote, A. (1994) Citizens’ Juries, London: Institute of Public Policy Research. United Nations (1993) The Global Partnership for the Environment and Development. A Guide to Agenda 21, post-Rio edition, New York: United Nations. Yon, D. and Tendron, G. (1981) Alluvial Forests of Europe, Nature and Environment Series 200, Strasbourg: Council of Europe.
Deliberation and economic valuation
225
10 Deliberation and economic valuation National park management Rosemary F. James and Russell K. Blamey
Introduction Among the wide variety of methods available for incorporating community attitudes and values into environmental decision-making are public hearings, calls for submissions, referenda and opinion polls. Each of these approaches has its limitations. Concerns also surround the use of stated preference techniques such as the contingent valuation method (CVM) and choice modelling as inputs to environmental decision-making, in part because responses are based on limited levels of information and deliberation. Techniques based on the theory of deliberative democracy have the potential to correct these adverse conditions. Citizens’ juries (CJs), deliberative polls, consensus conferences and related techniques offer participants time, information and opportunities for social interaction and deliberation, which may result in the development and articulation of robust preferences.1 Under the tenets of participatory democratic theory, public participation is functionally, logically and ethically central to the operation of democracy (Webler and Renn 1995). As Thomas-Slatyer (1995: 11) noted (in relation to community participation in project development in developing countries): ‘Central to these approaches is the belief that ordinary people are capable of critical reflection and analysis and that their knowledge is relevant and necessary.’ Public participation is required in order to ensure that the aspirations and needs of citizens are heard by decision-makers and is considered to produce many benefits including definition and communication of the public will (Rousseau 1968), popular sovereignty (Rosenbaum 1978), political equality (Rosenbaum 1978, Van Valey and Petersen 1987), protection of citizens’ interests (Rosenbaum 1978), definition of the public will (Dienel 1989) and enhancement of personal (Daneke et al. 1983) and social (Rosenbaum 1979) development. An experimental application of the CJ method to economic valuation is described in this chapter. The theoretical basis of the method is first described, followed by a detailed description of this application of the method. Some of the methodological and theoretical issues which arose from this work are discussed.
226 Rosemary F. James and Russell K. Blamey
Public participation through cost–benefit analysis There are numerous methods commonly used in developed countries to allow public input into environmental decision-making, including methods in which respondents are wholly self-selected (such as calls for submissions) and those in which potential respondents are identified by some other agency (such as opinion polls). Another form of public participation is provided by the market, where people’s preferences are reflected in their market transactions. In the absence of a market, surrogate markets and economic valuation surveys may provide information regarding the preferences of citizens. An assessment of the value (benefits or costs) accruing to the community from an actual or proposed course of action is an important component of environmental cost–benefit analysis (CBA), of which the objective is to inform decisions on the allocation of resources so as to maximize the excess of benefits over costs to society as a whole. The basis of the practice of environmental CBA lies in a form of preference utilitarianism in which it is assumed that preferences reflect utility functions, and willingness to pay (WTP) or willingness to accept (WTA) is taken as an indicator of strength of preference. The democratic basis of environmental CBA and associated estimates of community value can thus be summarized as ‘one dollar, one vote’. Environmental CBA involves an assessment of the implications for community environmental preferences, typically in WTP terms, of proposals under investigation, which may have impacts on the natural environment. Stated preference methods alone are potentially capable of estimating the full range of environmental values, while revealed preference methods only provide information on direct use values (see Chapter 1). Stated preference surveys involve eliciting information pertaining to strength of preference through the use of community surveys in which respondents are requested to state what they would do under prescribed circumstances. Respondents are questioned in such a manner as to elicit their WTP for a programme involving some potential environmental improvement (or prevention of some potential environmental deterioration) or WTA compensation for a programme involving some environmental deterioration or lack of environmental improvement. The two most commonly used stated preference methods are the CVM and choice modelling. The CVM, and to a lesser extent choice modelling,2 have been criticized on several grounds. One argument is that ‘stated preference surveys represent a highly constrained environment for preference construction’. Whilst some respondents to stated preference questionnaires may reveal their pre-existing preferences for the attributes and/or goods and services under examination, it is probably more common for respondents to construct their preferences at the time of completion of the questionnaire. Environmental goods and services are complex and often unfamiliar, and the method chosen to elicit WTP should allow respondents the opportunity to articulate their values for the goods and services under consideration and then to construct their preferences. Contingent valuation in particular is not seen as a method which permits this to occur (Gregory et al. 1993; Gregory and Slovic 1997;
Deliberation and economic valuation
227
Sagoff 1998). CVM respondents are required to make choices in the face of necessarily limited time and information and in the absence of the capacity to seek clarification of many of the issues of concern, which may severely constrain the construction of informed and deliberated preferences. Stated preference surveys, in common with other types of surveys, are subject to coverage, sampling, non-response and measurement error. Much, if not most, of the controversy surrounding techniques such as CVM and choice modelling focuses on measurement errors, particularly those attributable to respondents and their interactions with the questionnaire. Sources of such error include yea-saying and nay-saying (when respondents answer positively or negatively to valuation questions without regard to the amount involved in the scenario), the nonidentification of protest responses (objections to the use of a tax or levy to fund environmental improvements), strategic responses and amenity misspecification biases. Methodological criticisms of the CVM include the production of unacceptably large estimates for total WTP or WTA and the insensitivity of WTP or WTA estimates to price and scope variations (Perman et al. 1999). There are a range of philosophical concerns associated with environmental valuation surveys. These relate both to the use of the methodology itself and to respondent motivations and decision processes (Blamey and Common 1999). The ‘one dollar, one vote’ assumption of environmental CBA raises the question as to whether the wealthy should have more say than others. Ethical concerns are raised regarding the valuation of the non-monetary components of environmental goods in monetary terms (Blamey and Common 1992; O’Neill 1997; Gregory and Slovic 1997). The potentially reductionist and atomistic nature of stated preference methods is seen as problematic. Alternative methods which are more pluralistic in terms of community perspectives and values may offer a more appropriate and legitimate input into environmental decision-making. Environmental issues may then best be addressed by individuals acting as citizens with a societal perspective, rather than as consumers acting on the basis of individual preferences (Sagoff 1998).
Deliberative and discursive democracy In the last 30 years or so, several public participation methods have been developed which may surmount some or all of these difficulties and thus provide for more effective environmental value assessment. These methods, founded in deliberative and discursive democracy,3 provide participants with the opportunity for the collection and analysis of information and for discussion and deliberation of the issue under consideration. Those participating in deliberative processes are characterized by Dryzek as ‘amenable to changing their minds and their preferences as a result of the reflection induced by deliberation’ (Dryzek 2000: 31). Deliberative methods include consensus conferences, citizens’ juries, deliberative polls, planning cells, citizen advisory committees, study groups and many more. Processes such as these which have greater claims to being reflective, evolutionary, empowering and conflict-resolving may offer significant advantages over stand-alone stated preference methods in the determination of environmental values. Deliberative
228 Rosemary F. James and Russell K. Blamey processes may also be more acknowledging and accommodating of the complexity and uncertainty inherent in environmental problems and decision-making. The recent application of one of these methods is the CJ. Peter Dienel in Germany and Ned Crosby in the United States independently developed similar techniques for deliberative input by members of the public to decision-making on matters of public policy significance. The Planungszelle, or planning cell, method, was developed in Germany in 1969 and the CJ method was developed in 1971 in the United States (Crosby 1996). The CJ method has since been used in many parts of the world in numerous types of application. Recent publications describing citizens’ juries include Jefferson Center (1995, 1996, 1997), Coote and Lenaghan (1997) and Aldred and Jacobs (2000). The principal features of the CJ method have been well detailed in Crosby (1991, 1996). Essentially, a small group representing a larger population is selected, most usually by quota sampling, and asked to spend several days as part of the CJ. The CJ hears and questions witnesses, discusses and deliberates on a pre-specified charge, and reaches a conclusion. The result is, under ideal circumstances, consensual. The CJ technique may be applied in a number of ways to enable public participation in decision-making on environmental and other matters. The charge may be structured to provide for: 1 2
3
4
assessment of support: the level of support for a proposed action is determined, e.g. Does the jury think the harbour area should be developed?; information gathering: the views of participants are sought on some matter, e.g. What does the jury think are the most important aspects of the harbour? or What does the jury think are the most important impacts of the proposed harbour development?; option choice: a choice is made between several options, which may include cost data, in regard to a proposed development, e.g. Does the jury think the harbour should be developed and, if so, which is the preferred development plan?; deliberative valuation: the valuation of some proposed change in resource allocation or access using stated preference or other means, e.g. for anticipated declines in resource allocation or access: What is the maximum amount the jury would be prepared to pay to prevent the proposed harbour development? or What is the minimum amount the jury would be prepared to accept as compensation for the proposed harbour development?4
The potential for use of the fourth approach, deliberative valuation, has been discussed under various names by Brown et al. (1995), Jacobs (1997) and Sagoff (1998), but no examples of the practice appear to be reported in the literature to date (although see other chapters in this volume). The conduct of a stated preference method in the context of a CJ has the potential to reduce some of the difficulties associated with stated preference methods. The use of a discursive rather than questionnaire-based stated preference valuation method in a CJ may provide additional benefit.
Deliberation and economic valuation
229
Design and conduct of this application The CJ reported in this chapter was conducted as a research application of the technique, rather than commissioned by a government agency in response to an existing environmental issue. The specific topic examined in the CJ was the management of national parks5 in the most populous Australian state, New South Wales. This CJ formed part of an examination of the potential for use of the technique in environmental decision-making in Australia. There are several ways in which the CJ technique can be applied to non-market valuation, i.e. used for deliberative valuation. These are illustrated in Figure 10.1. The suite of work undertaken in this project is shown in italics. The CJ reported herein included the following elements: a conventional option choice charge which did not involve deliberative valuation; two approaches to deliberative valuation, involving a further charge in which the jurors were asked to determine societal WTP for a specified programme involving environmental improvement; and a stated preference (choice modelling) survey, completed by the individual jurors before and after the CJ proceedings. The principal focus of this chapter is on the first of these deliberative valuations, in which a process of group discussion and decision-making was used to arrive at a valuation (WTP) for a posited programme involving environmental improvement. A scenario based on consideration of five major national park management activities was developed and then refined using focus groups. The scenario specified options differing in terms of the levels of effort allocated to the following key national park management activities: fire management, weed control, feral animal control, maintenance of visitor facilities and management of historic sites.6 The relationships between park management activities and biophysical attributes such as wildlife and native plants were detailed in an explanatory document provided to the jurors. The same scenario format was used for both the conventional and deliberative valuation charges and the choice modelling questionnaire. Further, the same base allocation of effort (Option 1), representing the current situation, was used in all three approaches. The charges are shown in Table 10.1 and Table 10.2. The jury was given a set of information and asked which of three options they preferred before the introduction of a fourth option. They were told that the National Parks and Wildlife Service manages about 5 million hectares of parks in New South Wales (6.3 per cent of the state) in 479 parks and reserves. This includes national parks, nature reserves, historic sites, state recreation areas and regional parks. They were told that the term ‘national park’ included all these different types of parks and reserves. They were told that budgets for national park management are always tight and that there would never be enough money to do all the things which might be required to meet the expectations of the public about national park management. A possibility was described as to continue spending the existing budget as shown in Figure 10.2 and in Option 1. Further information about these management activities was provided in a pamphlet. Another possibility described was to keep the total amount of money spent on
230 Rosemary F. James and Russell K. Blamey Citizens’ Jury
With monetary valuation
Using conventional stated preference methods
Without monetary valuation conventional citizens' jury (CJ constructions (i) - (iii))
Using other valuation methods (CJ construction (iv))
Framing of charge Jury to estimate WTP for a specified option Jury to select preferred option from several given; each option includes a cost per potential payer
Framing condition Citizen Ambiguous/not specified Consumer
Framing condition Citizen Ambiguous/not specified Consumer
Response basis Group response Individual response
Response basis Group response Individual response
Figure 10.1 CJs applied to non-market valuation
national park management the same, but change how it is spent. For example, some money could be taken away from maintaining facilities and spent on feral animal control. This is what is proposed in Options 2 and 3 of the charge. The jurors were told that another possibility was to raise more money to spend on national parks. State governments could introduce a park management levy to help pay for increased levels of current activities. The levy would be paid by all
Deliberation and economic valuation
231
Table 10.1 Charge 1: CJ group preference among three options Outcomes of national park management
Option 1 current situation
Option 2
Option 3
Number of national parks with good fire management
100
40
160
Area of feral animal control each year (hectares)
50,000 ha
100,000 ha
30,000 ha
Area of weed control each year (hectares)
3,000 ha
1,000 ha
10,000 ha
Proportion of facilities that are well maintained
35%
45%
25%
Number of well-protected historic sites
7,000
6,000
7,500
Outcomes of national park management
Option 1 current situation
Option 4
Number of national parks with good fire management
100
400
Area of feral animal control each year (hectares)
50,000 ha
160,000 ha
Area of weed control each year (hectares)
3,000 ha
9,600 ha
Proportion of facilities that are well maintained
35%
100%
Number of well-protected historic sites
7,000
7,900
Recurrent levy on income tax
$0
$??
Table 10.2 Charge 2: introducing Option 4
those who live in NSW and pay income tax, and would be a set amount per year, paid as part of the income tax return system. They were then told: ‘Another way of raising extra funds would be to introduce entrance fees for more national parks or to increase those fees that already exist. Unfortunately, opportunities to significantly increase the funds raised from these fees are very limited. The costs of collecting entrance fees are often higher than the money collected. For this reason, entrance fees are not considered in this charge. We have put together a new option, Option 4, that involves a substantial increase in expenditure for all five park management activities.’ The results of the straw preference poll were used in establishing the priorities for this option. The jury was told that what Option 4 would cost to implement is unknown. However, it would cost a lot more money than the NSW National Parks and Wildlife Service currently has to spend on these activities. For the purposes of this scenario, jurors were asked to assume that the only way that Option 4 could be achieved is via the introduction of a park management levy. They were then asked ‘How high would a park management levy have to be before the jury would recommend Option 1 rather than Option 4?
232 Rosemary F. James and Russell K. Blamey Protection of historic sites and items (13%) 479 parks covering 5 million hectares Feral animal control (6%)
Maintenance of facilities (60%)
Fire management (15%)
Weed control (6%)
Figure 10.2 Spending on parks excluding salaries
In other words, how high would the levy have to be before the NSW public would be no better off under Option 4 than Option 1?’ The population relevant to the scenario was identified as the 3.8m population of the Australian state of New South Wales between 20 and 69 years of age. The jurors were selected using a two-phase telephone survey, in which various demographic and attitudinal data were collected. A sample was selected to be as representative as possible of the population on the basis of the latest available Australian Bureau of Statistics data on the following criteria: gender, age, place of residence, ranking of the environment in relation to other social issues, occupation, income, income source and education. One selected juror failed to attend. Jurors were provided with written instructions regarding both their roles and the roles of witnesses and with training in note-taking and strategic questioning. All travel, meal and accommodation costs were met and the jurors were paid a small sum (A$100 per day) to compensate them for out-of-pocket expenses incurred. Five witnesses with expertise in particular aspects of the charges were selected (i.e. fire, weeds, feral animals, facilities and historic sites), in addition to two general witnesses on national park management. Witnesses were provided with written instructions regarding the content of their presentations, their roles and the roles of the jurors in the jury process. The jury met over the period 5–7 October 1999. The first day of the process was devoted to preparation of the jurors and the jury for their roles over the following two days. The second day was largely occupied by presentations from and questioning of the witnesses; on the third day, the jury consulted two further witnesses and reached decisions on the two charges. The second charge was presented to the jury only after a decision had been reached on the first charge. Option 4 in the second charge was developed from the results of an initial confidential identification by the jurors of their individually preferred option from the first charge. The jurors were aware that Option 1 in each choice set represented the current situation. They were also informed that the results of this experimental application of the CJ method would be relayed to the NSW National Park Service for use in budgeting and in the design of public participation programmes.
Deliberation and economic valuation
233
Little has been documented about appropriate decision-making methods for use in citizens’ juries; there does not appear to be a method consistently applied as part of the CJ process. It was necessary, therefore, to develop a consistent, robust method for facilitating group consensus, which would prevent false or forced consensus and would allow for alternative strategies if a true consensus could not be reached. In this work, different decision-making methods and decision rules were used for the two charges. The processes followed are outlined at Figure 10.3 and Figure 10.4. For the first charge, jurors were asked to choose their preferred option privately and to provide written reasons for this choice. The results of this work were then displayed without citation of the source of the individual comments. This confidential approach was adopted in order to ensure the jurors first determined their own preferences and then considered the reasons cited by others for their choices without consideration of their authorship. This was done in order to avoid contamination of the ideas by jurors’ perceptions of the status of and personal relationships with the author. As anticipated, this was effective for a short time, after which the individual jurors in the minority (those who did not select Option 1) declared their choices and the discursive process commenced. Assessment of support for each option was then discussed openly by the group and consensus reached. When the second charge was presented, the approach adopted for the first charge was modified in anticipation of the need for the jurors to express their views regarding both the proposed introduction of some type of payment and the payment type proposed. Following considerable discussion by the jurors, public polling on the options was conducted in several stages.
Results After their initial and independent consideration of the options, nine jurors chose Option 1, one chose Option 2 and three chose Option 3 as their preferred option. After consideration of the reasons for the choice and discussion amongst the jurors, the jury reached consensus and selected Option 1 (the status quo) as the preferred option. The outcome was stated by the jury as follows: We reached a consensus on Option 1. It involved the four people who in the straw poll had expressed a preference for either Option 2 or Option 3. We listened to their points of view about the issues they had, and they agreed that, for the purpose of consensus, they would be prepared to move across to Option 1. All of us shared some of the concerns that were identified by those four people. For that reason, as part of the consensual process, we sought to add what we call a ‘rider’, or ‘tag’ on to the consensual decision, which comes in two parts: 1 that the feral animal, weed and fire experts be told that we can see benefits, based on their separate presentations to us, in their getting together. Some of the programs have some common benefits, and we think they could all be more efficient as a result of that liaison.
234 Rosemary F. James and Russell K. Blamey Framing statements presented by facilitator and project team members Silent ballot and listing of reasons for choice by jurors
Results of ballot and lists of reasons put up on noticeboards by facilitators
Consideration and discussion by jury
Open discussion of individual juror's views by jury
Consensus reached by jury
Figure 10.3 Process for decision making under ‘charge 1’
2 that we are all very concerned about the adequacy of the programs in Option 1, and that the level of funding which permits that type of combination of programs is inadequate. We would like to send a strong message to the politicians that, as we are a representative jury in the community, there ought to be more funding, which could be put to better use in developing these programs. 3 A sub-issue under that (relates) to the National Parks and Wildlife Service – it is possibly a little unfair because we have not heard from anyone from the National Parks and Wildlife Service – is that we can see from what the experts told us that better management would possibly enhance the way in which these programs work.
Deliberation and economic valuation
235
Framing statements by the facilitator and project team members Discussion re tax – occurrence methods of collection by jury and facilitator
Open discussion of individual juror's views by jury
Vote then conditional consensus reached by the jury
Figurer 10.4 Process for decision making under ‘charge 2’
The second charge required the jury to undertake a deliberative valuation. The jury entered into considerable debate, in stages as follows: (i) discussion of the charge and its interpretation – some jurors had initial difficulty with the idea of maximum willingness to pay; (ii) clarification regarding the timing of collection of the proposed levy (weekly or annual) and the eventual use of the funds collected; (iii) discussion about the way in which the levy should be determined; one juror put a contribution model7 in the jurors’ minds by stating ‘If we don’t know how much it is going to cost for each one of those (activities), how do we know the amount of the levy to impose to cover the costs?’;8 (iv) discussion about the acceptability of the levy to jurors, the type of levy (flat or scaled rate) and other potential sources of money to fund Option 4; and (v) discussion amongst pro-levy jurors about an appropriate type and amount of levy. At this point, in a public poll, the jury voted nine to four in favour of Option 4 (the ‘with levy’ option). Further discussion followed regarding the nature of the levy, during the course of which one additional juror came to support a levy. A lengthy discussion then followed about appropriate fixed levy amounts. The jurors who now supported a levy eventually voted ten to zero in favour of a proportional rather than a fixed price levy. The levy was to be calculated as a percentage of gross income. Following a discussion of appropriate percentage figures, a vote was taken to determine support for the two figures which had been proposed by the jurors, 0.1 per cent
236 Rosemary F. James and Russell K. Blamey and 0.25 per cent of gross income. A majority rule preference of a 0.1 per cent levy on income tax resulted, by a vote of eight to two. Three jurors maintained their objection to the use of a levy. Using the proposed levy of 0.1 per cent, an additional A$109.7m (1997/8 income data), equivalent to €70.2m, would be collected annually for use on national park management. Total annual expenditure on the five programmes was A$15.2m in 1999–2000.
Conclusions The conduct of this deliberative valuation exercise has raised a number of significant methodological and theoretical issues which require resolution in order that the method may be rendered more rigorous. These include methodological issues covering consensus formation, decision rules for polling or voting, equality of juror impact, and provision of information. Theoretical issues which arose concern framing, representation, non-consensual outcomes, and the economic interpretation of the results. The development of a consensual decision is considered the ideal outcome of a CJ, although alternative approaches can be accommodated. The selection of an appropriate concluding process is necessarily strongly influenced by the wording of the charge; as is illustrated by the case studies discussed by Coote and Lenaghan (1997) it may involve voting, listing of information or some other method. The importance accorded to consensus in the CJ method suggests a need for clarity and exactitude in identifying consensus. When is a consensus a consensus? Consensus requires that the eventual decision is agreed to by all members of the group (Center for Conflict Resolution 1981). As this jury considered the first charge, it was apparent that jurors differed markedly in their degree of participation in the discussion of the options. This is hardly surprising; the same situation pertains to most circumstances where a group is given a complex task to complete. However, the task in a CJ is to reach a consensual decision; thus there should be clear evidence that the group decision was willingly accepted by all and that apparently consensual outcomes are not false. Factors which may lead to false consensus include: compliant behaviour by jurors whose initial views are in the minority, driven by, for example, a need to accord with a group norm or with the views of particular jurors (conformity bias); and failure to participate in the process by individual jurors (due to, for example, lack of communication or interpersonal skills, or to irritation or boredom with the process). A protocol to reach consensus and robust methods of testing the validity of apparently consensual outcomes are required in order to prevent acceptance of false consensus. Such testing should necessarily be capable of distinguishing between the ‘workable agreements’ referred to by Dryzek (2000: 48), arrived at by participants agreeing to a common outcome for different reasons (i.e. content-related) and an apparent but false consensus developed in response to some failure within the process of a particular CJ in response to factors such as juror boredom (i.e. process-related). One possible approach would be to examine the decision-making processes of individual jurors by means of a post-decision questionnaire. Further, juries may fail to arrive at a consensus. At what point should
Deliberation and economic valuation
237
the possibility of a non-consensual decision be revealed to the jury? If this is done too early in the deliberation phase, abandonment of the consensus-seeking process may result. If done too late, a false consensus may have been reached. In addition, at what point in the deliberation and under what conditions should the jury be encouraged to implement other decision rules? What alternative decision rules should be used? As noted above, if a jury fails to reach a consensus, alternative decision rules are necessary. If these involve voting or polling, care is required to establish an appropriate procedure, as the polling process may affect the outcomes of a CJ significantly. There are four major methodological aspects of polling which require consideration: the form of the question posed, the polling method used, the structure of the deliberative and polling processes and the decision rule employed. The question to be considered may be framed in one of several forms, dependent in part on the construction of the charge. Possible approaches include: a choice between ‘yes’/‘no’/‘abstention’; forced choice ballots, in which all answers but ‘yes’ are taken as ‘no’; and various forms of rating and ranking (Gastil 1993: 56–60). The polling method used – confidential or public polling – may also influence poll results. The use of confidential straw or preliminary polls (as used in initial consideration of the first charge in this CJ) would seem to offer advantages, by allowing jurors to make their initial choice uninfluenced by the choices of others. The use of a subsequent anonymous tabling of reasons for these preliminary choices permits analysis of those reasons unclouded by social dynamics associated with the status of and personal feelings towards the source. The structure of the deliberative and polling processes may also influence the poll outcome. For example, multiple polls separated by deliberative phases are more likely to lead to consensus on controversial issues than a single poll, by allowing those holding differing views to consider and reflect on both their own views and those of other jurors. The decision rule to be applied also requires consideration. Should a simple majority vote, some proportional threshold or some other measure be used? The decision on this aspect of polling will clearly exert a strong influence on the poll results. Bohman noted that deliberation may be considered to have been successful when ‘participants in the joint activity recognize that they have contributed to and influenced the outcome, even when they disagree with it’. He considered that equality, at a minimum, required ‘equal standing and effective voice’ for every citizen, both in terms of access to deliberative processes and participation within specific processes (Bohman 1996: 33, 35). A lack of within-process equality was evident in this CJ. Preliminary analysis shows that, of the thirteen jurors, one generally spoke only when directly questioned and several others made minimal contributions to the discussions. Whilst it might be that these jurors were not distressed by their low profile in the jury and their minimal impact on the outcome (and may have in fact agreed with the decisions reached), their lack of contribution suggests a less than ideal deliberative process. The sources of this within-process inequality lie in the wide range of social and cognitive skills and capacities, and of prior knowledge, found in a jury derived from effective sampling of a diverse
238 Rosemary F. James and Russell K. Blamey population. To what extent it is desirable and possible to reduce this inequality is a fundamental issue affecting the legitimacy of both deliberative and nondeliberative public participation methods. Two key aspects of the provision of information to jurors are: the appropriate level of juror access to witnesses; and how best to deal with inaccurate or misleading information. Witnesses will vary both in their personal appeal to the jury and the level of interest engendered by their subject matter. Hence some witnesses may be sought after by the jury to provide additional information. Equality of access appears reasonable on procedural grounds. However, the purpose of the witness presentations is to provide the jurors with the information they consider necessary to address the charge. Hence, one could argue for unlimited jury access to each witness. In adversarial citizens’ juries, however, interested parties who consider themselves disadvantaged by relative lack of access to the jury could claim the outcome was biased. On the other hand, refusing the jury access to additional information from witnesses could be considered to subvert the jury process, by forcing jurors to reach a decision based on potentially inadequate information. Further, witnesses will vary in their capacity to provide information to the jury, in terms of accuracy, relevance and pedagogic effectiveness. Means are required to deal with these potential distortions in information delivery. The provision of a neutral expert to assist the jury in assessing information provided by witnesses and the use of some type of discovery process (as used in the legal system) have been proposed as possible mechanisms for dealing with inaccurate, misleading or irrelevant information (Crosby 1995). The effectiveness of these approaches has yet to be assessed in practice. A citizen rather than a consumer frame is considered appropriate for most citizens’ juries, although it may produce results differing from the estimates of Hicksian measures of welfare change sought for environmental CBA. A citizen frame, for example, encourages participants to think about costs and benefits in aggregate societal terms which may not fit well with the notion of individual WTP. Further, to the extent that individuals do consider costs and benefits at the individual level, they may adopt a contribution model rather than the purchase model typically assumed in environmental economics. This is problematical for environmental CBA, as are several other features of citizen responses to valuation questionnaires (Blamey 1996, 1998). In the study reported in this chapter, the charges were framed so as to have jurors provide WTP estimates according to a purchase model and hence potentially of use to environmental CBA, whilst maintaining a citizen rather than consumer perspective. A CJ engaged in deliberative valuation may provide either a maximum WTP or a minimum WTA figure for some course of action or programme involving altered levels of environmental goods and services. How should that figure be extended to reflect the societal valuation? The most obvious approach to this issue is to extrapolate the figure across the relevant population, as was done for this CJ. However, the question of the degree of representation of the sample arises in this context. If the jurors are acting fully as citizens – behind a Rawlsian veil of ignorance – and have perfect knowledge of the full range of relevant societal circumstances
Deliberation and economic valuation
239
and views, then representation is of limited importance in terms of determination of the societal view on the issue under consideration.9 That is, if the construction of the citizen frame of reference has been successful and the jurors are clearly deliberating and deciding in a citizen rather than an individual consumer context, the influence of juror characteristics may be low, if sufficient information about the relevant circumstances of other citizens is available to them. Relevant knowledge would include details of the costs of substitute and complementary goods and services and the amount of discretionary expenditure possible for all those potentially liable to pay for the posited environmental change. This is of particular importance where the jury is not representative of the population in terms of factors such as income which may exert a strong influence on WTP or WTA.10 If this is not the case, however, and a jury is selected which is not representative of the wider population in terms of income and other characteristics and has less than perfect knowledge of societal circumstances, biased WTP or WTA estimates may result. These estimates may not reflect the WTP/WTA which would apply to the population under the same conditions. The importance of this effect will be determined by several factors, including: the degree to which the jurors are acting as citizens; their knowledge of the relevant circumstances of the population; and the impact of any ignorance of relevant circumstances on the resultant collective estimate of WTP or WTA. Decisions reached via voting are unlikely to satisfy the tenets of neoclassical economics because the strength of each juror’s preference is not taken into account in reaching an outcome. An argument can perhaps be made for the use of some kind of averaging process under these circumstances. In this CJ, three jurors were not in favour of any payment and ten were in favour of a payment of 0.1 per cent of taxable income. Should an average WTP be determined in those circumstances?11 Or should the majority view of 0.1 per cent be accepted and used in estimation of the societal WTP? The interpretation of WTP recommendations reached by consensus raises some issues also associated with results from applications of stated preference methods. These include: the transferability of results; the consensual WTP determined by a jury represents the maximum levy any individual would agree to after having participated in a CJ on the topic under the same conditions of information, deliberation, decisiveness and responsibility; miscounting of benefits and costs; altruistic motivations and inclusion of bequest values by jurors may result in biased estimates of WTP (see Quiggin 1998 and Milgrom 1993). For reasons such as these, the results of deliberative valuation exercises may be better interpreted in terms of social welfare functions and social optimality than individual utility functions and Pareto optimality. Using this approach, the results of a deliberative valuation WTP exercise should be interpreted as the maximum amount of money which could be charged to members of the public using a given payment vehicle in order to obtain the posited environmental improvement programme without leaving social welfare as a whole worse off than prior to the change. This is an advantage of deliberative valuation over stand-alone applications of stated preference methods such as the CVM.
240 Rosemary F. James and Russell K. Blamey The application of the CJ method to environmental decision-making and specifically to deliberative valuation of environmental goods and services has been considered in this chapter. The use of deliberative valuation clearly has potential to overcome some of the deficiencies evident in valuation methods such as the CVM. The provision of information, and of time and opportunity for deliberation, allow for the development and expression of more robustly-held preferences than is considered to be the case in both CVM and choice modelling. However, some of the difficulties associated with the CVM and, to a lesser extent, choice modelling, still remain with deliberative valuation. Further, the use of the CJ method in deliberative valuation raises some additional methodological and theoretical issues. Aspects including representation, management of the jury process to ensure equity of participation, robust and effective decision rules and minimization of various forms of bias, all require the development of appropriate protocols. It is likely that research into the management and behaviour of legal juries will prove to be a rich source of procedural guidance in regard to these matters, particularly in relation to the conduct of adversarial citizens’ juries. There is also no doubt that potential future litigation surrounding the results of adversarial citizens’ juries will provide much impetus to the development and refinement of the method.
Acknowledgements Mick Common of the University of Strathclyde originated this project whilst at the Australian National University and has provided valuable advice during its course. Michael Lockwood of Charles Stuart University and Rebecca Smith of the Australian National University assisted in the conduct of the CJ. Mick Common, David Stern and Jonathan Aldred provided useful comments on this material.
Notes 1 Writing with reference to aid projects, Slocum and Thomas-Slatyer (1995: 3) defined public or popular participation to be ‘… active involvement of people in making decisions about the implementation of processes, programmes and projects which affect them’. The definition applies equally well to projects in developed countries. 2 Whilst choice modelling may provide a means to address some of the flaws cited with the CVM, the extent to which this is so remains to be established. To date, choice modelling in environmental applications has not been subject to the same degree of rigorous examination and litigious critique as CVM. 3 Whilst these terms tend to be used interchangeably, Dryzek’s (1990) discursive theory has stronger roots in the critical theory of Habermas (Dryzek pers. com.; Blaug 1999). Dryzek (2000) discusses the distinction in some detail. 4 Constructions (i), (iii) and (iv) may result in adversarial proceedings, in which the witnesses and perhaps the jurors perceive advantage from strategic behaviour and act accordingly. Hence careful management of the jury process is necessary in order to minimize bias. 5 The term was used to describe several categories of land managed by the NSW National Parks and Wildlife Service including national parks, nature reserves, state recreation areas and regional parks. 6 Fire management in this context referred to the preparation for wildfires; this included training of staff, controlled burning and the maintenance of fire trails and fire breaks. Weed and feral
Deliberation and economic valuation
7
8 9 10
11
241
(or pest) animal control are necessary in order to protect the natural biophysical resources of the national parks from damage. Repeated framing was necessary to ensure the jurors were operating on a purchase rather than a contribution model. The distinction, as enunciated by Kahneman et al. (1993), is critical; respondents must focus on the amount they are WTP or WTA when purchasing the posited environmental change, in order to enable estimation of societal WTP or WTA. If respondents instead adopt a contribution model, under which they nominate their willingness to contribute to achieve environmental change, the resultant amounts cannot be aggregated to provide estimates of the usual Hicksian measures of welfare change. The jurors were unaware of the current levels of expenditure on the five management activities during their consideration of the two charges. It may, nonetheless, be of political importance in gaining widespread acceptance of the results of the jury’s deliberations. For example, this jury was under-representative in the lowest age (20–29 years) and income (less than A$10,400 per annum) categories. Can the WTP figure from this jury be validly extrapolated to the population? In this case, it is perhaps considered legitimate to do so, as the amount to be paid by each individual was set at a percentage of taxable income. In situations where juries opt for absolute payments rather than payments scaled in relation to income, however, the development of societal WTP or WTA by extrapolation from a jury value is more questionable. Using this approach, a per-taxpayer contribution of 0.077 per cent of taxable income resulted. The resultant WTP for the improvements posited under Option 4 would be A$84.5m.
References Aldred, J. and Jacobs, M. (2000) ‘Citizens and wetlands: evaluating the Ely CJ’, Ecological Economics, 34 (2): 217–32. Blamey, R.K. (1996) ‘Citizens, consumers and contingent valuation: clarification and expression of citizen values and issue opinions’, in W. Adamowicz, P. Boxall, M.K. Luckert, W.E. Phillips and W.A. White (eds) Forestry, Economics and the Environment, Wallingford, UK: CAB International: 103–33. Blamey, R.K. (1998) ‘Contingent valuation and the activation of environmental norms’, Ecological Economics, 24: 47–72. Blamey, R.K. and Common, M. (1992) ‘Sustainability and the limits to pseudo-market valuation’, in M. Lockwood and T. De Lacy (eds) Valuing Natural Areas: Applications and Problems of the Contingent Valuation Method, Albury, Australia: Johnstone Centre, Charles Stuart University: 117–46. Blamey, R.K. and Common, M. (1999) ‘Valuation and ethics in environmental economics’, in J. van den Bergh (ed.) Handbook of Environmental and Resource Economics, Northhampton: Edward Elgar: 809–23. Blaug, R. (1999) Democracy, Real and Ideal: Discourse Ethics and Radical Politics, Albany, NY: State University of New York. Bohman, J. (1996) Public Deliberation, Cambridge, MA: MIT Press. Brown, T.C., Peterson, G.L. and Tonn, B.E. (1995) ‘The values jury to aid natural resource decisions’, Land Economics, 71(2): 250–60. Center for Conflict Resolution (1981) ‘Building united judgement’, Madison, WI: Center for Conflict Resolution. Coote, A. and Lenaghan, J. (1997) ‘Citizens’ juries: theory into practice’, London: Institute for Public Policy Research. Crosby, N. (1991) ‘Citizens’ juries as a basic democratic reform’, Minneapolis, MN: Jefferson Center.
242 Rosemary F. James and Russell K. Blamey Crosby, N. (1995) ‘Citizen’s juries: one solution for difficult environmental questions’, in O. Renn, T. Webler and P. Wiedemann (eds) Fairness and Competence in Citizen Participation: Evaluating Models for Environmental Discourse, Dordrecht: Kluwer: 157–74. Crosby, N. (1996) ‘Creating an authentic voice of the people’, paper presented to the Panel on Deliberation on Democratic Theory and Practice, Chicago, IL: Midwest Political Science Association. Daneke, G.A., Garcia, M.W. and Delli Priscoli, J. (eds) (1983) Public Involvement and Social Impact Assessment, Boulder, CO: Westview Press. Dienel, P.C. (1989) ‘Contributing to social decision methodology: citizen reports on technological projects’, in C. Vlek and G. Cvetkovich (eds) Social Decision Methodology for Technological Projects, Dordrecht: Kluwer: 133–51. Dryzek, J.S. (1990) Discursive Democracy: Politics, Policy and Political Science, Cambridge: Cambridge University Press. Dryzek, J.S. (2000) Deliberative Democracy and Beyond: Liberals, Critics, Contestations, Oxford: Oxford University Press. Gastil, J. (1993) Democracy in Small Groups: Participation, Decision Making and Communication, Philadelphia, PA: New Society Publishers. Gregory, R. and Slovic, P. (1997) ‘A constructive approach to environmental valuation’, Ecological Economics, 21: 175–81. Gregory, R., Lichenstein, S. and Slovic, P. (1993) ‘Valuing environmental resources: a constructive approach’, Journal of Risk and Uncertainty, 7: 177–97. Jacobs, M. (1997) ‘Environmental valuation, deliberative democracy and public decisionmaking institutions’, in J. Foster (ed.) Valuing Nature? Ethics, Economics and the Environment, London, UK: Routledge: 211–31. Jefferson Center (1995) Citizens’ Jury on Hog Farming: Final Report, Minneapolis, MN: Jefferson Center. Jefferson Center (1996) Citizens’ Jury on Comparing Environmental Risks, Minneapolis, MN: Jefferson Center. Jefferson Center (1997) Citizens’ Jury on Dakota County’s Comprehensive Land Use Plan, Minneapolis, MN: Jefferson Center. Kahneman, D., Ritov, I., Jacowitz, K.E. and Grant, P. (1993) ‘Stated willingness to pay for public goods: a psychological perspective’, Psychological Science, 4–5: 310–5. Milgrom, P. (1993) ‘Is sympathy an economic value? Philosophy, economics and the contingent valuation method’, in J.A. Hausman (ed.) Contingent Valuation: A Critical Assessment, Amsterdam: North-Holland: 417–40. O’Neill, J. (1997) ‘Value pluralism, incommensurability and institutions’, in J. Foster (ed.) Valuing Nature? Ethics, Economics and the Environment, London: Routledge: 75–88. Perman, R., Ma, Y., McGilvray, J. and Common, M. (1999) Natural Resource and Environmental Economics, New York: Pearson Education. Quiggin, J. (1998) ‘Existence value and the contingent valuation method’, Australian Economic Papers, 37(3): 312–29. Rosenbaum, N. (1978) ‘Citizen participation and democratic theory’, in S. Langton (ed.) Citizen Participation in America, Lexington, KY: Lexington Books: 43–54. Rosenbaum, W. (1979) ‘Elitism and social participation’, in Lincoln Filene Center for Citizenship and Public Affairs Citizen Participation Perspectives, proceedings of the National Conference on Citizen Participation, Tufts University, Washington DC, 28 Sept.–1 Oct. 1978. Rousseau, J. (1968) The Social Contract, Harmondsworth: Penguin.
Deliberation and economic valuation
243
Sagoff, M. (1998) ‘Aggregation and deliberation in valuing environmental public goods: a look beyond contingent valuation’, Ecological Economics, 24: 213–30. Slocum, R. and Thomas-Slatyer, B. (1995) ‘Participation, empowerment and sustainable development’, in R. Slocum, L. Wichhart, D. Rocheleau and B. Thomas-Slatyer (eds) Power, Process and Participation – Tools for Change, London: Intermediate Technology Publications: 3–8. Thomas-Slatyer, B. (1995) ‘A brief history of participatory methodologies’, in R. Slocum, L. Wichhart, D. Rocheleau and B. Thomas-Slatyer (eds) Power, Process and Participation – Tools for Change, London: Intermediate Technology Publications: 9–16. Van Valey, T.L. and Petersen, J.C. (1987) ‘Public service science centers: the Michigan experience’, in J. DeSario and S. Langton (eds) Citizen Participation in Public Decision Making, Westport: Greenwood Press: 39–63. Webler, T. and Renn, O. (1995) ‘A brief primer on participation: philosophy and practice’, in O. Renn, T. Webler and P. Wiedemann (eds) Fairness and Competence in Citizen Participation: Evaluating Models for Environmental Discourse, Dordrecht: Kluwer: 17–33.
244 Katja Arzt
11 The challenges of stakeholder participation Agri-environmental policy Katja Arzt
Introduction Some environmental economics approaches based on neo-classical theory attempt to reveal people’s values related to the natural environment by aggregating single statements, e.g. the contingent valuation method (CVM). Such approaches are based on the assumption that a person has complete information about the natural good at stake and that they already have formed preferences. However, people tend to form values and attitudes over time and may change their value perception of the natural environment as their knowledge and experiences increase (e.g. Jiggins and Röling 1998; O’Hara 2000). The case study presented in this chapter is based on the hypothesis that values for nature emerge from interactions between people (see O’Connor 1998) and could therefore be linked to Sahlins’s (1976) remark that valuation approaches should be recognized as a subjective, cultural and contextual phenomenon. This study is based on the observations that ecosystems are dynamic and that many of their changes are unpredictable (see Hanna et al. 1996; Berkes and Folke 1998; Holling et al. 1998). Therefore an optimal solution for the management of different ecosystems cannot be presumed beforehand (see Munda 2000). Complexity and unpredictability in physical and social systems has thus to be considered when valuing environments. An ‘interactive valuation process’ is here defined as different stakeholders concerned with a particular environmental issue coming together to exchange their views and values in order to decide together about future programmes, plans or targets for managing the environment. This kind of decision making pays attention to ‘plural rationality’ (Söderbaum 1999) and thus could be termed a participatory valuation process as different orientations and attitudes are to be considered beside monetary assumptions. A discursive process will enable individuals, in groups, to construct values rather than express prior preferences and is hence a deliberative approach which leaves it to the respondents to clarify among themselves explicitly what they are valuing and why. The deliberative group can define precisely which aspect(s) of the public good it values and by how much (Sagoff 1998: 224). According to Sagoff (1998: 227): ‘If individuals … do not come with predetermined preferences but must construct them, then the process of construction may legitimately involve social learning, since this is precisely what
The challenges of stakeholder participation 245 occurs in other contexts in which people work out their values’. Hence, it is hypothesized in this chapter that a mutual learning process is central for a successful decision making process. If different stakeholders learn about each others’ perceptions and sets of values it is more likely that their actions can be co-ordinated towards mutual agreements. As Ostrom (1998a: 71) has already explained for common-pool resources: ‘When all appropriators of a common-pool resource share a common set of values and interact within a complex set of arrangements, there is a much greater probability that they will develop adequate rules and norms to manage resources’. This study supports that learning refers to the accumulation of insights into system causes (see Meppem and Gill 1998). The intention is to promote the development of a more unified and shared impression of the underlying processes of cause and effect that describe the issues and problems under investigation. ‘Experts’ and citizens alike can be instrumental in the general learning process. This general consultative process is proposed as an alternative to the more exclusive, fundamentally ‘expert’ driven, and certainly less learning-oriented, decision-making framework of neo-classical economics (Meppem and Gill 1998; Jiggins and Röling 1998). Learning to communicate about values needs a process which goes beyond the supply of information. DeBono (1993, quoted in Meppem and Gill 1998: 118) states: ‘There are people who believe that if you get enough information then the information will do your thinking for you, however, it is the concept through which we perceive the information that gives it any value’. The interactive valuation process brings different orientations together; some participants may join the process because of individual interests while others may do so for altruistic reasons. Hence, it is the aim of the ‘learning’ institution that the limited number of participants in that deliberative approach act as citizens as opposed to individuals who try to choose an outcome most benefiting their individual interests. In other words, consumer preferences reflect conceptions of the good life that individuals seek for themselves, while citizens’ preferences reflect conceptions of the good society offered for the consideration and agreement of others (Sagoff 1998: 215). A well-performed learning process is one where the citizens’ preferences are emphasized; decision in the learning process may depend more on the force of the better argument about public interest (Habermas 1982). Discourse processes attempt to bring such different orientation together, and constitute a conscious decision process, in which individuals may reconsider their way of thinking (O’Hara 1999, 2000). A decentralized institution for valuation may therefore have the following positive impacts compared with those valuation methods which are based on neo-classical theory: it is a ‘learning institution’ in which new experiences and knowledge as well as local knowledge can be integrated continuously; decision results of such an institution are based on norms and rules which correspond to the local, social and natural environment; results can therefore be better accepted and transformed into action. An interactive valuation process which aims for sustainable solutions and sustainable actions needs an institutional design, capable of achieving a mutual learning process. At present different institutional settings for interactive decision
246 Katja Arzt making are being discussed. Different tools – such as citizens’ juries, consensus conferences, roundtable groups, and focus groups – are used for different purposes; some exist only for a short time to solve specific problems within communities while others persist over time in order to be established as a new democratic institution. The aim of this study is therefore to find the most significant circumstances aiding the creation of an institutional arrangement for decision making at the local level. The first section draws some theoretical assumptions for a successful institutional design. The second section presents a research project in northeast Germany, which established a ‘roundtable group’ in order to decentralize decisions about agri-environmental policies and schemes. In the conclusion the interactive valuation process in this project is argued to have been only partially successful due to an inappropriate institutional setting.
An analytical framework for interactive valuation processes A successful institutional design for interactive valuation based on mutual learning is defined here as one which exists over a period of time, is able to create sustainable decisions which are also accepted at higher levels of authorities and leads to actions to improve the environment. Such an arrangement should be flexible in order to react towards changing social, economic and environmental conditions. It is successful when it meets the needs for a democratic institution in which different stakeholders are involved deliberately. Thus success means airing and assessing (and possibly combining) different views rather than finding an ‘optimal’ solution of one underlying perception or view (for instance a result of a cost-benefit analysis, or a naturalistic view). In this respect the process which leads to the decision is important. This definition of success corresponds with Munda (2000: 12), who says that processes for environmental valuation should be considered under a procedural rationality, in which the process is evaluated: ‘Since an optimal solution is non-existent, the important factor is the quality of the process leading to the decision. Optimal solutions are hence replaced by satisfactory solutions’ (Simon 1983). In order to create an analytical framework, a theory is needed which makes assumptions about the way a group can reach mutual learning to attain satisfactory decisions and actions. Here Ostrom’s (1998b) development of ‘second generation models of rationality for social dilemma situations’ is used which is a behavioural approach to the rational choice theory of collective action. Although the situation discussed in this chapter differs from a social dilemma situation, the assumptions made by this theory help to understand when co-operation in small groups can take place, hence a mutual learning process is initiated which will lead to a successful institutional design. Ostrom analysed different laboratory research results of behaviour in social dilemma situations and concluded that the core elements of co-operative behaviour are trust, reputation and reciprocity. Only if those elements are developed positively within the group is co-operation possible. Ostrom defined trust as the expectation
The challenges of stakeholder participation 247 of one person about the actions of others that affects the first person’s choice, when an action must be taken before the actions of others are known. In the situation of decision making for environmental goods, individuals may agree upon certain programmes or measures only when they are sure that other members will also make compromises. For example, farmers might agree on special measures which will be good for certain birds. However, they may only agree upon a programme if they have the commitment from the government for monetary compensation. In situations where trust and reciprocity among the participants is difficult or impossible to create, then it is likely that people will not share information with others in order to gain advantages. This phenomenon is quite common during the implementation of agri-environmental policies. Farmers, for instance, are knowledgeable about their land and its vegetation. Also, they often realize that administrators know very little about it and use this fact to accept payments for extensions, although they would never seriously consider using this land for intensive cultivation (Hagedorn 2000). Trust affects whether an individual is willing to initiate co-operation in the expectation that it will be reciprocated. According to Ostrom (1998b), rational individuals enter situations with an initial probability of using reciprocity based on their own prior training and experience. At the core of a behavioural explanation are the links between the trust that individuals have in others, the investment others make in trustworthy reputations, and the probability that participants will use reciprocity norms. This mutually reinforcing core is affected by structural variables as well as participants’ past experiences. This study concentrates on the following structural variables which feed into that mutually reinforcing core: continuous discourse, participating stakeholders, flow of information, property right structure of the issues discussed and decision procedure. One important structural variable is the possibility for the participants to have face-to-face communication (Ostrom 1998b; Ostrom et al. 1994). With a repeated chance to see and talk with others, a participant can assess whether they trust others sufficiently to try to reach a simple contingent agreement regarding the level of joint effort and its allocation (Ostrom 1998b). Continuous discourse therefore raises the possibility to co-operate, in order to exchange mutual commitments, creating and reinforcing norms and developing group identity. Given the complexity of the physical world that individuals frequently confront, they are rarely ever able to get the rules ‘right’ on the first or second try (Ostrom 1990). In highly unpredictable environments a long period of trial and error is needed before individuals can find rules that generate substantial positive returns over a sufficiently long time-horizon (Ostrom 1998b). Another variable which feeds into the mutually reinforcing core is certainly the mixture of stakeholder representatives participating in the process and what knowledge they have from past experiences (see also Ostrom 1998b). Participants for the interactive valuation process should come from different spheres of society in order to be able to meet the interests of different stakeholders. On the one hand it is possible to involve stakeholders who are already elected by parties, clubs or hold certain administration positions; on the other hand stakeholders can be people who may not be linked to any constituency but are closely working with nature
248 Katja Arzt and are affected by ecological changes. Representatives should act for the group they present; however, this could make them quite inflexible in changing their point of view or in their decision competence, because they may come with only limited decision abilities. Nevertheless, representatives have the respect of the people they stand for, and it might be more likely that their decisions are accepted by others. Yet, a learning process with many elected representatives tends to consolidate the hegemony of the expert in decision making (Meppem and Gill 1998; Röling and Wagemakers 1998). On the contrary, stakeholders who are not representatives of an organization but are interested in discussing different issues might have a greater personal interest in the problem at hand, and might be better informed or interested in others’ perceptions. However, it is essential for an interactive valuation process that decisions are made by those who will bear the main consequences (O’Hara 2000; O’Conner 1998; Meppem and Gill 1998; Jiggins and Röling 1998). Most studies of deliberative institutions so far claim that only a limited number of stakeholders can be involved in an interactive decision-making process; a group of over twenty people is difficult to moderate for the purpose of a learning process. In this respect the selection of the ‘right stakeholders’ can be difficult, and so far no institutionalized methods (e.g. election process) exist to choose them. Many organization teams working with deliberative approaches try to inform as many people as possible about the process and then leave it to ‘the community’ who will participate. However, some thought should be given to the ability of the stakeholders to be involved in a process. Ostrom (1998a) stated that getting the institutions right is a difficult, time-consuming, conflict-invoking process. Time availability is certainly a limiting criteria; people who have to work long hours are unlikely to join the process over a long period, or only if the issue in question is in their own interests or of strong concern. Gender aspects are to be considered too, as for instance mothers with young children have disadvantages in joining processes which often take place in the evenings and during weekends when public child care is unavailable. People from big organizations or from governmental authorities, on the contrary, are likely to join the process, as it may correspond with their duties; they do not tend to use their spare time for the process. Still, their presence seems to be important because it assures that the process of decision making is already linked to existing institutions and there is a greater chance that the outcome of the deliberative process will be accepted at higher levels if authorities are involved from the beginning. The process can also be driven by the participation of ‘innovative’ local persons who are active and well respected by different stakeholders (Kolloge 2000). Generally, stakeholders only participate when they gain benefits from doing so. In the case of an interactive valuation process the benefits are nonmonetary in the short term; on the contrary, the participants have to invest time. However, most people will join the process when they are dissatisfied with the present situation, hope to gain a better image or would like to improve their knowledge. Once the process shows positive results more sceptical people may join the process. The kinds of benefits people expect when joining the process will feed considerably on the core relationship of trust, reputation and reciprocity.
The challenges of stakeholder participation 249 The flow of information is a key aspect of a learning organization and should be organized in a way that it internalizes different value aspects and makes information transparent for all stakeholders. Professional moderation and visualization techniques may be one helpful method, another is the use of language. In this respect, O’Hara (2000) states: Fair and competent discourse must mean giving particular attention to the differences in verbal skill among discourse participants. The distribution of verbal skill can become just as much a source of power and distortion as the distribution of wealth in price based valuation. The broader rationality of discursive reason may not at all be supported by featuring culturally rewarded skills like articulated speech, deductive logic, or abstract thinking to the exclusion of less articulate, empathetic, or traditional thinking. The more concrete the objects of decisions and the more obvious the consequences of action, the better is a mutual learning process initiated. Environmental problems which are worked out in a learning process should be obvious for all participants. Only then will these instruments be better than existing institutional arrangements. The source of information is also important for the learning and trust-building process. Some sociological studies in the UK (see Clark and Murdoch 1997) showed that in the past farmers lost trust in scientific results and ideas. However, those findings cannot be generalized; in other areas it may be observed that farmers depend on and use scientific research. In any case, while implementing a valuation process, researchers should be aware of the information used, as Bernabo (1998b) observed: ‘When researchers apply assessment models as mechanistic predictive tools, they risk producing results that are not sufficiently relevant and are prone to misuse in decision making (Bernabo 1998a). Policy-relevant assessment is a more organic process involving researchers and stakeholders in an interactive exercise that builds understanding rather than trying to define “truth”. The emphasis must be on providing practical information in an easily usable form.’ We may find that participants decide easily on one issue, but fail on another. Property rights theory certainly gives useful interpretations as to why people value items the way they do, as the owner of rights of a special natural good will have the pleasure of the corresponding utility, or, on the contrary, if the corresponding right is a duty, the owner has to carry the negative effects and costs (Hagedorn 2000; O’Neill and Spash 2000). For example, property rights on natural goods such as soil or water do not have homogeneous property right structures, but are differentiated by their different ecological characteristics; correspondingly the cost and benefits will vary, leaving room for different institutional designs of duties and rights (see Hagedorn 2000; Bromley 1997, 1998; Ostrom 1998a). The learning process should clarify through discourse which property right structures are connected to which participants and how much room is left to change the institutions. Lawrence (2000: 16) suggested to distinguish between ownership, control and use of property resources in environmental policy definition and applications. Soil erosion, for example, is different for farmers who actually are
250 Katja Arzt the owners of land as opposed to being tenants. Owners may have an interest in keeping erosion as small as possible to maintain the fertility of the land for their children or for future sales, whereas tenants may tend to reduce the fertility in the short run for the time of the contract. However, if arable land should be converted into meadow for ecological reasons, the owner of the land may not agree to convert the land, because the rent received will be less for meadows. In this case the property right structures are difficult to overcome and a decision will not be easily obtained. In order to build reciprocity within the group it seems necessary that the property right structures are clarified. This also corresponds with the control mechanisms. Only if participants are convinced that breaking the rules about the decisions made by the group (e.g. to reduce nitrogen fertilizer) can be controlled, is there then a chance that the group formulates rules and norms which consolidate trust, reciprocity and reputation. The urgency of a problem can also be an important factor for participants reaching consensus. For example, if farmers and environmentalists of a certain region have problems with the water regime, the farmers would like to reduce the water level in order to keep on farming, whereas the environmentalists would like to have a higher water level to increase biodiversity. For both, the situation is urgent and needs a solution in order to plan future activities. In this case the participants may invest more time and money to reach a consensus compared with a situation where nobody’s activities are considerably harmed. Apart from the learning process the decision procedure itself strongly influences the outcome. In the case of a roundtable group, decisions are made by consensus, and not by voting or other mechanisms. Consensus means that all participants have to agree upon the results by accepting the arguments addressed during discussion. In this case verbal skills and power structures may have an important impact. Also analytical methods such as multi-criteria analysis which recognize the plurality of values that inform environmental choices can be combined with more deliberative methods that enable citizens to articulate values and which are sensitive to concerns about fairness in procedures and outcomes (O’Neill and Spash 2000). In Table 11.1 the above-mentioned variables and their characteristics are summarized. The next section will analyse a project that was carried out in Germany to see how far it meets the idealized model presented here. It will do so by evaluating the variables described above. The data for the criteria were collected from qualitative interviews before and during the process and from meeting observations.
Conducting Brandenburg’s roundtable The GRANO1 project established a roundtable group to carry out an interactive valuation process in order to decentralize a decision about agri-environmental schemes and measures. The project’s aim is to test institutional approaches for interactive valuation in order to make statements about their ability to be transferred to other regions. The idea for the project arose from different stakeholders in the region of Brandenburg who were dissatisfied with the present situation of topdown policy approaches, because the existing agri-environmental schemes were
The challenges of stakeholder participation 251 Table 11.1 Variables feeding into trust, reciprocity and reputation Continuous discourse
Stakeholder
Information flow and information
Issue
Decision procedure
Frequency
Representatives Time availability
Property rights considered
Considered aspects
Duration
Expected benefits Gender aspects Information about past
Adequate level of information generated Credibility of information Language Moderation and presentation techniques
Control mechanism Urgency of the problem to the stakeholders
Participation Fairness
only partially in agreement with the practice of farming or were argued not to include enough ecological benefits (Müller et al. 2000). A roundtable group operates on the basis of seeking consensus between people with different viewpoints on the issues discussed. In this approach each participant has an equal opportunity to voice their option. Those involved should speak and listen to others with a view to seeking areas of consensus. Agreements come out of the discussion process. This approach works mainly over a series of meetings. In the case of the AgriEnvironmental Forum (AEF) the discussion for each meeting was prepared by a project team and led by a professional moderator. For the project team it was important to find an institutional arrangement that is quite simple to handle and is time- and cost-efficient, because the idea of the interactive valuation institution should be transferable to other regions and should be adopted by organizations within the region. On the basis of the common regulation of the European Union (EU), Germany developed different agri-environmental schemes at the Länder (county) level. Compared with other EU countries, where schemes are developed by central governments (e.g. the Countryside Stewardship Scheme in the UK), Germany already has a more decentralized approach. However, the weight given to different environmental measures and the types of environmental consideration that are important are clearly influenced by political priorities, policy objectives and cultural attitudes to the farmed environment (see also Curry and Winter 2000). Local farmers’ or local conservationists’ opinions about the different measures for the schemes are not taken into consideration when establishing new programmes, yet different experts are consulted by the local governments when setting up schemes. In Brandenburg, two environmental schemes for farmers exist: one is called the KULAP (cultural environmental landscape programme), the other is the Contract Nature Conservation Programme. In KULAP farmers can choose among different measures and apply for them. However, due to limited funds, the scheme operates on a ‘first come, first served’ basis. KULAP is mainly financed by the EU and at a national level (see EWG VO 2078/92 [Agri-environment Regulation (EEC) No. 2078/92] and VO 2052/88 [Council Regulation (EEC) No. 2052/88]). The Contract Nature Conservation Programme is financed completely at the Länder
252 Katja Arzt level. Farmers enter into individual contracts for their farms based on discussions with environmental officers. Only farmers whose land lies within a protected area can apply for the scheme (Hampicke et al. 1998). However, decisions about the measures are not taken in consensus with different stakeholders, and the targets for the scheme are fixed by experts. For the implementation of the project, a region in the northeast of the Uckermark (Brandenburg county, 180 km north of Berlin) was chosen. The selection was made in arrangement with the local department for agriculture and environment. The region contains 11,000 ha of agricultural land, is one of the most thinly populated districts of Germany (53 inhabitants/km) and has a high unemployment rate. An important factor for the selection of this particular region was its strong agricultural character, the variety of landscapes and the low density of protected areas. The project team is interdisciplinary and consists of several natural scientists and socio-economic scientists as well as a moderator. These professionals come from different universities and institutions. The different backgrounds of the project team made it difficult to find a general understanding of the definition of ‘value’ and ‘learning processes’ beforehand. The team could not create a common understanding about the institutional arrangement as discussed in this chapter. However, they already had agreed upon a phase- and process-oriented process. The strategy was to choose relevant topics with the stakeholders and subsequently each problem was to be discussed with regard to (a) reasons for the appearance of the problem and the expected long-term consequences; (b) solutions that should be achieved, including justifications; (c) measures to achieve these solutions, including a monitoring-process; (d) financing of the measures; and (e) action planning. However, this procedure could only be partially realized, because most stakeholders considered it to be too abstract and too long. It became increasingly evident by the third meeting that the stakeholders’ main interest was to talk chiefly about measures and their financing. Therefore the project team had to plan each meeting separately. After the region was chosen the project team contacted different stakeholders for initial interviews, in order to get to know the region and its problems, and also to voice the intention of the project. The results of those interviews were presented at a community meeting, to which over 100 different stakeholders had been invited (selected using telephone books and contacts), but only 43 stakeholders attended. During that meeting the project team asked for volunteers from different stakeholder groups who would like to participate in the roundtable group. At the end 23 stakeholders wanted to attend the AEF. From October 1999 to April 2000 six roundtable meetings took place (Figure 11.1), one each month, except in the summer months when farmers are too busy on the farm to engage in other activities. The process continued until December 2001, during which time 13 meetings took place. The project team found another regional organization who organizes the meetings until the present time but with a different focus. Todays meetings are more seen as ‘information meetings’, to inform farmers and environmentalists about agropolitical subjects.
The challenges of stakeholder participation 253 35 Agricultural institutions
30
Farmers
25
Environmentalists
20
Conservation association
15
Tourism
10
Politics Administration
5
Trade
0 6
Scientists
M
ee t in g
5
4
ee t in g M
3
ee t in g M
2
ee t in g M
ee t in g M
M
ee t in g
1
Others
Figure 11.1 Participants at the first six meetings of the AEF
The actual proceedings are outlined in Table 11.2. During the first meeting much time was spent on gettting to know each other and clarifying expectations. The different stakeholders revealed their interest in starting mutual discussions in order to find solutions for regional ecological problems, but most of them were unsure about the process itself and about the functioning of a roundtable discussion. At the first meeting, the stakeholders identified the topics that should be discussed over the following months. In order to have a better understanding of the current environmental problems in the region, the scientists were asked to prepare a general lecture about the environmental situation and a more detailed lecture on landscape characteristics for the second meeting. However, it became clear in the following meeting that the different participants including the project team had different intentions in mind or understood the topics differently. Accordingly, during the second meeting – which started with a scientific lecture about the reasons for structuring the landscape (e.g. biodiversity, biotope network, and other ecological functions) – several farmers complained that the financial aspects were of major concern, but were not included in the lecture. They said for instance: ‘Well as a human I find all those aims important, but as a farmer, I cannot accept all those targets, because I must work on my fields in order to survive’, and ‘I would do anything if I got enough cash’. The participants wanted to concretize the topic (‘We should not talk so scientifically, otherwise this evening does not mean anything to me.’) and decided to talk about Sölle (water bodies developed during the ice age, which are a typical element of the regional landscape) during the next meeting. As a result of this ‘unsuccessful’ meeting in terms of a mutual learning process about the topic, the project team had to decide how the topic of Sölle could be introduced and how stakeholders could come to a consensus about a regional programme. Prior to the next meeting the organization team sent information materials about Sölle to the stakeholders.
Agreement on an agricultural programme for Sölle
Presentation and reflection of expert knowledge about soil erosion Discussion about experiences
Agreement on an agricultural programme
4. AEF 12.1.00
5. AEF 24.2.00
6. AEF 11.4.00
3. AEF 2.12.99
2. AEF 3.11.99
To become acquainted with the participants Find relevant subjects Presentation of expert knowledge Problem analysis of landscape structures To find targets Ecological information about ‘Sölle’ Agreement on an agricultural programme for ‘Sölle’
1. AEF 20.10.99
What have been the expectations of the organizers?
Questioning about attitudes about ‘Sölle’ Expert presents knowledge within a dialogue Open discussion Visualization A farmer presents his farm Presentation of different measures and their costs by the organization team Discussion about transformation of those results to other farms Discussion about a regional programme Visualization Scientific lecture A farmer presents his experiences with different measures Open discussion about measures Visualization Lecture about a method to find out the erosion potential Presentation of practical experiences Structured discussion
Questioning of expectations and backgrounds, visualization of answers Point evaluation Scientific lecture Structured discussion Visualization
What methods were used during the meetings?
Table 11.2 Proceedings at the agri-environmental forum
Method was not accepted Discussion about the usefulness of an erosion programme No agreement about a regional soil erosion programme
No comments about the lecture Confusion about proceeding
Agreement on different measures for ‘Sölle’ Scientists were asked to present a regional programme for ‘Sölle’
Good information flow about ecological targets Discussion was about proceedings at the AEF Farming examples are wanted (experiences)
Participants want a priority catalogue for the most important environmental problems Structuring of the landscape as the main subject No discussion and questions about the lectures No structured discussion about problems and targets Concretize the subject: the group wants to talk chiefly about ‘Sölle’
What came out of the discussion process?
254 Katja Arzt
The challenges of stakeholder participation 255 At the start of the third meeting all participants were asked to write down what they knew about Sölle and why they thought it was important to protect them. Then, one expert on Sölle answered questions as required and explained their ecological functions. Contrary to the expectations of the organization team, the stakeholders did not have detailed knowledge about the Sölle in their region. Instead the stakeholders expected the natural scientist to know which treatment is the best. The following discussion about a regional scheme for Sölle was unsatisfactory because different scientists dominated the discussion while debating a general proceeding (what it is important to know in order to be able to decide on a Sölle programme). This confused those participants who were only interested in the measures and actions for Sölle. The fourth meeting was very structured, dealing with different measures for Sölle as prepared by the scientists. Each measure was subsequently discussed by the experts and stakeholders. The successful connection between practice and theory corresponded especially to practitioners. At the end they had a list of different measures for Sölle and how much payment should be given to a farmer when applying each of them. The scientist for agricultural policy put those measures into a programme for Sölle which they presented at the ministry of agriculture. The group decided to close the discussion about Sölle. The next meetings dealt with soil erosion, because the natural scientists in the team found that to be the most relevant ecological problem for the region. The participants agreed to discuss this problem, although they had never mentioned its urgency or importance before. During the fifth meeting a new scientific ‘expert’ was introduced who explained why soil erosion is a common problem and how important it could be to reach consensus about that topic among the stakeholders. However, instead of creating an atmosphere of dialogue, the expert just presented results and measures and left little room for the stakeholders to talk about their problems. Instead of a mutual learning process by scientists and stakeholders it was a lecture by one expert to the ‘others’. There was also a farmer who presented his experiences with some soil erosion measures, but his findings were commented on only by the expert. The sixth meeting was attended by eleven stakeholders, with only three farmers amongst them. It dealt with measures and if and how payments for soil erosion should be directed. In this case, the scientists tried to direct the discussion towards the regional risk of soil erosion in different parts of the landscape. They showed with different techniques that soil erosion is not a problem on all farms and suggested therefore that payments should be directed to high-risk farms only. The discussion which followed the presentation of different techniques was highly controversial and dominated by the people from the agricultural administration, who discussed the pros and cons of such directed payments. The farmers themselves were quiet during the discussion and avoided direct statements. However, they complained that not enough farmers had attended this meeting, and they feared that their decisions would not be accepted. Yet, this discussion left room to express different perceptions of the problem. Although a structured discussion was planned in order to reach consensus about a regional programme for soil conservation, the stake-
256 Katja Arzt holders talked about the general usefulness of soil erosion programmes and no consensus was reached.
Analysis of the agri-environmental forum In this section I discuss how the AEF performed with respect to stakeholders and representation, issues raised, information flow, and decision procedure. The roundtable group consisted of different stakeholder groups: representatives from clubs (e.g. the farmers’ union, nature conservation club), officers from different institutions (agricultural and environmental ministries), private farmers and people interested in the subject (e.g. tourist industry). There was a great fluctuation in numbers and in backgrounds of participants during the process. A minority of the representatives did not join the discussion in a creative way but just observed the meeting(s) or made some sarcastic comments; it seemed that they were sent by their constituency to observe the process but had no abilities or consent to make decisions. Other participants, however, enjoyed the interesting mix of different stakeholders: ‘It is good that many different people come together’, ‘it is important that agriculture and conservationists come together, there is not such a mixture in other institutions’. From the local agricultural ministry two different representatives attended alternately but had different opinions about the topics discussed. The stakeholder group of farmers complained that not enough farmers joined the process and that their number was decreasing during the process. Also, not all types of farmers joined the process. At the AEF only three farmers from the former Agricultural Production Co-operative (LPG) took part, as well as newcomers to the region, who have smaller farm units. Typical for the region are farmers who worked formerly in the LPG and started their own business after the transformation process. Their farms are usually smaller in size and run by family labour, whereas the former LPG are usually very big (up to 2000 ha). However, it was very difficult for the organization team to invite farmers from those ‘smaller’ farms to the AEF. They complained about time limits. One farmer said that he was sceptical about sitting together with farmers from the big co-operatives, because he knew that those big co-operatives have more power and were always dominant in consensusbuilding processes in the past; therefore he was unwilling to come to the AEF. Another imbalance to representation was the small number of women contributing to the discussion process (3 women out of 23 participants). Women suffered most from the agricultural institutional change, leaving most of them unemployed. However, women are seldom in leading positions (in administration, on farms, in trade, etc.) and were therefore not able to speak for a constituency. Even those women who participated in the process hardly joined in the discussion. The organization team (scientists) were the dominating stakeholder group not only in number, but also in activity. As they prepared the meetings they had a broad influence on the process. However, it became clear during the process that the scientists themselves did not share a common vision about the process, because they all had different interests in the outcome.
The challenges of stakeholder participation 257 The topics discussed were not urgent problems for most stakeholders even though they chose the topics themselves, as some participants admitted during the interviews. They did not identify with the problems of Sölle. This point was difficult to handle for the organization team, because the stakeholders did not contribute their own ideas, interests and knowledge, but expected the scientists to do so. The topic of soil erosion was introduced by natural scientists but was not seen as an urgent problem by the participants. Some said that it is an issue for an information service and need not be discussed in a big group. While soil erosion control mechanisms were mentioned during the discussion, no consensus could be reached. Topics which seemed to be of great interest for most stakeholders, such as hedge planting, were not chosen in the first round of meetings because the participants mentioned the difficult property right structures. If someone wants to plant a hedge the owner and the farmer both have to be consulted; both of them might reject the planting proposal because the farmer will lose land for crop growing and the owner’s rent income will go down. Although hedges can help reduce soil erosion and improve biodiversity – important for both owner and farmer – a solution is difficult to find, unless property right structures are changed. The stakeholders criticized the lack of a clear structure; they were confused when the scientists started to discuss the general proceeding and project aims. The participants liked the fourth meeting the best, because the organization team presented a clear structure. They called it ‘effective’ working; the results were clear. The mutual learning process only began to develop during the third meeting. The stakeholders could first express their own knowledge and then the experts formulated their knowledge on demand; it was not a lecture about the ‘right’ knowledge but a communication about the problem at hand. During the second and fifth meetings, however, scientific lectures were given, leaving little room for the stakeholders to formulate their problems, but giving insight into the concept of one scientist (or scientific community). However, some participants mentioned that for them the scientific input was very important: ‘The scientist should tell us what is good for the environment’, ‘scientists should show us how to solve problems’. In particular, natural scientists seem to have a very high position to formulate environmental targets in the perception of most participants. Nevertheless, when talking about economic data, the knowledge of economic scientists was criticized. Some of the farmers complained that some of the economic data presented by the scientists for the different measures were unclear: ‘You never know what kind of data they use, … each farm is so different.’ During the meeting it could be observed that frequently scientific experts (‘professors’) dominated the discussion and if they could not attend a meeting, then people from the administration dominated the process. The participants did not complain about the language used during the process. However, people who were ‘used to talking’ at meetings/in large groups were found to be dominant. A decision procedure at the AEF was performed during the fourth meeting, when participants decided on measures for Sölle. Decisions were taken on a consensus basis, yet during the discussion it did not become sufficiently clear as to which criteria the stakeholders divided the measures into those for direct payments
258 Katja Arzt for farmers, and others classified as ‘codes of good agricultural practice’ without direct payments. Usually they considered costs during the discussion. One administrator from the environmental board, for example, said: ‘the long transportation of the cut grass is very costly and a farmer could hardly pay for that’ and ‘I do not think that a farmer is interested in cutting the grass in that place, therefore I agree to a payment’. The decisions exclude consideration of long-term consequences for the region’s ecology and different stakeholder groups.
Conclusions Certainly, six meetings are insufficient to acknowledge if the AEF will be institutionalized as an interactive valuation process in which learning processes are initiated. Within four meetings the participants agreed upon a list of different measures for Sölle. The list was transformed into an agri-environmental programme for Sölle and presented by the organization team to the local government of agriculture, but it has not yet been accepted as part of a new agri-environmental scheme in the future. For the second subject, soil erosion, no consensus could be reached after three meetings within the group. Overall, it was the first time that local stakeholders together with scientists decided and reached consensus about agricultural measures, and on this basis the process can be called a success. They exchanged their value perceptions on different issues. However, the decline in numbers of participants over the series of meetings signalled that not all participants were satisfied with the process itself. This study hypothesized that a good learning process is a result of wellimplemented variables (e.g. selection of stakeholders, issues discussed, flow of information, etc.). In the case of Sölle it can be concluded that the participants learned about them during the meetings and were helped to create a value perception about the issue. It also became clear that they needed different information sources. The third meeting, for instance, was successful in stimulating learning about the ecology of the subject under consideration, but participants failed to decide about a local programme because more information was needed. After the presentation of practical experiences from farmers a decision procedure was possible. In the case of soil erosion, however, a decision process could not be stimulated. The issue may not be of any relevance to the participants or the information given to them did not correspond to their needs. However, it was shown that the presentation of scientific lectures alone neither stimulated a dialogue among the participants nor helped to create an atmosphere of trust. What could be improved in future settings in order to stimulate a mutual learning process? With regard to stakeholder selection it should be ensured that the representatives of different organizations back-check with and can speak on behalf of the constituency group, because it happened that two representatives from the same organization had opposite opinions. This fact did not help to build reciprocity among the participants. The group tended to exclude some stakeholder groups, like farmers from family businesses, women, or people who do not belong to any organization. More thoughts should be given how those groups could be integrated
The challenges of stakeholder participation 259 or how the selection of stakeholders should be organized in order to have an adequate representation of the county community. The organization team should seek to be more independent of the other stakeholder groups. Scientists therefore should be restricted to an advisory function, and should only make a contribution when requested. The chosen topics did not correspond to the interests, knowledge or urgent concerns of the participants. Therefore a better method for choosing topics seems to be necessary. However, the method still has to be quite simple so that it can be transferred to other regions. This requirement precluded a careful analysis of needs of the participants through social scientific methods (e.g. interviews) prior to the process. Conversely, to start with conflicting problems may impede a constructive discussion and may not help to build trust among the participants in the beginning. Therefore it might be best to start with an issue which is relatively uncontroversial. Property right structures (ownership, control and use) have not been exclusively mentioned during the discussion but this should be changed so that action can easily follow the decision process and reciprocity among the participants can emerge. The information flow during the process was strongly influenced by scientists. There is unmistakably a big gap between a scientist’s way of thinking and that of practitioners. Scientific training emphasizes the hierarchical organization of an issue: analysis of the situation, targets, instruments, implementation and testing. A practitioner, on the other hand, tends to be more interested in concrete decisions about actions. The organization of information flow is central for the learning process. It helps people if a discussion can be followed easily in order to contribute fruitfully to the debate. Scientific lectures should only be held on urgent request. Further it was observed that participants with good verbal skills dominated the discussion process; a more balanced level of participation by all stakeholders could be achieved by using better moderation techniques or a different procedure.2 The decision procedure about the Sölle programme was inefficient. It lacked the participation of all stakeholder groups and did not consider enough different aspects. A well-performed MCA might improve the procedure and also stimulate a learning process. However, institutional difficulties within the decision procedure may be overcome by better institutional design or by learning about the process itself. Certainly more information and perceptions are exchanged in such deliberative institutions compared with monetary evaluations, but at the same time it is costly and timeconsuming. Yet, getting the institution right is a time-consuming process. The stakeholders are dissatisfied with the present situation of agri-environmental policy decisions but at the same time sceptical about the success of such new valuation institution(s). So far none of the stakeholder groups or the organization team know the direction that a change of the existing institutional setting in agricultural policy may be taking. The decision procedure may improve over time and with more experience, but the acceptance of those procedures from existing institutions can still threaten the whole process. Therefore it is important to fit such new valuation institutions in with existing institutions right from the beginning. The framework used in this study certainly helped to analyse the process with its different aspects. However, at this early stage in the deliberative process of the
260 Katja Arzt case study it is difficult to draw any conclusions on the relevance of each element to the process and whether there are other important elements which might lead to a successful performance of interactive valuation processes. Only the comparison of different case studies will clarify this question.3 Certainly more information is needed about the reasons why people are involved in the process and their expectations in order to understand their needs. This information might shed more light on the adequate design of the interactive valuation process. Another question is whether all environmental problems of agriculture can be solved with these kinds of participatory methods. More research has to be done on which coordination structures correspond to which kinds of transaction and property right structures. In this respect a more hierarchical decision procedure might address some problems better when transaction costs, and also fairness, are considered. Finally, more knowledge is needed about the underlying assumptions of the core elements of trust, reciprocity and reputation and how these stimulate a learning process among different stakeholder groups. Further investigations are needed as to how institutions enhance or restrict the building of mutual trust, reciprocity, and reputation. In this respect future investigations should also be interdisciplinary in their design.
Acknowledgements This chapter arose from work commissioned by the German Federal Ministry of Research and Education (Bundesministerium für Bildung und Forschung); it is part of a scientific co-operative project called GRANO. The author is grateful to the steering group for helpful discussions, and also to the Chair of Resource Economics and Prof Dr K. Hagedorn, the scientists at the Countryside and Community Research Unit in Cheltenham, UK, and especially to John Powell. The author would like to thank Simon Andrews, Claudia Carter and the colleagues who joined the discussions during the ESEE Conference 2000 in Vienna. The author alone is responsible for any errors and omissions.
Notes 1 GRANO project stands for ‘Approaches for a sustainable agricultural production: Application for North-eastern Germany’. GRANO is a scientific co-operative project supported by the Federal Research Ministry. GRANO chose the districts of Uckermark, Barnim, and Elbe-Elster to search for, develop, and test new ideas and solution strategies. The ‘round table’ project is one among other projects carried out by GRANO. 2 The organization team discussed that problem, together with some participants, and decided that bilateral interviews about the programme measures (in this case measures against soil erosion) should precede the next meeting. Consequently the organization team prepared a questionnaire and asked all participants separately during the summer months when no meeting took place, and presented the results during the next meeting. This procedure was very time and resource consuming. 3 At the moment a PhD student is working on that question.
The challenges of stakeholder participation 261
References Berkes, F. and Folke, C. (1998) Linking Social and Ecological Systems. Management Practices and Social Mechanisms for Building Resilience, Cambridge: Cambridge University Press. Bernabo, J.C. (1998a) ‘Improving integrated assessments for application to decision making’, in S. Lee and T. Schneider (eds) Risk Assessment: Linking Science to Policy, Amsterdam: Elsevier Science. Bernabo, J.C. (1998b) ‘Perspectives and principles from other assessments’. Online. Available at: http://www.agci.org/publications/EOC97/eoc97session2/Bernabo.html (accessed 16 June 2004). Bromley, D.W. (1991) Environment and Economy: Property Rights and Public Policy, Cambridge, MA: Basil Blackwell. Bromley, D.W. (1997) ‘Property rights in environmental economics’, in H. Folmer and T. Tietenberg (eds) The International Yearbook of Environmental and Resource Economics 1997/ 1998: A Survey of Current Issues, Cheltenham: Edward Elgar. Clark, J. and Murdoch, J. (1997) ‘Local knowledge and the precarious extension of scientific networks: a reflection on three case studies’, Sociologia Ruralis, 37(1): 37–60. Curry, N. and Winter, M. (2000) ‘The transition to environmental agriculture in Europe: learning processes and knowledge networks’, European Planning Studies, 8(1): 107–21. DeBono, E. (1993) Parallel Thinking, Ringwood: Penguin Books. Habermas, J. (1982) Theorie des kommunikativen Handelns, Frankfurt/Main: Suhrkamp. Hagedorn, K. (2000) ‘Umweltgenossenschaften aus institutionenökonomischer Sicht’, in M. Kirk, W. Jost and S. Kramer (eds) Genossenschaften und Kooperation in einer sich wandelnden Welt. Festschrift zum 65. Geburtstag von Prof. Dr. Hans-H. Münkner, Münster: LIT Verlag: 267–91. Hampicke, U., Müller, K., Meyer-Aurich, A. and Kachele, K.U. (1998) ‘Agrarumweltpolitik nach dem Subsidiaritätsprinzip. Teil III Fallstudie Brandenburg/Uckermark’, working paper of the Universität Greifswald und Zentrum für Agrarlandschafts- und Landnutzungsforschung e.V. Hanna, S., Folke, C. and Mäler, K.-G. (eds) (1996) Rights to Nature, Washington, DC: Island Press. Holling, C.S., Berkes, F. and Folke C. (1998) ‘Science, sustainability and resource management’, in N.G. Röling and E. Wagemakers (eds) Facilitating Sustainable Agriculture, Cambridge: Cambridge University Press. Jiggins, J. and Röling, N. (1998) ‘Interactive valuation’, paper presented at the Conference of the European Society for Ecological Economics, University of Geneva, Switzerland, 5 and 6 March 1998. Kolloge, S. (2000) ‘Die Wirkung der Agenda 21. Eine institutionentheoretische Wirkungsanalyse am Beispiel von sechs Kommunen im ländlichen Raum Großbritaniens und Deutschlands’, unpublished PhD thesis, Landwirtschaftliche-Gärtnerische Fakultaät der Humboldt-Universität zu Berlin, Germany. Lawrence, R. (2000) ‘Property, rights and fairness’, in C.L. Spash and C. Carter (eds) EVE Policy Research Brief Series, no. 6, Cambridge: Cambridge Research for the Environment. Meppem, T. and Gill, R. (1998) ‘Planning for sustainability as a learning concept, survey’, Ecological Economics, 26: 121–37. Müller, K., Bork, H.-R., Dosch, A., Hagedorn, K., Kern, J., Peters, J., Petersen, H.-G., Nagel, U.J., Schatz, T., Schmidt, R., Toussaint, V., Weith, T., Werner, A. and Wotke, A. (eds) (2000) Nachhaltige Landnutzung im Konsens: Ansätze für eine dauerhaft-umweltgerechte Nutzung der Agrarlandschaften in Nordostdeutschland, Giessen: Focus Verlag.
262 Katja Arzt Munda, G. (2000) ‘Conceptualising and responding to complexity’, in C.L. Spash and C. Carter (eds) EVE Policy Research Brief Series, no. 2, Cambridge: Cambridge Research for the Environment. O’Conner, M. (1998) ‘Walking in the Garden(s) of Babylon: an overview of the VALES project’, Paris: C3ED Rapport de Recherche. O´Hara, S. (2000) ‘The challenges of valuation: ecological economics between matter and meaning’, in C. Cleveland, R. Costanza and D. Stern (eds) The Nature of Economics and the Economics of Nature, London: Edward Elgar. O’Hara, S., Shandas, V. and Vazquez, J. (1999) ‘Communicating sustainable development options – who evaluates the trade-offs?’, in I. Ring, B. Klauer, F. Waetzold and B. Mansson (eds) Regional Sustainability. Applied Ecological Economics Bridging the Gap between Natural and Social Sciences, Heidelberg: Physica Verlag. O’Neill, J. and Spash, C.L. (2000) ‘Conceptions of value in environmental decision-making’, in C.L. Spash and C. Carter (eds) EVE Policy Research Brief Series, no. 4, Cambridge: Cambridge Research for the Environment. Ostrom, E. (1990) Governing the Commons: The Evolution of Institutions for Collective Action, Cambridge: Cambridge University Press. Ostrom, E. (1998a) ‘The institutional analysis and development approach’, in E.T. Loehman and D.M. Kilgour (eds) Designing Institutions for Environmental and Resource Management, Cheltenham: Edward Elgar: 68–90. Ostrom, E. (1998b) ‘A behavioral approach to the rational choice theory of collective action’, presidential address, American Political Science Association, American Political Science Review, 92(1): 1–22. Ostrom, E., Gardener, R. and Walker, J. (1994) Rules, Games, and Common-Pool Resources, Ann Arbor: University of Michigan Press. Röling, N. and Wagemakers, E. (eds) (1998) Facilitating Sustainable Agriculture, Cambridge: Cambridge University Press. Sagoff, M. (1998) ‘Aggregation and deliberation in valuing environmental public goods: a look beyond contingent pricing’, Ecological Economics, 24(2–3): 213–30. Sahlins, M.D. (1976) Culture and Practical Reason, Chicago, IL: University of Chicago Press. Simon, H.A. (1983) Reason in Human Affairs, Stanford, CT: Stanford University Press. Söderbaum, P. (1999) ‘Values, ideology and politics in ecological economics’, Ecological Economics, 28: 161–70.
Preference transformation through deliberation 263
12 Preference transformation through deliberation Protecting world heritage Simon Niemeyer
Introduction Preferences are prone to change. Yet much effort in environmental economics has been applied to the measurement of static preferences for use in policy. This ‘endogenous preference’ orthodoxy has recently been challenged on at least two fronts. One criticism stems from findings that support the idea of ‘constructed’ preferences in view of exogenous influences (for example Slovic 1995; van den Bergh et al. 2000). A second criticism, with implications for environmental policy, is the emerging tension between the normative goal of sustainability and aggregation of sovereign preferences (for example Common and Perrings 1992). These have led to arguments within ecological economics for policies that will lead to so-called ‘sustainable’ preferences (Norton et al. 1998), although it is uncertain what is needed to achieve such a normative goal. The fact that preferences can change invokes notions of preference quality – that some preferences might be ‘better’ than others. Most economists do not make such distinctions: a stated preference is a legitimate preference where consumers are sovereign and free to choose as they wish. Others, such as O’Neill (1997) and Elster (1983) are less sanguine. For them, the process of producing ‘better’ (or reasonable, in the sense that they have been through a process such that chosen means match justified ends) preferences is more important than simply realising given preferences – the reason for a preference is as important as the preference itself. Deliberation is one (if not the) process for reasonable preference formation. It is the process of deliberation, be it interactive deliberation with others, or deliberation within oneself (e.g. Goodin 2000) whereby one recursively constructs those means that suit desirable ends. In the case of environmental policy, the most vaunted end is sustainability although it is unclear and possibly indefinable. Many authors are turning to deliberation in the search for mechanisms capable of elevating environmental considerations in policy making (for example Sagoff 1998). An important element of this ‘deliberative turn’ in environmental policy is the tendency for deliberation to transform preferences in the direction of ‘natureloving’, or sustainability. The relationship between preference for nature and sustainability is partially accounted for by the simple premise of ‘more nature
264 Simon Niemeyer preserved is more nature sustained’; but sustainability is inherently more complex and infused with a specific meaning depending on context. One danger in attempting to produce ‘sustainable’ preferences is to pre-empt outcomes that are nominally designated sustainable when that goal itself is not well defined. Just as Elster (1983: 9) states that utility maximisation as a single endeavour is futile because happiness eludes those who actively strive for it, the same may also be true when actively seeking to achieve sustainability. Instead of defining a sustainable end, a better approach is to focus on those processes that tend to produce ‘sustainable’ outcomes; better still if those same processes help clarify the debate concerning definitions thereof. Fortuitously, deliberation is also identified as such a process through which sustainable outcomes are defined and achieved as part of the same process (Rydin 1999). In other words, ‘deliberators’ formulate those ends deemed sustainable and form judgements about which policy means are best suited. This chapter seeks to explore the ‘deliberative alternative’ and its ability to transform preferences to suit so-called sustainable ends. The case study concerns a controversial road constructed through World Heritage Listed rainforest in Far North Queensland, Australia, on which a Citizens’ Jury (CJ) was held in January 2000. CJs are seen here as an example of a formal deliberative process that has some similarity to a legal jury where decisions are made collectively by a randomly selected group of citizens based on evidence presented (see other chapters in this volume and also Smith and Wales 2000). The focus of this chapter is on the margins of ‘preference shift’, or those mechanisms through which preferences are transformed by the deliberative process. The aim is to measure both changes to (rank ordered) preferences and the underlying subjectivity for those preferences. Subjectivity is measured using Q method, a form of ‘quantitative discourse analysis’, which is also compared with actual jury discourse and interview data. The following discussion begins by considering tentative mechanisms whereby preferences are transformed through group deliberation in an information-rich environment. This is followed by a description of the case study, its history and politics, considered important because, as will be shown, issue context is a major factor in preference dynamics. The impact of the deliberative processes on both preference and subjectivity is then discussed with particular mechanisms explaining the processes that contribute to preference shifts. Conclusions are then drawn about the role of deliberative processes in improving environmental policy outcomes.
Discourse and preference shift The following discussion attempts to identify mechanisms whereby preferences change through discourse, particularly during participation in a formal deliberative process. The use of the term preference varies according to disciplinary perspective. Here preference refers simply to the process of ranking choices, such as in the act of voting. Preferences then reflect the tendency to favour one option over another. Aldred (this volume, Chapter 8) argues that the use of the term preference is often
Preference transformation through deliberation 265 confused with judgement. According to Elster (1983: 16) judgement is defined as ‘the capacity to synthesize vast and diffuse information that more or less clearly bears on the problem at hand, in such a way that no element or set of elements is given undue importance’. A relationship between judgements and preferences exists to the extent that judgements represent the reasoning behind a preference (Keat 1994: 335). It is posited here that preferences are shaped during deliberation by three inter-related phenomena: (i) information (knowledge acquisition and understanding of consequences); (ii) attitude (perspective adopted during judgement, attitude transformation); and (iii) process (information absorbed and attitudinal lens adopted by virtue of process, degree of preference construction). Each of these impacts is briefly discussed in turn below. Information is a widely accepted influence on preference. Elementary economic theory is premised on ‘full information’; that is, the consumer is assumed to be aware of all costs and benefits, including opportunity cost, when making choices about consumption. In reality no individual has recourse to the entire suite of information when making choices, uncertainty is implicit to the act of choice (values also play an important role in information acquisition and impact on preference). Where provision of a good requires social cooperation the problem of information deficit is particularly strong.1 The problem is exacerbated for environmental goods, which are characterised by complexity and uncertainty, making the ‘cost’ of acquiring information particularly high. That political choices are poorly informed increases scope for preference manipulation. To reduce the burden of information acquisition, individuals may adopt simplifying strategies in order to make sense of the issue. Intuitively appealing conclusions may be drawn from the little information available, which is in no small part a product of the political environment. If the issue is polarised, with interest groups adopting extreme positions, individuals may choose to follow these external cues, resulting in issue polarisation being reflected in individual preference. The mechanism relies on the tendency for interest groups to adopt extremes to elevate their cause within the political sphere; and for individuals to replicate these extremes in their own discourse and preferences rather than the messy middle ground (Elster 1983: 19). Environmental issues in particular are often dominated by polarised interests which play an important role in defining the lines of discourse surrounding issues within the public sphere (for example Harrison et al. 1999). This results in preferences that reflect symbolism rather than substance, an outcome referred to as symbolic preference (Sears 1993). Where symbolic preferences dominate an issue, an information-rich deliberative process may improve the ability of the individual to judge the merits of policy alternatives, or at least decouple them from the posturing of external parties. The result is that improved information about the public good in question then changes perception regarding the outcome of a particular policy. The individual is armed with sufficient knowledge, giving them the fortitude to make an independent assessment of the issue at hand, rather than defaulting to intuitively appealing arguments proffered by entrenched political interests. Even where issues are not dominated by symbolic politics, information will tend to transform individual preferences by improving the understanding of consequences attached to certain choices.
266 Simon Niemeyer Preferences are more than a function of information. At least as important are attitudes, or values. These help to frame both the desired end and the way that information is received and processed. Attitude refers to the belief system, or subjectivity, that comes into play during the process of judging alternative environmental policies as far as they relate to subjective values, or value orientations. In other words, attitude is the value attached to a particular outcome, rather than a judgement regarding its likelihood. There are a number of attitudinal perspectives that may coexist in relation to an issue, even within the same individual (Etzioni 1986). In relation to environmental attitudes, the most celebrated example of this is that of Sagoff ’s (1988) citizen–consumer schism, which describes in very simple terms how the same problem can be conceptualised from the position of consumer or citizen, each yielding entirely different outcomes. Keat (1994) argues that, because environmental goods are public goods requiring collective action, (sustainable) environmental decisions require that individuals address them from the perspective of the citizen. This implies other-regarding behaviour and explicit consideration of ethical dimensions in contrast to the self-interested consumer (see Spash and Hanley 1995). Some authors contend that deliberation tends to foster citizenry in making choices (Ward 1999). Goodin (1996) posits that deliberation can extend other-regardedness to that of Nature itself, an otherwise silent partner in democratic processes. If individuals change their perspective during deliberation, adopting different sets of attitudes through which they judge the process, it is anticipated that preferences will change accordingly to reflect adjustments in the desired end. Should group deliberation produce changes to attitudes such that they tend to favour collectively desirable sustainable outcomes (over the satisfaction of individual wants) then these should be reflected in the preferences of individuals. Construction process concerns both the cognitive act of turning information and attitudes/values into preferences and the context in which these are formed. There has been much discussion in recent literature, particularly concerning contingent valuation, regarding the construction of preference (Slovic 1995; Schkade and Payne 1994). It is posited that an extensive process of information acquisition, evaluation and judgement – such as for a formal deliberative process – leads to better-constructed preferences than those based on superficial (or symbolic) assessment of consequences.2 Information availability forms an important part of this context because it facilitates the judgement process by filling gaps in knowledge and providing linkages between cause and effect, improving translation of attitudes regarding consequences into preferences. During the construction process, information and attitude conjointly influence preferences as well as each other: information transforms attitudes and attitudes impact on the acquisition of information. With respect to ecology, Callicot (1982) argues that increased knowledge ‘changes our values by changing our concepts of the world and of ourselves in relation to the world’ (which resonates with the ‘biophilia hypothesis’, see Kellert and Wilson 1993). Changes to attitudes, or subjectivity, will also bear upon the acquisition of information for the judgement process (Elster 1983: 19). If this contention is correct, a positive feedback occurs where increased valuation of nature stimulates the desire for more knowledge about nature, and this in turn
Preference transformation through deliberation 267 increases environmental values. Such processes occur over a longer time-frame than a deliberative forum, but may nonetheless be observable during deliberative processes.
Case study: the Bloomfield Track The case study for deliberations concerned policy options for the Bloomfield Track in the Daintree Region in Far North Queensland, Australia. Figure 12.1 shows the location of the Track, which traverses 30 km of mostly lowland coastal rainforest
Figure 12.1 The Bloomfield Track
268 Simon Niemeyer from Cape Tribulation north to the Bloomfield River near the Wujal Wujal township. It lies almost entirely within the Wet Tropics World Heritage Area, the boundaries of which are shown as the continuous white line circumscribing the region. The map also shows the Great Barrier Reef passing along the coast. Smaller inshore reefs abut the coast along the region, creating a unique situation where tropical rainforest adjoins coral reef. The Bloomfield Track provides a link, albeit a rudimentary one, between the remote Bloomfield region which is home to approximately 1000 residents, mostly located in the Wujal Wujal Aboriginal Community, and major southern centres such as Mossman and Cairns. Prior to its construction, local residents were restricted to northern access via a rough track to the more remote township of Cooktown approximately 60 km away – and from there a three-hour-plus inland drive south to reach Cairns. The satellite image in the background for Figure 12.1 provides some indication of the ruggedness of the area, most of which is covered by dense tropical rainforests and steep terrain. The intended community-access Bloomfield Track is currently used mostly for tourism. From the beginning of construction in 1982 the Bloomfield Track has been highly controversial. Despite stated objectives to increase community access, many attribute its construction to the political dynamics of the time. The period was one in Australian politics where environmental issues were beginning to peak, in electoral terms, at the federal level (Papadakis 1993: 142). In Queensland, however, state politics remained dominated by development interests and strong state sentiment against federal intervention. Tension between the federal government and the developmentoriented Queensland state government was reaching a particularly strong climax (see Kellow and Niemeyer 1999: 210). Significantly, there were no formal processes established for public input into the decision.3 Construction was undertaken by the Douglas Shire Council with Queensland state government backing, amidst pitched battles between environmentalists and the state police. The Track was constructed in haste; a single bulldozer cleared a trail, sometimes swerving around protesters buried along its path to prevent construction. Even during construction, the green movement intensely lobbied the federal government to nominate the area for World Heritage and intervene in the road issue, as it had recently done in the state of Tasmania, without success. The thrust of public debate against the track focused on highly emotive and symbolic issues such as the intrusion into a significant wilderness (for example, Wilderness Action Group 1983). Pictures of pristine rainforest ‘ruined’ by bulldozers dominated the campaign. Environmentalists cited damage caused to the onshore reefs by sediment run-off from the Track. Seven years after the construction of the Bloomfield Track a state election in Queensland replaced the incumbent government. This resulted in politically aligned governments at both state and federal levels and provided the opportunity to declare the Daintree region a World Heritage Area managed by a joint federal–state body, the Wet Tropics Management Authority. The Bloomfield Track has remained a thorny issue – a product of a different political era – one that has yet to be properly resolved.4 There are strong, and opposing, interests, and interconnected pressures
Preference transformation through deliberation 269 within the region make the formulation of policy in the region extremely difficult. Although at first indifferent, there is strong support for the Track from within the vocal local community groups, who express resentment at what they view as ‘outside’ interference in their own affairs by those who call for its closure. The Track has also become an important, if symbolic, link between the indigenous Wujal Wujal community and their southern kinship groups, historically separated by forced resettlement. There is consequently intense pressure from local residential groups to keep the Track open and upgraded, though some residents also resent the increasing intrusion by tourists. The combined forces of geographic isolation and a sense of disenfranchisement have enhanced suspicion of outside interference in local affairs. Environmental groups have consistently maintained their calls that the Track be closed, and down-graded to a walking trail. Polarisation of interest groups is, in part, an artefact of the political history of the area and the absence of mechanisms for consultation and mediation. The continued policy stagnation of the Bloomfield Track issue is a constant reminder of its bitter political history. The Far North Queensland CJ was conducted as part of research into the use of CJs for environmental policy in Australia. The jury, which was recruited on a random-stratified basis from the region, was given the task of considering management options for the Track over a four-day period. In broad terms the jury process consisted of four stages: site familiarisation; presentations by expert witnesses and questions; presentations by community witnesses and questions; and deliberation and report writing. Site familiarisation consisted of a field trip to the Track.5 This was followed by two days of presentations during which short talks were given by witnesses. Expert witnesses gave evidence pertaining to technical aspects under the jury remit. Themes which these witnesses were recruited to address included tourism; environmental impact to both reef and rainforest; engineering; and economics and regional planning. Community witnesses were recruited to give voice to the needs and aspirations of the community and to provide the deliberations with a social context. Community representatives included the mayors of the Douglas and Cook Shire Councils and a local Bloomfield councillor to provide a community perspective. A local indigenous perspective was provided for as far as was possible.6 Report writing took place on the final day of the process, during which jurors developed policy recommendations. Information about the manner in which participants in the CJ changed attitudes towards the Bloomfield Track issue and their policy preferences was obtained by surveying participants at the outset of deliberations, at the midway point (in the case of attitudes),7 and immediately after the proceedings. The empirical tool used to probe subjectivity was Q method. The method provides insights into predominant attitudes through the analysis of discourse (Barry and Proops 1999). It is a qualitative, but statistical approach and differs from more conventional methods of statistical analysis because it compares subjectivity across individuals rather than comparing predefined variables (Brown 1993).8 The basic approach to Qmethodology involves the identification of a concourse, or ‘running together of ideas’, that emerge around an issue. Sample statements (which can also involve pictures or some other medium relevant to the issue) are drawn from this concourse.
270 Simon Niemeyer Respondents are asked to order statements from ‘agree’ to ‘disagree’, by ranking them under a forced distribution (Brown 1993: 1–11).9 Individuals are then correlated according to their responses and factor analysis is applied to draw out dominant discourses among the group. These factors are then related back to the original statements so that major characteristics, similarities and differences between factors can be investigated. The Q-sort for the CJ was developed from a variety of sources, both directly and indirectly related to the Bloomfield Track.10 The statements were printed on individual cards, each with nine boxes labelled from –4 to +4. Jurors were first asked to run through the cards and score each statement. They were then asked to sort the cards from ‘most agree’ to ‘most disagree’ using the original scores as a guide, although they were free to change the orderings where they felt appropriate. The 42 cards were subsequently sorted into a quasinormal distribution (3, 4, 5, 6, 6, 6, 5, 4, 3 cards from disagree to agree). Analysis of the Q data revealed a number of characteristic patterns. The resulting factors each represent a particular type of subjectivity. The degree to which each juror conformed to that factor is expressed in terms of a factor loading (a loading of 0 representing no conformance, 1 for complete conformance and –1 for complete non-conformance). The factors were extracted by analysing all three sorts for all individuals together, as though they were one group. Doing so facilitated comparison between different sorts so that changes to each individual’s loading on a factor could be examined. Measurement of preference shifts during the CJ was done using a rank ordering of policy options. The options used for the preference rank were obtained from information about the policy issue from which five distinct alternatives were chosen, broadly representative of the major policy options. The options included bituminisation of the Track (Bituminise); upgrading the Track so it was passable by all vehicular types, but with a gravel surface (Upgrade); stabilisation of the steeper slopes to reduce run-off (Stabilise); leave the Track as it is under the current management regime (Status quo); and close the Track and revegetate, with the possibility of using it for a walking track (Close).11
Results In the following discussion the results from the CJ are briefly outlined. This is followed by more substantial analysis and discussion to establish the implications of deliberation for preferences and attitudes, and mechanisms for change thereof. Attitudes Measurable changes to attitudes did occur during the deliberative process. Yet there were a number of constant aspects – in terms of their relative weighting. Most significant of these was the tendency for jurors to place considerable emphasis on long-term solutions to the issue, and its impact on future generations. For example, jurors consistently agreed that a short-term perspective was inappropriate for deciding the issue (see statement 30 in Table 12.1, x1 x3 = –2.08, s1 = 1.44, x3 = –2.25, s3 = 1.22); and that inaction would damage the opportunities of future
Preference transformation through deliberation 271 generations (statement 25, x1 = 3.00, s1 = 0.74, x3 = 2.83, s3 = 0.94). There was also a strong, though variable, commitment to conservation in the Daintree irrespective of cost (statement 20, x1 = 2.67, s1 = 1.97, x3 = 2.42, s3 = 2.27). All jurors tended to agree that the Bloomfield Track issue was an important one (statement 12, x1 = 1.58, s1 = 1.00, x3 = 1.67, s3 = 0.78). One of the most intractable issues among the jurors throughout the process was the level of permissible development within the World Heritage area (statement 10, x1 = 0.92, s1 = 2.39, x3 = 1.25, s3 = 3.39).12 There was also widespread disagreement regarding whether the track should stay, given that it had already been constructed, although there tended to be a net shift away from this position (statement 39, x1 = 0.75, s1 = 1.54, x3 = –0.83, s3 = 2.66). As will be seen in the preference analysis below, this was reflected in a division among jurors over the preferred policy option. Using conventional Q analysis, described above, it is possible to establish an overview of the attitudinal changes within the jury as a result of deliberation. By tracking changes to juror’s loadings on each factor – representing the degree to which they ‘agreed’ or ‘disagreed’ with that subjective position – we can draw insights into changes to attitudes during the deliberative process. The combined data for all three Q sorts by the jurors produced four distinct factors: Preservation, Optimism, Pragmatism and Symbolism. Statement scores for each factor are provided in Table 12.1. The following four paragraphs briefly describe the contours of each factor and the degree to which their role in the attitudes of jurors changed during the deliberative process. The description of each factor takes the combined form of narrative regarding its defining features and includes a summary of factor scores for relevant statements. The first and most dominant factor – accounting for 40 per cent of variation between sorts – is Preservation. Based on the statement scores for the factor given in Table 12.1, Preservationists feel a personal commitment to protection of the Daintree area. Statements with high scores for this factor tended to reflect an anti-development sentiment towards the area (such as responses to statement numbers 10 and 32). There is also a perception that the Bloomfield Track is a contributing factor to future development (statement 2). (This is significant given, as previously mentioned, that there was an increasing sense among the jury during deliberation that the Bloomfield Track was part of an interconnecting web of regional impacts, which cannot be properly considered in isolation.) Also evident is the belief that the use of vehicles on the Bloomfield Track is detrimental to both rainforest (statement 21) and, to a lesser extent, the reef (statement 7). Preservationists also feel they would be better off, or at least no worse off, because of protection of the area (statement 11). Figure 12.2 shows how jurors changed their loadings on Factor 1 during deliberation. Each of the bars represents the loading of each juror on the factor before, during and at the conclusion of the CJ. (The names in the figure are the abbreviated codenames chosen by jurors.13) The left-most bar represents the initial loading at the beginning of deliberation. The middle (thinner) bar shows loadings midway and the darker righthand bar shows the loading of jurors at the end of the process. From Figure 12.2, most jurors’ loadings on Preservation have changed substantially.14 Eight of the 12 jurors are significantly loaded at the beginning of deliberation (based on a level of
18
17
10 11 12 13 14 15 16
9
7 8
5 6
4
3
2
1
No.
Laying bitumen on the Bloomfield Track would be beneficial for the environment. It may even help reduce fuel usage and the greenhouse effect. I don’t know if improving the Bloomfield Track would lead to a rapid acceleration of development in the area to the detriment of the environment. In deciding on what to do with the Bloomfield Track I don’t know whether it’s more important to meet the needs of the community or the environment. Whilst impacts on locals in the Bloomfield area are a concern, it is the broader community that should carry more weight when deciding what to do with the Bloomfield Track. I don’t know what the people of Bloomfield think about the Bloomfield Track. The road is just the ‘thin edge of the wedge’. Further improvement of the road will lead to more development in the area resulting in environmental damage. This may not happen for a long time, but it will happen. Erosion from the Bloomfield Track is permanently damaging the coral reefs that fringe the beaches below. When it comes to the Bloomfield track, people living in Cairns are in no position to judge what the interests of the local residents of Bloomfield are. If the Bloomfield Track is sealed (bitumenised) there will not be a rapid increase in environmentally damaging development in the Daintree area in the future. It may even benefit the environment there. No development should be permitted in World Heritage areas such as the Daintree. I would be worse off if more of the Daintree rainforest is protected. The Bloomfield Track issue is important for Queensland. The Bloomfield Track is bad for World Heritage values. I’m not sure if the future of the Bloomfield Track should be determined by locals, outsiders or both. The fate of the Bloomfield Track is of no concern to me. Economic development associated with the Bloomfield Track will provide more opportunities for future generations in North Queensland. The future of the Bloomfield Track should be determined by everyone and not just by those who live in the Bloomfield area. I don’t know how, but I think that there must be some way in which everybody benefits from protecting the rainforest near the Bloomfield Track.
Statement
Table 12.1 Statement scores for Q attitudinal factors
3
2
3
0
0
1
4
3 0
2
3
4
2
1
3
2
1
3 –3 1 0 –4 –2 –1 –1 2 1 1 0 0 0 1 1 –1 1 –1 –1 –4 –2 –3 –4 –1 1 –1 1
–3 –1 –2 –2
1 –1 –4 0 1 0
2 –2 4 3
2
2 –2
3 –1
0 –3
2
1 2 3 –1
2
–2
–2
–3
1
Factor Score
272 Simon Niemeyer
43
40 41 42
39
35 36 37 38
33 34
29 30 31 32
27 28
26
25
19 20 21 22 23 24
The Bloomfield Track is important because it allows quick access to remote areas of the North. Conservation in the Daintree area is worthwhile at whatever cost. Using cars on the Bloomfield Track is bad for the rainforest. Any decision about the Bloomfield Track will greatly affect people like me. I have no idea what the people in the Bloomfield area think about the Bloomfield Track. Erosion from the Bloomfield Track does not cause siltation or damage to the fringing inshore reefs between Cape Tribulation and Cooktown. If we don’t take steps to protect the Daintree rainforest, future generations will miss out on the opportunity to experience the area as we do now. We don’t need to worry too much about environmental damage in the Daintree region because future generations will be better able to deal with these problems than we are. There is no reason to believe that the Daintree Rainforest is under threat. If future generations could have their say about the Bloomfield Track, they would be less concerned about the environmental impacts than many people make out. The protection of plants and animals in the Daintree is OK so long as it doesn’t affect me. Let’s fix the problems in the Daintree just for now. The future will take care of itself. The more that it is possible for the average North Queensland resident access the Bloomfield Track the better. I don’t like how development is creeping further and further North into the Daintree and beyond because of its effect on the environment. The coral reefs along the foreshore below the Bloomfield Track are not badly affected by the road. Native animals in the Daintree need protection because they have a right to life which cannot be traded against economic considerations. The Wujal Wujal community is better off now that the Bloomfield Track has been built. The most important use of the Bloomfield Track is for tourism. I’m concerned that I will be made worse off by any decision about the Bloomfield Track. I think that both short and long term perspectives are needed in deciding what should be done with the Bloomfield Track, but I don’t know which one is more important. The Bloomfield Track may not have been the best idea, but I guess there is probably little point in closing it now that it has been built. I don’t really know who benefits most from the protection of rainforest in the Daintree. A long term perspective on the Bloomfield Track is essential. When it comes to the Bloomfield Track, it’s not important to worry about what the future will hold. We need to worry about now. Everyone in Queensland is better off for having a road like the Bloomfield Track. 4
3
4
0 3
2 0 –4 3 4 2 –1
4
3 2 0 –2 0 –2 2 2
4 –3 1 1
–1 –2 –3 –2
0 1 4 4 –1 –3
–1 –1
1 0 1 –2 0 –1 0 3
0 3
–2 –3 –4 –4 –3 –4 –3 –2 –2 –2 –2 –1 2 0 2 0
–4 –3 0 –4 –2 –4 –2 –2
–3 –4 –4 –4
4
0 2 0 2 4 3 –2 4 2 0 –1 0 1 0 –1 2 1 1 1 0 –1 –1 0 –3
Preference transformation through deliberation 273
274 Simon Niemeyer 100 90
Factor loading (x100)
80 70 60 50 40 30 20 10 0 ADV ASW BOA JAN
First sort
JUL
KEI KOD MAT PEA Juror Second sort
RAS SNO TAM
Third sort
Figure 12.2 Juror loadings on factor 1 ‘preservation’
significance of 0.4).15 During deliberation there has been a net increase in factor loadings ( x1 = 0.57, x3 = 0.61), with the exception of Adventure, Boat, Julie and Keith, who have all decreased their loadings. The Optimist archetype is characterised by someone who seeks an outcome where everybody wins – environment and humans (exemplified by statement 18). Though optimistic, they are very uncertain about a number of issues, including damage to the area because of development (statement 2). They feel that the continued existence of the Track will probably not lead to more development (statement 6), or destroy the fringing reef (statement 7). Possibly in view of this, they are more predisposed to development in the region than others (statement 10). Optimists are not inured to future consequences. Indeed, they are less concerned about the present than others (statement 42), but they are uncertain about whether now or the future is the more important (statement 38). This could be a result of the level of uncertainty attached to future outcomes, consequently greater significance is attached to the present – a crude kind of discounting of future benefits.16 The Optimist is anthropocentric in orientation.17 They feel that assessment of environmental problems should be based on a perception of the efficacy of technological solutions.18 Greatest emphasis is on how to achieve the best outcome. When questioned after deliberation, jurors most strongly loaded on Optimism felt that closure of the Bloomfield Track should only be undertaken once the costs are known to outweigh the benefits. These benefits should be given proper account (statement 19). They feel that the Daintree could in fact be under threat, but its conservation may not require the closure of the Bloomfield Track (statements 20 and 27). Optimists tended to prefer technologically based ‘win– win’ solutions. Optimism accounts for 9 per cent of variation between sorts. Net loadings on the factor have steadily declined ( x1 = 0.26, x2 = 0.21, x3 = 0.17). As
Preference transformation through deliberation 275 90 80 70 Factor loading (x100)
60 50 40 30 20 10 0 -10 ADV
AS W BO A J AN
J UL
KEI
KO D MAT
P EA RAS
S NO TAM
-20 J uror First sort
Second sort
Third sort
Figure 12.3 Juror loadings on faction 2 ‘optimism’
shown in Figure 12.3, only a few of the jurors subscribed to Optimism at any stage of deliberation. Notably, Julie and Keith – the main dissenters at the end of deliberation – have increased their loadings on the factor. Pragmatism has many similarities to Optimism (with a correlation of 0.59). Dominant features that distinguish Pragmatists from Optimists include scepticism about anything but firm facts; confidence in their held beliefs; and less emphasis on ‘win–win’ solutions. In contrast to the ecologically cautious approach of the Preservationist – where the burden is to prove that activities do not damage the area – Pragmatism focuses on the available facts. For example, because existing research could not sustain the contention of reef damage, it is dismissed almost entirely (statements 33, 7 and to a lesser extent 24). The pragmatic view of science reflects this. Science is considered as the definitive input into decisions, rather than a conservative and heuristic process that yields probabilistic outcomes. At its extreme the logic is simply ‘no evidence, therefore no damage’ (and therefore, no roadclosure). As discussed below, there is a strong correlation between this position and those jurors who prefer the Status quo option, provided that there is strict regulation of Track use. The Pragmatist attaches more weight to present consequences than future ones (statement 42), although they believe prudent decision making should also benefit the future (statement 30). They are less ecocentric than Optimists (statement 34); but they are not against preservation per se, believing that development in World Heritage areas should be curtailed (statement 10). Because of the conservatism of the Pragmatist they tend to place the burden of proof onto those who wish to change the Status quo. Like Optimism, because damage to the Daintree area does not appear significant (statement 27), conservation should be weighed against benefits that development of the area may bring (statement 20). (However,
276 Simon Niemeyer unlike the Optimist, Pragmatists are prepared to accept winners and losers (statement 18).) Now that the Track is built and no definitive evidence says that it should go, it may as well stay (statement 39); after all, it provides a benefit to the community north of the Bloomfield River (statement 35). Pragmatists tend not to be in favour of upgrading the Track (statements 6 and 9). As can be seen in Figure 12.4, pragmatism saw a marginal overall increase in factor loadings, although these were far from uniform. Most of the increase can be attributed to three jurors (Adventure 11 to 37, Janine –6 to 36 and Snoopy 7 to 33). The last factor, Symbolic, resonates with the symbolic preferences discussed earlier. Symbolists emphasise the impact of the road on fringing reefs, as well as its benefit to the local community. There are only four statements that define the Symbolic factor. Two of these concern strong belief in damage to the reef caused by the Bloomfield Track (statements 7 and 33). There is a good deal of confidence among Symbolists in ascribing benefits of protection (statement 40) as well as the feelings of locals in the Bloomfield area (statement 5). Apart from these statements, there is a high degree of affinity with factor 1, the Preservationist – with a high level of correlation between the two (r = 0.63).19 Figure 12.5 shows changes to loadings for Factor 4, Symbolism. Of all factors, it suffered the biggest decline during deliberation. Beginning with five jurors loaded at the 0.15 level, by the end of deliberation not one juror was significantly loaded – except for Keith and Tamarra who were negatively loaded. Preference shift Attention is now turned to how preferences changed during deliberation. Table 12.2 shows the rankings given for each of the policy options before and after the CJ process. To facilitate comparison of attitudes to preference, preference rakings were subjected to factorial analysis similar to the attitude Q sort. This resulted in four preference factors: Close Road, Stabilise, Bitumenise and Status quo. Archetypal preference rankings for each preference factor position are provided in Table 12.3. Figure 12.6 shows the loadings of jurors on each preference factor before and after the deliberation. It can be clearly seen from both Table 12.3 and Figure 12.6 that preferences changed substantially during deliberation. Two jurors in particular – Snoopy and Aswad – experienced a near complete reversal of their original orderings. During deliberation, preferences among jurors converged, although they did not reach consensus.20 At the beginning there was a small preference for minor road works to stabilise the Track and reduce run-off, or to maintain the status quo. By the end of deliberation, road closure was clearly the most preferred option. Correlating preferences and subjectivity It was evident from jury discourse even before the empirical results from the CJ were analysed that jurors had experienced dramatic changes in their attitudes concerning the Bloomfield Track issue. Jurors related during interviews, and in
Preference transformation through deliberation 277 80 70
Factor loading (x100)
60 50 40 30 20 10 0 -10 -20
ADV
AS W
BOA
J AN
J UL
KEI KOD J uror
First sort
MAT
P EA
Second sort
RAS
S NO
TAM
Third sort
Figure 12.4 Juror loadings on factor 3 ‘pragmatism’ 70 60
Factor loading (x100)
50 40 30 20 10 0 -10 -20 -30
ADV
ASW BOA
JAN
JUL
KEI
KOD MAT
PEA
RAS
SNO TAM
-40 Juror First sort
Second sort
Third sort
Figure 12.5 Juror loadings on factor 4 ‘symbolism’
questionnaires, how they felt they had changed their opinions and the emphasis of issues they felt were most important. Although this supports the hypothesis that preferences change with attitudes, alone it does not reveal a causal relationship. One way of detecting possible relationships between preference and attitude is to compare whether those individuals correlated to a preference factor also tended to load to a corresponding attitude factor. These overall correlations before and
Bitumenise
5 4 5 1 2 4 2 5 4 4 1 2
3 3.25 1.54
Codename
Adventure Aswad Boat Janine Julie Keith Koda Matilda Pearl Rastus Snoopy Tamarra
Overall rank Average rank St dev
4 3.33 1.15
4 3 3 4 1 3 5 4 3 3 2 5
Upgrade
Rank before CJ
1 2.17 1.03
3 1 1 3 4 2 1 2 2 1 3 3
Stabilise
2 2.42 1.24
2 2 2 2 4 1 4 1 1 2 4 4
Status quo
Table 12.2 Rank ordering of preferences for management options
Close
5 3.92 1.56
1 5 4 5 5 5 3 3 5 5 5 1 5 4.92 0.29
5 5 5 5 5 5 4 5 5 5 5 5
Bitumenise
Rank after CJ
4 3.92 0.51
4 4 4 4 3 3 5 4 4 4 4 4
Upgrade
3 2.50 0.80
3 3 2 3 1 1 2 3 3 3 3 3
Stabilise
2 1.92 0.67
1 1 3 2 2 2 3 2 2 2 1 2
Status quo
Close
1 1.75 1.14
2 2 1 1 4 4 1 1 1 1 2 1
278 Simon Niemeyer
Preference transformation through deliberation 279 Table 12.3 Factor scores for preference ranking Factor Description
1 2 3 4
Close road Status quo Bitumenise Stabilise
Rank preference for factor Bitumen
Upgrade
Stabilise
Maintain
Close
5 5 1 2
4 3 4 5
3 2 3 1
2 1 2 4
1 4 5 3
Figure 12.4 Correlation between preference and Q factors loading Preference factor Stage of deliberation
Close Before
After
Status quo Before After
Stabilise Before After
Preservation Attitude Optimism factor Pragmatism Symbolism
0.77 –0.51 –0.15 0.05
0.84 –0.26 –0.80 –0.09 0.16 0.32 0.00 0.43 0.24 0.07 –0.67 0.65 0.64 –0.31 0.23 0.04 0.04 –0.02 0.51 –0.05
Bitumenise Before After 0.41 0.06 –0.44 –0.09
–0.35 –0.20 0.16 –0.14
after deliberation are shown in Table 12.4. The following discussion explores these relationships and mechanisms that might explain such correlations. It does so by considering each preference factor in turn and those attitude factors that are significantly correlated to it. From Table 12.4 it can be seen that loading on the Close preference factor is strongly correlated with the Preservation attitude. It is also negatively correlated with both Optimism (decreasingly so) and Pragmatism (increasingly so). The nature of the changes in correlation between attitude and preference can be illustrated by plotting each juror’s loading on both factors at the stages before and after deliberation. Figure 12.7 shows loadings on the Preservation and Close Road factor spectrum. The asterisks represent jurors’ initial loadings (before deliberation), the solid dots represent the comparative loading of jurors on each factor at the conclusion of the deliberative process, and the slender arrows indicate the ‘migration’ of preference and attitude during deliberation. The more significant shifts (those that are subject to further scrutiny herein) are highlighted using bold arrows. The two heavily dashed regression lines (the lighter line indicates the relationship at the beginning of deliberation and the darker after deliberation) indicate the relationship between preference and attitude before and after deliberation respectively. From Figure 12.7, most jurors have migrated north-east, increasing loadings on both factors indicated by the broad arrow. Others (Keith, Julie and Boat) have moved north-west, increasing loading on Close, but decreasing on Preservation. Adventure’s shift is particularly notable: beginning the process at the top of the Preservation–Close spectrum, but retreating somewhat from this position as deliberation has progressed. That Preservation and Close should be so strongly correlated is intuitively appealing: given the way they feel that the track impacts on the region in concert with multiple pressures, a Preservationist would feel it to be the only solution to
280 Simon Niemeyer Before deliberation 100 80 60 40 20 0 -20 -40 -60 -80 -100 ADV1 ASW1 BOA1
JAN1
Close
JUL1
KEI1
Status quo
KOD1 MAT1
PEA1
RAS1
Bitumenise
SNO1 TAM1 Stabilise
After deliberation 100
80
60
40
20
0
-20
-40 ADV3
ASW3 BOA3 Close
JAN3
JUL3
KEI3
Status quo
KOD3 MAT3
PEA3
Bitumenise
RAS3
SNO3 TAM3 Stabilise
Figure 12.6 Changes to preference factor loadings before and after deliberation
Preference transformation through deliberation 281 JAN3
100 First Sort Final Sort
BOA3
TAM MAT PEAADV1RAS ASW SNO3 KOD3 ADV3
80
Factor loading (close road) × 100
60 40 KEI3
JUL3
TAM1 BOA1MAT1
20 0 KEI1
PEA1
-20
KODA1
RAS1ASW1
-40 JAN1 -60 -80 SNO1
-100 0
10
20
30
JUL1
40 50 60 Factor loading (preservation) ×100
70
80
90
100
Figure 12.7 Changes to loadings on ‘preservationist’ and ‘close road’ factors
protecting the area. However, while Preservation is the most significant attitude factor throughout, track closure only reached prominence by the end of the process. Why did Preservationists not vote more strongly for track closure from the outset? This anomaly might be partially accounted for by the possibility that jurors also initially loaded on other attitude factors which ‘polluted’ their preferences. The role of these other factors is discussed below. Another related possibility is that the preferences were not fully ‘constructed’ at the outset, the correspondent preference position crystallising during deliberation. There was some evidence to support the claim that jurors did not initially vote in a manner true to their convictions. One juror in particular expressed that she would have liked to rank track closure as her first preferences, but did not think that it was a realistic outcome – her opinion, and conviction, on this matter changed dramatically during deliberation. As revealed in Table 12.4, preference for the Status quo is positively correlated with Pragmatism throughout deliberation. During the process it has also become correlated with Optimism and strongly negatively correlated with Preservation. The position of jurors on the Pragmatism–Status quo spectrum is shown in Figure 12.8. From the figure it appears that only Adventure and Snoopy were influenced by Pragmatism such that they increased preference for the Status quo – the rest having more or less vertical or horizontal shifts. Keith and Julie did not increase their loadings on Pragmatism, but remained the most significantly loaded of the jurors within this spectrum. There is evidence here regarding how different attitudinal factors interact in forming preferences. Those individuals who increased their loadings in this spectrum were also significantly loaded on the Preservationist. However, while they clearly sympathised with the Preservationist attitude, they hesitated in whole-heartedly advocating road closure because they felt more proof was needed beyond that presented to prove the case beyond doubt.21 In Adventure’s case in particular, this is
282 Simon Niemeyer consistent with a decline on loading on both Preservation and Close (see Figure 12.7), as well as the jury discourse. Like Adventure, the impact of Pragmatism also marginally increased Snoopy’s preference for the Status quo, but migrating from an entirely different overall attitudinal position. From Figure 12.7 it can be seen that Snoopy experienced the most dramatic shift of all jurors, having begun the process at the low end of the Preservation–Close Road spectrum. Figure 12.9 shows the relationships between Optimism and Status quo. Snoopy began deliberation as an Optimist, decreasing loading significantly by the end. Based on the changes observed in Figure 12.9, increases in loading on the Optimist attitudinal factor appears to have affected the preferences of Keith and Julie at the end of deliberation. Both felt very strongly that the track should remain open to permit access to the area by anyone who wished to visit. Like other jurors, they also agreed with the substance of the preservationist position that the area is important and under threat, but sought to find outcomes whereby everyone could benefit. Keith in particular was keen to explore other solutions outside the mainstream of discourse among the jurors – advocating the use of alternative approaches such as hovercraft, or even a monorail. One plausible explanation for their position is that Keith and Julie were the elder members of the jury. They were concerned that Closure would effectively exclude access to those who, like them, were not able to use the walking track that would replace it. They were optimistic that a solution could be found to the satisfaction of all users. Due to restriction of time and evidence, the merits of these alternative options were not fully deliberated – with view to their cost and feasibility. Optimism alone does not account for the attitude of those jurors who loaded significantly on the Status quo. It appears instead to act in concert with Pragmatism. As can be seen in Figure 12.8, Keith and Julie were also highly loaded
100
J UL3
BOA1 AS W1
RAS1
KE I3
PE A 80
KE I1
Factor loading (Status quo)
60 MAT
40
SNO3ADV3 AS W
20
BOA3
TAM3 0
RAS3
ADV1 MAT3
KODA1 PE A3
J AN3
J UL1
J AN1 SNO1
KOD3
–20
–40
–60
–80 –20
TAM1
–10
0
10
20 30 40 Factor loading (Pragmatist)
50
60
70
Figure 12.8 Changes to loadings on ‘pragmatist’ and ‘status quo’ factors
80
Preference transformation through deliberation 283 BOA1 KEI3
JUL3
ASW1
RAS1
90 PEA1 KEI1 70
Factor loading (Status Quo)
50
MAT1 ADV3
SNO
30
ASW3
10
JUL1 TAM3 ADV1 MAT
PEA3
BOA3 JAN3 KOD1 RAS JAN1
–10
SNO1
KOD3 –30
–50 TAM1 –70 –10
0
10
20
30 40 Factor loading (Optimist)
50
60
70
80
Figure 12.9 Changes to loadings on ‘optimist’ and ‘status quo’ factors
on the Pragmatism factor. Optimism and pragmatism shared many features but were distinguished by emphasis on different facets of the issue. Stabilise and Bituminise are considered together here because of both their similarity in preference structure and interaction with attitudes. Most strongly associated with either stabilising the track, or bituminising is Symbolism (Table 12.4). The correlation between Symbolic and Stabilise is significant at the outset, but not by the end of deliberation. The explanation lies in the ability stabilisation or bituminisation (which is second preference for the factor, see Table 12.2) to address symbolic issues of reef damage and community access. For a Symbolist, choosing other options would place the competing goals of preservation and community access in tension with one another. As deliberation progressed and symbolic concerns dissipated, Symbolism was supplanted by emphasis on indirect regional and long-term impacts, epitomised by the Preservationist factor. Subsequently the loadings on Symbolism declined, along with a statistically detectible relationship with the Stabilise preference factor.
Conclusions Overall there appears to be a significant shift in attitudes resulting from discourse in the CJ. Jurors adopted a ‘big picture’ perspective of the issue and generally increased their predisposition toward preservation of the Daintree area. When questioned about these changes, many jurors cited that improved knowledge helped them better understand the implications of different actions (information effect). The decline of symbolic attitudes, with concomitant changes to preference ranking was one of the most striking outcomes of the process. This appears to reflect the ability of deliberation to decouple attitudes from extreme positions transmitted to
284 Simon Niemeyer the public domain through the political discourse of interest groups and issue polarisation. The evidence presented to the jury was based on a very small amount of research directly relevant to the road. Damage to the fringing reefs is one of the few impacts of the Bloomfield Track that has been studied to date. However, there was enough information to address many of the myths and symbols that had dominated the issue to that date. Most notable was the effect of evidence on the impact of the road on the fringing reefs. Findings from this research, as presented to the jury, do not support this contention. Rather than focus on these issues, jurors considered the broader and longer-term implications of the future of the Bloomfield Track. The fact that jurors were encouraged by the process to look at the issue from the perspective of the citizen, rather than what benefited them individually (as consumer) was also cited by jurors as a significant feature in changes to attitudes and preferences. Based on observations and interviews with jurors, this perspective change was most important insofar as it served as a pretext for the manner in which information was weighted and used in the judgement process. There is evidence in the Q sort for an attitudinal change that impacted on preferences. There is also evidence of ‘better construction’ of preferences, to the extent that individuals are better informed about what policy means best suit their desired ends and a concomitant increase in correlations between attitudes and preferences. Another lesser factor is increased fortitude of jurors to vote for those options they most preferred. Preferences tended to converge during deliberation, although discernible differences remained. Contours of difference emerged around different interpretations of evidence. These reflected individual predispositions, such as whether they believed science could provide certain answers, or whether technology could solve otherwise intractable problems. All jurors agreed on the importance of the area. However for some, the fact that science could not prove beyond doubt that the track caused irreparable damage led some jurors to adopt a slightly more conservative and pragmatic approach, deferring closure until more evidence becomes available. Two other jurors saw opportunities for ‘win–win’ in the evidence presented: that technically sophisticated options might satisfy competing ends. Differences also reflected a form ‘starting point bias’, where jurors who began loaded on one extreme of an attitude–preference spectrum, maintained a residual affinity with that position. The contours of difference, though still significant, became more clearly identifiable and therefore arguably easier to address. The question remains what role further deliberation directly addressing these difference may have. It is possible that directly addressing these issues might have the same impact as evidence by dissipating symbolic concerns, leading to acceptance by the rest of the jury. Based on the evidence from the CJ on the harmonising capacity of deliberation, it is less likely to lead to increasing polarisation, but this is dependent on a number of factors, not least being willingness of jurors to tolerate scrutiny of their formed position by others. In either case it seems that deliberation is never complete. Despite emergent differences among the jury, the overwhelming trend was toward increased concern for the longer-term and interconnected issues with
Preference transformation through deliberation 285 the Bloomfield Track, and a concomitant preference for track closure. This outcome is very different to preferences and attitudes elicited at the beginning of the process, which varied widely but tended to support maintaining the status quo. This chapter is part of a broader project of ecological economics to organise human institutions to elevate the cause of ecological sustainability. It is specifically concerned with processes that shape preferences to be consistent with this goal. Within this aim, formal deliberative processes such as CJs are an arguably better foundation for public input into environmental policy than aggregation exercises such as referenda or the contingent valuation method (see Niemeyer and Spash 2001, and also Aldred this volume, Chapter 8). The deliberative process helps to improve judgement as far as preferences tend to match means to desired ends better than when elicited without subjecting individuals to the deliberative process. The matching of preferred means to ends suits the goal of sustainability if, because of deliberation, each individual desires that collective outcome. Indeed, it appears that the attitudes of participants changed in a manner conducive to the provision of the common good. Deliberative approaches may also help to avoid a potential trap for ecological economics as the ‘science of sustainability’ where predefining outcomes subverts both democratic process and the achievement of sustainability. Although the trend favouring preservation was a strong feature of the CJ, the detail of jury discourse has revealed more subtle differences on the margins of debate about what is the most preferred outcome. The main driver of these differences was perceptions among the jurors about the nature of the scientific evidence and the ability of technical solutions to produce optimal outcomes. If unanimity is the benchmark for deliberative success, that the CJ produced multiple options could be considered a less than optimal result. An alternative view is that democratic process should help to identify substantive differences (for example Benhabib 1996). Within this view, the CJ marks an important step in defining substantive issue contours to be addressed by policy. These results are arguably more ‘democratic’ because the information and deliberative process provided individuals with the necessary tools to make their own judgements, rather than defer to information they might normally receive through conventional channels of political discourse. Where conventional democratic processes might normally provide individuals with predefined policies deemed by experts or elites to be ‘sustainable’, the deliberative process has produced positive contributions by identifying a number of possible paths. The result places the onus back on elites to establish better information from research, or better educate the public about the use of science in policy. Even if this was the only outcome of formal deliberative processes, they provide a surer footing for the development of sustainable policies. There are important caveats to the conclusions drawn here. First, the results are only from one deliberative process. Others have conducted similar processes, with mixed results (for example see Pelletier et al. 1999 and also the previous chapter in this volume). If anything, this proves that the outcome of deliberative forums is highly sensitive to context and design. Therefore, they are amenable to problems of political manipulation and legitimacy. An important question also concerns
286 Simon Niemeyer whether changes to preferences observed within the CJ could be replicated in the broader public if the political environment was more ‘discursive’ (in the sense of Dryzek 2000). If so, rather than rely exclusively on manufacturing formal deliberative process, a better approach is to facilitate critical deliberation in the public sphere. It is possible that adopting formal deliberative approaches in the medium term will facilitate this transformation, over the long term empowering citizens in the democratic process through improved and relevant information. However, with scope for manipulation, the adoption of deliberative approaches will depend on a genuine desire on the part of organisers and elites to replicate discursive ideals. An important conclusion is that the ability of deliberation to transform preferences cannot be divorced from an appreciation of political context. This provides even greater impetus for ecological economics to embrace the political dimension of the science of sustainability.
Notes 1 For example, in the case of voting, Downs (1957) observed that voters had little incentive to acquire information about alternatives because of the small marginal gain and individual impact on election outcome. 2 There is a possibility that deliberation may not always lead to better preferences. Symbolic preferences may not dissolve in the process of acquiring information during the deliberative process, but become entrenched. Depending on an individual’s ability to assimilate information, it is entirely possible that too much detail can lead to information overload and reduce capacity for judgement, for that is what contributes to symbolic preference in the first place. Within the deliberative process there is an ongoing tension between the acquisition and assimilation of information, which if not properly balanced can lead to dramatically different results (e.g. Pelletier et al. 1999). 3 There was a statutory obligation under state law that an environmental impact assessment be prepared for major public works, often treated as no more than a symbolic requirement in an administrative environment geared for development. To satisfy this requirement, the shire engineer who oversaw the construction of the road prepared a short impact assessment, which failed to find against road development. 4 A combination of factors has meant that the Bloomfield Track has effectively been in policy stasis since construction. This was partially broken in 1998 when, due to prohibitive cost of maintenance of a dirt track following steep contours in a heavy rainfall region, the Douglas Shire Council applied to the Wet Tropics Management Authority to upgrade the steepest section and lay gravel along major sections. They granted an approval for the works on the condition that the local council prepared a comprehensive environmental impact assessment with recommendations for providing for long-term management. To date the impact assessment has yet to be commissioned. 5 Jurors were accompanied on this trip by an engineer from the Douglas Shire who was on hand to answer questions about construction of the road, and a well-known local and tour operator in the Daintree area who gave the issue a historical and local context as well as pointing out major features. 6 Representation of indigenous interests was mainly achieved by the fortuitous recruitment of a Kuku-Ylanji (traditional owners of the Daintree area) woman as part of the jury. 7 Preferences were not surveyed at this midway point because of concern that, because of the simpler nature of the questionnaire, preferences surveyed too close together might be biased by jurors feeling that they should be consistent with previous preference orderings.
Preference transformation through deliberation 287 8 Although for this research there is a mixing of methodologies by assessing changes within individuals over time. 9 This approach, referred to as ‘forced distribution’ is commonly used in Q method to facilitate analysis by keeping means equivalent, but does not affect the actual result so long as the rest of the study is well designed (Brown 1980: 198). Some of the jurors found it difficult to conform to this imposed regime and were given the freedom to depart from the template where they felt doing so better represented their position. 10 Statements were grouped into whether they pertained to beliefs about the consequences or the fundamental beliefs about the importance of those consequences. Statements were also grouped according to whether they involved a level of knowledge about the issue, the way in which the issue was framed or conceived – a citizen or a consumer, for example – and the time frame over which the implications of decisions about the issue are considered. Statements within each category were phrased either as a positive statement, a negative statement, or a statement concerning the level of confidence in a particular belief. Expression of certainty is an important element of the concourse and including it in the survey assists the researcher to detect changes in confidence as a result of the deliberative process. 11 Importantly, when making recommendations at the end of the process the jury was not limited to these options. The ranking of preferences was simply used to follow the way in which their ordering of these particular options had changed. 12 The loading on statement 12 tended to reflect the juror’s predisposition to finding technical solutions (of which one was the construction of a mono, or sky rail) or regulatory solutions to the Bloomfield Track issue. 13 Because of the controversial nature of the issue, jurors were given the option of choosing codenames to protect their identity in publications arising from this research. 14 Significantly, most revealed markedly higher shifts during the first half of the process, in some cases ‘settling back’ toward their original position, though still substantially changed. Due to space restrictions, the implications of this cannot be discussed here. 15 Based on a level of significance of 1/ n 0.15 all jurors are significantly loaded on the factor at the outset, with only Keith defecting. However, although statistical significance is the commonly accepted test for difference, Q method is based on a measure of ‘psychological significance’, which is a phenomenologically rather than statistically determined level (Brown 1980: 198). Using the statistically accepted level for significance suggests a higher level of agreement among jurors than was actually observed during the CJ proceedings. For the purposes here, because of the high level of correlation between factors and significant loading of jurors on more than one factor, a level of 0.4 is adopted. 16 This type of response is consistent with the ‘pioneering’ ethic ascribed to remote areas of Northern Australia. Many authors portray early pioneers in terms of rapacious developers with little concern for the environment and future consequences (e.g. Lines 1992). Passmore (1974) illuminates the issue from a different perspective whereby actions were driven precisely because of concern for future generations, but unconditioned by awareness, or belief in, negative environmental consequences. If there is certainty that exploitation of environmental resources has negative consequences, then there is a case for changing course. 17 There are a number of similarities with the Individualist position identified in Steg and Sievers (2000). 18 This is also consistent with the pioneering ethic, as was epitomised by the Premier of Queensland at the time the Bloomfield Track was constructed (Passmore 1974: 77; Fitzgerald 1984). 19 Symbolic contrasts most strongly with the Pragmatic (r = 0.38), with disagreement on the question of the reef (7, 24, 33) and impact of the road on the rainforest (27). 20 One participant expressed pleasure that consensus was not reached because: ‘the fact that we did not come to a definite solution, I think, shows how successful the jury process is at tackling any problem … this problem does not have a definite solution’. 21 Many of the technical witnesses themselves pointed out the poor quality of the evidence directly relevant to the road gathered to that date.
288 Simon Niemeyer
References Barry, J. and Proops, J. (1999) ‘Seeking sustainability discourses with Q methodology’, Ecological Economics, 28: 337–46. Benhabib, S. (1996) ‘Toward a deliberative model of democratic legitimacy’, in S. Benhabib (ed.) Democracy and Difference: Contesting the Boundaries of the Political, Princeton, NJ: Princeton University Press. Brown, S.R. (1980) Political Subjectivity: Applications of Q Methodology in Political Science, New Haven: Yale University Press. Brown, S.R. (1993) ‘A primer on Q methodology’, Operant Subjectivity, 16: 91–138. Callicot, J.B. (1982) ‘Hume’s is–ought dichotomy and the relationship of ecology to Leopold’s land ethic’, Environmental Ethics, 4: 163–73. Common, M. and Perrings, C. (1992) ‘Toward an ecological economics of sustainability’, Ecological Economics, 6: 7–34. Downs, A. (1957) An Economic Theory of Democracy, New York: Harper & Row. Dryzek, J.S. (2000) Deliberative Democracy and Beyond: Liberals, Critics, Contestations, Oxford: Oxford University Press. Elster, J. (1983) Sour Grapes: Studies in the Subversion of Rationality, Cambridge: Cambridge University Press. Etzioni, A. (1986) ‘The case for a multiple utility conception’, Economics and Philosophy, 2: 159–83. Fitzgerald, R. (1984) From 1915 to the Early 1980s: A History of Queensland, St Lucia, Australia: University of Queensland Press. Goodin, R.E. (1996) ‘Enfranchising the earth and its alternatives’, Political Studies, 44: 835–49. Goodin, R.E. (2000) ‘Democratic deliberation within’, Philosophy & Public Affairs, 29: 81– 109. Harrison, C.M., Burgess, J. and Clark, J. (1999) ‘Capturing values for nature: ecological, economic and cultural perspectives’, in J. Holder and D. McGillivray (eds) Locality and Identity: Environmental Issues in Law and Society, vol. 3, Dartmouth: Ashgate: 85–110. Keat, R. (1994) ‘Citizens, consumers and the environment: reflections on “The economy of the earth” ’, Environmental Values, 2: 333–49. Kellert, S.R. and Wilson, E.O. (1993) The Biophilia Hypothesis, Washington, DC: Island Press. Kellow, A. and Niemeyer, S. (1999) ‘The development of environmental administration in Queensland and Western Australia: why are they different?’, Australian Journal of Political Science, 34: 205–22. Lines, W.J. (1992) Taming the Great South Land: A History of the Conquest of Nature in Australia, North Sydney: Allen & Unwin. Niemeyer, S.J. and Spash, C.L. (2001) ‘Environmental valuation analysis, public deliberation, and their pragmatic syntheses: a critical appraisal’, Environment and Planning C, 19: 567–86. Norton, B.G., Costanza, R. and Bishop, R.C. (1998) ‘The evolution of preferences: why ‘sovereign’ preferences may not lead to sustainable policies and what to do about it’, Ecological Economics, 24: 193–211. O’Neill, J. (1997) ‘Managing without prices: the monetary valuation of biodiversity’, Ambio, 26: 546–50. Papadakis, E. (1993) Politics and the Environment: The Australian Experience, St Leonards, NSW: Allen & Unwin.
Preference transformation through deliberation 289 Passmore, J. (1974) Man’s Responsibility For Nature: Ecological Problems and Western Traditions, London: Duckworth. Pelletier, D., Kraak, V., McCullum, C., Uusitalo, U. and Rich, R. (1999) ‘The shaping of collective values through deliberative democracy: an empirical study from New York’s North Country’, Policy Sciences, 32: 103–31. Rydin, Y. (1999) ‘Can we talk ourselves into sustainability? The role of discourse in the environmental policy process’, Environmental Values, 8: 467–84. Sagoff, M. (1988) The Economy of the Earth: Philosophy, Law and the Environment, Cambridge: Cambridge University Press. Sagoff, M. (1998) ‘Aggregation and deliberation in valuing environmental public goods: a look beyond contingent pricing’, Ecological Economics, 24: 213–30. Schkade, D.A. and Payne, J.W. (1994) ‘How people respond to contingent valuation question: a verbal protocol analysis of willingness to pay for an environmental regulation’, Journal of Environmental Economics and Management, 26: 88–109. Sears, D.O. (1993) ‘Symbolic politics: a socio-psychological theory’, in S. Iyengar and W.J. McGuire (eds) Explorations in Political Psychology, vol. 5, Durham: Duke University Press. Slovic, P. (1995) ‘The construction of preference’, American Psychologist, 50: 364–71. Smith, G. and Wales, C. (2000) ‘Citizens’ juries and deliberative democracy’, Political Studies, 48: 51–65. Spash, C.L. and Hanley, N. (1995) ‘Preferences, information and biological diversity’, Ecological Economics, 12: 191–208. Steg, L. and Sievers, I. (2000) ‘Cultural theory and individual perceptions of environmental risks’, Environment and Behaviour, 32: 250–69. van den Bergh, J., Ferrer-Carbonell, A. and Munda, G. (2000) ‘Alternative models of individual behaviour and implications for environmental policy’, Ecological Economics, 32: 43–61. Ward, H. (1999) ‘Citizens’ juries and valuing the environment: a proposal’, Environmental Politics, 8: 75–96. Wilderness Action Group (1983) The Trials of Tribulation, Mossman: Wilderness Action Group.
290 Index
Index
Aarhus Convention (1998) 4 agri-environmental forum (AEF) see GRANO project agri-environmental schemes 14 Agricultural Production Co-operation (LPG) 256 Ajzen, I. 52, 75; et al. 9; and Fishbein, M. 75; and Madden, T.J. 75 Aldred, J. 26, 192, 201, 204; and Jacobs, M. 197, 201, 205, 209, 228 analytic hierarchy process (AHP) 9, 103, 140–3, 148; application of 109; check logical consistence of pairwise comparisons 108–9; consistency 108–9, 130 (4n); construct suitable hierarchies 106–7; development of 106; establish priorities between hierarchies 107–8; in practice 109; random index 109, 130 (5n) Anand, P. 205 Andreoli, M. and Tellarini, V. 99, 129 Andreoni, J. 46 Aristotle 196 Armour, A. 201, 204 Arrow, K.J. 46, 161; et al. 23, 33, 45, 209 attitudes, CJs/Bloomfield Track 270–6, 283–4; group influences upon 76–8; information 15; as predictors of WTP within CVM 52–4; New Environmental Paradigm (NEP) 52–3; towards the environment 75; towards money as measure of value 60; towards nature/ climate change 57–9 Austen-Smith, D. 195, 196; and Riker, W. 196 Australia 15; forest regions 9–10, 134–56; parkland study 229–36, 240–1 (5n, 6n); wet tropics area 267–86 Austria 7, 24, 32–43
Ballestero, E. and Romero, C. 122 Bamberg, S. 52; et al. 65 Bandura, A. 78 Barry, J. and Proops, J. 269 Bateman, I.J. and Langford, I.H. 51, 54 Bayesian theory 198 Becker, G.S. 30, 72 behaviour see human behaviour Benhabib, S. 285 Berkes, F. and Folke, C. 244 Bernabo, J.C. 249 bid curve analysis 7 Bingham, G. et al. 31, 32 biodiversity 24, 26, 43, 45 (4n) biophysical systems 4 Bishop, R.C. and Heberlein, T.A. 53; and Welsh, M.P. 29 Blackburn, M. et al. 28 Blamey, R.K. 31, 209, 238; and Common, M. 227; et al. 51 Blaug, R. 240 Bloomfield Track (Australia), attitudes/ responses 270–6, 287 (14n–19n); conclusions concerning 283–6; construction of 268–9, 286 (3n, 4n); correlating preferences/subjectivity 276–83, 287 (21n); deliberation on 284–5; geographical location/ background 267–8; optimism concerning 274–5, 279, 281–3; polarisation of interests in 269; pragmatism concerning 275–6, 279, 282–3; preference shift 276, 287 (20n); preservation 271, 274, 279, 281; results of study 270–83; study methodologies 269, 286–7 (7n–11n); symbolism 276, 283; use of CJs 269–70, 286 (5n, 6n)
Index 291 Bohman, J. 192, 200, 204, 237; and Rehg, W. 204 Borders Forest Trust (BFT) 212 Borders Region (Scotland) 12, 211–12 Boulding, K.E. 27, 28, 29, 45 bounded rationality 197–9 Boyce, R. et al. 27, 31 Boyle, K.J. and Bergstrom, J.C. 212, 213; et al. 213 Brandenburg (Germany) 250–60 Breusch–Pagan test 63–4 Bromley, D.W. 249 Brouwer, R. et al. 201; and Slangen, L.H.G. 55 Brown, S.R. 269 Brown, T.C. et al. 5, 201, 209, 210, 228 Buchanan, J. 192 Burgess, J. et al. 29, 32 Burney, J. 2 Callicot, J.B. 266 Cameron, J.I. 31 Cardinal Consistency Index 144 Carinthia (Austria) 24; descriptive overview of empirical results 33–5; design of survey 33; empirical results of survey 32–43, 44, 46–7 (18n–23n); explaining WTP-bids 35–7, 43; importance of extended valuation framework 35–7, 43; social context 32 Carson, R.T. 26; et al. 212 Carter, J.R. and Irons, M.D. 29 Castells, N. and Munda, G. 99, 129 CHAID (Chi-squared Automated Interaction Detector) 60, 61–3 Chang, N.-B. et al. 124 charitable giving 7, 213 choice modelling 226, 240 (2n) Churchman, C.W. et al. 102 citizens’ juries (CJs) 14, 15, 17, 204–5 (3n), 225; advantages of 209; application of 228; Bloomfield Track 269–83, 285; cf contingent valuation method 11–12; design/conduct of 229–33; development of 228; Ettrick Valley project 214–22; issues concerning 240; park management 229–40, 241 (8n–10n); and pioneering ethic 287 (16n); results of deliberations 233–6, see also CVM/CJs comparison civil rights 4 Clark, J. and Murdoch, J. 249 Climaco, J. 104 climate change 57–9
Common, M. et al. 27, 31; and Perrings, C. 31, 263 commons dilemma 83–6 communicative influence 87; and attitude position 88; and biased processing 88; external input variables 87; interventions 89–91; personal/actual (state) values discrepancy 87; and persuasiveness of communicator 89; and self-responsibility 87–8 Comprehensive, Adequate and Representative (CAR) Reserve system 135 Comprehensive Regional Assessment (CRA) (Australia) 9, 134, 135, 136–7 Conlisk, J. 31 consumer boycotts 16 consumer–citizen dichotomy 24, 30–2, 43, 46 (12n, 15n) context 16 contingent valuation method (CVM) 2–3, 5, 7–8, 17, 51, 65, 209, 212–14, 225, 244, 285; and CBA 43; criticisms of 43, 226–7; (dis)agreements concerning 23; extended model 24, 26–32; and home-grown values 29; and hypothetical bias 23, 28; and individual willingness to pay 23; institutional context 28–30; lexicographic preferences/willingness to exchange 24–7, 46 (8n); and respondents’ beliefs/perceptions 24; social context 24, 30–2; standard model 24, 45 (2n); in theory 205 (9n); usefulness of 44; and valuation questions 29; and value construction 29; and welfare economics 43, see also CVM/CJs comparison Contract Nature Conservation Programme (Germany) 251–2 Coote, A. and Lenaghan, J. 192, 201, 204, 209, 210, 228, 236 cost–benefit analysis (CBA) 1, 23, 43, 75, 102, 160, 226–7; and attitudes 52–4; and benevolence/malevolence 27; counter-critiques 17; criticisms of 1–2, 3–4, 27; and heroic ethic 27; popularisation of 2; and preferences 26; validity in applying 4 Crosby, N. 201, 204, 210, 228 Cummings, R.G. et al. 28 Curry, N. and Winter, M. 251 CVM/CJs comparison 11–12; advantages/weaknesses 211;
292 Index agreement/deliberative valuation 200–3; background 187–8; bounded rationality/new perspectives 197–9; decision-making processes 189–90, 205 (4n, 5n); deliberation/discussion 190–1; economic appraisal methods 209; epistemic merits 191–2; equity/ distributional issues 210; feasibility constraint 203–4; policy implications 204; pooling private information 195–7; questionnaires 209–10; and rationality 210–11; restoration project 211–22; and value construction 210 Daneke, G.A. et al. 225 de Dios, E.S. 26 De Montis, A. et al. 100, 125, 129 DeBono, E. 245 Dechamps, H. 211 decision making 31–2, 112, 139, 157 (2n), 211; and AHP 106–9, 109; and ELECTRE III 113–15; and Evamix 109–12; by lexicographic ordering 27; and MAUT 102–3, 102–6, 105; and MOP/GP 121–4; and NAIADE 118–21; processes 189–90; quality of 4; and Regime 115–18; social/ institutional context 24, 45 (3n); software packages 102–3, see also multiple criteria decision analysis (MCDA) deliberative democracy 187, 190–1, 205 (6n), 225, 227–8, 240 (3n); and agreement 200–3; as alternative to aggregative models 188, 204 (2n); appealing to principles 193–5, 205 (8n); avoiding appearing selfish 194–5; and better dicisions 191–2; and ‘cheap-talk’ literature 196; and credible communication 196; critics of 195–7, 199; diagnostic/constructive distinction 191; and divergent interests 198–9; empirical/theoretical 193; and rational choice/game theories 191; strategic reasons 193–4; theories of 188, 204 (1n); and trust 198, see also discursive democracy deliberative monetary valuation (DMV) 4, 5–6, 11–14, 15, 17, 201–2, 204 deliberative process 263–4, 266, 284–6 (2n) dichotomous choice (DC) format 213 Diekmann, A. 58, 69, 72; and Franzen, A. 93 Dienel, P.C. 225, 228
Dillon, J.L. and Perry, C. 102, 103, 104, 105 discursive democracy 227–8, 240 (3n) Dovers, S. 139 Downs, A. 286 Doyle, T. and Kellow, A. 139 Dryzek, J.S. 227, 236, 240 Dubourg, W.R. et al. 45 Dunlap, R.E. and van Liere, K.D. 52 Dwyer, W.O. et al. 93 Eagly, A.H. and Chaiken, S. 52, 58 ecological systems 4 economic utility theory 26 economic values, focus on CVM 6; and psychology of behaviour 6–9; specific validity issues 6–7 economics 28–30 Edwards, S.F. 26 Ehrlich, P. and Ehrlich, A. 26 Elaboration Likelihood Model 87–9 ELECTRE III 9, 113–15, 117 Elster, J. 190, 192, 193, 210, 211, 263, 264, 265, 266 Encarnación, J. 26 Environment and Heritage Assessment (Australia) 136 environmental assets 55 environmental behaviour, assumptions 75; and attitudes 92; background 69–70; commons dilemma 83–6; discussion of findings 91–3; experiments/results 76–91; framework model 71–6; group influences upon attitudes 76–8; internal/external input variables 72, 73; and low susceptibility to influence 91; pioneers 69–70, 78, 287 (16n); processing communicative influence 87–91; and role models 91; simulation procedure 70–1; and social influence 92–3; social learning of behaviour 78–83; and state of the resource 91; sub-models 74–5; turn-arounds in 70 ethics 9, 27, 287 (16n) Ettrick Valley Forest floodplain project 12, 211–12; access/future management concerns 215–16; benefit of 214; charitable donation payment/openended elicitation format 213; CJ survey 214–16, 223 (5n); CVM survey 212–14, 222 (1n), 223 (3n); decisionmaking process 221–2; deliberations 214–15; funding for 220–1; and policy-relevant information 219–20,
Index 293 222; positive/negative views on 218, 219; qualititative/quantitative data on 217–18; questionnaires used 221, 223 (7n); recommendations concerning 215; research conclusions 219–22, 223 (7n); valuation workshops 216–19; and willingness to pay 213–14 Etzioni, A. 30, 266 evaluation matrix (Evamix) 9; appraisal scores 111–12; development of 109; dominance scores for ordinal/cardinal criteria 110–11; ordinal/cardinal distinction 109–10; overall dominance scores 111; in practice 112; standardized dominance scores for ordinal/cardinal criteria 111 existence value 3 expanded rational choice model 72–6; attitude 72; behavioural intentions 75–6; commons dilemma 83–6; conclusions concerning 91–3; group influences upon attitudes 76–8; knowledge 74; motives 74; persuasiveness 74; processing communicative influence 87–91; state of resource 74; self-responsibility 74; situational factors/incentives 74; social learning of behaviour 78–83; status 74; sub-models 74–5; use 72; values 74; and volition 75–6 Exxon Valdez 2 Farrell, J. and Gibbons, R. 195 Fearon, J. 191, 192, 193, 195, 197, 198 Ferguson, I. 139 Fischhoff, B. 31, 53; and Furby, L. 31 Fishbein, M. and Ajzen, I. 45, 52 Fishburn, P.C. 102 Fisher, E. and Harding, R. 167, 171 Fitzgerald, R. 287 focus groups 5, 16, 201, 213 Forest and Resource Management System (FRAMES) 137 Forest Resource Use Model (FORUM) 136, 137–8 forest resources see Southern Forest Region (New South Wales) Fransson, N. and Gärling, T. 65 Frey, B.S. 31 Frey, D. 76; et al. 75, 76, 77 Funtowicz, S.O. 4; and Ravetz, J. 99, 162 Gambetta, D. 192 Garrod, G.D. and Willis, K.G. 51
Gastil, J. 237 Generalized Algebraic Modelling System (GAMS) 136 genetically modified (GM) crops 10; application of MCM to 172–80; criteria 177–8; findings 175–80; options 176; pilot study on 172–5; ranking of options 177–80; UK debate over 172; use of MCM 172–82 Geographic Information System (GIS) 137 Georgiou, S. et al. 24, 52 Germany 14, 250–60 Getzner, M. 28, 32 Gonzales, M.H. et al. 90 Goodin, R.E. 266 Goodman, S. et al. 57 governments 10, 16 Gowans, C. 195 Gowdy, J. 26; and Olson, P.R. 26 GRANO project 260 (1n); analysis of agri-environmental forum 256–8; criticisms by participants 256; development of 250–2; discussion of measures 255–6; financial aspects 253; implementation of 252; inefficent decision-making 259–60; information flow 253, 259; interdisciplinary team 252; knowledge concerning 255; learning process 253, 258–9; meetings/proceedings 253–6; relevance of topics 253, 259; results 258; stakeholders 252 Green, C. 65; et al. 64; and Tunstall, S. 65 Green Party 16 Greene, W.H. 63 Greenhouse Effect 2 Gregory, R. et al. 210, 226; and Slovic, P. 226, 227 Grove-White, R. et al. 172 Gutmann, A. and Thompson, D. 192 Habermas, J. 240, 245 Hagedorn, K. 247, 249 Hahnemann, W.M. 46 Hampicke, U. et al. 252 Hanley, N. et al. 26, 29; and Spash, C.L. 1 Hanna, S. et al. 244 Harrison, C.M. et al. 265 Harrison, G.W. 28 Harsanyi, J.C. 69 Hayden, F.G. 24 Hayek, F.A. 164 hedonic pricing 5
294 Index Hinloopen, E. 115; et al. 115; and Nijkamp, P. 115, 116, 117 HIPRE 3+ 103 Hodgson, G.M. 24 Hoehn, J.P. and Randall, A. 46 Hofer, D. 57 Hogwood, B. and Gunn, L. 139 Holland, A. 32 Holling, C.S. et al. 244 Holm-Müller, K. et al. 29 homo oeconomicus 23, 29 Howlett, M. and Ramesh, M. 139 human behaviour 3–4; altruism 30; assumptions concerning 71; CBA component 8; and cognition 69; economic/non-economic motives 7–8; and environmental consciousness 69–93; and ethics 9; expanded rational choice model for 72–6; free-rider 29; importance of attitudes 7–8; personal/ social factors 69; and ‘pioneers’ 8, 287 (16n); predictability of 65; and pressure to conform 77–8; and processing of information 8–9; psychology of 6–9; rational 30; and self-concept 77; and self-confirmation 81; social comparison processes 76–7; social learning of 78–81; social/ personal factors 8; and status 81; sustainability motives 8 Hwang, C. and Yoon, K. 103 Hylland, A. 192 hypothetical bias 28 Ignizio, J.P. 121 incertitude 163–5 indifference curve 26 interactive valuation process, analytical framework for 246–50; background 244–6 Jacobs, M. 5, 31, 188, 201, 204, 209, 210, 228 Jakobsson, K.M. and Dragun, A.K. 28 James, R. and Blamey, R.K. 201, 202, 203, 209 JANIS criteria 135 Janssen, R. 99, 129 Jeffrey, R.C. 26 Jiggins, J. and Röling, N. 244, 245, 248 Johansson, P.-O. 26 Johnson, J. 193–4 Johnson, S. 196 Jorgenson, B.S. et al. 55
Jorgenson, D.O. and Papciak, A.S. 81 Journal of Economics Perspective 29 Kahneman, D. et al. 52; and Knetsch, J.L. 46; and Tversky, A. 3 Kaiser, G.F. et al. 53 Kapp, K.W. 1 Keat, R. 46, 266 Keeney, R.L. and Raiffa, H. 102 Kellert, S.R. and Wilson, E.O. 266 Kenyon, W. 215; et al. 5 Klandermans, B. 75 Kmietowicz, Z.W. and Pearman, A.D. 116 Knetsch, J.L. 3; and Sinden, J.A. 4 Knight, F. 129 Korhonen, P. 121, 122 Kosz, M. 31 Kotchen, M.J. and Reiling, S.D. 52, 53, 59, 64 Krehbiel, K. 195 KULAP (cultural environmental landscape programme) 251 Kuper, R. 204 Lane, R. 192 Lara, P. and Romero, C. 122 Latané, B. 79, 80, 92; and Wolf, S. 79 Lazo, J.K. et al. 29 learning process 245, 249–50 Lee, S.M. 121 lexicographic preferences 26–7, 43, 44, 45 (7n), 46 (8n–10n) Liebrand, W.B.G. 83; et al. 83, 192 Loasby, B. 162 Lockwood, M. 26, 27 Lothian, J. 139 Luce, R.D. 105 Luhmann, N. 26, 46 McDaniels, T.L. and Roessler, C. 29 Macedo, S. 192 Mackie, G. 196 Mann–Whitney test 56 Mardle, S.J. et al. 123 Martinez-Alier, J. et al. 4 Marwell, G. and Ames, R.E. 29 Massam, B. 138 Mayer, S. et al. 172 Mendoza, G.A. and Prabhu, R. 99, 129 Menegolo, L. and Guimarães Pereira, A. 118, 120 Menger, C. 26 Meppem, T. and Gill, R. 245, 248
Index 295 Milgrom, P. 239 Miller, D. 192, 202 Mitchell, R.T. and Carson, R.C. 33, 213 Moisseinen, E.M. 53, 66 monetary valuation 60; concerns about 4; as essential 1; methods used 1 Montreal Process 135 Mosler, H.-J. 71, 72, 81, 83; et al. 89; and Gutscher, H. 83, 93; and Tobias, R. 75 Müller, K. et al. 251 multi-criteria mapping (MCM) 10–11, 14, 171; application of 172–80 Multi-Objective-Programming/Goal Programming (MOP/GP) 9, 121; additions to 123; application 123; choosing most preferred solution 121–2; criticisms of 124; development of 124; expectations 122; nondominated solutions 121; and problems of reliability 122–3 multiple attribute utility theory (MAUT) 102; based on utility theory 103; development of 105; and interdependence of attributes 105–6; key assumptions 105; preparation for 103–4; software 102–3; and stakeholder participation 106; submethodology of 104; summary of 106 multiple attribute value theory (MAVT) 9, 104–5 multiple criteria analysis (MCA) 9–11, 45, 134, 250, 259; and AHP 140–3; as aid to decision-making process 139–40; different types of 140; forest study 143–56; in practice 138–9; and WTP determinants 60–4 multiple criteria decision analysis (MCDA), background 100; comparison of methods 124–8; concluding remarks 128–9; outranking methods 112–21; programming methods 121–4; quality assessment of methods 100–2; single synthesizing criteria 102–12; and stakeholder participation 134–56, see also decision making multiple regression analysis 63–4 Munda, G. 99, 118, 128–9, 130, 246; et al. 118 Munro, A. and Hanley, N. 209 Nagel, T. 195 National Estate Assessment (Australia) 136
National Oceanic and Atmospheric Administration (NOAA) 3, 52, 213 natural goods 23, 29 nature 59 New South Wales 13; forest region 134–56; parkland study 229–36, 240–1 (5n, 6n) Nijkamp, P. and Voodg, H. 116 non-use values 54–5, 64–5 Norton, B.G. et al. 263 novel approach to imprecise assessment and decision environments (NAIADE) 9; aggregation of criteria 119–20; development of 118; evaluation of alternatives 120; fuzzy numbers 118–20, 130 (10n, 14n); linguistic variables 119, 130 (11n); pairwise comparisons 118–19; stakeholder participation 121 Nowak, A. et al. 79; and Latané, B. 79, 81 O’Connor, M. 46, 188, 244, 248; et al. 4 O’Hara, S. 32, 244, 245, 248, 249 Olson, D.L. 99, 129 O’Neill, J. 1, 11, 13, 16, 196, 204, 263; and Spash, C.L. 249, 250 Orbell, J. et al. 192 O’Riordan, T. and Cameron, J. 167, 171; and Jordan, A. 167, 171 Ostrom, E. 13, 245, 246, 247, 248, 249 outranking methods 112–13; ELECTRE III 113–15; NAIADE 118–21; Regime 115–18 parkland study (New South Wales), choice of jurors 232; and consensual decisionmaking 236–7; deliberative valuation 235–6; design/conduct of CJs 229–33; discussion/decision-making processes 232–3; issues raised 236; options chosen 233–5; options given 229–32; and provision of information 238; results 233–6; and sample representation 238–9, 241 (10n); scenario for 229; structure of deliberative/polling processes 237; and within-process (in)equality 237–8 participatory approach 4, 5, 11–15, 16; forest resources 134–56; GRANO project 250–60; in MAUT 106; park management 225, 226–7, 240 (1n); stakeholders 5, 16–17, 106, 112, 121, 134, 247–8, 250–60; through cost– benefit analysis 226–7
296 Index Passmore, J. 287 Pate, J. and Loomis, J.B. 63 Payne, J.W. and Bettman, J.R. 210 Pelletier, D. et al. 285, 286 Perman, R. et al. 227 Peterken, G.F. and Hughes, F.M.R. 211 Pettit, P. 205 Petts, G. 211 Petty, R.E. and Cacioppo, J.T. 87 Peukert, H. 26 planning cell method (Planungszelle) 228 policy 16, 44–5 Prato, T. 140 precautionary approach 180, 182; implementation of 170–1; and regulatory appraisal of environmental risk 167–70 precautionary principle 159, 167 PREFCALC 102 preferences 13, 15; and attitudes/values 266–7; background 263–4; case study 267–83; and CBA 26; and construction process 266; convergence of 284; and deliberation 263–4, 266, 284–6, 286 (2n); endogenous 263; and hypothetical bias 28; individual 106; and information 265, 286 (1n); lexicographic 26–7, 43, 44, 45 (7n), 46 (8n-10n); quality of 263; shift in 264–7, 276; and subjectivity 276–83; sustainable 263–4; symbolic 265; use of term 264–5 Pressy, R. et al. 137 Pretty, J. et al. 215 principle of compatibility 54 probability 198 Proctor, W. 138 Productive Capacity of the Forest and Biodiversity 144 programming methods 121–4 prominence hypothesis 27 prospect theory 3 psychology 3–4, 6–9, 71, 72, 76 public goods 29, 30 Queensland 15 Quiggin, J. 239 Raffensberger, C. and Tickner, J. 167, 171 Randall, A. 65 rational choice theory 161, 192, 198, 205 (7n) Ready, R. et al. 213 reasoned action 52–3
Regime 9, 115–18 Regional Forest agreements (RFAs) 134, 135, 143–4, 155 Regional Forest Forum (Australia) 143–4 regulatory appraisal of risk 10, 159; applying MCM technique 172–80; conventional reponses 160; divergent approaches 160–1; implementing precautionary approach 170–1; incertitude concerning 162–5; practical consequences of assessment 165–7; science/precaution in 167–70, 180 Reise, K. 51 Renn, O. and Klinke, A. 171 representation 14, 16 Resource and Economic Assessment (Australia) 136 resource environment valuation 83–6 Rietveld, P. 116 rights-based beliefs 4, 27 Riker, W. 195 Rippe, K. and Schaber, P. 205 risk 103, 129 (1n), 198; evaluative criteria 171; and ignorance 162–3; implementation of precautionary approach 170–1; and outcomes 162; practical consequences of assessment 165–7; precautionary approach 167–70, 180, 182; and probability 162–3; science-based approach 159–61, 164, 167–70, 180, 182; and uncertainty 162–5 Rosenbaum, N. 225 Rottumeroog island 54 Rousseau, J. 225 Roy, B. 100, 113, 115; and Bouyssou, D. 104 Royal Commission on Environmental Pollution (1998) 16 Rydin, Y. 264 Saaty, T.L. 106, 107, 130, 140, 144; and Alexander, J.M. 106; and Forman, E. 106; and Vargas, L. 141 Sachs, L. 60 Sagoff, M. 27, 30, 191, 203, 209, 210, 227, 228, 244, 245, 263, 266 Sahlins, M.D. 244 Saimaa seal 53 Sally, D. 192 Salminen, P. et al. 99, 129 Salo, A. 164 Sanders, L. 204
Index 297 Scarelli, A. 113 Schkade, D.A. and Payne, J.W. 31, 46, 210, 266 Schulze, W. et al. 28 Schumann, S. 58 Scotland 12, 211–12 Sears, D.O. 265 Sen, A. 192, 202, 205 sensitivity analysis 150–2 Silberman, J. and Gerlowski, D.A. 55 single synthesizing criteria 102; AHP 106–9; Evamix 109–12; MAUT 102–6 Slocum, R. and Thomas-Slaytyer, B. 240 Slovic, P. 263, 266 SMART (Simple Multi-Attribute Rating Technique) 102, 104, 129 (2n) Smith, G. and Wales, C. 200, 203, 205 Smithson, M. 162 social choice theory 161 social dilemma situations 246–8 social/institutional context 7 socio-economic systems 4 Söderbaum, P. 244 Solow, R.M. 26 Southern Forest Region (New South Wales) 10; background 134, 157 (1n); case study 143–4; forest assessments 135–8; forum preferences 144–6; merits/flaws of study 156; outcomes 152–6; performance of options against criteria 146–50; problems 156; results 144–50; sensitivity analysis 150–2; study method 138–43 Spada, H. and Ernst, A.M. 83 Spaninks, F.A. et al. 55 Spash, C.L. 2, 4, 7, 8, 9, 12, 123; and Carter, C. 1; et al. 4, 7; and Hanley, N. 7, 26, 27, 266; and Simpson, I.A. 4 SPSS-CHAID analysis 61–3 Stahlberg, D. and Frey, D. 87 Stalder, J. 78 stated preferences surveys 226–7 Steuer, R.E. and Gardiner, L.R. 122 Stevens, T.H. et al. 27, 31, 35, 46 Stewart, J. et al. 209 Stirling, A. 164, 165, 166, 169, 171, 172, 174; and Mayer, S. 140, 171, 172, 174, 176 Stork, P. and Viaene, J.-M. 26 substitutability 24, 26 Sugden, R. 205 sustainability 4, 8, 75, 83–6, 263–4 Sutherland, R.J. and Walsh, R.G. 63 system theory 46 (13n)
Taha, H.A. 117 Tajfel, H. 76 Tamiz, M. et al. 122, 123; and Jones, D. 123 Tanford, S. and Penrod, S. 77 tax 1, 18n Theory of Planned Behaviour 75 Theory of Resource Mobilization 75 Thibaut, J.W. and Kelley, H.H. 76 Thomas-Slatyer, B. 225 Toman, M.A. 32 trade-offs 4, 26, 27, 43, 201 travel cost method 4 trust 13–14, 247, 249 Turner, B. 137 Tversky, A. et al. 27 Uckermark (Germany) 14, 252 uncertainty 4, 162–5, 198 utilitarianism 30, 226 valuation theory 1–6, 11–15 Van Delft, A. and Nijkamp, P. 116 van den Bergh, J. et al. 263 Van Valey, T.L. and Petersen, J.C. 225 Vatn, A. and Bromley, D.W. 27, 205 Vincke, P. 104, 113, 122, 123, 129 Von Neumann, J. and Morgenstern, O. 105 Von Schomberg, R. 172 Voogd, H. 102, 109, 114 Wadden Sea 7, 54; background 51–2; discussion/conclusions of survey 64–5; influence of attitudes on WTP 57–60; multiple analysis of WTP determinants 60–4; and protest bids 55–6; study design/main results 4–6, 66 (1n-4n) Ward, H. 5, 201, 202, 203, 205, 266 Water and Soil Resorces 144 Webler, T. and Renn, O. 225 welfare theory 13 Wet Tropics World Heritage Area (Australia) 15, 268 wetlands see Bloomfield Track (Australia) Wierstra, E. 51, 57, 60, 64 Wilderness Assessment (Australia) 136 Williams, B. 195 willingness to accept (WTA) 3, 7 willingness to pay (WTP) 2, 6, 7–8, 13, 23, 24, 31, 65, 213–14, 222–3 (2n), 226, 241 (10n, 11n); attitudes as predictors of 52–4; decisions concerning 239, 241 (11n); deliberative evaluation
298 Index process 239; empirical results of survey 32–43; explaining WTP-bids 32, 35–7, 43; influence of attitudes on 57–60; logit model estimated coefficients for 39–42; multiple analysis of determinants 60–4; regression model variables 38; and use values 55
Wujal Wujal Aboriginal Community 15, 268 Wynne, B. 162; and Mayer, S. 172 Yon, D. and Tendron, G. 212 Zionts, S. and Wallenius, J. 122