The International Yearbook of Environmental and Resource Economics 2004 2005: A Survey of Current Issues (New Horizons in Environmental Economics)

The International Yearbook of Environmental and Resource Economics 2004/2005

NEW HORIZONS IN ENVIRONMENTAL ECONOMICS Series Editors: Wallace E. Oates, Professor of Economics, University of Maryland, USA and Henk Folmer, Professor of General Economics, Wageningen University and Professor of Environmental Economics, Tilburgh University, The Netherlands This important series is designed to make a significant contribution to the development of the principles and practices of environmental economics. It includes both theoretical and empirical work. International in scope, it addresses issues of current and future concern in both East and West and in developed and developing countries. The main purpose of the series is to create a forum for the publication of high quality work and to show how economic analysis can make a contribution to understanding and resolving the environmental problems confronting the world in the twenty-first century. Recent titles in the series include: Environmental Policy Making in Economics with Prior Tax Distortions Edited by Lawrence H. Goulder Recent Advances in Environmental Economics Edited by John A. List and Aart de Zeeuw Sustainability and Endogenous Growth Karen Pittel The Economic Valuation of the Environment and Public Policy A Hedonic Approach Noboru Hidano Global Climate Change The Science, Economics and Politics James M. Griffin Global Environmental Change in Alpine Regions Recognition, Impact, Adaptation and Mitigation Edited by Karl W. Steininger and Hannelore Weck-Hannemann Environmental Management and the Competitiveness of Nature-Based Tourism Destinations Twan Huybers and Jeff Bennett The International Yearbook of Environmental and Resource Economics 2003/2004 A Survey of Current Issues Edited by Henk Folmer and Tom Tietenberg The Economics of Hydroelectric Power Brian K. Edwards Does Environmental Policy Work? The Theory and Practice of Outcomes Assessment Edited by David E. Ervin, James R. Kahn and Marie Leigh Livingston The International Yearbook of Environmental and Resource Economics 2004/2005 A Survey of Current Issues Edited by Tom Tietenberg and Henk Folmer

The International Yearbook of Environmental and Resource Economics 2004/2005 A Survey of Current Issues

Edited by

Tom Tietenberg Mitchell Family Professor of Economics, Colby College, USA and

Henk Folmer Professor of General Economics, Wageningen University, The Netherlands and Professor of Environmental Economics, Tilburg University, The Netherlands

NEW HORIZONS IN ENVIRONMENTAL ECONOMICS

Edward Elgar Cheltenham, UK • Northampton, MA, USA

© Tom Tietenberg, Henk Folmer 2004 Chapter 2 © Shell International Exploration and Production B.V. 2004 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical or photocopying, recording, or otherwise without the prior permission of the publisher. Published by Edward Elgar Publishing Limited Glensanda House Montpellier Parade Cheltenham Glos GL50 1UA UK Edward Elgar Publishing, Inc. 136 West Street Suite 202 Northampton Massachusetts 01060 USA

A catalogue record for this book is available from the British Library

ISSN 1460 7352 ISBN 1 84376 681 7 (cased)

Typeset by Cambrian Typesetters, Frimley, Surrey Printed and bound in Great Britain by MPG Books Ltd, Bodmin, Cornwall

Contents List of figures List of tables List of boxes List of contributors Preface Editorial board

vi vii viii ix xi xii

1. Fifty years of contingent valuation V. Kerry Smith 2. Environmental policy, induced technological change and economic growth: a selective review Wolfgang K. Heidug and Regina Bertram 3. Land use decisions and policy at the intensive and extensive margins Ian W. Hardie, Peter J. Parks and G. Cornelis van Kooten 4. Indicators of sustainability Eric Neumayer 5. Value transfer and environmental policy Ståle Navrud 6. Joint implementation in climate change policy Suzi Kerr and Catherine Leining 7. Environmentally harmful subsidies Jean-Philippe Barde and Outi Honkatukia Index

1

61

101 139 189 218 254

289

v

Figures 2.1 2.2 2.3 2.4 2.5 2.6 3.1 3.2 4.1 4.2 4.3 4.4 7.1 7.2 7.3 7.4 7.5 7.6 7.7

Effects of knowledge growth and shadow cost on optimum abatement schedule Optimal carbon tax in the static benefit–cost case Welfare gains resulting from abatement cost reduction due to technological innovation Firm-level incentives to adopt under pollution control Steady-state solution for the endogenous and exogenous growth case Impact of environmental policy on long-term environmental quality and interest rate Relationship between stand age and amenity value, various amenities Non-convexities and optimal Hartman–Faustmann rotation age Genuine savings rates in per cent of GNP Genuine savings rates for Saudi Arabia in per cent of GNP United States GDP versus GPI per capita Index of GDP, ISEW and corrected ISEW for the United Kingdom Market and government intervention failures Subsidies in OECD countries (most recent years) Linkages between support measures and environmental effects Composition of producer support estimate (1986–2002) Producer support estimate by country Support to coal in selected OECD countries (US$ million) Support to coal in selected OECD countries (US$/tce)

vi

68 70 72 77 84 88 120 122 144 148 153 158 257 262 267 278 279 280 281

Tables 1.1 1.2 1.3 1.4 2.1 2.2 4.1 4.2 5.1 7.1 7.2 7.3 7.4

Behavioral models linked by use and non-use motives A comparison of the features of four surveys Tests for scope with Southern California survey Results for joint estimation with complementary samples Summary of results of the Goulder and Mathai (2000) model Ratio of welfare gains from innovation to welfare gains from pollution control, WH/WP Ecological footprint and deficit of selected countries and the world Relative change (%) in MF and MF intensity in 15 EU member states, 1980–2000 Classification of environmental valuation techniques Overview of subsidy measurement approaches Subsidies in OECD countries Environmental impacts of subsidies removal Some examples of possible environmentally harmful consequences of water-related subsidies

vii

10 18 22 39 71 75 161 166 194 260 263 269 283

Boxes 5.1 Damage function approach applied to emissions to air and water 6.1 Key ‘Kyoto’ jargon for the uninitiated 7.1 Relative potential impacts of agricultural producer support measures on the environment

viii

193 221 276

Contributors Jean-Philippe Barde, Head of the National Policies Division, OECD Environment Directorate, Paris, France Regina Bertram, Institute of Economic Theory, Department of Economics, University of Hagen, Germany Ian W. Hardie, Department of Agricultural and Resource Economics, University of Maryland, USA Wolfgang K. Heidug, Shell International Exploration and Production B.V., Rijswijk, The Netherlands Outi Honkatukia, Principal Administrator at the Private Office of the OECD Secretary General G. Cornelis van Kooten, Department of Economics, University of Victoria, Canada Catherine Leining, Center for Clean Air Policy, USA Suzi Kerr, Motu Economic and Public Policy Research, New Zealand Ståle Navrud, Department of Economics and Resource Management, Agricultural University of Norway, Norway Eric Neumayer, Department of Geography and Environment, London School of Economics and Political Science, UK Peter J. Parks, Department of Agricultural, Food, and Resource Economics, Cook College, Rutgers University, USA V. Kerry Smith, University Distinguished Professor, North Carolina State University and University Fellow, Resources for the Future, USA

ix

Preface As a discipline, Environmental and Resource Economics has undergone a rapid evolution over the past three decades. Originally the literature focused on valuing environmental resources and on the design of policy instruments to correct externalities and to provide for the optimal exploitation of resources. The relatively narrow focus of the field and the limited number of contributors made the task of keeping up with the literature fairly simple. More recently, Environmental and Resource Economics has broadened its focus by making connections with many other subdisciplines in economics as well as the natural and physical sciences. It has also attracted a much larger group of contributors. Thus the literature is exploding in terms of the number of topics addressed, the number of methodological approaches being applied and the sheer number of articles being written. Coupled with the high degree of specialization that characterizes modern academic life, this proliferation of topics and methodologies makes it impossible for anyone, even those who specialize in Environmental and Resource Economics, to keep up with the developments in the field. The International Yearbook of Environmental and Resource Economics. A Survey of Current Issues was designed to fill this niche. The Yearbook publishes state-of-the-art papers by top specialists in their fields who have made substantial contributions to the area that they are surveying. Authors are invited by the editors, in consultation with members of the editorial board. Each paper is critically reviewed by the editors and by several members of the editorial board. The editors would like to thank Wallace Oates for his help in getting the project started. We also very much appreciate the assistance of Arie Oskam, Eirik Romstad, Jacqueline M. Geoghegan, John Pezzey, Cees Withagen, Michael Finus, Sjak Smulders, Richard Carson and Bengt Kristom in shaping up this collection of papers. Tom Tietenberg Henk Folmer

xi

Editorial board EDITORS Henk Folmer, Wageningen University and Tilburg University, The Netherlands Tom Tietenberg, Colby College, USA

EDITORIAL BOARD Kenneth Arrow, Stanford University, USA Scott Barrett, London Business School, UK Peter Bohm, Stockholm University, Sweden Lans Bovenberg, Tilburg University, The Netherlands Carlo Carraro, University of Venice, Italy Partha Dasgupta, University of Cambridge, UK H. Landis Gabel, INSEAD, France Shelby Gerking, University of Wyoming, USA Lawrence Goulder, Stanford University, USA Michael Hoel, University of Oslo, Norway Per-Olov Johansson, Stockholm School of Economics, Sweden Bengt Kriström, Swedish Agricultural University, Sweden Karl-Gustav Löfgren, University of Umeå Karl-Göran Mäler, The Beijer Institute, Sweden Mohan Munasinghe, The World Bank, USA Wallace Oates, University of Maryland, USA David Pearce, University College London, UK Charles Perrings, York University, UK Rudither Pethig, University of Siegen, Germany Alan Randall, Ohio State University, USA Michael Rauscher, Kiel University of World Economics, Germany Kathleen Segerson, University of Connecticut, USA Mordechai Shechter, University of Haifa, Israel Horst Siebert, Kiel Institute of World Economics, Germany V. Kerry Smith, Duke University, USA Robert Solow, Harvard University, USA Olli Tahvonen, Finnish Forest Research Institute, Finland Alistair Ulph, University of Southampton, UK Aart de Zeeuw, Tilburg University, The Netherlands

xii

1. Fifty years of contingent valuation V. Kerry Smith* I.

INTRODUCTION

Econometric analyses and testing of economic models began during the last 50 years. Contingent valuation (CV) developed over the same time period.1 CV uses survey questions to elicit information that allows economic values for non-market resources to be estimated. Today contingent valuation occupies a strange position in economics. A significant component of the CV research, in both the USA and Europe, has been conducted to evaluate policy alternatives (see Carson, 2003). Little of this work has been associated with litigation. Yet in considering the relevance of CV estimates, the US Office of Management and Budget’s (2003) draft guidance for preparing benefit–cost analysis as part of the evaluation of regulations reiterates several of the criteria for a reliable study that were recommended by the National Oceanic and Atmospheric Administration’s (NOAA) Panel. The NOAA Panel was charged, in the early 1990s, with responsibility for evaluating contingent valuation for use in natural resource damage assessments.2 Current research on contingent valuation has pursued methodological issues in the structure of choice and open-ended valuation questions (Cameron et al., 2002), experimental design (Kanninen, 2002), interrelationships between public goods (Bateman et al., 2004), as well as numerous policy issues ranging from estimating the willingness to pay for reductions in risk of premature death (Krupnick et al., 2002 and Alberini et al., 2001) to managing waste from large-scale hog operations (Mansfield and Smith, 2002). Many of the issues raised in early contingent valuation surveys have been ‘rediscovered’ under the label of behavioral economics. Nonetheless, despite a spreading of survey methods to a wide range of applications (Blinder, 1991; Berry et al., 1998), there remains ‘discomfort’ among most economists about using the estimates from contingent valuation to measure consumers’ willingness to pay for changes in non-market goods. This concern is especially pronounced in applications involving resources that might be associated with non-use values. In this respect, then, the empirical research sponsored by Exxon (see Hausman, 1993) continues to haunt CV researchers. 1

2

Environmental and resource economics 2004/2005

Fortunately, this anxiety has not impeded a robust research program from using CV, or CV-like, methods in at least four areas: • refinement of econometric approaches for estimating parametric, semiparametric, and non-parametric models using discrete and other types of censored responses to survey questions; • application of repeated choice, preference scaling, or ranking questions with different sequences of hypothetical variations in the attributes of environmental resources; • investigation of the incentive properties of different modes for eliciting respondents’ preferences for market and non-market resources using theoretical, experimental, and survey techniques; and • integration of revealed and stated preference data in the joint estimation of individual preferences where there exist opportunities to collect both types of information from a common group of individuals. Moreover, in recent years the Association of Environmental and Resource Economists has recognized with their distinguished research awards several of the early research contributions relating to CV, including Randall et al. (1974), Bishop and Heberlein (1979), Hanemann (1984) and Mitchell and Carson (1989). Overall, then, the current activities in environmental economics are not consistent with the mainstream concerns. Clearly one motivation for the highly visible controversy in the United States – large-scale damage assessments – has been removed. A great deal of additional research has been developed, but CV and CV-like applications for the purpose of measuring economic values have reverted to a ‘tolerated fringe’ within environmental economics. There is not a strong presence in mainstream applications or among faculty in leading departments in the USA.3 This chapter has three objectives: (a) to describe what we actually know about the performance of survey approaches for inferring individuals’ values for changes in non-market resources; (b) to outline several new, or under-recognized, reasons that assure, in my view, CV or CV-like methods will remain a significant part of efforts to understand consumer preferences for non-market (and new) goods; and (c) to suggest ways in which environmental economists can enhance acceptance for measures of the economic values of non-marketed goods in general and CV in particular by the mainstream body of the profession. The chapter is developed in four sections after this introduction. Section II compares CV questions with others commonly used in economics and acknowledges the limitations in the information that is actually available for most revealed preference methods. Section III describes the evolution of CV research as well as the methodological and empirical issues the method helped to resolve. In the process it summarizes where we stand and why some

Fifty years of contingent valuation

3

economists in other fields remain skeptical about CV. Section IV describes reasons assuring that CV will not ‘go away’. The last discusses some practical steps that may help to increase confidence in the method.

II.

QUESTIONS AND ECONOMIC DATA

A recent referee’s report, commenting on some proposed travel cost recreation demand research, identifies a problem in much of empirical economics. The referee noted that data obtained from questions about a set of individuals’ past fishing trips were not consistent with ‘revealed preference methods’. A market record had not provided them. These comments reflect a difficulty in much of the empirical research in economics. There seems to be a belief that the information available from public sources is collected using a neutral external process that registers outcomes of people’s market choices. This assumption is generally incorrect. The 2000 Census of the United States provides a tangible reminder that much of what we commonly accept as revealed preference data arises from surveys of individuals, describing their own (or their household’s) activities or summarizing information developed from other sources (for example a firm’s records) in response to a set of instructions. Established US economic surveys, whether the Consumer Expenditure Survey, the Current Population Survey, Panel Study on Income Dynamics or, more recently, the Health and Retirement Study, all result from interviews with samples of respondents. All require people to report information about their activities, financial circumstances, family, or health status. They often involve some type of judgment by the respondents involved. Using the implicit standards in the comment I cited at the outset of this section, none of the economic data in these sources would qualify as consistent with the ‘revealed preference standard’. The reason is direct. There is no reason for public records of most market transactions. Only in situations where there are taxes or regulations does one find detailed economic data on prices and quantities.4 Until recently the same statement would have applied to private records as well. However, the dramatic advances in micro-computers has enabled record keeping and expanded the prospects for maintaining and analyzing private data that are the result of market transactions by individuals.5 While these new sources offer intriguing opportunities for research, they have not been the norm for revealed preference research to date. Instead, most economic data are derived from recall questions. These can include requests for information about the respondent’s past behavior or about members of his/her household or extended family. The time spans involved can extend as long as a year or require periodic reports, through diaries or repeated interviews to form a panel.

4

Environmental and resource economics 2004/2005

At least two limitations in these types of questions are important to their role as a performance standard in evaluating the reliability of stated preference questions. First, there is an established literature on the effects of: time span, questioning mode, extent of aggregation of events by size or type, as well as the influence of demographic and cognitive factors on the recall bias in these responses. Westat, Inc.’s (1989) panel study of the effects of time span for the accuracy of responses to the US Fish and Wildlife Service’s National Survey of Fishing, Hunting, and Wildlife Associated Recreation is one of the most widely recognized evaluation of these effects on data used in recreation models. This study found substantial overstatement of proportion taking fishing trips as well as the average number of fishing trips, days, and trip-related expenditures using an annual recall question in comparison to shorter time periods. Earlier recall studies with surveys involving consumer expenditures and earnings confirm the presence of substantial biases with lengthening of the recall period and changes in the character of expenditures involved in the questions. Both over- and understatements have been reported. Second, and equally important for panel databases, the process of collecting information from respondents should be considered an economic choice. Both the early recall studies and the more recent experience with the panel surveys used in damage assessment cases suggest respondents may well ‘learn’ that their answers can increase the time commitment expected of them. As a result, it has been suggested that observed declines in recreation participation in an ongoing panel are the result of survey incentives and not actual behavior (see Hanemann, 1995 and Desvousges et al., 1996 for further discussion).6 Conventional sources of microeconomic data can require tasks in addition to or instead of simple recall.7 One especially important example involves the estimation of the rent and the home values for owner-occupied housing. Adjustment of this index to account for the rent attributed to owner-occupied housing has been an important component of the reforms made in recent years to the Consumer Price Index. What is used to meet this need is the homeowners’ rent estimate? It is based on a hypothetical judgment responding to the following question: If you were to rent out your home today, how much do you think it would rent for monthly, unfurnished and without utilities?

The Census of Population uses a comparable question to estimate the property value. Owners are asked: What is the value of this property, that is, how much do you think this property would sell for if it were for sale?

Fifty years of contingent valuation

5

Data derived from both of these questions have considerable impacts on economic policy. In non-market valuation the home value question continues to serve as a basis for estimating the marginal willingness to pay for reducing air pollution (see Kahn, 1997 and Chay and Greenstone, 2000 as examples).8 The point of this overview of existing sources of micro-data is not one of discrediting all data describing individual consumption and prices (or other resource allocation choices associated with market exchanges). Rather, the lesson to be drawn from these examples is that data collection itself is an economic activity. What is collected is the outcome of an economic choice (for example a respondent allocates time, effort, and so on to provide the information requested). The debates over the reliability of contingent valuation have seemed to suggest that these issues are only relevant to stated preference data. I believe they are relevant in varying degrees to most commonly available public data sources.9 These concerns are the same as the first of the two questions often raised with CV – to what extent do respondents take the time and make the effort to seriously consider the hypothetical situation being posed? For most economic data the process of reporting the information imposes costs with few obvious benefits for the individual respondent. These disincentives provide a partial explanation for why survey research emphasizes the importance of making the context and framing of questions salient to those being surveyed. Given this shared problem of creating incentives for respondents to expend time and effort, the key distinction between conventional surveys and CV surveys (on this issue) must be whether the hypothetical nature of the choice being offered (or the lack of experience in making these choices) makes it differentially more burdensome. When interpreted in this context, the efforts since the Carson et al. (1992) Exxon-Valdez survey and the NOAA panel recommendations (as discussed in more detail below) can be seen as responsive to this concern. Unfortunately, despite the extensive research on the effects of information and other efforts to make CV choices tangible and salient, there have not been comparable efforts evaluating the sensitivity of reports provided about revealed preference behaviors as the character of the questions is altered. Thus, at this point all we know is that information can matter to CV responses (and we would expect that it should), but we cannot evaluate whether this effect is any different from what would happen if a comparable effort was made to evaluate the data about ‘real’ behavior or actual prices collected with more conventional economic surveys. The second issue raised with stated choices stems from the role of financial incentives. It is argued that there is no reason to interpret tradeoffs reflected in stated choices as comparable to those where a choice cannot be made without an explicit financial commitment. This concern, in contrast to the first, has received attention and some of the findings are discussed below.

6

Environmental and resource economics 2004/2005

III.

THE EVOLUTION OF CV

A.

Early Efforts and Shared Mistakes

After Davis’s (1963) path-breaking research, initial applications of contingent valuation can be classified into three groups. Some discussion of each provides an important context for both the controversies that developed over CV’s reliability and the lines of research present in the current literature. The first was a dominant initial component of the literature. The applications of CV in this group were interpreted at the time as ‘experimental’. Interviews were often conducted by graduate students with convenience samples, and a key focus of the studies was to evaluate the influence of different ways the survey questions were asked, considering especially the importance of strategic responses.10 A wide array of ‘biases’ was defined and questioning formats proposed to minimize their impact.11 The experimental flavor of the use of surveys may have stemmed from Davis’s use of the term ‘bidding games’ to describe the iterative questioning of respondents, asking the maximum amount they would pay for some specified ‘commodity’. The object of choice (or commodity) was a change in one or more dimensions of environmental resource that was described to each individual. The process involved a starting point and progressive adjustments by the interviewer up or down from it, based on the respondent’s answers. Analogies to auctions and bidding were often used in the early literature (despite the fact that there was no interaction among consumers). The size of the increments, refinements to them, explanations for changes (once a response was given), and so on were vaguely described. In some cases, it was not clear whether individual interviewers (for example economics graduate students) adjusted survey scripts to try to make them more understandable to respondents. Questions were developed by researchers without apparent field testing in advance of the surveys involved. None of these comments is intended to be critical of this work. The research efforts were regarded by all involved as experimental and intended to evaluate the contingent valuation method in relatively small-scale applications. The second group begins with Randall et al. (1974). This study is credited with doing the first, serious, professionally administered, population survey to collect CV responses.12 It also used the so-called ‘bidding game’ framework, but adhered to the conventions of professional survey research with training of interviewers. The survey questionnaire was evaluated and revised in two pretests. This close adherence to the protocols of survey research is the hallmark of the second line of research and is generally associated with all of the Mitchell–Carson surveys. While their work was conducted in several different stages (with a 1981 pilot survey), all of their findings remained in unpublished

Fifty years of contingent valuation

7

EPA reports prior to their 1989 book. Bill Desvousges and I had access to the early versions of their results and worked cooperatively in exchanging results and questionnaires. Their final water quality survey, conducted in November 1983, reflected both the experience of their pilot survey (conducted in 1981) and the findings of our survey for the Monongahela study conducted in November–December 1981 (see Desvousges et al., 1983 and Smith and Desvousges, 1986). That research, along with virtually all of the work conducted in the first two groups, shared another attribute. Because they were generally funded by mission agencies, projects intended to be primarily methodological had to have a link to a specific policy issue. In many cases, they were never envisioned to be ready for use in specific policy analyses. Rather the objective was to illustrate how the CV approach might offer estimates to meet policy objectives, provided a more complete evaluation of the survey alternatives was conducted. For example, the Monongahela study was regarded as a methodological effort to evaluate CV questioning modes and conduct a comparative study (following Brookshire et al., 1981, 1982) of CV and revealed preference (for example travel cost recreation demand) estimates of the value of water quality changes.13 To my knowledge, Randall et al. (1974) and Mitchell and Carson’s final survey (1984) were the primary exceptions. Nonetheless, after the fact, many of the studies’ results have been used in policy analyses. Later critiques of CV have described the work as if those involved designed it to serve the policy objectives identified in each study. This misunderstanding is unfortunate and does not reflect the actual intentions of the researchers involved in those early analyses. A third group of research efforts seems to have been lost in discussion of this early CV research. The first of these, predating Randall et al. (1974), was the 1969 Hammack and Brown survey of hunters’ willingness to pay for hunting published in 1974 (see Brown and Hammack, 1973 and Hammack and Brown, 1974). This group of studies focused on specific populations (usually involved in some form of outdoor recreation) that were identified through onsite surveys, licenses or, in Hammack and Brown’s case, through required federal waterfowl hunting stamps. Following the Hammack and Brown lead in 1971–72, Charles Cicchetti and I conducted the first (to my knowledge) CV survey to investigate what was later described as the ‘scope’ of change in environmental quality using a sample of wilderness recreationists.14 The sample was derived from an earlier record of their entry to the Spanish Peaks Primitive (later Wilderness) Area in Montana (see Cicchetti and Smith, 1973). By varying the potential congestion it was suggested each respondent would experience during a proposed trip to the same area and then asking for their maximum willingness to pay for each trip, we estimated how different aspects of congestion affected their valuation of wilderness recreation trips at this site.

8

Environmental and resource economics 2004/2005

Research falling into either of the first two groups generally shared characteristics that I believe influenced both the development of subsequent applications and the degree of acceptance experienced later for the method. It appeared (to those doing it) that CV’s ability to control the object of choice presented removed the need to impose prior restrictions (from economic theory) on how the results were interpreted. This statement is a complex way of saying the estimates were usually summarized with univariate statistics – sample means and variances – computed for the treatment groups comprising the sample. The treatments were usually attributes of the framing of the valuation question, the starting point used in the iterative bidding question, or the use of another question format. Several studies did report regression summaries of responses, and Desvousges and I (see Desvousges et al., 1983) attached considerable importance to the role of income as the only variable we hypothesized economic theory implied should be relevant. With the benefit of hindsight, more effort at linking the sources of use and non-use value would have provided a wider range of such variables to investigate. What was missing was a link between the functions estimated and the constrained preference models economic theory implied should be linked to economic tradeoffs. Some years later Xiaolong Zhang and I (1997) proposed a framework that explicitly defined the structure of use and non-use values with an individual’s preference function and then used travel cost and CV data to attempt to estimate the model. To illustrate the issues, we considered a specific functional form that conforms to Hanemann’s (1988) proposed definition for non-use values. His suggestion requires separability between the resource(s) contributing to the non-use enhancement to well-being and the use-related motives. Consider an additively separable component with the quality of one of two types of beach resources providing the source of the benefits and the second providing the source for the non-use. Further, assume the use-related benefits arise from a trip to a recreation site, such as a beach. In the example here, we selected a specification for the indirect utility function that is consistent with a linear demand (and thus a finite choke price). The non-use contribution is assumed to stem from a CES function in the qualities (q1 and q2) for the userelated and the non-use resources. The latter might be an estuarine reserve, precluded from use to protect fragile habitat. Equation (1.1) illustrates the function. It would be consistent with a linear Marshallian demand (that is, t = trips = a + bp + gm + dq1 with b < 0). 1 1/ V = exp(– gp) m + — (bp + a + dq1 + b g) + d . (qh1 + qh2) h g

(

/

)

(1.1)

The point of our argument was to indicate that the specification of a

Fifty years of contingent valuation

9

preference structure allows an interrelationship between CV questions for non-use choices with either other non-use resources or with quality attributes that might be related to use of another resource. Thus, there should be crossequation restrictions for the parameters associated with models for each data source. Table 1.1 illustrates these possibilities with the travel cost demand function and a variety of quality changes and payment mechanisms. Notice the shared parameters across the different willingness to pay definitions. These relationships become the source of the links I referred to earlier. We were unable to exploit fully this framework due to data limitations. More recently, Herriges et al. (2004) have used structural logic to consider the importance of identification conditions in relating use and non-use values. With a modified Stone Geary specification, they use a Kuhn–Tucker corner solution model to explicitly introduce weak complementarity together with the prospect for corner solutions to attempt to estimate use and non-use values. They observe that tests of weak complementarity and the use/non-use value distinctions are conditional in an important way on the maintained assumptions associated with the preference specification. Moreover, they suggest, based on their analysis of this specification, that: The demand systems imply different welfare estimates depending upon what one assumes to be the underlying form of preferences, with no behavioral footprint available to distinguish among competing alternatives. (p. 12)

However, as they conclude, this limitation arises when analysts are unwilling to use stated preference data to further discriminate among restrictions. This issue is one point of the algebra in Table 1.1. With multiple sets of choice information and a preference specification, we may be able to discriminate among some aspects of the alternative treatments of non-market goods. We will not escape all of the maintained restrictions associated with a parametric specification of preferences. They are what provide the linkages necessary to use multiple types of information in the tests of the role of non-market resources in preferences. A second feature was the presumption that respondents would accept the hypothetical object of choice without asking how the implied change would be accomplished. The use of focus groups to evaluate CV wording came later in the evolution of CV practice (see Desvousges et al., 1984). As a result, some studies varied both the baseline condition of a resource and the new environmental conditions offered as part of the CV question without specific attention to the actual state of the resources involved. Finally, the analysis rarely collected, and never (to my knowledge) used, information about the behavioral changes that an individual would make if the change in the resource was introduced. For example, when we asked about

Table 1.1 Behavioral models linked by use and non-use motives Travel cost demand

trips = a + bp + gm + dq01 with b < 0

10

Willingness to pay for an improvement in q1 from ]q0 to q1 (use-related) 1 1

WTP = d(q11 – q01) + d . (egp)[((q11)h + (q02)h) /h – ((q01)h + (q02)h) /h]

Willingness to pay for an improvement in q2 from q02 to q12 (non-use-related)

WTP = d . (egp) [((q01)h + (q12)h) /h – ((q01)h + (q02)h) /h]

User fee (F) and improvement in userelated q from q01 to q11

1 1 1 (bp + a + dq0 + b ) m+— /g ((q01)h + (q02)h) /h – ((q11)h + (q02)h) /h 1 1 g F = – — ln ———————————————— + —————————————————-—— 1 (bp + a + dq1 + b ) 1 (bp + a + dq1 + b g m+— m+— /g /g 1 1 g g

1

1

[

1

1

]

Fifty years of contingent valuation

11

improvement in water quality conditions in the Monongahela River from its current (at the time) boatable conditions to swimmable conditions, we asked if those respondents would visit sites along the river more frequently. Unfortunately, we did not use their responses as integrated components of the analysis. Today, the response to these follow-ups asking about the number of additional trips and their location would be integrated into the models and estimates used to analyze the data.15 The composite of the Bishop–Heberlein (1979) simulated market analysis, along with their introduction of discrete choice take-it-or-leave-it questions and Hanemann’s (1984) explicit link between the econometric analysis and preference functions, set in motion changes that have brought the current analyses of CV data more directly into line with the practice of microeconometric research in other applications. Now there is extensive use of restrictions from economic theory in the analysis of CV data. What has not been appreciated in both the reviews of the CV literature and by critics of its findings is that the last category of CV survey did not have the attributes of the first two. Beginning with Brown and Hammack economic functions were specified to be consistent with the behavior being modeled. Changes in environmental resources were closely linked to existing conditions and behavioral responses were often assumed to take place. The result, I believe, is a less critical judgment on the findings from these studies on the part of mainstream economists.16 This distinction has been interpreted as simply reflecting the experience of the respondents with the resource changes asked about and the focus on use values. However, it could equally well have been due to the distinctive differences in the CV questions and in the ways they were analyzed. B.

Discrete Response CV

Bishop and Heberlein’s (1979) evaluation of CV versus travel cost methods introduced the discrete response CV question where a respondent states a purchase or voting decision. It is the intellectual ‘parent’ of most of the research focused on refining the econometric methods used with contingent valuation. In addition it motivated a specific focus on what economic theory implies can be learned about an individual’s preferences from the answers given to discrete response questions. Hanemann’s (1984) seminal paper transformed the theory and the practice of CV for all single and multiple discrete response questions.17 An important byproduct of Hanemann’s explicit use of theory in interpreting CV data was the need to be precise about the description of how people answer stated preference questions. All of this subsequent literature (including the research on the incentive properties of different types of CV questions, see

12

Environmental and resource economics 2004/2005

Carson et al., 1999), assumes that they are equivalent in the incentives they create to actual choice questions.18 Carson et al. describe this condition as a requirement of consequential choice. This requirement means not only must the objects of choice be presented in a way that they are believable, but respondents also must believe their answers have an impact on provision of the good.19 The attention given to linking the analysis of CV responses to the theory has not been uniform in all applications. There remains substantial confusion about the differences between discrete response and open-ended questions. Some of it arises from differences in maintained assumptions about consumer preferences for the open-ended and discrete response questions. Most estimates for mean or median willingness to pay with discrete response questions begin with a specification for preferences, including unobserved heterogeneity (that is, a distribution for the error). Comparisons of their results with other question modes cannot distinguish the effect of mode from these maintained assumptions, unless it is built into the design of the tests (see Huang and Smith, 1998, 2002 and Poe and Vossler, 2002). To my knowledge, Cameron et al. (2002) offer the first test of different question modes, including an ‘actual’ program that allows evaluation of the stated preference with a real solicitation for participation.20 The program was a green energy plan offered by an electric utility that involved using renewable sources for energy and planting trees. Their study included: an actual choice using a dichotomous format, a hypothetical discrete choice with multiple prices, an open-ended question about willingness to pay (WTP), a payment card format, a multiple bounded elicitation, and a conjoint format.21 They conclude that: ‘The two methods which appear to be the least consistent with the others are the only two methods that attempt to elicit WTP directly, rather than inferring WTP from choices’ (p. 422). In terms of pairwise tests of the mean willingness to pay, these two methods – open-ended and payment card – are distinctly different from the others.22 Hypothetical dichotomous, with one price or with multiple prices, was not significantly different from the actual choice results. Conjoint and multiple bounded questions were also not significantly different from the actual. Of course, there were significantly different patterns in the variance of the estimate for WTP, with the one price hypothetical dichotomous choice exhibiting considerable variability. Once multiple prices were introduced, this spread was dramatically reduced. C.

Pre-NOAA Panel Damage Assessments and CV

The 1989 Exxon Valdez oil spill in Alaska changed the professional and public attention given to contingent valuation. Three studies conducted in support of the natural resource damage litigation associated with the spill

Fifty years of contingent valuation

13

have been especially influential on CV research. Two of these were intended to evaluate the plausibility of CV estimates of the non-use (or passive use) values people have for environmental resources.23 Desvousges et al. (1993) used a survey to elicit people’s willingness to pay for avoiding the loss of wildlife (birds) in waste oil holding ponds. By varying the proposed number of birds that would be lost in the event action was not taken, their surveys investigated the sensitivity of CV estimates to the ‘size’ of the object of choice. The baseline population of birds, target population of respondents, and other dimensions of the CV survey were held constant across three different levels of bird losses. Their findings suggested no significant difference in the estimated mean (or median) willingness to pay for avoiding the different sized bird losses constructed with three independent samples.24 Diamond et al. (1993) also reported survey results testing several consistency features they argued were implied by well-behaved preferences. The two that have had the most impact on the subsequent literature were: their tests of CV estimates of the responsiveness to size of object of choice and of the adding-up conditions across different changes posed individually versus as a composite to independent samples. Following the same practice as the early experimental and general population surveys, the primary reports of this research focused on univariate descriptive statistics for their tests. The clearest case (in terms of the object of choice) questioning consistency of these CV estimates with conventional theory seems to arise with the Desvousges et al. ‘bird’ loss surveys.25 Diamond et al. (1993) considered different numbers of specific wilderness areas and these, as a result, are harder to reduce to a simple quantity metric.26 The last CV study in this litigation group is the Carson et al. (1992) effort to estimate the loss to US households as a result of the spill. In contrast to the earlier two studies, this CV survey was intended to develop an estimate of the welfare loss. While never published, the report to the State of Alaska has been widely available (http://www.oilspill.state.ak.us/pdf/econ5.pdf). I believe it had a dramatic effect on research practice. This was accomplished in a least three ways. First, the CV question is about a plan to change environmental resources, not a specified change. The plan describes what it is anticipated will happen and is quite specific about the process. As a result, people’s choices are about the plan, not the resource. Use of the results to estimate people’s willingness to pay for the resource change requires the analyst to consider whether respondents believed the plan would work (see Smith, 1997). Second, the survey was among the most detailed in terms of documenting the steps used in developing a framing of the object and circumstances of the choices posed. Finally the analysis distinguished the estimation of a mean willingness to pay (WTP) from these censored responses from the multivariate analysis of the determinants of respondents’ stated choices. As with the Desvousges et al.

14

Environmental and resource economics 2004/2005

(1993) and Diamond et al. (1993) studies, the estimates minimized the maintained restrictions imposed on the stated responses in the process of estimating WTP. D.

NOAA Panel and CV

The Exxon Valdez case was settled before anyone learned how CV would fare in court.27 However, the Superfund legislation and Oil Pollution Act required that regulations specify how the economic damages from natural resource injuries would be measured. Court rulings to challenges to draft rules suggested that CV could be used in this process. The concerns associated with future liabilities likely contributed to Exxon releasing a set of the research it had sponsored in a highly visible conference in 1992. The Hausman (1993) edited volume includes those papers. It had a marked impact on rulemaking and the review panel chaired by Kenneth Arrow and Robert Solow was NOAA’s General Counsel’s response to the concerns about CV.28 The Panel’s report provided an extensive set of guidelines for CV survey construction, administration and analysis. In the Panel’s view, ‘the more closely the guidelines are followed, the more reliable the result will be’ (Arrow et al., 1993, p. 4609). In addition, the Panel distinguished a subset of items from their guidelines for special emphasis and described them as burden of proof requirements. In describing the elements with this special focus, the Panel stated: if a CV survey suffered from any of the following maladies, we would judge its findings ‘unreliable’: • a high non-response rate to the entire survey or to the valuation question • inadequate responsiveness to the scope of the environmental insult • lack of understanding of the task by the respondents • lack of belief in the full restoration scenario • ‘yes’ or ‘no’ votes on the hypothetical referendums that are not followed up or explained by making reference to the cost and/or the value of the program’. (Arrow et al., 1993, p. 4609)

The second item in this list, ‘inadequate responsiveness to the scope of the environmental insult’, or the scope test, attracted the most attention and was regarded at the time as an acid test for CV studies. Given these guidelines and burden of proof requirements, the Arrow–Solow Panel concluded its report noting that: under those conditions (and others specified above), CV studies convey useful information. We think it is fair to describe such information as reliable by the standards that seem to be implicit in similar contexts, like market analysis for new and

Fifty years of contingent valuation

15

innovative products and the assessment of other damages normally allowed in court proceedings. . . . CV [contingent valuation] produces estimates reliable enough to be the starting point of a judicial process of damage assessment, including passive-use values [i.e., non-use values]. (Arrow et al., 1993 p. 4610)

Their recommendations focused attention on survey and questionnaire design. They also created expectations that there were some economic properties we should expect to observe in WTP estimates. Indeed, Diamond (1996) suggested that the scope test could be made more specific – offering a bound on the WTP estimates for different sized objects of choice. For example, using Dq1 and Dq2 as two different changes in an environmental resource q, with Dq1 > Dq2), Diamond’s bound is given in equation (1.2): WTP(Dq1) ≥ (Dq1/Dq2) . WTP(Dq2)

(1.2)

His result follows from three assumptions: (a) Dq1 and Dq2 represent losses in a base level of q to be avoided; (b) the utility function is quasi-linear, so the marginal utility of income is constant; and (c) the plan described as providing the means to avoid the losses is perceived to provide outcomes that are perfect substitutes for the environmental resource q. The first two assumptions influence the specific form of the WTP function and, as Diamond has argued, seem plausible as descriptions of a number of CV applications. The last is not as plausible and plays a central role in Diamond’s specific bound for responsiveness to scope as well as in the adding-up test. That is, Hicksian WTP measures differences in the ‘spacing’ of indifference curves in monetary terms. This can be appreciated when the WTP to obtain the change is written as equation (1.3), (that is, with the initial income m0, an unchanged price vector, and improved q1, a higher utility level, u1 can be realized). This equation leads to the informal characterization of WTP as a monetization of the change in utility from u0 to u1. WTP = e(p, q1, u1) – e(p, q1, u0)

(1.3)

Measures for WTP follow from how the utility function influences the derived relationship (given the effects of budget-constrained utility maximization) between changes in q, income, and the spacing in the feasible indifference curves. Normally we describe these distinctions as akin to substitution and income effects, but in fact they are interrelated. Unless we select specifications for preferences that impose specific constraints (for example quasi-linearity), we can expect that the curvature and spacing (or substitution and income) effects will appear to be separable only at a point. Their interrelationship

16

Environmental and resource economics 2004/2005

changes as we change either the level of the environmental resource or the level of well-being. When the specification is simplified to permit us to abstract from the role of the income–utility link, it will also have implications for the effects of substitution. Without Diamond’s perfect substitution assumption all we can expect from a scope test is that a large amount of the same object should have a greater measured WTP than a smaller amount (provided they are perceived this way by the respondents involved). Closer scrutiny of the literature over the past five years has revealed in early CV studies that the estimates were in fact consistent with the properties implied by the scope test. Carson et al. (1997), Smith and Osborne (1996), Smith et al. (1997) and Takeuchi and Ueta (1996) (for Japan) among others report summaries of either past evidence or new small-scale surveys that satisfy the broad version of the NOAA Panel’s scope test. Most of these studies relate to resources that support uses. As a result, it might be argued that responsiveness to scope effects would be more likely. That is, respondents evaluating changes in resources or their quality would respond to proposed alternatives considering both potential use and any non-use related reasons for their choices. Overall, the NOAA Panel’s report surprised the mainstream of the economics profession, who seemed at the time to have expected a complete repudiation of CV. By contrast, their summary appeared to endorse it. However, there is another interpretation that seems to have been overlooked. This view would hold that the panel’s report established such stringent guidelines for CV surveys that it raised the ‘price’ for ‘reliable CV’ above the maximum willingness to pay for that information.29 Thus, it didn’t have to mandate an end to CV, but instead ‘priced the practice out of the market.’30 Some authors have argued the report had little overall merit. Rather than pricing ‘reliable’ CV surveys out of reach, they suggest it distracted practitioners’ attention from the real problems and how to improve CV. For example, Harrison (2000) notes that: The NOAA report was a great disappointment. The CVM literature was slowly emerging from several decades of intellectual darkness, in which one conventional wisdom after another became established by mere repetition. The literature is full of ‘truth by assertion’ in which opinions substitute for careful and scientific research; . . . Given the inability of CVM practitioners and consumers to weed out sense from nonsense in the extent literature, it is unlikely that there will be much progress as the result of the NOAA report. (p. 2)

This observation is a strong criticism. As I noted earlier, Harrison also documents differences between the Panel’s judgments and experimental or conceptual papers on related questions that had been completed at the time their work was undertaken. Nonetheless, it is important to return to the Panel’s

Fifty years of contingent valuation

17

charge. It was specific to court cases involving damage assessments. Its guidance was to the legal community and potential experts as to the procedures to be used and issues addressed to satisfy a burden of proof judgment. This context is definitely not the same as should be used in outlining a research agenda or in developing guidance for ‘best’ research practices. An expert provides an opinion. To be sure, it should be informed by the relevant information, but it remains an opinion. As a result, it would seem to me that Harrison’s criticism is not so much directed at the Panel, but rather at subsequent researchers who use the report as a litmus test or set of guidance principles for their research. E.

The Montrose Study

To my knowledge, the only CV effort to meet the NOAA guidelines was undertaken as part of litigation. The analysis was associated with the alleged natural resource injuries due to PCB (polychlorinated biphenyl) and DDT releases in the Southern California Bight. Named for one of the chemical companies (Montrose) associated with some of the releases of these pollutants, the economic analysis for this case involved three components: (a) tests for three parts of the NOAA guidelines; (b) estimates of the willingness to pay for accelerated recovery of injuries assumed to be linked to the releases of the PCB and DDT; and (c) scope effects to evaluate whether willingness to pay measures were sensitive to the ‘size’ of the injuries. The tests of NOAA Panel recommendations used an exact replication of the Carson et al. (1992) Exxon Valdez questionnaire (updated to reflect the date of the new survey). Aspects of this work have been published (see Carson et al., 1997, 1998, as well as Krosnick et al., 2002). However, the distinguishing feature of the study – the scope test – while available as a government report in 1994, has never been published. Table 1.2 summarizes the features of the three surveys used in the research in comparison to the characteristics of the original Alaska/Exxon Valdez analysis. All four surveys used discrete response (take-it-or-leave-it, voting-based) framing for the CV valuation question as well as in-person interviews. The survey to test three NOAA Panel recommendations was conducted by the National Opinion Research Center at the University of Chicago and is labeled here as the NORC study. It consisted of four separate questionnaires including: (1) a complete replication of the original Alaska instrument modified only slightly to reflect the timing of the new interviews in relation to the Exxon Valdez oil spill (termed the replication version); (2) a version in which the respondent votes on a paper ballot that is placed in a sealed box and the interviewer does not know the decision (the ballot box version); (3) a version where the respondent is told there are three options – ‘for’, ‘against’, and

Table 1.2

A comparison of the features of four surveys

Alaska

NORC

Date of survey Population

1991 US

Object of choice

Plan to provide two Coast Guard ships to escort oil tankers in Prince William Sound to prevent future accidents and avoid future injuries due to oil spills 1043 One-time addition to federal income taxes

1993 12 PSUsb Same as Alaska

18

Attribute

Sample size Nature of payment

Tax amounts Focus groups Pre-tests Pilots Response rate

$10, $30 $60, $120 7 2 (12, 18) 4 (105, 195, 244, 176) 75.2%

1182 One-time addition to federal income taxes $10, $30 $60, $120 – 2e (64, 26) – 73.0%

Southern Californiaa base

Southern Californiaa scope

1994 California

1994 California

Plan to accelerate recovery of reproduction problems of four species by 45 yearsc

Plan to accelerate recovery of reproduction of two species by 10 yearsd

1857 One-time addition to California State income taxes $10, $25, $80 $140, $215 5 4 (44, 57, 49, 116) 4 (332, 460, 324, 473) 72.1%

953 One-time addition to California State income taxes $10, $25, $80 $140, $215 9 4 (44, 54, 40, 44) – 73.8%

Notes: a Sample was intended to represent the population of English-speaking Californians, age 18 or older, living in private residences they own or rent (or whose rent or mortgage they contribute to). Thirteen primary sampling units were selected with probabilities proportional to their 1990 Census population counts, including: Del Norte and Humboldt; El Dorado, Placer, Sacramento and Yolo; Alameda, San Mateo, San Francisco, Marin and Contra Costa; San Joaquin; Santa Clara; Fresno; Santa Barbara; Ventura; Los Angeles County; Los Angeles City; Orange; Riverside and San Bernardino; and San Diego. Within the selected PSUs, 652 segments (city blocks, groups of blocks, or Census equivalents in rural areas) were selected with probabilities proportional to their 1990 Census counts of housing units. b The 12 PSUs selected from NORC’s master area probability sample were: Baltimore, MD; Birmingham, AL; Boston, MA; Charleston, SC; Harrisburg, PA; Ft. Wayne, IN; Manchester, NY; Nicholas County, KY; Portland, OR; Richmond, VA; Seattle, WA; and Tampa, FL. c The four species include two birds: bald eagles, peregrine falcons; and two fish: white croaker and kelp bass. See Carson et al. (1994) for the description of the injuries. The time period for natural recovery was described as 50 years. d For the scope scenario only the two fish species were described as injured, and the time period for natural recovery was reduced to 15 years. e These pre-tests were conducted to evaluate the instructions used with the design variations, e.g. ballot box, and would-not-vote and composite versions of the questionnaire, not to evaluate framing. Source:

Carson et al. (1996c).

19

20

Environmental and resource economics 2004/2005

‘would not vote’ (the no-vote version); and (4) a version with the three choice options and the ballot box (the no-vote/ballot box version). Tests of the effect of including a would-not-vote option in a discrete choice/referendum CV question, as well as for the potential of a social desirability effect with in-person interviews (that is, respondents reporting support for a program to please the interviewer) suggest these possibilities were not concerns that require modifications to the framing or implementation of CV questions.31 The primary objective of the Montrose study was to estimate the damages associated with injuries to four species thought to have been impacted by DDT and PCB. As noted earlier, the final survey included a test for whether CV estimates of the willingness to pay for resource changes with little apparent userelated motivations would be responsive to the ‘size’ of the change. The objects of choice in the base and scope surveys were distinguished by the number of species that were identified as injured by the DDT and PCB deposits and the time span required for natural recovery to eliminate the source for these injuries. In the base scenario, four species – two birds (bald eagles and peregrine falcons) and two fish (white croaker and California sea bass) – were described as having reproductive problems in the areas affected by the DDT/PCB deposit. If the recovery was left to natural events, the base scenario suggested their recovery would require 50 years. The proposed plan being evaluated with the surveys assured an accelerated recovery of the species. Implementing the plan was to take five years. At the end of the implementation of the plan, the description associated with the CV question indicated that the recovery would be complete. Thus, in the base situation households were asked about a plan that would reduce the recovery time by 45 years. In the scope sample, the number of impacted species was reduced to two – the two fish – and the time for natural recovery was described as 15 years. The time described for the plan to be completed remained at five years. Several aspects of the development of these scenarios and implementation of the scope test are not necessarily apparent from this summary. They arise because the survey design stressed the importance of logical consistency. It also recognized that when data collection involves ‘in-person’ interviews, variations in the conditions described in the interview must consider both the assignment of interviewers to versions of the survey and the task of explaining the reasons for these variations to those interviewers. Concerns for logical consistency arise because of the mechanism linking the deposit of DDT and PCB was off the coast of California (under 100 feet of water) to the injuries had to be detailed in a way the respondents could understand. The survey’s explanation used the food chain linking benthic organisms in direct contact with the deposits of DDT and PCB to larger organisms, small fish, and ultimately to the larger fish and then to the birds identified as having documented

Fifty years of contingent valuation

21

injuries. In the base scenario it was explained the birds became contaminated by eating contaminated fish. A smaller scenario could not confine the injury to the birds without also explaining how these birds were exposed to the substances that were described as the source of the problem. As a result, the scope was confined to the two fish species.32 Both injuries were limited to the South Coast of California. To avoid confounding interviewer and version (base versus scope) effects, interviewers were randomly assigned to each version. This process implied each interviewer would be administering both questionnaires. While each respondent received one version, the interviewers administered the questionnaires in a random sequence. This process raised a separate issue. The training of interviewers was conducted to convince them the problem was real and the proposed plan feasible. This process assured they would be convincing in describing it to respondents. This strategy requires some explanation for the scope injury. It was described as reflecting potential scientific uncertainty about the injuries. This explanation was intended to avoid subtle differences in the interviewers’ levels of confidence with each version of the survey. Three separate tests of the responsiveness of WTP to the scope of the injuries presented in the Montrose survey were undertaken. The least restrictive tests are the simple contingency tests comparing the stated choices in the base and scope samples. Table 1.3 reproduces these tests at each of the tax amounts used in the discrete response CV questions (see Carson et al., 1994 for more details). There is a clear difference in stated choices between the two samples with more votes for the plan in the base (larger) injury description. Table 1.3 also reports two further tests. The first of these compares the Turnbull lower bound mean for WTP in both the base and scope samples (see Haab and McConnell, 1997). Both a simple test for differences in these means and a likelihood ratio test for differences in the non-parametric estimates of the distribution confirm significant differences in the base and scope distributions, with respondents willing to pay more for more significant programs. It is also possible to evaluate whether, following their stated ‘votes’, the respondents in the base and scope samples reported different perceptions of the seriousness of the injuries in each case. This difference is important because at the time of their vote, the respondents in each sample (that is, base and scope) were not aware of the alternative (larger or smaller) injury description. A severity evaluation was based on an attitude question asked after the CV choice. For the base questionnaire it was: All things considered, would you say the fish and bird reproduction problems I told you about in the South Coast were not serious at all, not too serious, somewhat serious, very serious, or extremely serious?

22

Environmental and resource economics 2004/2005

Table 1.3 Tests for scope with Southern California survey A. Contingency tests tax amount For $10 Base (%) 209 (55.9) Scope 72 (35.6) $25 Base 163 (46.3) Scope 45 (24.7) $80 Base 120 (32.9) Scope 35 (17.9) $140 Base 102 (26.5) Scope 29 (14.9) $215 Base 85 (22.3) Scope 19 (10.7) B. Turnbull lower bound mean (LBM) LBM Base $63.24 Scope $34.02 Z-test 7.17

Likelihood ratio

83.46

C. Survival model test results (1) Weibull Z-test for Location parameter Likelihood ratio test for location and scale parameter (2) Log normal Z-test for location parameter Likelihood ratio test for location and scale parameters

Against 165 (44.1) 130 (64.4) 189 (53.7) 137 (75.3) 245 (67.1) 161 (82.1) 283 (73.5) 166 (85.1) 296 (77.7) 159 (89.3)

Carson et al. (1996c).

p-valuea 0.001 (R)

23.50

0.001 (R)

14.39

0.001 (R)

10.00

0.002 (R)

10.85

0.001 (R)

Std dev. 2.54 2.82 (reject null hypothesis of equality of LBM p-value <0.001) (reject null hypothesis of equality of distributions pvalue <0.001)

(reject null hypothesis p-value <0.001) (reject null hypothesis p-value <0.001) (reject null hypothesis p-value <0.001) (reject null hypothesis p-value <0.001)

Note: a R – reject null hypothesis at most conventional levels for p-value. Source:

c2 21.50

Fifty years of contingent valuation

23

For the scope questionnaire this question replaced fish and bird with fish. A simple two-way chi square test indicated a significant difference in stated seriousness of the problem (that is, c2 = 148.90 with a p value < 0.001). There have been two types of responses to the Montrose findings. The first suggests that scope is not the ‘real’ hurdle for CV, but rather distinguishing different values for different types of environmental resources.33 A second comment suggests the difference between the two estimates is not ‘large enough’. The argument here implicitly calls for a decomposition of the injuries and an effort to evaluate whether respondents treat the difference in natural recovery time and the two versus four species as simple economic models might hypothesize. Most of this concern arises from a simple interpretation of the Diamond (1996) argument discussed earlier. In this case, two dimensions of the objective of choice vary between the base and scope surveys – the number of species affected and the time savings based on natural recovery. Judgments about the size of the difference in willingness to pay require assumptions about how each factor enters individual preferences. Conventional theory (as well as the empirical literature on how people evaluate the importance of different types of natural and environmental resources) offers little explicit guidance on how to address these types of questions.34 Without exogenous information that allows the analyst to determine plausible restrictions to preferences there appears to be no straightforward resolution of these types of questions. F. Joint Estimation In 1988, Trudy Cameron outlined how revealed and stated preference information, collected from a single sample, can be used in joint estimation of preferences.35 Morikawa (1989) independently proposed stacking revealed and stated preference data to recover estimates of consumers’ preferences for existing and ‘new’ features of transportation systems as part of using random utility models to describe consumers’ decisions among different transportation modes.36 There are several important differences between the Cameron framework and the work by Ben-Akiva, Morikawa and McFadden as well as the related activities of marketing researchers. The latter group has relied on a single response outcome – a choice of one type of program or policy from a set of different alternatives. It relies on extreme corner solution models whose behavioral foundation was described with simple (usually linear) preference functions. However, the earliest contributions did not attempt to impose the restrictions implied by constrained utility maximization. Cameron, by contrast, derived travel cost demand models from a quadratic utility function and then interpreted the CV question within the context of that constrained optimization framework, using the restrictions implied by theory.

24

Environmental and resource economics 2004/2005

This strategy has at least three important implications. First, and perhaps most important, it requires an explicit connection to be developed between the CV question and individual preferences. This description interprets how the object of choice posed in a CV question is introduced into an individual’s utility function. The reason for being explicit arises because we wish to take advantage of the efficiency gains from estimating the same parameters with information from different choices. Models derived to describe the revealed preference and CV responses must share some structural parameters for the estimation to offer efficiency advantages. As a rule, the process requires selecting a specific functional form for the preference (or indirect utility) function and defining the observable outcomes from each approach in terms of that function (see Smith et al., 2002 for an example reversing that logic for preference calibration). While it may not require defining a quantity metric for the environmental resource, the specification does imply the role of the resource for other types of behavioral outcomes must be described.37 Thus, as Larson et al. (1993) demonstrated in later research, the specification of use and non-use values attributed to the change in environmental quality becomes explicit with the description of how ‘q’ (from earlier discussion of Diamond) enters the preference (or indirect utility) function. There is a potential ‘downside’ to this approach when a CV question has not been designed to consider explicitly how the choice that is posed to each respondent is to be linked to a behavioral model. Under such circumstances it may be possible, ex post, to offer several different connections. Each would impose different restrictions on the parameters to be estimated because each corresponds to a different interpretation of how people answered the questions. The Cameron (1992) and Englin and Cameron (1996) questions are in this category and thus different restrictions linking the stated and revealed preference models could be maintained. Without specifics about how respondents interpreted the questions, we cannot determine which interpretation is correct. The second implication is that the analysis allows for evaluation of factors that influence the level of utility and those that influence how preferences are assumed to change with changes in the parameters of constraints (that is, prices or the levels of quasi-fixed goods). Take-it-or-leave-it choices are modeled as corner (extreme or generalized) and to describe them we must deal explicitly with how the full choice set contributes to individual well-being.38 The level of demand for a linked private commodity is a second-order response (that is, Roy’s identity reveals features of the first derivatives of the indirect utility function). Without the stated choice information we would not be able to recover the separable contributions to well-being. This argument underlies Larson et al.’s measures of non-use values. Third, there is, of course, no reason to require that joint estimation be

Fifty years of contingent valuation

25

confined to a single revealed and a single stated response, or that both always need to be present (see Zhang and Smith, 1997).39 Indeed, as my discussion of conjoint analysis in the next section suggests, the primary innovation here is the stacking of multiple responses that requires a preference specification describing how attributes of the object of choice contribute to individual decisions. Eom and Smith (1994) illustrate some of the difficulties when multiple commodities convey the potential quality measure (or in their case a risk of cancer). Modeling can require development of consistent price and quantity indexes that reflect how the set of choices and amounts would change with the proposed policy. Joint estimation has seen an increased number of applications. Many have adopted the basic logic proposed by Cameron (1992). The primary limitation seems to be the increased burden on data collection. It requires that the questionnaire carefully connect what is asked to a set of plausible behavioral responses.40 G.

Conjoint Analysis

Conjoint analysis (CJA) refers to a variety of different methods. Widely used in marketing research to evaluate consumer preferences for private goods, conjoint analysis has been proposed by a number of these researchers as providing ‘the answer’ to the problems posed by CV. These problems stem in part from the desire to evaluate the properties of willingness-to-pay functions. Calls for scope tests have increased interest in evaluating how the willingness to pay responds to the amount of the change in the object of choice offered and in how it is offered. When each respondent is only offered one alternative, the cost of investigating these issues becomes prohibitive. The conjoint format seeks to decompose the object of choice into a set of attributes and to design choices (or rankings) in a framework that presents multiple alternatives to the same individual. Unfortunately, conjoint methods do not solve these problems. To understand why, some distinctions need to be drawn between the ways CJA is implemented. Two of the most common methods are ratings and choices.41 I will discuss the issues that arise with each in turn. Roe et al. (1997) illustrate the distinctions between CV and ratings by posing (in a very simplified form) a CJA question in comparison to their description of a CV question as follows: CV: Would you pay an additional $p to fish for salmon if fishing conditions were changed, from {q01, q02, . . ., q0k} to {q11, q12, . . . , q1k}? CJA: On a scale from 1 (very undesirable) to 10 (very desirable) how would you rate a salmon fishing trip with the attribute and price levels listed below?

26

Environmental and resource economics 2004/2005 Trip

qi1, q i2, . . ., qik

Price

Your rating

A

q01, q02, . . ., q0k

p0

——

B

q11,

p1

——

q12,

. . .,

q1k

(Roe et al., 1997, p. 147).

They argue this comparison highlights the importance of asking each respondent to rate the baseline conditions. Moreover, they suggest that conventional Hicksian welfare measures can be recovered from models that estimate the ratings differences (that is, between each situation in comparison to the baseline). There is a problem with their characterization of both the CV and the CJA cases. Conventional measures for WTP cannot be defined from either question. The problem arises because the default state if a respondent answers ‘no’ is not clear. One interpretation is that the number of trips is assumed to be invariant to the change in quality and prices. Another is that you could change the number of trips taken but only in the new condition. The specific models that are estimated in their analysis and in most conjoint studies assume that the demand for trips does not change. Each individual’s demand is always one.42 Recently, Layton and Lee (2002) have suggested an innovative recoding of the ratings that would provide an interpretation consistent with a choice model. Using the random utility model, they convert the rating responses into a set of ranks. Basically, their logic follows from interpreting the ratings as a censored ranking. They recognize that the probability of any censored ranking can be represented as the sum of probabilities of some set of complete ranks. The information provided by creating the model of censored ranks based on three complete ranks depends on the number of alternatives evaluated by each respondent and the number of ties (which are endogenous in their description). Nonetheless, their strategy would seem to offer a promising way of using information previously considered inconsistent with conventional economic choice models. A second issue arises with the rating issue itself. Even if the baseline condition is included in the listing (and each person’s individual travel costs (per trip) were included in the prices), a request to provide a rating that ranges from ‘undesirable’ to ‘desirable’ asks that the respondent make a comparison to something that is not mentioned. Including the baseline condition in the list does not free the comparison from being associated with a specific standard. We can derive a net comparison but this process does not remove concerns about exactly what the ‘price’ is in that default condition. Only in the case of constant marginal utility of income (and thus no adjustment in desired trips)

Fifty years of contingent valuation

27

will CJA be able to estimate incremental WTP. It cannot recover WTP per day because the ‘no-trip’ alternative is not identified by the model.43 The second approach involves a description of a set of choice alternatives with specified attributes, including the cost as well as the non-consumption of any type alternative (Adamowicz et al., 1994, 1997 offer examples of this type of CJA).44 In these applications the model used to analyze the CJA data relies on the IIA (independence of irrelevant alternatives) assumption. By offering a randomized set of alternatives usually along with a constant, ‘no-consumption’, baseline, and pooling answers to these multiple questions across respondents, it is possible to estimate a conventional random utility model (RUM).45 However, welfare measurement, including the per-trip measures commonly used with RUM applications involving revealed preference data, generally are not feasible. The reason follows directly from the expression for the unconditional indirect utility function – the choice set for each individual is not known.46 Thus the conventional willingness-to-pay measures derived with RUM analyses from revealed preference responses are not available. Both of these conceptual issues ignore the econometric questions raised by pooling multiple responses across individuals. With the exception of Layton and Lee, much of the literature assumes responses are independent. This is probably incorrect given the experience with doublebounded CV questions. Overall, then, CJA does not appear to have fully ‘earned’ the status claimed for it! The ability to collect larger amounts of information cannot be denied. It also seems reasonable to expect that the interactive process, with each respondent making repeated choices, does help them in evaluating situations where they may not have had experience. Nonetheless, the record to date seems to suggest the methods must face most of the questions posed with CV as well as some potentially difficult identification questions in recovering sufficient information to estimate the WTP with either the ratings or choice versions of CJA.47 H.

‘Homegrown Values’ and CV

As noted earlier, Bishop and Heberlein’s research introduced the discrete responses approach to eliciting individuals’ values and was the first to propose a simulated market approach for evaluating methods for non-market valuation. A simulated market refers to a situation where the investigator conducts real sales or purchases of commodities. Their research design was a purchase format and involved sending checks (in different amounts) to those recreationists who had received hunting permits. They asked the individuals to return the permit (and keep the check) or to return the check.48 The primary objective was experimental – to implement ‘real’ transactions for an object of choice that matched what CV and travel cost methods were describing. This

28

Environmental and resource economics 2004/2005

study also sought to use the results from the simulated market to evaluate the performance of both of these methods. Their framework was imaginative and has generated a new type of experimental research to evaluate the reliability of CV. However, in many subsequent applications of the logic, results from simulated markets have been described as completely comparable to those from conventional experimental markets used in other types of economic applications. They are not. When insights are derived about the performance of the economic institutions from experimental economics, they rest on three important assumptions: (1) the economic properties of rules governing exchanges between people are not affected by the size or character of the incentives people have to engage in those exchanges;49 (2) these properties are not dependent on the specific individuals’ attributes (for example experience, cognitive ability, and so on); and (3) the properties are insensitive to other aspects of the context in which they are undertaken. Conventional laboratory experiments, as they are conducted for most economic applications, evaluate the properties of the rules governing transactions. Many applications rely on the fact that participants devote equivalent effort regardless of the size of the monetary incentive at stake. As Harrison (1989) has noted, sometimes the difference in rewards between an optimal response, requiring considerable effort to understand the incentive structure, and an approximately optimal one (but easier to understand) may be very small. The second assumption arises because the participants are usually students, but the results are assumed relevant to other (potentially older and more experienced) individuals who might face the experimental manipulation in the ‘real world.’ Simulated markets reverse the logic of the first assumption.50 That is, instead of imposing controlled monetary incentives and changing the exchange rules to evaluate their properties, they select exchange rules with known properties and infer participants’ underlying values. These studies then proceed to argue that given the knowledge of one group’s ‘values’, these measures can be used as a standard for evaluating another group’s stated values, elicited with CV. Unfortunately, the logic relies on all people having the same values (or the same functions describing their values) and, perhaps more importantly, that people’s attributes and the exchange context are unimportant to the insights derived for CV in general. Do people respond the same way to specific, small budget, private goods as to more public goods? While much of the most recent literature has sought to consider more public goods, these activities are recent and we have too limited experience to draw general conclusions One notable example by Vossler and Kerkvliet (2003) exploits local referenda and follows the early proposals made by Carson et al. (1986). In these

Fifty years of contingent valuation

29

types of studies, CV survey is conducted in advance of an actual referendum and the results compared. In some situations where the action is financed with a change in local property tax rates, the tax ‘price’ experienced by each homeowner will depend on their assessed valuation. As a result, it is possible to observe a difference in anticipated prices for the proposed public action. The Vossler and Kerkvliet study reports the findings for one of the most careful and detailed of these efforts. It relates to a riverfront improvement project in Corvallis, Oregon. The authors find that survey votes and actual votes are not significantly different. They used actual precinct-level votes to estimate a probit, together with information from a post-election survey to collect demographic information. The estimated average for the actual and hypothetical willingness-to-pay measures were remarkably close – $51.75 and $52.27, respectively. Moreover, tests of the common parameters of the respective choice models could not reject the null hypothesis that they were equal. Nonetheless, the authors do recognize the limitations in using these results for evaluating CV in other settings. They note that respondents to CV surveys may not be familiar with the policy and do not know they can vote in the future. Both conditions are different in their study. Equally important, as they note, the respondents to their hypothetical survey are unlikely to believe their responses will be consequential because they know ‘the real referendum’ is coming. Overall, then, while this research has certainly enhanced our understanding of the many influences on people’s choices, we are nonetheless left with a collection of special cases that is difficult to use to evaluate CV (or conjoint) when there are not plausible mechanisms for observing how individuals ‘choose’ the environmental resources they wish. To know how to use them requires answers to the same types of questions we had about CV. As a result, there is little point at this stage in providing a detailed review of all of the simulated market studies (see Schulze et al., 1996; Smith, 1997; Smith and Mansfield, 1998 for recent reviews and discussion of them). In some respects, the conclusion that the findings from simulated markets would be difficult to generalize was predetermined by the issues that motivated use of CV. CV is most useful where there is an absence of observed behavior that could be linked to variations in the amounts of the environmental resources whose economic values are to be measured. If homegrown value experiments (that is, relying on each participant’s preferences as compared to induced preferences through a monetary incentive program) could be designed to provide the resources, then it should also be possible to offer something paralleling that mechanism for actual provision. We would then not need the valuation measure being sought. Thus, simulated markets can be designed for commodities that are always going to be questioned as more ‘private’ or ‘useoriented’ than the CV application under study.51

30

IV.

Environmental and resource economics 2004/2005

A BOTTOM LINE ON CV

The majority of research on the reliability of CV seems to rest on the presumption that there exists a crucial experiment (or set of experiments) that, once conducted, will allow us to decide in favor of or against the method. I believe this is a strategy that can never succeed. Just as there is no single experiment that discredits the method, there is none that unambiguously supports it. In most science, the accumulated evidence of repeated and verifiable experiments testing (and failing to reject) some hypothesis corresponding to what might be described as a ‘stationary theoretical principle’ ultimately changes the beliefs of the scientific community. Accumulated evidence provides the basis for revisions in what are taken to be the principles governing the behavior that is observed. This strategy will not work with CV because the economic values to be measured (and even their relationship to observable characteristics of the background context) are unlikely to appear to be stationary. One might ask if the economic values implied by the choices of non-market goods are less stationary than private goods. This question implicitly asks about individual preferences and the information set available as choices are made. It seems reasonable to expect that market goods and services with a large number of substitutes, as well as a reasonably large share of the individual budget, would be less stationary. Small price changes would induce large changes in the quantity demanded. By contrast, goods with few substitutes and small fractions of the budget have inelastic demands and would likely have relatively stationary choice patterns. These are basic concepts underlying the reasoning used to characterize price elastic and inelastic demands. Converting them to judgments about economic values requires a more formal statement of the choice problem. For example, if we consider the lumpsum willingness to pay for a lower price of a marketed good, this measure of economic value likely follows the insights derived from elasticities. For example, consider the case of a constant elasticity demand function. Using Hausman’s (1981) logic to derive the quasi-indirect utility function, equation (1.4) describes the Hicksian willingness to pay for such a price change (that is, p1 < p0 ). 1

— (1 + g) WTP = m – ——— e–a (p11+ b – p10+ b) + m1–g 1–g 1+b

(

where m = income p = commodity price b, g = price and income elasticities (b<0)

)

(1.4)

Fifty years of contingent valuation

31

In general cases, the degree of responsiveness of WTP to the price elasticity will depend on the other features of demand. Nonetheless, for this example they do follow the basic pattern described in terms of elasticities. As we move to non-market goods, the issues in answering the issue of stationarity concern how we define the choice giving rise to the measure of economic value. Most definitions of non-use motives for environmental resources follow Hanemann’s (1988) definition and treat them as arising because the resource makes a completely separable contribution to well-being. No private goods contribute to how it enhances individual utility. Under these circumstances, the willingness to pay for increases in such a resource should not be affected by the prices of other private goods and services. It would, of course, be bounded by available income. Overall, then, stationarity of the economic values for environmental goods depends on the choices used to characterize the values and on what is assumed about how they contribute to individual preferences. As a rule, we do not know much about this second issue. Indeed, this issue is usually among the reasons analysis is undertaken. Contingent valuation estimates are primarily questioned because those providing them never actually pay. While this is important, it misses a general characteristic of how valuation measures are actually used in all policy analyses. That is, when any set of benefit measures is used to evaluate the net gains realized from a new policy, no one ever pays! If they did, the measured gains would be different simply as a result of the availability of less income for other things.52 Thus acceptance of CV may well be improved if the results from such studies were used to calibrate preference functions. The estimates of benefits for any policy could then be derived from these functions with the analytical restrictions implied by a budget constraint imposed through the maintained preference function.53 This strategy shifts attention from the mechanism used to acquire information about people’s preferences to how that information is used to compute the benefits attributed to specific policies. There are several important precedents for this type of approach for policy analyses. Harberger’s (1971) proposed approximate measures of the welfare gains or losses from price changes imposed a budget constraint on how these changes could be computed.54 This restriction has been lost in the analyses of non-market changes. Such a ‘Harbergerian calibration’ would remove one aspect of the concerns raised with CV responses. That is, ‘they are no longer payments that would never be made’. It is, as developed below, a logical consequence of the joint estimation strategy that has progressively developed from Cameron’s important advance in the practices used to analyze CV information.

32

Environmental and resource economics 2004/2005

V.

CV AND THE NON-MARKET ECONOMY

A.

Sources of Continuing Demands for CV

A lack of alternatives provides the most compelling reason contingent valuation is unlikely to ‘go away’. This motivation becomes especially important once we acknowledge that even in situations where there is some prospect for revealed preference approaches, there may not exist a continuum of environmental resource alternatives. As a result, we may not observe any behavioral response to changes in environmental quality. As Bockstael and McConnell (1999) observed: Individuals will not change their behavior if they cannot adjust at the margin and if their next best alternative generates less utility than their current choice, even with environmental degradation. A localized water quality incident may not provoke a change in behavior if the next best alternative recreation site is still less desirable. . . . the individual may, instead suffer in (behavioral) silence (p. 26)

What this implies is that observed behavior (especially relying on models that assume marginal adjustments) may not capture losses (or gains) incurred by some individuals. Combining this case with the primary ones that are usually cited for CV (that is, no available choice mechanisms to reveal behavior), we have three motivations for continued use of CV: (a) provide people with the opportunity to make (state) choices for changes in environmental objects of choice not available on a discretionary basis; (b) provide analysts greater resolution in the shape of benefit functions for different characteristics or components of environmental resources; and (c) identify the heterogeneity in adjustment thresholds across people for different amounts of environmental resources. None of these ideas is especially new. In some respects they bring us back to one of the motivations for public and environmental economics – the belief that there are some commodities people want (or want to avoid) that cannot be accommodated within conventional market exchanges. There may exist no means of making a choice (even among a limited number of alternatives). Nonetheless, all would agree that the commodity defining the choice could be extremely important. Under these conditions the commodity definition would seem to preclude (tautologically) revealed preference. Most environmental resources do not fit this extreme case. There are ways people do make choices that can reveal something about their values. The trick remains to integrate these sources of choice information with what could be

Fifty years of contingent valuation

33

derived from stated preference surveys. The link between choice and economic value is clearest when people undertake activities that in some way rely on the services of the non-marketed resource. Improving the water flow rate in a river that supports whitewater rafting or the water level in a lake that provides boating and fishing opportunities should increase the likelihood these types of recreationists would decide to use the improved resources. These increases in likelihood (or rate) of use provide information about the importance of the respective improvements to each type of uses. Thus, if analysts search for choice information related to non-market resources, recognizing that it would need to be combined with contingent valuation information, we would then expect that stated preference findings would rarely be the exclusive basis for measuring people’s values for improving environmental resources. Instead insights from these methods would be sources of complementary information. B.

CV as a Source of Complementary Data

Ebert’s (1998) overview of the conditions for consistently recovering welfare measures provides a general statement of the factors that underlie joint estimation of revealed and stated preference models. His synthesis suggests it is possible with incomplete conditional demand models to recover consistent welfare measures.55 Given sufficient information about the conditional demands for private goods, along with the virtual price (or marginal willingness-to-pay) functions for non-market goods, he defines a set of weak integrability conditions. When these are satisfied, they assure it is possible to construct welfare measures for changes in the observed goods’ prices together with changes in the observed non-market goods. Joint estimation using revealed and stated preference data for unified models, especially in cases where the process serves to isolate otherwise unobservable parameters, offers an example of how weak integrability would work. In these cases the analysis has typically begun with a prespecified direct or indirect utility function. His Proposition 2′ implies that provided weak integrability conditions are satisfied we could relax these assumptions as well. Thus, parametric pricing is not a requirement for recovering sufficient information to develop consistent welfare measures. Choice information is. Another way of interpreting professional objections to CV is to suggest people’s responses are at best motivated by a simple integral of one of the stacked components of Ebert’s incomplete conditional demand and should not, on theoretical grounds alone, be expected to offer accurate welfare measures. All prices and other public goods (or conditioning quantities) are implicitly held constant. Cameron’s joint estimation addresses this concern, albeit to date the

34

Environmental and resource economics 2004/2005

applications have not been at the scale implied by Ebert’s logic. While the Cameron logic has attracted a number of comparable replications, it has not been a ‘transforming methodology’ for non-market valuation. I believe the reason is that it appears to imply all (or at least a reasonably large number of the closely linked private goods with connections to environmental resources) consumption choices – market and non-market – must be collected simultaneously in order to use the logic. The remainder of this section is a suggestion that joint estimation does not require joint data collection. As a result, we can recast the role of CV surveys. They can be treated as small-scale and complementary data sources that are designed to search individual preferences in regions not observed with ordinary market choices. My proposal is for estimation of joint models with ‘case-controlled’ or complementary independent samples. It is not a new practice in epidemiology, or even in some applications in economics, but to my knowledge has not been recognized as a strategy for using CV in non-market valuation.56 The logic is straightforward. We begin analysis of consumer (or household) choices using a framework that results from constrained optimization. The outcomes we observe – whether the level of demand, a change in demand, a bid or a stated choice – are all defined explicitly in terms of that framework. This is the logic used by Cameron (1992) and is what underlies the ability to define weak integrability conditions in Ebert (1998). None of this requires samples from the same individuals for all the outcomes. It does require that they have the same preferences. Moreover, when there is incomplete information in any one sample about the factors influencing behavior they must be invariant across the individuals in that sample. The framework used must also contain sufficient information to identify the preference parameters to be estimated jointly with other samples. Finally, when the samples are independent there must also be shared parameters to assure any structural restrictions used in identifying parameters have an effect on the estimation. Of course, it is important to acknowledge that the proposed ability to use complementary samples relies on the information an analyst is willing to add to the process. In this case, that corresponds to the detailed specifications for preference functions and constraints that underlie the parametric restrictions. As one reader has suggested, this approach seems to be the antithesis of using non-parametric strategies with a single, well-specified, change described in a survey. This limitation is certainly true. However, the information required is not more than is already required for the joint models where revealed and state preference information are collected from the same individuals. To make this point more tangible, consider a simple example. Suppose the objective was to estimate the willingness to pay to avoid losses associated with an oil spill on recreational beaches. If we select California beaches as

Fifty years of contingent valuation

35

the location (recognizing the history of spills on the central coast of California, including the most recent associated with the American Trader case),57 then we would have two sets of information available. The first includes travel cost data collected as part of NOAA’s contribution to the Public Area Recreation Visitors’ Survey (PARVS), an on-site survey of beach visitors.58 The second is a large-scale contingent valuation survey sponsored by the State of California and the US Department of Justice. It was intended to serve as a basis for developing estimates of the per household ‘ex ante economic value for a program to prevent a specified set of natural resource injuries to those species of birds and inter-tidal life that are consistently affected by oil spills along California’s central coast’ (Carson et al., 1996b, p.iii). The first data set reflects on-site use values from beach recreation. It seems reasonable to expect this behavior would be influenced by concern for injuries from oil to ‘birds and inter-tidal life’ impacted by oil spills along the central coast. Similarly it would seem respondents evaluating plans to avoid future oil spills would consider not only the injuries avoided, but also the avoided disruption to planned beach trips because of oil spills required beach closings.59 One might also like to target the valuation estimates to specific beaches. The two studies were independently planned. The PARVS survey does not consider the values people might place on related species and the California Oil Spill (COS) study does not collect sufficient information to estimate travel cost recreation demand models. This situation is typical of most of the existing data sets available for non-market valuation. Complementary analysis allows both to be used in estimating a single set of household preferences and the desired specific benefit measures developed. It requires explicit specification of how the object of choice enters consumer preferences. Because this exercise is intended to be an example, I will assume a linear travel cost demand with the implied gains from a plan to protect beaches from oil spills – both in assuring more consistent access and in avoiding injuries as an additive effect on the trip demand. This is designated as q in equation (1.5), with x the beach trips, tc the round-trip travel cost, and m income measured on a per household member basis. x = a0 + btc + gm + a1q

(1.5)

Without loss of generality we can assume this function resulted (via Roy’s identity) from an indirect utility function given in (1.6).60 1 n = exp(–gtc) (m + — (btc + a0 + b g + a1q)) g

/

(1.6)

36

Environmental and resource economics 2004/2005

The COS study presents a plan to escort oil tankers that could impact California’s central coast paid for through a one-time income tax surcharge on California households.61 For the purpose of this analysis, I assume the plan changes q from a baseline of q0 to a new level of q*. An individual’s willingness to pay (WTP) for the plan can be derived using the definition in equation (1.6). A little algebra yields a simple expression in equation (1.7). WTP = (a1 g) (q* – q0)

/

(1.7)

Notice that all other aspects of the recreation behavior cannot be recovered from the choices about the plan, but they would affect the Marshallian consumer surplus measured for the plan based on the demand function given in equation (1.5).62 Because there is only one variation in the plan we cannot recover estimates that isolate how changes in q* influence an individual’s value. Moreover, we cannot identify a1 separately from g with the CV data alone. As equation (1.7) suggests, it is the ratio of the two parameters that scales the increment of q. With variation in a fixed fee for each increment in q, we do not observe enough information to recover a1 separately from g. Likewise the travel cost information does not include variation in q – all individuals face the same level of protection for coastal resources from oil spills. As a result we cannot consider the value of changes in the plan. This issue is one of the key problems posed with CV (and a motivation for the efforts to use an attribute-based format such as conjoint analysis). To make the CV question tangible, the proposed change in the resource (or the plan) is defined in specific terms. Without variation in the change (q1 – q0) across people, we learn nothing about how WTP responds to q. Thus, the available literature rarely provides estimates that exactly correspond to the alternatives considered in policy evaluations using benefit information. As a consequence, analysts must adjust the available results to fit the features of each proposed resource change. This process is usually described as a benefits transfer. CV estimates for a single, well-defined, change in a specific resource provide very limited information to help guide this activity. Efforts to develop these adjustments can be considered an identification problem. That is, does the available research offer sufficient information to allow those structural parameters relevant for describing people’s choices to be identified? If the answer is yes, then they can be used to compute the required benefit measure (see Smith et al., 2002). This concern is the reason why I focus on what can be identified from the specific estimating framework. A limited benefit transfer would require that we have a measure of a1. Complementary sample analysis recognizes that joint estimation with the two samples allows recovery of a1 (q* – q0) because g affects both decisions. More specifically, equations (1.5) and (1.7) can be used to define two choices.

Fifty years of contingent valuation

37

Trip demand for a given q

x = (a0 + a1q0) + btc + gm

(1.8a)

Vote for the COS plan at a one-time tax surcharge of T

WTP – T = a1 g (q* – q0) – T

(1.8b)

/

Joint estimation is accomplished (even with a take-it or leave-it CV question) by treating the score from the likelihood function for a respondent’s vote as a moment condition. Stacking the moment implied by (1.8a) with this score, and applying a generalized method of moments estimator, it is possible to incorporate the cross-equation restrictions. More specifically, for my example using a logit framework to describe the take-it or leave-it discrete choices, the resulting first-order condition for the log-likelihood function, (L), is given in equation (1.9) a ∂L exp(( 1 g) (q* – q0) q – qTi) — = Σ yi – —————————— ——— ———————— Ti i ∂q (1 + exp((a1 ) (q* – q0) q – qTi)) g

( )

[

]

/

/

(1.9)

In this framework q is the reciprocal of the scale parameter and yi is the discrete response variable for the choice each respondent makes. Combining (1.9) with (1.8a) we have a linear and a non-linear moment condition. The discrete nature of the responses implies that we estimate the structural parameters up to the scale parameter, q. Nonetheless logit or probit estimates of the model, as formulated by Cameron (1988), yi = a0 + a1T yields an estimate for WTP. Equation (1.10a) defines a0 and (1.10b) a1 in terms of the structural parameters. Equation (1.10c) suggests that their ratio provides an estimate of WTP: a0 = (a1 g) (q* – q0) q

/

(1.10a)

a1 = – q

(1.10b)

WTP = – (a0 a ) = (a1 g) (q* – q0) 1

/

/

(1.10c)

With the joint estimator we can also identify the WTP and recover estimates for both a1 (q* – q0) and g using the demand function. This would be important to the estimation of the willingness to pay to avoid an oil spill on the beaches of California’s Central Coast if we had a more detailed set of design variations in the CV survey or some independent information allowing us to describe changes in q. The joint estimator stacks data from two different surveys as if these

38

Environmental and resource economics 2004/2005

responses were provided by the same person. In implementing the model, a variety of strategies can be used. In some cases the two samples would be used in the joint estimation. Alternatively, a random sample can be selected from the larger COS sample to ‘match’ one or more specified exogenous characteristics of the respondents that are assumed relevant to the model.63 The COS sample includes individuals who use beaches as well as those who do not. The framework suggested here does not imply the structural parameters determining responses to the CV question for non-users will be different from users. The model implicitly assumes non-users have travel costs corresponding to their choke price.64 Table 1.4 provides some illustrative estimates for the model’s parameters. Three sets of samples from the COS samples were considered. The first is a simple random sample from the full COS survey. The second matches the reported beach use distribution of the set of respondents to the COS survey from the central coast area of California to the distribution in the PARVS sample for the two beaches in the impact area for the spill. The last limits the random sample from the COS survey to households with members who traveled the coastal highway (Highway 1) in the last five years.65 Two sets of estimates are reported with each set of samples. Separate estimates are given for the CV choice model with logit and the joint estimation with the complementary samples. The estimates of the travel cost demand model with the PARVS data (using OLS) do not change with the different samples from the COS survey and are reported in the first column. Independent logit analysis of the CV survey will change with the sample and the separate estimates with each sample are reported in the first column under each sample grouping. The logit choice model alone cannot identify the effect of beach quality and the oil spill plan. This limitation is indicated in the labeling of the parameters for each sample. Because PARVS and COS surveys were undertaken five years apart, the Consumer Price Index (CPI) was used to adjust the travel cost and income levels reported in 1989 for PARVS to 1994 dollars. The second column in each ‘sample-grouping’ reports joint estimates with a generalized method of moments estimator, using as instruments the exogenous variables from each of the complementary samples.67 The estimates across the three samples are quite comparable, with a small increase in the size of the jointly estimated parameter (that is, 0.074 versus 0.063) for the quality/oil spill plan with the sample drawn from users in the central coast area. This difference is not statistically significant. Thus, at this stage, the best one can conclude is that there is suggestive evidence that the complementary analysis would help to target plausible individual effects. The primary advantage gained in this example is the ability to identify separate estimates for a1. Such identification would allow benefit measures for a wide range of use and quality changes when it is applied to a CV survey with greater

Fifty years of contingent valuation

39

Table 1.4 Results for joint estimation with complementary samples Parameter

a0 (intercept) b (travel cost) g (income) a¯1 (quality) q (reciprocal scale) q(a¯1/g)

Simple random Use matched Traveled Highway 1 ————————————— —————————— ————— —————— OLS/LOGIT JECS LOGIT JECS LOGIT JECS 26.413 (4.71) –0.492 (–3.10) 0.081 (0.62)

0.009 (3.48) 0.789 (2.71)

Proportion of yes responses 0.50 Mean number of trips 18.44 No. of observations 118 R2 0.077

26.413 (4.77) –0.492 (–3.14) 0.081 (0.63) 0.063 (0.61) 0.009 (3.21) –

118 –

0.007 (2.87) 0.928 (3.01)

26.422 (4.77) –0.492 (–3.14) 0.080 (0.62) 0.074 (0.61) 0.007 (2.82)

0.008 (2.89) 0.605 (2.18)

0.56

0.50

18.44 118 –

18.44 118 –

26.412 (4.77) –0.492 (–3.14) 0.081 (0.63) 0.048 (0.60) 0.007 (2.70)

118 –

variation in the quality conditions included in the design variations presented to respondents. More generally, the primary advantage of joint estimation is in the ability to recover an estimate of the effects of the plan (that is, a1 (q* – q0)) on individual preferences. If the index people used to evaluate q were known, then we could estimate changes in q different from what was offered in the CV survey. Moreover, this ability is sustained without the need to have wide variation in q in the CV scenario. I do not want to understate the difficulty in knowing the index or set of values people use to characterize q. In most realistic applications q is likely to be multidimensional and to arise in bundles of different services or characteristic ties with technical restrictions on how the individual components can vary. My discussion of the issues associated with designing a scope test in the analysis conducted for the Montrose case is an example of those relationships. If the goal is to link the source of an injury to an impacted species, then the

40

Environmental and resource economics 2004/2005

intervening connections must be established and changes must be plausible within that context. Technical measures of q are unlikely to be on the same scale as is used by people to make their own judgments about resources. All of these issues create significant challenges for any activity that proposes to parameterize a role for q in individual preferences. Clearly, more work is needed on the impact of other types of sampling criteria. Nonetheless, the prospect of being able to investigate these alternatives with modest research resources highlights the potential importance of designing CV studies to serve as complementary samples that can be used to enhance the insights available with larger existing data sets. Under these circumstances, the structure of the CV object of choice and choice context can be designed to explore the role of q in preferences and its potential for interaction with priced commodities that may be linked to it.

VI.

DISCUSSION

Contingent valuation has prompted the most serious investigation of individual preferences that has ever been undertaken in economics. These efforts have used teams of social scientists, drawing insights from cognitive and social psychologists, sociologists, and survey researchers into the approaches environmental economists now routinely use to elicit consumer choices. The success in this integration of research methods across these disciplines has, to some degree, been overlooked in the controversy about CV. CV research has also transformed the framework used in experimental economics. Conventional experimental methods relied on using controlled incentives to study the effects of information and exchanges rules. Beginning with Bishop and Heberlein (1979) environmental economists changed the research framing to one that relies on known properties of exchange rules (and information sets) to study preferences and methods for eliciting them. The framing of questions used in other economic surveys remains primitive by comparison to the methods designed by contingent valuation researchers. Borrowing from marketing research and psychology, a linked sequence of focus groups, cognitive interviews and pre-tests is now routine practice in designing the key questions for CV surveys. The laboratory research under way at the Bureau of Labor Statistics and the Bureau of the Census, intended to improve conventional economic questions, is a decade or more behind standard practice in CV research. These two activities, together with the insights derived from hundreds of applications, have offered a much better understanding of what is actually revealed about preferences from an individual’s choices. Fifty years ago Houthakker (1950) demonstrated that Samuelson’s (1938) argument that

Fifty years of contingent valuation

41

revealed preference analyses freed demand analysis from the need to specify a utility function was not correct. As Samuelson (1950) later acknowledged, the requirements for integrability are not so easily dismissed. Unfortunately none of this work ever considered objects of choice that were not ‘supported by’ a budget constraint with constant (per unit) prices. Houthakker’s proof demonstrated how we could vary prices and income so as to reveal non-integrable preference fields. This ‘decade plus’ sequence of research on what we really know about the properties of preferences from behavior has focused particular attention on the early theory associated with rationed goods. As Carson et al. (2001) suggest, intuition about what properties can be expected for income responsiveness or substitution that is formed based on insights using priced goods does not readily transfer to the case of rationed goods. Moreover, it is the rationed commodity case that is most directly relevant to non-market valuation. Contingent valuation has provided both motivation for developing these insights and a methodology grounded in survey design and econometric analysis that allows economists to learn about consumer preferences without the separating hyperplane defined by income and prices. As in the case of the revealed preference axiom, CV does not free the analyst from the use of assumptions. This recognition has been slow to come to CV applications. Restrictions from economic theory have been viewed as ways to discipline ‘wayward’ choices so they appear to be ‘economic’ decisions. This strategy is too narrow. Theory offers the opportunity for designing a constructive program where the economic model of choice can be used as an integral part of the design of understandable decisions for non-market resources. The primary limitation to a strategy that collects both CV questions and more conventional economic data has been the cost of data collection. However, the prospect of using complementary samples to estimate jointly preferences may relax this significant constraint. The focus on using the theory of individual choice along with the insights from psychology and survey research in the design of stated preference surveys has another advantage. It offers the potential to ‘build in’ questions that elicit confirmatory behavioral responses. Measures of willingness to pay can never be confirmed but observable behavior can. By specifying explicitly the full dimensions of the economic choice process as a part of the design of CV questions it is possible to evaluate the performance of the maintained theory and the CV questions. This approach is the non-market equivalent of Varian’s (1982, 1983) generalized axiom of revealed preference and may offer the opportunity, together with complementary sample analysis, to change the economic profession’s view of stated preference research.67

42

Environmental and resource economics 2004/2005

APPENDIX A: TEXT OF THE CALIFORNIA OIL SPILL STUDY’S CONTINGENT VALUATION QUESTION (Source: Carson et al., 1996b, Appendix A, pp. 8–14) Recently, the federal government passed a new law to help reduce the number of oil spills. Ten years from now, all oil tankers and barges will be required to have two outer hulls instead of the single hull most of them have now. Double hulls provide much more protection against oil leaking after an accident. However, it will take ten years before all single-hulled tankers and barges can be replaced. Until then, spills are expected to happen every few years along the Central Coast, just as they have in the past, unless something is done. In the next ten years: Scientists expect that a total of about 12 000 birds of various types will be killed by oil spills off the Central Coast. In addition, about 1000 more birds are expected to be injured but survive. Also, many small animals and saltwater plants are likely to be killed along a total of about ten miles of shoreline. The harm from an oil spill is not permanent. Over time, waves and other natural processes break down the oil in the water and on the shoreline. Typically, within ten years or less after a spill, there will be as many of the affected birds as before the spill. The small animals and saltwater plants in the affected area recover somewhat faster, in about five years or less. If taxpayers think it is worthwhile, the state could prevent this harm by setting up a prevention program for this part of the coast. This program would be similar to those successfully used by other states, such as the State of Washington. It would last for ten years, until all tankers and barges have double hulls. This program would do two things. First, it would help prevent oil spills from occurring. Second, if an oil spill does occur, it would prevent the oil from spreading and causing harm. Here is how a Central Coast program would prevent spills from occurring. Oil spill prevention and response centers would be set up in three different locations along this part of the coast. Specially designed ships, called escort ships, would be based at each center. An escort ship would travel alongside every tanker and barge as it sails along the Central Coast. This would help prevent spills in this area by keeping the tankers and barges from straying off-course and running into underwater rocks, other ships, or pipelines. If any oil were spilled, here’s how the program would keep it from spreading and causing harm.

Fifty years of contingent valuation

43

The crew of the escort ship would quickly put a large floating sea fence into the water to surround the oil. To keep it from spreading in rough seas, this fence would extend 6 feet above and 8 feet below the surface of the water. Then skimmers, like the one shown here, would suck the oil from the surface of the water into storage tanks on the escort ship. Other ships would be sent from the nearest prevention and response center to aid in the oil recovery and clean-up. The money to pay for this program would come from both the tax-payers and the oil companies. Because individual oil companies cannot legally be required to pay the cost of setting up the program, all California households would pay a special one-time tax for this purpose. This tax money would pay for providing the escort ships and setting up the three oil spill prevention and response centers along the Central Coast. Once the prevention program is set up, all the expenses of running the program for the next ten years would be paid by the oil companies. This money would come from a special fee the oil companies would be required to pay each time their tankers and barges were escorted along the Central Coast. Once the federal law goes into effect ten years from now, all tankers and barges will have double hulls and this program would be closed down. We are interviewing people to ask how they would vote on this Central Coast prevention program if it were put on the ballot in a California election. There are reasons why you might vote for setting up this program and reasons why you might vote against it. The program would prevent harm from oil spills in the Central Coast area during the next ten years. Specifically, the program would: • Prevent the deaths of about 12 000 birds as well as the deaths of many small animals and saltwater plants along about 10 miles of shoreline, and • Prevent 1000 more birds from being injured. On the other hand, • The number of birds and other wildlife it would protect is small in comparison to their total numbers, and none are endangered. • Your household might prefer to spend the money to solve other social or environmental problems instead. • Or, the program might cost more than your household wants to spend for this.

44

Environmental and resource economics 2004/2005

If the Central Coast prevention program were put into place, it would cost your household a total of $——. You would pay this as a special one-time tax added to your next year’s California income tax. If an election were being held today, and the total cost to your household for this program would be $——, would you vote for the program or would you vote against it? The proposed tax amounts were randomly assigned as either $5, $25, $65, $120 or $220. The study reports an estimate of the Turnbull lower bound mean for the ex ante economic value of the oil spill prevention program of $76.45 (standard error of $3.78) in 1994 dollars.

NOTES *

1.

2.

An earlier draft of this chapter was presented at the Kobe Conference on Theory and Application of Environmental Valuation, organized by the Japan Forum of Environmental Valuation, 22–23 January 2000, at Kobe University, Rokkoudai Campus, Japan. Thanks are due to participants at that conference for helpful comments, to Richard Carson, Rick Dunford and Pierre du Vair for providing me with the two data sets used in the application described in section IV, and to Richard Carson, Bengt Kriström and one anonymous referee for very detailed and constructive comments on a revised draft of the 2000 paper. Thanks are also due to Susan Hinton, who continues to make sense of my unending drafts. Partial support for this research was provided by the US Environmental Protection Agency under CR824861-01-0 and R-82950801 as well as from the NC Agricultural Research Service Project # NCO 5889. Most economists attribute the first suggestion to use survey techniques to value non-market resources to Ciriacy-Wantrup (1947). In a market context with the objective of valuing attributes of fresh produce Waugh (1929) may well have been the first economist to use survey techniques. Often credited with the first application of hedonic price indexes, Waugh also conducted a survey to elicit valuation scales for these attributes and suggests values could be elicited directly. Robert Davis’s famous Ph.D. thesis in 1963 marks the start of formal empirical efforts to use surveys to measure willingness to pay for non-market goods. Hanemann (1992) notes other less formal efforts as early as 1958, citing Mack and Meyers’s (1965) contingent valuation survey. NOAA was designated a trustee for some types of natural resources under the Oil Pollution Act of 1990 and assigned the responsibility to develop regulations for natural resource damages as part of the liability for these damages established in the United States by the Superfund legislation and the Oil Pollution Act. The Panel NOAA composed to assist in the rulemaking process included economists and other social scientists, including two Nobel Laureates. The Panel’s was report published in the Federal Register in 1993, with follow-up comments in 1994 (see Arrow et al., 1994). The exact sequence of events leading to this panel and its activities is a complex story beyond the scope of this chapter. See Portney (1994), Diamond and Hausman (1994) and Hanemann (1994) for further perspectives on them. The Office of Management and Budget (OMB) guidance recommends that the NOAA Panel’s scope test be conducted in an external or split sample format. It also calls for inperson interviews, ties judgments about reliability to response rates, and recommends reminders of substitutes and alternative expenditure possibilities as well as deflection of transaction values. Each of these elements is presumed to increase the reliability and quality of a CV study. However, as Harrison (2000) suggests, there is not a set of systematic tests

Fifty years of contingent valuation

3.

4. 5.

6.

7.

8.

9.

45

of the effects of these choices that supports this recommendation. They certainly ‘make sense’. The issue is documentation of the basis for suggesting them (OMB, 2003, p. 5519). This judgment may be too extreme. Some reviewers of an earlier draft have suggested that the differences in economists’ evaluations of contingent valuation are not as marked. It would be interesting to systematically compare the relative frequency of articles using laboratory experiments (also a new methodology) in the top economics journals to the same relative counts for those papers using contingent valuation or related survey methods in the top journals in the post-1992 period. My conjecture is that there would be a significant difference favoring applications of experimental methods. Of course, a very detailed process is used to assemble the information necessary to construct the Consumer Price Index. But the price information must be collected. See Bureau of Labor Statistics (1997) for an overview. Scanner data on sales in supermarkets have been available to marketing research firms for some time (see Cotterill, 1994). Because they are on privately held databases, these data are only recently becoming available to economists. Similarly, large-scale databases on housing sales are becoming available to economists and are transforming hedonic modeling. See Beron et al. (2001) and Sieg et al. (2002). Without an experimental test evaluating the importance of these behavioral responses it is hard to judge the size of their effects. The logic for their existence parallels the concerns raised by Heckman (1998) about evaluating the general implications of policy interventions with social and natural experiments. One indirect set of information indicates the effects of incentives in a survey context do exist. Smith and Mansfield (1998) found differences with the time of the week a survey was conducted as part of an experimental evaluation of the reliability of CV. This analysis used a question offering to pay respondents to participate in a second interview. The sample was randomly assigned to different treatments – one offered ‘real’ payments and the other ‘hypothetical’ payments. The participation and implied opportunity cost of time depended on the timing of their initial interviews. The experimental question suggested the new interviews would be conducted under comparable conditions. It appears that respondents interpreted this framing to include the timing of the interviews. If we assume this interpretation of their expectations is correct, then they revealed a significantly different responsiveness to financial incentives depending on when the interviews would be conducted. When the first systematic evaluation of contingent valuation was undertaken (see Cummings et al.,(1986), I reviewed a range of economic surveys. It was possible to identify six tasks that were asked of respondents in economic surveys: recall, partitioning, judgment of a status, reports of sensitive information, evaluation of attitudes, and responses to hypothetical circumstances. See Smith (1986) for a report of the specific wording of some of the questions illustrating how these tasks were framed. Early research by Kain and Quigley (1972) identified the biases in these reports when compared to market prices. While their analysis suggests that the averages from these reported values across people are reasonable as approximations of the average market prices, the individual values were found to be responsive to the census respondent characteristics. Ideally, we might expect a new homebuyer to report accurately the market price he (or she) paid for the house involved. One referee for this chapter has suggested this is not obvious. My point is simply that I would not expect these reports to vary with the socioeconomic characteristics of the people providing them. Home sales prices are reported to county tax departments for assessing property taxes and these records are in the public domain. As a result, there does not appear to be a clear incentive to misrepresent the data. For many market goods, where there is not scope for bargaining, it would seem the same argument might apply. Even private databases developed from market transactions can involve a selection effect. These are likely to become increasingly important as consumers become aware that their choices are being saved and linked to other databases describing their demographic and financial characteristics. I do not have direct evidence that this recognition is growing. However, I do have several types of anecdotal information consistent with this conclusion. Recently, I attempted to obtain information from preferred customer cards from a large US

46

Environmental and resource economics 2004/2005

retail grocery chain. The managers explained they could not cooperate because their customers expressed concerns about the record keeping and tracking of consumption. A small percentage of respondents to a recent mailed survey about the use of freshwater recreation sites in North Carolina also expressed similar concerns about the identification numbers on the surveys. About half the respondents who called to ask about the survey indicated they were aware of the tracking or asked how they were identified to be recruited for the survey. Finally, news accounts about tracking the usage of Internet sites by private consumers have caused many consumers to become increasingly sophisticated in the recognition of how their behavior is being recorded. 10. Some examples of studies in this category include: Brookshire et al. (1976), Rowe et al. (1980) and Schulze et al. (1981). 11. The biases initially identified included: strategic, starting point (for iterative ‘bidding’ elicitation questions), vehicle (for the payment mode), and information. Mitchell and Carson (1989) extended this framework to provide a fairly exhaustive description of the potential sources of bias in their ‘Typology of Potential Response Effect Biases’ (see Table 11-1, pp. 236–7). 12. In many respects this paper’s descriptions of the key features of survey design closely parallel the subsequent guidelines attributed to the NOAA Panel almost two decades later: Some desirable characteristics of such surveys have been identified . . . The hypothetical situation presented should be realistic and credible to respondents. Realism and credibility can be achieved by satisfying the following criteria for survey instrument design: Test items must have properties similar to those in the actual situation; situations posited must be concrete rather than symbolic; and test items should involve institutionalized or routinized behavior, where role expectations of respondents are well defined. (Randall et al., 1974, p. 136) 13.

14.

15.

16. 17.

18.

The site originally proposed for the work was to be water quality conditions in specific sites in North Carolina. Once the award for the research was made to the Research Triangle Institute, EPA staff indicated a preference for considering the Monongahela River as the study location. We learned much later after the research was completed that the Agency was involved in litigation concerning emission standards for steel firms in the area and there was some belief the results of the survey might have value in that case. However there is an important distinction between the NOAA Panel’s focus and our study. Scope was used as a gauge of the reliability of CV when it was used in natural resource damage assessments to estimate non-use values. In this study we were measuring changes in use values with the attributes of the conditions at the site involved in the recreation. Indeed, in a recent conjoint analysis of smokers’ responses to different types of hypothetical filters (to evaluate the importance of the health effects of cigarette smoking to smokers), a team including Reed Johnson and I asked about both the amount of filters after the choice and whether respondents would use them for every cigarette that they smoked. We conjectured that smokers might try to adapt to the price of these filters by adjusting the risks of health effects with less than a 100 per cent utilization pattern (for example use filters with a subset of the cigarettes smoked to save money). Readers, I am sure, can appreciate the ‘nightmare’ we created for subsequent analysis. I should add that I cannot cite specific evidence that this research had more influence on mainstream economists. Taken together with Cameron (1988) and McConnell (1990), this research led to a much closer examination of the properties we should expect of Hicksian welfare measures. It was the large disparities between willingness-to-pay and willingness-to-accept measures and the subsequent speculation that they resulted from failures of CV that stimulated significant advances in economists’ understanding of the properties of these measures. See Hanemann (1991, 1999) and Hanemann and Kanninen (1999). The interest in incentive characteristics arises because the questions separate the individual’s choice from the mechanism that describes how the collection of ‘all’ individuals choices is used to determine whether the stated offering would be provided. Thus the incentive analysis is really about provision mechanisms and not whether respondents take CV questions

Fifty years of contingent valuation

19.

20. 21.

22.

23. 24. 25. 26.

27. 28. 29.

47

seriously. Taking them seriously is a maintained assumption that permits the subsequent analysis. An issue raised independently by Shläpfer (2002) and Nicholas Flores concerns whether respondents adjusted the stated payment to reflect their personal circumstances – those in high income brackets answering as if the stated payment they would be required to make were lower than what they believe it would actually be. As Shläpfer notes, ‘there clearly is a strategic reason for the respondent to do other than answer truthfully however, when payment levels do not correspond with the actual cost incurred by the respondent . . .’ A few recent related studies considering the form of the question include Balistreri et al. (2001), Geraud et al. (2001), Poe et al. (2002) and Reaves et al. (1999). The administration of the surveys across mode varied. The actual participation question was a telephone survey with a specific number of participants identified. The one price hypothetical discrete choice also identified the total number of desired participants for the program and was mailed. All other modes were mailed. There is one potentially important limitation in their model. While the theoretical development of their joint model and estimator begins with a common preference function and error specification, their implementation introduces a subtle problem. One of the conditioning variables for their varying parameters across individuals is income. They explain this potential inconsistency as a reflection of eliciting income in classes and as an effort to use the midpoint of the class as discrete valued indicator of socio-demographic status to capture heterogeneity in preference parameters (p. 409). While this seems plausible, most sources of income from surveys convert interval responses to continuous measures using mid-points and adjustments for open-ended intervals. The expression for willingness to pay will be different with their non-linear form for preferences and this variable interpreted as the measure of income. Ideally, their comparisons would have reported other than income to provide the varying parameters. Given the importance of the functional form for preferences to their comparison across modes, their decision to use income as a proxy for demographics raises inevitable questions about whether they achieved the consistent comparison they sought. As I have suggested throughout this discussion, the economic values can only be measured for a well-specified object of choice – which in this case is some change (positive or negative) in one or more environmental resources. Their research also included consideration of discrete response versus open-ended questions and a second independent analysis of scenarios that evaluated plans to avoid injuries from different-sized oil spills. There are a number of reasons, aside from the commodity, for questioning the relevance of the Desvousges et al. survey. What is important from the perspective of the evaluation of CV is the impact of the findings on the recommendations of the NOAA Panel and new research. Indeed, in unpublished research Richard Carson and Nicholas Flores (1996) found that the estimates for different wilderness areas could be ‘rationalized’ by the sizes of the areas involved. That is, by simply regressing the mean WTP reported by Diamond et al. (1993) on the number of acres preserved for each of the three areas, they reject the null hypothesis that WTP does not increase with size at a 0.01 level. Their results were based on summary statistics and not the original data, so the sample size was limited to the number of reported sample means. The Exxon settlement for the oil spill called for $1 billion in natural resource damages. The Carson et al. (1992) estimated median was $30.91 (in 1991 dollars using the Weibull survival specification) per household. The Panel also included Edward Leamer, Paul Portney, Roy Radner and Howard Schuman. For related discussion, see also Arrow et al. (1994). One referee pointed out that perhaps the NOAA Panel sought to prevent the government from ‘cutting corners’. That is, under the legal framework that the Panel was asked to take as guidance, the government gets reimbursed for all the costs of doing the research studies (including interest costs), provided that they were done in a cost-efficient manner, and that the government prevails on the claim. Under this framework, it may seem unlikely there would be disincentives to perform a state-of-the-art study. However, this is only part of the

48

30.

31.

32. 33.

34.

35. 36.

37.

Environmental and resource economics 2004/2005 story. There are limits – in the time of public attorneys to bring cases and in the legal record of decisions to guide practice. So, state-of-the-art research does not necessarily assure success. The final regulations for damage assessments changed completely in 1996, recommending a physical criteria ‘habitat equivalency’ (see Smith, 1999 for discussion), and CV research largely returned to the role of small-scale research studies. This change took place despite the fact that the rules do allow CV and other choice-based research to be used in these cases. The NOAA Panel’s concern about temporal reliability of CV responses seems to be more closely linked to highly visible damage assessment cases like the Exxon Valdez. Comparison of the 1991 and 1993 responses suggested it was not important for either the distributions of the discrete responses or the willingness-to-pay estimates derived from the Exxon Valdez survey. The details of each test are reported in Carson et al. (1997, 1998). When I have described this research in workshops and to critics of CV, it has been suggested that more variations in species and time should have been pursued. This requirement for consistency in the linkages limited the design alternatives that could be undertaken. Kahneman and Riktov (1994) rediscovered a critique of CV first raised by Cummings et al. (1986) evaluation – respondents seem to be willing to pay about $25 for anything. As a result, one might ask whether people’s responses are simply reflections of a desire to ‘do good things’, or are specific efforts to improve the specific aspects of environmental resources that are posed in CV questions. See Carson et al. (1996a) and Smith (1996) for further discussion of this line of research. Hanemann (1996) has made the same point quite forcefully in commenting on the earlier ‘bird loss’ experiments that began the line of research on scope tests. He observed that: My view is that economic theory per se provides no guidance about what people should care about or how much they should care. As in Diamond’s case, one can always generate specific predictions by introducing some assumptions. But, those predictions are no more valid than the assumptions they rest on. As Simon [1986] has noted, ‘almost all the action [in economics], all the ability to reach non-trivial conclusions, comes from the factual assumptions and very little from the assumptions of optimization. Hence, it becomes critically important to submit the factual assumptions to careful empirical test’. What I find disquieting is the apparent reluctance to do this. Thus, in the case of the birds or the wilderness areas, when CV data disconfirm his additivity assumption, Diamond chooses to reject the data and believe the assumption. I think that many of the assertions about what types of preferences are consistent with economic theory are merely expressions of personal opinion masquerading as statements of economic theory. The critics of CV are saying ‘I don’t feel that way about bird deaths, and I don’t think that anybody else should’. (Hanemann, 1996, p. 53) Her unpublished paper circulated for several years before publication in 1992; see Cameron (1992) for the final version. A number of applications followed in the early 1990s. See Morikawa et al. (1990), BenAkiva and Morikawa (1990) and Ben-Akiva (1992) as examples. Market researchers (see Swait and Louviere, 1993 and Swait et al., 1994) then used the framework to recover estimates of the ratio of the scale parameters for the errors assumed to be present for the two types of data – describing it as a type of calibration factor. See Louviere et al. (1999) for a fairly recent summary of the marketing applications in this literature. The treatment of the amount of the environmental resource in functions linked to the responses provided by individuals depends both on whether scope effects are incorporated in the CV design and on whether the form of preferences allows the effects of q to be identified. In some situations, we can also assume there are differences in the baseline conditions of q relevant to each respondent. In this situation, even if the CV question offers the same increment, it may be possible to identify the parameters of the resource measure with measures of the different baselines. A composite of early discussions of the modeling of responses from discrete response questions (especially McConnell’s 1990 evaluation) and Cameron’s paper triggered this line of research. I attempt to make this connection specific in Table 1.1 describing how behavior relates to WTP functions in Smith (1997).

Fifty years of contingent valuation

49

38.

This is a complex issue that offers the potential to relate the modeling of joint systems (revealed and stated preferences) to the Kuhn–Tucker approaches for modeling recreation demand. It requires only the translation of questions into a format consistent with the preference specification being used. 39. Double-bounded questions offering interval estimates of willingness to pay are early reflections of the use of these interconnections. 40. As discussed below, the joint estimates do not have to involve responses from the same individuals. These types of applications have been prevalent in labor economics for some time (see Angrist and Krueger, 1992 and Arellano and Meghir, 1992), but have not found their way into environmental economics. There are several ways to implement these strategies and I illustrate one below. Others involving matching actual survey responses to pilot surveys or developing joint estimates are discussed in Sloan et al. (2003) and Smith and Pattanayak (2002), respectively. 41. Rankings have also been used in conjoint surveys. This questioning format corresponds more directly to a choice format comparable to the random utility model. 42. In the linear case they specify a ratings function as (omitting demographic effects) r(p, q, m) = f (v (p, q, m)) where p = price, q = quality, and m = income. They further specialize this in two cases: (A) r ( ) = f (v(q) + a (m – p)) or (B) r ( ) = f (v(q) + a ln (m – p))

43. 44.

45. 46.

47.

48.

In both cases the Marshallian demand for trips is unity. There is a further issue in the comparisons. If we return to the case they ask about quality change in the CV question, the specified ‘p’ is a price increment, not the price of the trip and pi in the conjoint version is presumably the price of a trip. Thus the price may not mean the same thing in both questions. The form of the payment in Johnson and Desvousges (1997), as the percentage of a respondent’s utility bill, poses even more serious problems when we attempt to link it to a Hicksian WTP. One of the earliest applications of the multinomial choice framework in a CV/conjoint setting was undertaken in Carson et al. (1990). Respondents were offered the choice of purchasing a Kenai King salmon stamp that allows them to take different numbers of salmon in fishing trips. The number of salmon available in different types of stamps is controlled by the government so the choice set issue was avoided. A no-purchase alternative was permitted using a nested logit framework. Thus, this early study avoided many of the criticisms of later conjoint research. This conclusion relies on the properties of the type I extreme value distribution and simple nested logit models in allowing for random selections of choice alternatives in estimating the preferences underlying an RUM description of individual choice. MacNair and Lutz (1998) have suggested a potentially important resolution of this problem. They call for using the initial choice set available to respondents and considering CJA as a supplement to a revealed preference analysis. In this context the revealed preference would be necessary to define the baseline choice set. An alternative is to use the estimated indirect utility function to evaluate baseline or status quo conditions as a single alternative relative to one new alternative (the policy option). This approach would not require an evaluation of the alternatives a person might consider and could be developed in a consistent way. However, this strategy abandons a central motivation for adopting an RUM in the first place, completely ignoring substitutes for the policy option. The experience reported in Johnson et al. (2000) does indicate that there is learning and fatigue in a range of different conjoint analyses. Most of the studies in the literature have not incorporated these effects. Moreover, there has been very limited research evaluating whether these effects can be distinguished from the incentive and informational influences on respondent choices. For an alternative view on the potential strengths of conjoint methods, see Adamowicz and Boxall (2000). Ninety-four per cent (22 people) returned either their check or the permit. Sixteen checks were not returned. The authors’ discussion notes that when payment was stopped on the

50

49.

50. 51.

52.

53. 54. 55. 56.

57.

58.

59.

60. 61. 62.

Environmental and resource economics 2004/2005 checks, five days after the deadline only one check had been cashed without returning a permit. This assumption refers to attributes outside the confines of incentives defined by the optimization models used to describe behavior. For example, one participant may ‘prefer’ to deal with another – they enjoy the conversation that accompanies the exchange. Ex post we can rationalize that preference using conversation as a utility-enhancing jointly consumed ‘product’. However, we could not have assumed it would be present based on the conventional optimization models used to describe individuals’ choices. A more detailed summary of their initial results as well as further experiments is provided in Bishop and Heberlein (1990). One of the results from Carson et al. (1999) is relevant to those applications where individuals are asked to contribute voluntarily to the provision of a public good. There is a clear prediction about the expected results from ‘real’ and hypothetical applications of this approach. Their analysis suggests a real contributions format will lead to choices that understate WTP and the hypothetical consequential format in the sense that it is perceived as influencing a contributions program, should lead to choices that overstate WTP. Thanks are due to a referee for suggesting this connection to their work. Of course we rely on having sufficient knowledge of preferences to compute a hypothetical case where the result is a gain with net benefits positive, even with such reallocations. My point is that this conclusion is a conceptual point and no more a ‘revealed’ outcome than CV is a revealed payment. See Smith et al. (2002) for an example using this logic with results from contingent valuation, travel cost recreation demand and hedonic methods. See Hines (1999) for further discussion. The primary contributions are Mäler (1974), Hausman (1981), LaFrance and Hanemann (1989), Willig (1978) and Bockstael and McConnell (1993). In their use of independent macro- and micro-data sets in estimating labor supply responses, Imbens and Lancaster (1994) describe several analogies to the use of independent samples in jointly estimated models. In environmental economics, there are also a number of parallels including the use of the varying parameter model, regional recreation demand models, meta analyses and others. Nonetheless, I could not find a discussion of using CV in ways that are analogous to the case-controlled or treatment samples used in epidemiology. On 7 February 1990, a tanker, American Trader, spilled 416 598 gallons of oil off the coast of Huntington Beach, California. Some of the oil was washed ashore and approximately 14 miles of beach were closed for 34 days. This area closed extended from Alanaitos Bay in Los Angles County to Crystal Cove State Beach in Orange County. Under the 1990 Oil Pollution Act, as well as California statutes, the trustees for resources injured as a result of the spill can seek damages for the injury, or loss of use of the affected resources. See Chapman et al. (1998) and Dunford (1999) for a discussion of the case from different perspectives. This example illustrates how existing data might be used to evaluate this type of loss. This on-site survey was undertaken by NOAA for ten beaches in the summer of 1989. Five were in California, three in Oregon and two in Washington. Two areas, San Onofre State Beach and Cabrillo-Long Beach, fall in an area close to the spill and were used for my example. See Leeworthy et al. (1990) for discussion. Independent analysis of these two along with Santa Monica beach was undertaken by Leeworthy and Wiley (1993). My argument is somewhat similar to an early discussion by McConnell and Duff (1976). They considered the travel cost demand when there was uncertainty about entry. In this case their first-order condition adjusted the travel cost by the inverse of the probability of entry, because they could incur the cost and not be admitted. Here we might assume they know in advance about availability of the site so they don’t have to incur the costs. See Hausman (1981) for details and Bockstael et al. (1984) for other examples. Von Haefen (2002) provides a complete catalog of restrictions for the case of multiple equation applications. The specific text of the COS contingent valuation question is given in Appendix A. This format is the same specification Larson (1993) used to argue that q represented the

Fifty years of contingent valuation

63.

64.

65.

51

effects of non-use or existence value because q did not appear in the Hicksian demand for recreation. This interpretation is arbitrary. My point here is to illustrate the use of complementary data sets in a straightforward way, not to consider the full implications of alternative linkages between q and x. Of course, that would be an important aspect of more detailed applications of the proposed methodology. Examples of the basic logic proposed here can be found in Arellano and Meghir (1992) and Imbens and Lancaster (1994). There is an important addition to my proposal over theirs. It is the use of a formal structural model of choice that implies specific restrictions linking the individual models’ parameters. It is also possible to consider matching to develop control samples for reduced form tests of treatments or manipulations to the information or context in stated preference analysis. This process requires a common set of questions across surveys. See Sloan et al. (2003) for an example. The indirect utility function was derived from the travel cost demand model. As a result this is the only interpretation it allows for non-users. The general framework does not require this type of structure. We could select more complex treatments (see Table 1.1 above). This simply requires more common information from the samples being used in a complementary joint analysis. The trips were reported in intervals with the COS study. The PARVS trip records were adjusted to conform to the COS format. The trip distribution with the numbers of respondents and the percentage of the sample for the three samples is given as: Trips

PARVS

COS

1 or 2 >3 and <=10 >10

47 (39.8) 37 (31.4) 34 (28.8)

236 (21.7) 349 (32.2) 283 (46.1)

COS/Central Coast 105 (26.4) 168 (42.3) 124 (31.2)

For the other sample (that is, a household member boating or fishing in the past five years) there were only 500 households satisfying this criterion in the COS sample. To match the number of observations in the PARVS sample we selected randomly 23.6 per cent of this group. 66. The instruments for the first and second models included: the travel cost, income, age, amount of the one-time payment in the COS sample, whether an individual had traveled the coastal highway, lived in the central coast and was familiar with birds likely to be injured. The last model drops the variable associated with using Highway 1 because it was a part of the criterion defining the sample and adds a qualitative variable based on whether a respondent identifies himself (or herself) as an environmentalist. 67. Crooker and Kling (2000) have taken a first step in this process in developing non-parametric bounds for CV estimates. My point is that once the objective of undertaking such behavioral ‘tests’ is built into the design of CV surveys we can expand the range of uses of revealed preference insights beyond the bounds they propose. As Houthakker proposed a sequence of price and quantity changes to uncover non-integrable preference fields, the CV research can also search the space of stated choices for consistent responses.

REFERENCES Adamowicz, V. and P. Boxall (2000), ‘Future directions of stated choice methods for environmental valuation,’ paper presented at The Kobe Conference on the Theory and Application of Environmental Valuation, Kobe University, Rokkoudai Campus, Japan, 22–23 January. Adamowicz, V., J. Louviere and M. Williams (1994), ‘Combining revealed and stated preference methods for valuing environmental amenities,’ Journal of Environmental Economics and Management, 26, 271–92.

52

Environmental and resource economics 2004/2005

Adamowicz, V., J. Swait, P. Boxall, J. Louviere and M. Williams (1997), ‘Perceptions versus objective measures of environmental quality in combined revealed and stated preference models of environmental valuation,’ Journal of Environmental Economics and Management, 32 (January), 65–84. Alberini, A., A. Krupnick, M. Cropper, N. Simon and J. Cook (2001), ‘The willingness to pay for mortality risk reductions: a comparison of the United States and Canada,’working paper, October. Angrist, J.D. and A.B. Krueger (1992), ‘The effect of age at school entry on educational attainment: an application of instrumental variables with moments from two samples,’ Journal of the American Statistical Association, 87 (418), 328–36. Arellano, M. and C. Meghir (1992), ‘Female labour supply and on-the-job search: an empirical model estimated using complementary data sets,’ Review of Economic Studies, 59 (3), 537–59. Arrow, K., R. Solow, E.E. Leamer, R. Radner and H. Schuman (1994), ‘Comment on NOAA proposed rule on natural resource damage assessments,’ ANPNM, comment no. 69, 7 January. Arrow, K., R. Solow, P.R. Portney, E.E. Leamer, R. Radner and H. Schuman (1993), ‘Report of the NOAA Panel on Contingent Valuation,’ Federal Register, 58 (10), 4601–14. Balistreri, E., G. McClelland, G. Poe and W. Schulze (2001), ‘Can hypothetical questions reveal true values? A laboratory comparison of dichotomous choice and open-ended contingent values with auction values,’ Environmental and Resource Economics, 18 (March), 275–92. Bateman, I.J., M. Cole, P. Cooper, S. Georgiou, D. Hadley and G.L. Poe (2004), ‘On visible choice sets and scope sensitivity,’ Journal of Environmental Economics and Management, 47 (January), 71–93. Ben-Akiva, M.E. (1992), ‘Incorporation of psychometric data in individual choice models,’ paper presented to American Marketing Association Advanced Research Techniques Forum, June. Ben-Akiva, M.E. and T. Morikawa (1990), ‘Estimation of switching models from revealed preferences and stated intentions,’ Transportation Research A, 24 (6), 485–95. Beron, K., J. Murdock and M. Thayer (2001), ‘The benefits of visibility improvements: new evidence from the Los Angeles metropolitan area,’ The Journal of Real Estate Finance and Economics, 22 (March/May), 319–38. Berry, S., J. Levinsohn and A. Pakes (1998), ‘Differentiated products demand systems from a combination of micro and macro data: the new car market,’ working paper 6481, National Bureau of Economic Research, March, Cambridge, MA. Bishop, R.C. and T.A. Heberlein (1979), ‘Measuring values of extra-market goods: are indirect measures biased?’, American Journal of Agricultural Economics, 61 (8), 926–30. Bishop, R.C. and T.A. Heberlein (1990), ‘The contingent valuation method,’ in R.L. Johnson and G.V. Johnson (eds), Economic Valuation of Natural Resources, Boulder, CO: Westview Press. Blinder, A.S. (1991), ‘Why are prices sticky?: preliminary results from an interview study,’ American Economic Review, Proceedings, 81 (May), 89–100. Bockstael, N.E. and K.E. McConnell (1993),. ‘Public goods as characteristics of nonmarket commodities,’ Economic Journal, 103 (3), 1244–57. Bockstael, N.E. and K.E. McConnell (1999), ‘The behavioral basis of non-market valuation,’ in J.A. Herriges and C.L. Kling (eds), Valuing Recreation and the Environment, Cheltenham, UK and Northampton, MA: Edward Elgar.

Fifty years of contingent valuation

53

Bockstael, N.E., W.M. Hanemann and I.E. Strand, Jr. (1984), ‘Measuring the benefits of water quality improvements using recreation demand models’, Department of Agricultural and Resource Economics, University of Maryland, report to US Environmental Protection Agency. Braden, J.B. and C.D. Kolstad (eds) (1991), Measuring the Demand for Environmental Quality, Amsterdam: North Holland. Brookshire, D.S., B.C. Ives and W.D. Schulze (1976), ‘The valuation of aesthetic preferences,’ Journal of Environmental Economics and Management, 3 (4), 325–46. Brookshire, D.S., R.C. d’Arge, W.D. Schulze and M.A. Thayer (1981), ‘Experiments in valuing public goods,’ in V.K. Smith (ed.), Advances in Applied Microeconomics, Greenwich, CT: JAI Press. Brookshire, D.S., M.A. Thayer, W.D. Schulze and R.C. d’Arge (1982), ‘Valuing public goods: a comparison of survey and hedonic approaches,’ American Economic Review, 72 (1), 165–77. Brown, G.M. and J. Hammack (1973), ‘Dynamic economic management of migratory waterfowl,’ The Review of Economics and Statistics, 55 (February), 73–82. Brown, T.C., P.A. Champ, R.C. Bishop and D.W. McCollum (1996), ‘Response formats and public good donations,’ Land Economics, 72 (2), 152–66. Bureau of Labor Statistics (1997), BLS Handbook of Methods, bulletin no. 2490, Washington, DC: US Department of Labor. Cameron, T.A. (1988), ‘A new paradigm for valuing non-market goods using referendum data: maximum likelihood estimation by censored logistic regression,’ Journal of Environmental Economics and Management, 15, 355–79. Cameron, T.A. (1992), ‘Combining contingent valuation and travel cost data for the valuation of nonmarket goods,’ Land Economics, 68 (3), 302–17. Cameron, T.A., G.L. Poe, R.G. Ethier and W.D. Schulze (2002), ‘Alternative non-market value-elicitation methods: are the underlying preferences the same?’ Journal of Environmental Economics and Management, 44 (November), 391–425. Carson, R.T. (1997), ‘Contingent valuation surveys and tests of insensitivity to scope,’ in R.J. Kopp, W.W. Pommerehne and N. Schwarz (eds), Determining the Value of NonMarketed Goods: Economic, Psychological, and Policy Relevant Aspects of Contingent Valuation Methods, Boston: Kluwer Nijhoff. Carson, R.T. (2003), Contingent Valuation: A Comprehensive Bibliography and History, Cheltenham, UK and Northampton, MA: Edward Elgar. Carson, R.T. and N.E. Flores (1996), ‘Another look at “Does contingent valuation measure preferences? Experimental evidence”: How evident is the evidence?’ discussion paper 96–31, Department of Economics, University of California, September. Carson, R.T. and R.C. Mitchell (1993), ‘The value of clean water: the public’s willingness to pay for boatable, fishable, and swimmable quality water,’ Water Resources Research 29 (July), 2445–54. Carson, R.T., W.M. Hanemann and R.C. Mitchell (1986), ‘The use of simulated political markets to value public goods,’ unpublished paper, Economics Department, University of California, San Diego. Carson, R.T., W.M. Hanemann and D. Steinberg (1990), ‘A discrete choice contingent valuation estimate of the value of Kenai King Salmon,’ Journal of Behavioral Economics, 19, 53–67. Carson, R.T., R.C. Mitchell, W.M. Hanemann, R.J. Kopp, S. Presser and P.A. Ruud (1992), ‘A contingent valuation study of lost passive use values resulting from the Exxon Valdez oil spill,’ unpublished report to Attorney General of the State of Alaska, La Jolla, CA: National Resource Damage Assessment Inc., 10 November.

54

Environmental and resource economics 2004/2005

Carson, R.T., W.M. Hanemann, R.J. Kopp, J.A. Krosnick, R.C. Mitchell, S. Presser, P.A. Rudd and V.K. Smith (1994), ‘Prospective interim lost use value due to DDT and PCB contamination in Southern California,’ report to National Oceanic and Atmospheric Administration, Natural Resource Damage Assessment Inc., La Jolla, CA, September. Carson, R.T., N.E. Flores, K.M. Martin and J.L. Wright (1996a), ‘Contingent valuation and revealed preference,’ Land Economics, 72 (February), 80–99. Carson, R.T., M.B. Conaway, W.M. Hanemann, J.A. Krosnick, K.M. Martin, D.R. McCubbin, R.C. Mitchell and S. Presser (1996b), The Value of Preventing Oil Spill Injuries to Natural Resources Along California’s Central Coast, Vol. I, Natural Resource Damage Assessment Inc., La Jolla, CA., 31 March 31. Carson, R.T., W.M. Hanemann, R.J. Kopp, J.A. Krosnick, R.C. Mitchell, S. Presser, P.A. Rudd and V.K. Smith (1996c), ‘Was the NOAA panel correct about contingent valuation?’ discussion paper 96-20, Resources for the Future, May. Carson, R.T., W.M. Hanemann, R.J. Kopp, J.A. Krosnick, R.C. Mitchell, S. Presser, P. Rudd and V.K. Smith (1997), ‘Temporal reliability of estimates from contingent valuation,’ Land Economics (May), 151–63. Carson, R.T., W.M. Hanemann, R.J. Kopp, J.A. Krosnick, R.C. Mitchell, S. Presser, P. Rudd and V.K. Smith (1998), ‘Referendum design and contingent valuation: the NOAA Panel’s no-vote recommendation,’ Review of Economics and Statistics (May), 335–7. Carson, R.T., T. Groves and M. Machina (1999), ‘Incentive and informational properties of preference questions,’ unpublished paper, Department of Economics, University of California, San Diego. Carson, R.T., N.E. Flore and N.F. Meade (2001), ‘Contingent valuation: controversies and evidence,’ Environmental and Resource Economics, 19 (June), 173–210. Champ, P.A. and R.C. Bishop (2001), ‘Donation payment mechanisms and contingent valuation: an empirical study of hypothetical bias,’ Environmental and Resource Economics, 19 (August), 383–402. Champ, P.A., N.E. Flores, T.C. Brown and J. Chivers (2002), ‘Contingent valuation and incentives,’ Land Economics, 78 (November), 591–604. Chapman, D.J., W.M. Hanemann and P. Rudd (1998), ‘The American Trader oil spill: a view from the beaches,’ AERE Newsletter, 18 (November), 12–25. Chay, K. and M. Greenstone (2000), ‘Does air quality matter? Evidence from housing markets,’ working paper, University of California, Berkeley. Cicchetti, C.J. and V.K. Smith (1973), ‘Congestion, quality deterioration, and optimal use: wilderness recreation in the Spanish Peaks Primitive Area,’ Social Science Research, 2, 15–30. Ciriacy-Wantrup, S.V. (1947), ‘Capital returns from soil-conservation practices,’ Journal of Farm Economics, 29, 1181–96. Cotterill, R.W. (1994), ‘Scanner data: new opportunities for demand and competitive strategy analysis,’ Agricultural and Resource Economics Review (October), 125–39. Crooker, J. and C.L. Kling (2000), ‘Nonparametric bounds on welfare measures: a new tool for nonmarket valuation,’ Journal of Environmental Economics and Management, 39 (March): 145–61. Cummings, R.G., D.S. Brookshire and W.D. Schulze (eds) (1986), Valuing environmental goods: an assessment of the contingent valuation method, Totowa, NJ: Rowman & Littlefield. Cummings, R.G., G.W. Harrison and E.E. Rutström (1995), ‘Homegrown values and

Fifty years of contingent valuation

55

hypothetical surveys: i,s the dichotomous-choice approach incentive-compatible?’, American Economic Review, 85 (1), 260–66. Davis, R.K. (1963), ‘The value of outdoor recreation: an economic study of the Maine woods,’ Ph.D. dissertation, Harvard University Department of Economics. Desvousges, W.H., V.K. Smith and M. McGivney (1983), ‘A comparison of alternative approaches for estimation of recreation and related benefits of water quality improvements,’ US Environmental Protection Agency report no. EPA-230-05-8301, Washington, DC. Desvousges, W.H., V.K. Smith, D.H. Brown and D.K Pate (1984), ‘Detailed summary: the role of focus groups in designing a contingent valuation survey to measure the benefits of hazardous waste management regulations’, EPA Contract No. 68-016595, June. Desvousges, W.H., F.R. Johnson, R.W. Dunford, K.J. Boyle, S.P. Hudson and K.N. Wilson (1993), ‘Measuring natural resource damages with contingent valuation tests of validity and reliability,’ in J.A. Hausman (ed.), Contingent Valuation: A Critical Assessment, Amsterdam: North Holland. Desvousges, W.H., S.M. Waters and K.E. Train (1996), ‘Supplemental report on potential economic losses associated with recreation services in the Upper Clark Fork River Basin,’ unpublished paper, Triangle Economic Research, Durham, NC, 1 February. Diamond, P.A. (1996), ‘Testing the internal consistency of contingent valuation surveys,’ Journal of Environmental Economics and Management, 30 (3), 337–47. Diamond, P.A. and J.A. Hausman (1994), ‘Contingent valuation: is some number better than no number,’ Journal of Economic Perspectives, 8 (Fall), 45–64. Diamond, P.A., J.A. Hausman, G.K. Leonard and M.A. Denning (1993), ‘Does contingent valuation measure preferences? Experimental evidence,’ in J.A. Hausman (ed.), Contingent Valuation: A Critical Assessment, Amsterdam: North Holland. Duffield, J.W. and D.A. Patterson (1992), ‘Field testing existence values: an instream flow trust fund for Montana rivers,’ working paper presented to W-133 Meetings, 14th Interim Report, R.B. Rettig (ed.), Department of Agricultural and Resource Economics, Oregon State University. Dunford, R.W. (1999), ‘The American Trader oil spill: an alternative view of recreation use damages,’ AERE Newsletter, 19 (May), 12–20. Ebert, U. (1998), ‘Evaluation of nonmarket goods: recovering unconditional preferences,’ American Journal of Agricultural Economics, 80 (May): 241–54. Englin, J. and T.A. Cameron (1996), ‘Augmenting travel cost models with contingent behavior data,’ Environmental and Resource Economics, 7 (2), 133–47. Eom, Y.S. and V.K. Smith (1994), ‘Calibrated non-market valuation,’ unpublished paper, Center for Environmental and Resource Economics, Duke University, August. Freeman, A.M. III (1993), The Measurement of Environmental and Resource Values: Theory and Methods, Washington, DC: Resources for the Future. Haab, T.C. and K.E. McConnell (1997), ‘Referendum models and negative willingness to pay: alternative solutions,’ Journal of Environmental Economics and Management, 32 (February), 251–70. Hammack, J. and G.M. Brown, Jr (1974), Waterfowl and Wetlands: Toward Bioeconomic Analysis, Baltimore: The Johns Hopkins University Press, for Resources for the Future. Hanemann, W.M. (1984), ‘Welfare evaluations in contingent valuation experiments with discrete responses,’ American Journal of Agricultural Economics, 66 (3), 332–41.

56

Environmental and resource economics 2004/2005

Hanemann, W.M. (1988), ‘Three approaches to defining “existence” or nonuse values under certainty,’ working paper Department of Agricultural and Resource Economics, University of California, Berkeley, July. Hanemann, W.M. (1991), ‘Willingness to pay versus willingness to accept: how much can they differ?’, American Economic Review, 81 (3), 635–47. Hanemann, W.M. (1992), ‘Preface,’ in Ståle Navrud (ed.), Pricing the European Environment, Oslo: Scandinavian University Press. Hanemann, W.M. (1994), ‘Valuing the environment through contingent valuation,’ Journal of Economic Perspectives, 8 (4), 19–44. Hanemann, W.M. (1995), ‘Review of Triangle Economic Research report on economic loss to recreational fishing in the Upper Clark Fork Basin,’ testimony in the Upper Clark Fork River case, 27 October. Hanemann, W.M. (1996), ‘Theory versus data in the contingent valuation debate,’ in D.J. Bjornstad and J.R. Kahn (eds), The Contingent Valuation of Environmental Resources, Cheltenham, UK and Brookfield, VT: Edward Elgar. Hanemann, W.M. (1999), ‘The economic theory of WTP and WTA,’ in I. Bateman and K. Willis (eds), Valuing Environment Preferences: Theory and Practice of the Contingent Valuation Method in the US, EC and Developing Countries, Oxford: Oxford University Press. Hanemann, W.M. and B. Kanninen (1999), ‘The statistical analysis of discreteresponse CV data,’ in I. Batemen and K. Willis (eds), Valuing Environmental Preferences: Theory and Practice of the Contingent Valuation Method in the US, EC and Developing Countries, Oxford: Oxford University Press. Harberger, A.C. (1971), ‘Three basic postulates for applied welfare economics,’ Journal of Economic Literature, 9 (September), 785–97. Harrison, G.W. (1989), ‘Theory and misbehavior of first price auctions,’ American Economic Review, 79 (September), 749–62. Harrison, G.W. (2000), ‘Contingent valuation meets the experts: a critique of the NOAA Panel report,’ working paper, Department of Economics, University of South Carolina, Columbia, SC, October. Hausman, J.A. (1981), ‘Exact consumer’s surplus and deadweight loss,’ American Economic Review, 71 (4): 662–76. Hausman, J.A. (ed.) (1993), Contingent Valuation: A Critical Assessment, Amsterdam: North Holland. Heckman, J.J. (2001), ‘Micro data, heterogeneity, and the evaluation of public policy: Nobel Lecture’, Journal of Political Economy, 109 (4), 673–748. Herriges, J.A., C.L. Kling and D.J. Phaneuf (2004), ‘What’s the use? Welfare estimates from revealed preference models when weak complementarity does not hold,’ Journal of Environmental Economics and Management, 47 (January), 55–70. Hines, J.R., Jr (1999), ‘Three sides of Harberger Triangles,’ Journal of Economic Perspectives, 13 (Spring),: 167–88. Houthakker, H.S. (1950), ‘Revealed preference and the utility function,’ Economica 17, 159–74. Huang, J.C. and V.K. Smith (1998), ‘Monte Carlo benchmarks for discrete response valuation methods,’ Land Economics, 74 (May), 186–202. Huang, J.C. and V.K. Smith (2002), ‘Monte Carlo benchmarks for discrete response valuation methods: reply’, Land Economics, 78 (November), 617–23. Imbens, G.W. and T. Lancaster (1994), ‘Combining micro and macro data in microeconometric models,’ Review of Economic Studies, 61, 655–80. Johnson, F.R. and W.H. Desvousges (1997), ‘Estimating stated preferences with rated-

Fifty years of contingent valuation

57

pair data: environmental, health and employment effects of energy programs,’ Journal of Environmental Economics and Management, 34 (September), 79–99. Johnson, F.R., K.E. Matthews and M.F. Bingham (2000), ‘Evaluating welfare-theoretic consistency in multiple-response, stated-preference surveys,’ paper presented at the Kobe Conference on Theory and Application of Environmental Valuation, Kobe University, Rokkoudai Campus, Japan, 22–23 January. Kahn, M. (1997), ‘Are the social benefits for combating Los Angeles smog rising or falling?’ working paper, Columbia University. Kahneman, D. and I. Ritov (1994), ‘Determinants of stated willingness to pay for public goods: a study in the headline method,’ Journal of Risk and Uncertainty, 9 (July): 5–38. Kain, J.F. and J.M. Quigley (1972), ‘Note on owner’s estimate of housing value,’ Journal of the American Statistical Association, 67 (December), 803–6. Kanninen, B.J. (2002), ‘Optimal design for multinomial choice experiments,’ Journal of Marketing Research, 39 (May): 214–27. Kopp, R.J. and V.K. Smith (1997), ‘Constructing measures of economic value,’ in R.J. Kopp, W. Pommerhne and N. Schwarz (eds), Determining the Value of NonMarketed Goods: Economic, Psychological and Policy Relevant Aspects of Contingent Valuation Methods, Boston: Kluwer Nijhoff. Krosnick, J.A., A.L. Holbrook, M.K. Berent, R.T. Carson, W.M. Hanemann, R.J. Kopp, R.C. Mitchell, S. Presser, P.A. Ruud, V.K. Smith, W.R. Moody, M.C. Green and M. Conaway (2002), ‘The impact of “no opinion” response options on data quality: non-attitude reduction or an invitation to satisfice?’, Public Opinion Quarterly, 66 (Fall), 371–403. Krupnick, A. Alberini, M. Cropper, N. Simon, B. O’Brien, R. Goeree, and M. Heintzelman (2002), ‘Age, health and the willingness to pay for mortality risk reductions: a contingent valuation survey of Ontario residents,’ The Journal of Risk and Uncertainty, 24 (2), 161–86. LaFrance, J.T. and W.M. Hanemann (1989), ‘The dual structure of incomplete demand systems,’ American Journal of Agricultural Economics, 71 (May), 262–74. Larson, D.M. (1993), ‘On measuring existence value,’ Land Economics, 69 (November), 377–89. Larson, D.M., J.B. Loomis and Y.L. Chien (1993), ‘Combining behavioral and conversational approaches to value amenities: an application to gray whale population enhancement,’ paper presented to American Agricultural Economics Association, August. Layton, D.F. and S.T. Lee (2002), ‘From ratings to rankings: the econometric analysis of stated preference ratings data,’ unpublished paper, University of California, Davis. Leeworthy, V.R., D. Schruefer and P.C. Wiley (1990), A Socioeconomic Profile of Recreationists at Public Outdoor Recreation Sites in Coastal Areas, Vol. 5, Washington, DC: National Oceanic and Atmospheric Administration. Leeworthy, V.R. and P.C. Wiley (1993), ‘Recreational use value for three Southern California beaches’, unpublished paper, Strategic Environmental Assessments Division, National Oceanic and Atmospheric Administration, Rockville, MD, March. Louviere, J.J. (1996) ‘Relating stated preference measures and models to choices in real markets: calibration of CV responses,’ in D.J. Bjornstad and J.R. Kahn (eds), The Contingent Valuation of Environmental Resources, Cheltenham, UK and Brookfield, VT: Edward Elgar.

58

Environmental and resource economics 2004/2005

Louviere, J.J., R.J. Meyer, D. Bunch, R.T. Carson, B. Delleart, W.M. Hanemann, D. Hensher and J. Irwin (1999), ‘Combining sources of preference data for modeling complex decision processes,’ Marketing Letters, 10, 205-17. Mack, R.P. and S. Myers (1965), ‘Outdoor recreation,’ in R. Dorfman (ed.), Measuring Benefits of Government Investments, Washington, DC: The Brookings Institution. MacNair, D. and J. Lutz (1998), ‘Combining revealed and stated preferences to evaluate in-kind transfers for natural resource damages: an application to saltwater fishing,’ presented at the Center for Environmental and Resource Economics Camp Resources Workshop, Wilmington, NC, 13 August. Mäler, K.G. (1974), Environmental Economics: A Theoretical Inquiry, Baltimore: The Johns Hopkins Press, for Resources for the Future. Mansfield, C. and V.K. Smith (2002), ‘Tradeoff at the trough: TMDLs and the evolving status of U.S. water quality policy,’ in J.A. List and A. de Zeeuw (eds), Recent Advances in Environmental Economics, Cheltenhen, UK and Northampton, MA: Edward Elgar. McConnell, K.E. (1990), ‘Models for referendum data: the structure of discrete choice models for contingent valuation,’ Journal of Environmental Economics and Management, 18 (January), 19–34. McConnell, K.E. and V.A. Duff (1976), ‘Estimating net benefits of recreation under conditions of excess demand,’ Journal of Environmental Economics and Management, 2 (February), 224–30. McFadden, D. (1994), ‘Contingent valuation and social choice,’ American Journal of Agricultural Economics, 76 (4), 689–708. Mitchell, R.C. and R.T. Carson (1984), A Contingent Valuation Estimate of National Freshwater Benefits: Technical Report to the U.S. Environmental Protection Agency, Washington, DC: Resources for the Future. Mitchell, R.C. and R.T. Carson (1989), Using Surveys to Value Public Goods: The Contingent Valuation Method, Washington, DC.: Resources for the Future. Morikawa, T. (1989), ‘Incorporating stated preference data in travel demand analysis’, unpublished Ph.D. thesis, Massachusetts Institute of Technology, Department of Civil Engineering. Morikawa, T., M. Ben-Akiva and D. McFadden (1990), ‘Incorporating psychometric data in econometric travel demand models,’ presented at Banff Symposium on Consumer Decision Making and Choice Behavior, May. National Oceanic and Atmospheric Administration (1996), ‘Natural resource damage assessments final rule for Oil Pollution Act of 1990,’ Federal Register, 61 (5 January), 440–510. Office of Management and Budget (2003), ‘Draft 2003 report to Congress on the costs and benefits of federal regulations: notice,’ Federal Register, 68 (February), 5492–527. Poe, G.L. and C.A. Vossler (2002), ‘Monte Carlo benchmarks for discrete response valuation methods: a comment,’ Land Economics, 78 (November), 605–16. Poe, G.L., J.E. Clark, D. Rondeau and W.D. Schulze (2002), ‘Provision point mechanisms and field validity tests of contingent valuation,’ Environmental and Resource Economics, 23 (September), 105–31. Portney, P.R. (1994), ‘The contingent valuation debate: why economists should care,’ Journal of Economic Perspectives, 8 (Fall): 3–18. Randall, A., B. Ives and C. Eastman (1974), ‘Bidding games for valuation of aesthetic environmental improvements,’ Journal of Environmental Economics and Management, 1 (August), 132–49.

Fifty years of contingent valuation

59

Reaves, D.W., R.A. Kramer and T.P. Holmes (1999), ‘Does question format matter? Valuing an endangered species,’ Environmental and Resource Economics, 14 October, 365–83. Roe, B., K.J. Boyle and M.F. Teisl (1997), ‘Using conjoint analysis to derive estimates of compensating variation,’ Journal of Environmental Economics and Management, 31 (September), 145–59. Rowe, R.D., R.C. D’Arge and D.S. Brookshire (1980), ‘An experiment on the economic value of visibility,’ Journal of Environmental Economics and Management, 7, 1–19. Samuelson, P.A. (1938), ‘A note on the pure theory of consumer’s behavior,’ Economica, 5 (February), 61–71. Samuelson, P.A. (1950), ‘The problem of integrability in utility theory,’ Economica, 17, 355–85. Schläpfer, F. (2002), ‘A note on the issue of incentive compatibility in referendumformat CVM,’ unpublished paper, Institut für Umweltwissenschaften. Schulze, W.D., D.S. Brookshire, E.G. Walther, K. Kelly, M.A. Thayer, R.L. Whitworth, S. Ben-David, W. Malm and J. Molenar (1981), ‘The benefits of preserving visibility in the national parklands of the Southwest,’ Methods Development for Environmental Control Benefits Assessment, vol. VIII, final report to the USEPA Grant R805059010, Office of Exploratory Research, Office of Research and Development. Schulze, W.D., G. McClelland, D. Waldman and J. Lazo (1996), ‘Sources of bias in contingent valuation,’ in D.J. Bjornstad and J.R. Kahn (eds), The Contingent Valuation of Environmental Resources: Methodological Issues and Research Needs, Cheltenham, UK and Brookfield, VT: Edward Elgar. Sieg, H., V.K. Smith, H.S. Banzhaf and R. Walsh (2002), ‘Interjurisdictional housing prices in locational equilibrium,’ Journal of Urban Economics, 52 (July), 131–53. Simon, H.A. (1986), ‘Rationality in psychology and economics,’ in R. Hogarth and M.W. Reder (eds), Rational Choice: The Contrast Between Economics and Psychology, Chicago: University of Chicago Press. Sloan, F.A., V.K Smith and D.H. Taylor, Jr (2003), The Smoking Puzzle: Information, Risk Perception, and Choice, Cambridge, MA: Harvard University Press. Smith, V.K. (1986), ‘To keep or toss the contingent valuation method,’ in R.G. Cummings, D.S. Brookshire and W.D. Schulze (eds), Valuing Environmental Goods: An Assessment of the Contingent Valuation Method, Totowa, NJ: Rowman & Littlefield. Smith, V.K. (1996), ‘Can contingent valuation distinguish economic values for different public goods?’ Land Economics (May), 139–51. Smith, V.K. (1997), ‘Pricing what is priceless: a status report on non-market valuation of environmental resources,’ in H. Folmer and T. Tietenberg (eds), The International Yearbook of Environmental and Resource Economics, Cheltenham, UK and Brookfield, VT: Edward Elgar. Smith, V.K. (1999), ‘Resource compensation and the restoration of injured natural resources: context and future research,’ unpublished paper, Center for Environmental and Resource Economics, Duke University, January. Smith, V.K. and W.H. Desvousges (1986), Measuring Water Quality Benefits, Boston: Kluwer Nijhoff. Smith, V.K. and L. Osborne (1996), ‘Do contingent valuation estimates pass a “Scope” test? A meta analysis,’ Journal of Environmental Economics and Management, 31 (November), 287–301.

60

Environmental and resource economics 2004/2005

Smith, V.K., X. Zhang and R.B. Palmquist (1997), ‘Marine debris, beach quality, and non-market values,’ Environmental and Resource Economics (October), 223–47. Smith, V.K. and C. Mansfield (1998), ‘Buying time: real and contingent offers,’ Journal of Environmental Economics and Management (November), 209–24. Smith, V.K., G. Van Houtven and S.K. Pattanayak (2002), ‘Benefit transfer via preference calibration: “prudential algebra” for policy,’ Land Economics, 78 (February), 132–52. Smith, V.K. and S.K. Pattanayak (2002), ‘Is meta analysis the Noah’s ark for nonmarket valuation?’, Environmental and Resource Economics, 22 (June), 271–96. Swait, J. and J. Louviere (1993), ‘The role of the scale parameter in the estimation and comparison of multinomial logit models,’ Journal of Marketing Research, 30 (August), 305–14. Swait, J., J. Louviere and M. Williams (1994), ‘A sequential approach to exploiting the combined strength of SP and RP data: application to freight shipper choice,’ Transportation, 21, 135–52. Takeuchi, K. and K. Ueta (1996), ‘Another scope test on nonuse value of the Shimanto River, Japan,’ working paper, Kyoto University, Japan, April. Varian, H.R. (1982), ‘The nonparametric approach to demand analysis,’ Econometrica, 50, 945–73. Varian, H.R. (1983), ‘Nonparametric tests of consumer behavior,’ Review of Economic Studies, 50, 99–110. Von Haefen, R.H. (2002), ‘A complete characterization of the linear, log-linear, and semi-log incomplete demand system models,’ Journal of Agricultural and Resource Economics, 27 (December), 281–319. Vossler, C.A. and J. Kerkvliet (2003), ‘A criterion validity test of the contingent valuation method: comparing hypothetical and actual voting behavior for a public referendum,’ Journal of Environmental Economics and Management, 45 (May), 631–49. Waugh, F.V. (1929), Quality as a Determinant of Vegetable Prices: Studies in History, Economics and Public Law, no 312, 1st AMS edn (from the Columbia University 1929 edition), ed. Faculty of Political Science, Columbia University, New York: AMS Press, 1968. Westat, Inc. (1989), ‘Investigation of possible recall/reference period bias in national surveys of fishing, hunting, and wildlife associated recreation,’ final report to US Department of Interior, Contract No. 14-16-009-87-008, December. Whitehead, J.C., G.C. Blomquist, R.C. Ready and J.-C. Huang (1998), ‘Construct validity of dichotomous and polychotomous choice contingent valuation questions,’ Environmental and Resource Economics, 11 (January), 107–16. Willig, R.D. (1978), ‘Incremental consumer’s surplus and hedonic price adjustment,’ Journal of Economic Theory, 17 (2), 227–53. Zhang, X. and V.K. Smith (1997), ‘An integrated model of use and nonuse values,’ unpublished paper, Center for Environmental and Resource Economics, Duke University, July.

2. Environmental policy, induced technological change and economic growth: a selective review Wolfgang K. Heidug and Regina Bertram 1.

INTRODUCTION

Technological change has long appeared to play a backstage role in economic thinking. Its impact was typically described in terms of variables that change exogenously with the progress of time. It is only with the advent of the new growth theory (reviewed in the monographs by Barro and Sala-i-Martin, 1995, and Aghion and Howitt, 1999), in which technological change is endogenously determined, that issues of technological change have become a focus of economic research. Specifically, the important role of environmental policy for inducing technological progress through creating constraints and incentives has been increasingly recognized. The discussion concerning the optimal time path of carbon taxes in the presence of induced technical change illustrates this. Wigley et al. (1996) argue for a policy that makes postponement of abatement attractive in order to optimally exploit the reduction of abatement cost resulting from technological progress. However, work by Grubb et al. (1996) indicates that a policy that favors more abatement in the short term is superior to a ‘wait-and-see’ approach when technological progress advances through learning-by-doing. This chapter reviews the relation between environmental policy, the technological change that it induces and the resulting consequences for economic growth. Through its inclusion of growth aspects it complements an earlier review by Jaffe et al. (2002) and the reviews by Clarke and Weyant (2002) and Grubb et al. (2002), which focus on climate change and energy policy. The review is selective in that it does not attempt to mirror the burgeoning literature on the subject but rather aims to highlight different routes of analysis and to portray in broad strokes the current state of discussion. Specifically, the review aims to give state-of-the-art answers to the following questions: does the presence of induced technical change lower the cost of achieving emission abatement targets? How does induced innovation affect the timing of optimal 61

62

Environmental and resource economics 2004/2005

abatement? To what extent might exogenous technology models understate the welfare gains from environmental policies? And what are the implications of endogenous technological change for environmental policy? To bring some structure into the diversity of models that analyze the relationship between environmental policy and technological change, it is convenient to organize them according to a scheme proposed by Clarke and Weyant (2002). The simplest types are cost-function models, in which the technological advances induced by environmental policy translate only into changes of the abatement cost function. One step up on the ladder of complexity are intertemporal partial equilibrium models. These models study the equilibrium consequences of policy interventions in the market for environmentally focused innovation over time by ignoring effects on other markets. The boundary between both groups is not sharp and to a large extent marked by the model’s emphasis rather than methodology. In fact, the first two models that we consider in this review can be seen as belonging to either category. The model by Goulder and Mathai (2000), which we review in section 3, focuses mainly on the determination of an optimal emission path for an economy in response to some policy criterion. It challenges the claim that the presence of induced technological change calls for a more cautious environmental policy. The welfare aspects that are attributable to induced technology innovation, and in particular their magnitude in relation to the Pigouvian welfare gains of an optimal emission tax are examined in section 4 using the model by Parry et al. (2002). This model shows that for a broad range of circumstances the Pigouvian welfare gains dominate those from technological innovation. More difficult both in conception and in the use of mathematical tools are growth models in which technological change is endogenized. We review in section 5 a series of rather general models of this type that explicitly include environment–economy interactions (Bovenberg and Smulders, 1995, 1996; Smulders 1995, 1998). The analytical tractability of these models makes them a good vehicle for gaining a conceptual understanding of the implication of technological change. A brief discussion of intertemporal general equilibrium models in section 6 follows. These models are typically computer-based, involve explicit representations of markets and their interactions over time, and as they shed light on the role market inefficiencies play for environmental technological innovation, they add a dimension to the analysis that is absent in the other types of models. As representatives for this model type we select the models by Goulder and Schneider (1999) and by Popp (2002), whose main field of application is the analysis of greenhouse gas abatement policies.1 We conclude in section 7 with a brief discussion of the implications of the models for environmental policy. To lay the ground for our discussion we continue in the next section with a brief outline of some of the salient features of technological progress.

A selective review

2.

63

KEY FEATURES OF TECHNOLOGICAL CHANGE

Technological progress as understood here manifests itself through its impact on the macroeconomic production function thereby changing the input–output combinations and society’s production possibilities. Induced technology progress occurs in response to policy intervention and, as such, is different from exogenous progress that is just a result of the passage of time. There is a multitude of channels through which public policy can affect technological advance. An incomplete list includes changes in the relative prices of polluting and non-polluting goods, which could result, for example, from the imposition of an emission tax; research grants to private firms; or subsidized research and development at national laboratories and universities. Technology, a form of knowledge, has properties of a public good. That is, technology knowledge is • nonrival, that is, the use of technology by one agent does not preclude others from using the same technology. Conventional private goods, in contrast, are rival. A piece of capital equipment, for example, can only be used in one place at a time. An immediate implication is that the production and allocation of technological knowledge cannot be completely governed by competitive market forces. Once technological knowledge has been created, the marginal cost of supplying it to an additional user is almost zero. It follows that a competitive market, in which private gains stem from marginal cost pricing, does not provide the economic incentives for the creation of knowledge, because the innovator will typically fail to appropriate most of the returns generated by the new knowledge. Some departure from the competitive model is therefore needed to explain why firms embark on knowledge creation; • partially nonexcludable, meaning that the creators of technological knowledge often have difficulties in preventing others from using it. This attribute of knowledge distinguishes it, to stay with the above example, from a piece of capital equipment, which is readily excludable. The degree of ‘excludability’ depends both on the nature of the knowledge itself – encoded TV satellite transmissions are highly excludable, while computer software is less excludable – and on the economic institutions governing property rights, and is therefore partly a function of policy choices. There is a variety of forces that govern the accumulation of technological knowledge, including (Romer, 1996): • Public funding of basic research. Society may deem it worthwhile to

64

Environmental and resource economics 2004/2005

incur costs for funding of basic research when no private firm would undertake it. Obvious examples are research on public health and the environment. Since the results of this research are given away for free and their cost is not reflected in the price that private firms using the results charge for their goods, this mode of knowledge accumulation has a positive externality. The basic scientific knowledge that it creates very often provides the technology paradigm for subsequent innovation by private firms. • Industrial R&D. Industrial R&D can be a channel to accumulate knowledge for private gains. The provision is that the creators of knowledge are able to appropriate (for instance, through patents, trade secrets, leadtime effect, and so on) at least some of the benefits from their inventive efforts. The part of knowledge that is not appropriated and that cannot be prevented from entering the public domain represents the technological spillover associated with private R&D. The accepted view is that knowledge spillovers, because of their generic nature, are extremely difficult to prevent (Arrow, 1962a). They will add to a pool of public knowledge serving as an input to future research. The accumulation of public knowledge will thus lower the cost of future generations to achieve some technological breakthrough. Such cost reduction suggests a mechanism whereby private investment incentives can be preserved despite the tendency for the private returns from innovation to fall as result of increases in the number of competing technologies. Knowledge spillovers can thus be a source of long-term economic growth (Grossman and Helpman, 1997). • Learning-by-doing. Technology can advance as a consequence of the production and use of technology (Arrow, 1962b). Technological progress is then ‘free’ in the sense that it does not require investment in R&D. Firms learn better ways to produce as an accidental by-product of the production process. For instance, the more photovoltaic cells a firm produces, the better it becomes at it. This type of knowledge accumulation is referred to as learning-by-doing. Common to the last two types of knowledge accumulation is that they regard, in one way or another, profit-optimizing decisions by firms as the major source driving technological advance.2 The difficulty profit-optimizing firms have in appropriating rents from knowledge they create leads to inefficient, monopolistic behavior and probably to an underinvestment in innovation. On the other hand, competition for patents can also generate an excessive amount of R&D, if firms do not take into account that their own research activities reduce the likelihood for others to win patents (Wright, 1983). These remarks show that in the presence of induced technological change

A selective review

65

environmental policy faces a twofold challenge. It needs to address the market failure resulting from environmental externalities, and it has to respond to the public-good character of environmentally focused innovations.

3.

COST-FUNCTION ANALYSIS

Lots of mileage for analyzing questions about the effect of environmental policy on induced technological change can be gained by concentrating exclusively on abatement costs. Such an approach was used by Goulder and Mathai (2000), who analyzed not only the optimal timing of CO2 abatement but also the optimal time path of an emission tax under both a cost-effectiveness and a benefit–cost criterion. In addition, they considered different specifications for the accumulation of knowledge, namely investment in R&D and learning-bydoing.3 We start our review by focusing on a situation in which R&D investment drives technology change, and the policymaker adopts a costeffectiveness criterion. Let C denote abatement costs, It the investment in knowledge, q the real price of investment, Ht the stock of ‘knowledge’ (or ‘technology’), and q the social discount rate. Then the mathematical problem for a social planner willing to implement a cost-effective environmental policy is to choose time paths At for abatement and It for investment that minimize the discounted sum of abatement costs and R&D expenditure into the infinite future, that is, ∞

min At,It

∫

(C(At, Ht) + q(It)It)e–qt dt,

(2.1)

0

provided that atmospheric CO2 concentrations St meet a target concentration S¯ at time T and remain after that always below the target (St ≤ S¯, ∀t ≥ T). Goulder and Mathai take abatement costs C to depend both on the amount of abatement At and the knowledge stock Ht. It appears natural to assume CA(·) > 0, CAA(·) > 0, CH(·) < 0, and CAH(·) < 0, expressing the idea that abatement cost and marginal abatement cost will increase with abatement, and that knowledge Ht decreases both cost and marginal cost of abatement.4 To keep the model analytically tractable Goulder and Mathai assume that abatement affects atmospheric CO2 concentration by the simple relation S˙t = –dS + E0t – At,

(2.2)

where d is the natural rate of removal of atmospheric CO2, and E0t denotes baseline CO2 emissions. Equation (2.2) states that changes in the atmospheric

66

Environmental and resource economics 2004/2005

CO2 concentration, S˙t, are equal to current net emissions E0t –At diminished by the amount of CO2 that is removed naturally from the atmosphere. In light of the discussion above, knowledge accumulation can be accomplished through various channels. Goulder and Mathai assume that technological change accumulates both exogenously and endogenously, and in the latter case via investment in abatement technology, that is, H˙t = atHt + kY(It, Ht),

(2.3)

where at denotes the rate of exogenous technological change and k is a constant parameter (k > 0) that indicates the presence of induced technological change. The knowledge accumulation function Y is endowed with the properties Y > 0, YI > 0, and YII < 0 to capture the idea that knowledge increases with investment but at a decreasing rate. Requirement (2.1) represents a dynamic optimization problem for the control variables At and It; St and Ht, whose time evolution is described by (2.2) and (2.3), play the role of state variables. The problem can be solved by employing the maximum principle (for example Chiang, 1992) using as the current-value Hamiltonian Ht = –(C(At, Ht) + q(It)It + pt(–dSt + E0t – At) + mt(aHt + kY(It, Ht)), (2.4) for times t < T, and the Lagrangian Lt = Ht + ht(S¯ – St),

(2.5)

for t ≥ T. The symbols pt and mt denote co-state variables, and ht is a Lagrange parameter. Restricting attention to t < T, the equations most interesting for the present discussion are the ones that are implied by the requirements ∂Ht/∂At = 0 and p˙t = ptd + qpt of the maximum principle, namely CA(·) = –pt, p˙t = (q + d)Pt,

(2.6) (2.7)

where –pt can be interpreted as the shadow cost of CO2 emissions, which in a decentralized, competitive industry equals the optimal CO2 tax. With this interpretation, equation (2.6) states that at all times the optimal level of abatement is achieved, when marginal abatement cost equals CO2 tax. The co-state equation (2.7) implies that the optimal CO2 tax increases over time exponentially with rate (q + d). Note that the effective discount rate is larger than q to account for the natural removal of CO2 along the optimized abatement path.

A selective review

67

The parameter k in equation (2.3) provides the means to analyze the consequences of induced technological change for the scenario considered here. Obviously, the larger k, the larger the increase in knowledge stock that results from investment It at fixed knowledge stock Ht. To quantify its impact on the optimal abatement profile Goulder and Mathai differentiate equation (2.6) with respect to k totally5 to obtain after rearrangement

d(–pt)

dHt

––––– – CAH(·) –––

dAt dk dk –––– = —————————————. dk CAA(·)

(2.8)

If one neglects for the moment the effect of technological change on the optimal carbon tax itself, that is, assumes that d(–pt)/dk = 0, then equation (2.8) shows that technological change stimulates abatement, dAt/dk > 0, provided that –CAH > 0 as assumed earlier. Goulder and Mathai refer to this as the ‘knowledge growth effect’. Figure 2.1 depicts the situation schematically. Assuming for simplicity a linear abatement schedule, A0, induced technological change leads to pivoting of the abatement curve upward. The new path coincides with the original optimal path at time t = 0 at which the technology stock has the same value in both scenarios. Clearly, the new path cannot be optimal as too much abatement is undertaken, and consequently atmospheric CO2 concentration at time T is less than the target concentration S¯ . It is the effect of k on the shadow price pt, which we have neglected so far, that corrects the situation. To prevent an overshooting of the total amount of abatement one would expect the sign of d(–pt)/dk, the first term in the numerator of equation (2.8), to be opposite of the second term, and the more detailed mathematical analysis of Goulder and Mathai does indeed confirm that the presence of induced technological change lowers carbon taxes.6 The intuition is that it is easier to achieve CO2 emission reductions with advance technology than without it. This will lower the shadow cost of CO2 emissions and, hence, the tax. As indicated in Figure 2.1, the additional effect of k on pt yields an optimal abatement path, A*, along which abatement levels are initially below those realized without technological change but the increase is steeper with time. Put succinctly, the prospect of technological change justifies delaying abatement until it is less costly. Next Goulder and Mathai consider the situation that the stock of abatement knowledge H does not grow by investment in abatement technology, but is rather accumulated via learning-by-doing. The optimization problem is then

68

Environmental and resource economics 2004/2005

A*

Abatement

Effect of carbon cost (2) Optimal path – induced technological change A0

Effect of knowledge growth (1)

Optimal path – no induced technological change

Time Source:

T

Goulder and Mathai, 2000.

Note: The ‘knowledge growth’ effect leads to pivoting of the abatement path upward (1). This is counteracted by the effect of technology on the shadow price of CO2 (2).

Figure 2.1 Effects of knowledge growth and shadow cost on optimum abatement schedule ∞ min At

∫

C(At, Ht) e–qt dt,

(2.9)

0

where now H˙t = atHt + kY(At, Ht),

(2.10)

and all other conditions remain unchanged. The problem (2.9) can again be solved with the usual mathematical tools, and we shall discuss here only the most pertinent results. First, as the co-state equation (2.7) remains unchanged, the optimal carbon tax grows, as before, at rate (q + d) for t ≤ T. However, instead of equation (2.6) we now have

A selective review

CA(·) – mtkYA(·) = –pt,

69

(2.11)

indicating that the marginal social cost of abatement is no longer CA but rather CA – mtkYA because of the cost reduction associated with the learning-bydoing stemming from that abatement. The impact of induced technological change is described by d(–pt) dHt –––––– + mtYA(·) – CAH(·) –––– dAt dk dk –––– = –––––––––––––––––––––––––––, dk CAA(·)

(2.12)

which is obtained by differentiating (2.11) totally with respect to k. On the right-hand side of this equation mtYA > 0 represents a positive learning-bydoing effect. This effect strengthens the knowledge growth effect –CAH(dHt/dk) > 0 on abatement but is opposed to the shadow price effect for which d(–pt)/dk ≤ 0. Thus the net effect of all three contributions is ambiguous, and it is certainly not clear that learning-by-doing always justifies lower initial abatement as was the case for R&D-induced technological change. The intuition is here that future abatement costs determine how much abatement should be moved to the future. However, as future abatement costs are determined by the near-term abatement effort and are the lower the more abatement is undertaken near term, it is clear that no general statement concerning the level of initial abatement can be made. In their benefit–cost models, Goulder and Mathai obtained similar results. Here, the objective of the social planner is to minimize both the costs of achieving the desired abatement target and the damages resulting from CO2 emissions over an infinite time horizon.7 As before, the impact of induced technological change on abatement is given by (2.8) and (2.12) respectively. Again, we are confronted with a positive knowledge-growth effect and a negative shadow-cost effect: as long as induced technological change reduces marginal abatement costs, abatement tends to rise. This is counteracted by the negative effect on the shadow costs of today’s emissions, which justifies postponing some abatement until marginal abatement becomes cheaper. Thus, under the benefit–cost criterion the presence of technological change induced by R&D implies that optimal abatement levels fall in early years, but rise later. When knowledge is accumulated via learning-by-doing, the overall effect of induced technological change on optimal abatement is again analytically ambiguous. The total amount of optimal abatement is higher in both settings than it would be without induced innovation. Since new technologies reduce the marginal abatement costs over time, a higher amount of abatement becomes

70

Environmental and resource economics 2004/2005

Tax MC

MC' p p' MB

0

A

A'

Abatement

Note: Innovation leads to a shift of the marginal abatement cost curve from MC to MC′. As a consequence the total abatement level increases from A to A´ while the optimal tax rate decreases from p to p′, if marginal benefits MB decrease with abatement. Source:

Goulder and Mathai, 2000.

Figure 2.2

Optimal carbon tax in the static benefit–cost case

optimal. Perhaps somewhat contrary to intuition, this increase in cumulative abatement is accompanied by a lower tax rate, provided the marginal benefits (MB) decrease with abatement level, as will typically be the case. In this situation, depicted in Figure 2.2, technological progress shifts the marginal abatement cost curve downwards (MC´), which increases the optimal amount of abatement (A´) and reduces the optimal environmental tax rate (p´). If, on the other hand, the marginal benefit curve slopes upward, as would be the case if damages were concave in the CO2 concentration, induced technological change will increase the optimal tax rate, and cumulative abatement will be lower than without induced innovation. The analytical results obtained by Goulder and Mathai are summarized in Table 2.1. A key finding is that under typical conditions the presence of induced technological change implies a lowering of the optimal tax path. A lower tax is all that is needed to achieve the abatement goal, even when the desired total extent of abatement is higher, as is the case when the government

A selective review

71

Table 2.1 Summary of results of the Goulder and Mathai (2000) model Channel

Optimal solutions Tax path

Abatement path

Induced technological change Tax path

Abatement path

Cost-effectiveness criterion: R&D exp. growth slopes upward LBD exp. growth ambiguous

falls falls

A0 falls; path is steeper ambiguous

Cost–benefit criterion: R&D slopes upward* slopes upward* LBD slopes upward* ambiguous

falls falls

A0 falls; path is steeper ambiguous

*Provided that damages are convex and optimized CO2 concentration rises over time. Note: The first column refers to the optimal tax and abatement paths in the presence of induced technological change. The second column summarizes the effects of induced innovation on the slope of the optimal paths.

employs a benefit–cost policy. Another key finding relates to the optimal timing of abatement. If abatement knowledge is generated through R&D, induced technological change makes it optimal to shift some abatement efforts from the present to the future. In contrast, no general statement pertains to the timing of abatement when learning-by-doing is the channel for knowledge accumulation.

4.

WELFARE ASPECTS

This section discusses the welfare gains resulting from environmental policy. Specifically, we will have a closer look at the claim that the welfare benefits from induced technological innovation outweigh the Pigouvian welfare gains resulting from correcting social excessive pollution from firms (Orr, 1976; Parry, 1998). This would seem to imply that the focus of environmental policy should be on environmental technology innovation rather than on pollution control. In addition, we will explore how the different instruments of environmental policy are suited to achieve social optimal emissions levels when the effect of induced technological change is accounted for. 4.1

Welfare Benefits from Environmental Regulation and Innovation

Parry et al. (2002) compared the welfare gains associated with a Pigouvian tax with the welfare benefits from induced technological change, using a dynamic social planning model in which investments in R&D enhance the stock of knowledge, thereby reducing future abatement costs. Their analytical and

72

Environmental and resource economics 2004/2005

Marginal cost/benefits

MC

MC' a p WP

c

f

WH

b

0

AP

A*

¯ A

Note: Area WP represents the Pigouvian welfare gain from abatement. Innovation causes the marginal abatement cost curve MC to rotate downwards (MC′), yielding welfare gain WH. Source:

Parry et al. (2002).

Figure 2.3 Welfare gains resulting from abatement cost reduction due to technological innovation numerical results indicate that the welfare gains from induced technological innovation are typically smaller than the welfare gains resulting from pollution control. Only in special cases will the welfare gains from innovation exceed those from the Pigouvian tax. Some of the insights obtained by Parry et al. can be explained with the help of Figure 2.3, which plots for one time period the marginal abatement cost curves before and after an increase in the ‘abatement knowledge stock’. To keep the discussion parsimonious it is assumed that marginal benefits from avoiding pollution are constant. The optimal abatement level is the point where marginal abatement costs, MC, equal marginal benefits of pollution control, p. In the absence of technological innovation this level is reached at AP in Figure 2.3. The corresponding total abatement costs are given by the area of the triangle 0aAP, while the total benefits corresponding to this level of

A selective review

73

abatement are given by the rectangle 0paAP. The difference between these two areas, triangle 0pa, is the welfare benefit WP of the Pigouvian tax in one period. Since the stock of knowledge is constant along the optimal abatement ˆ P of the Pigouvian welfare gain over the entire time path, the present value W horizon, defined as the discounted sum of the net benefits in each period, is ∞

ˆP = W

∑ t=1

WP WP –––––– = –––––, (1 + q)t q

(2.13)

where q denotes the social discount rate. The welfare gain from innovation WH is represented by triangle 0ac in Figure 2.3, and consists of the reduction in marginal abatement costs (area 0ab) and the additional benefits from higher abatement (area acb). If one assumes that R&D investments leading to innovation are undertaken only in ˆ H of all future benefits from innovation is period 0, the present value W ∞

ˆH = W

H

H

W W ∑ (1–––––– = –––––, + q) q t=1

t

(2.14)

ˆ H, Parry et al. compared the discounted welfare gain from innovation, W ˆ P, using the ratio W ˆ H/ W ˆ P. To with the discounted Pigouvian welfare gain, W get a first feeling for the magnitude of this ratio we notice that for the welfare gain from innovation to be large relative to the Pigouvian welfare gain, ˆ H/Wˆ P must be greater than unity. Since in our simplified analysis W ˆ H/W ˆ P is W ˆ H/Wˆ P equal to WH/WP, we can easily establish a bound on the magnitude of W by using Figure 2.3: assume that R&D in period 0 reduces (marginal) abatement costs in period 1 by a factor b with 0 ≤ b < 1. Then MC′ = (1 – b)MC, and8 ˆH W WH b –––– = –––– = –––––. ˆP W WP (1–b)

(2.15)

According to this equation, even an abatement cost reduction of as much as 50 per cent (b = ½) would still yield an innovative welfare gain that is smaller ˆH<W ˆ P. We see already in this highly than its Pigouvian counterpart, that is, W simplified analysis that the net benefits from induced technological change are limited by the maximally feasible reduction in abatement costs. In most situations, they are likely to be smaller than the Pigouvian welfare gains. ˆ H to exceed W ˆ P: consider the arrival of a new, However, it is possible for W

74

Environmental and resource economics 2004/2005

‘disruptive’ technology that completely and instantaneously eliminates all abatement costs. Then, the marginal abatement cost curve MC′ in Figure 2.3 ˆ H/W ˆ P corresponds geometricoincides with the horizontal axis and the ratio W ¯ cally to the area of the trapezoid 0afA divided by the area of the triangle 0pa. ˆ H/W ˆ P of In this case, Parry et al. calculated an upper bound on the ratio W respectively 19, 4, 2.3 and 1, when the initial Pigouvian abatement level is 10 per cent, 40 per cent, 60 per cent and 100 per cent. Thus, only when innovation substantially reduces abatement costs and the Pigouvian abatement level is fairly modest can the welfare gain from induced innovation be significantly greater than the Pigouvian welfare gain. The intuition here is that for a pollution problem that is already severe enough to warrant a high level of abatement without R&D, additional welfare gains from innovation will be relatively small. Conversely, if abatement is initially too costly to justify major emission reductions, the gain from innovation could be more substantial. A more thorough discussion has to take into account that it needs time, perhaps decades, to accumulate the knowledge required to substantially reduce abatement costs. During this intermediate time the marginal abatement cost curve in Figure 2.3 will only gradually rotate downward, so that the benefits from innovation will be smaller than the benefit that is realized when knowledge accumulation is instantaneously complete. In addition, the direct costs of R&D that were neglected in the argumentation so far need to be taken off the discounted stream of benefits to obtain the true discounted welfare gains from innovation. As far as the timing of these R&D expenditures is involved, Parry et al. note that a balance must be struck between the gain of immediate increases in the knowledge stock and the cost-saving from gradual adjustment. This is of course the same type of dynamic optimization problem that we encountered earlier and discussed in the context of the benefit–cost policy criterion of Goulder and Mathai (2000). Parry et al. choose to tackle the problem by means of a numerical simulation. They explore different scenarios for the time it takes for knowledge accumulation to produce a 50 per cent reduction in abatement cost. Table 2.2 summarizes the results of the base case simulations by Parry et al. that were carried out for a discount rate of 5 per cent. For different values for the initial Pigouvian abatement level in the first column, the table specifies ˆ P ratios corresponding to these ˆ H/W in the three subsequent columns the W initial levels depending on the various time periods in which abatement costs reduction of 50 per cent are achieved. In line with the earlier observation, the simulation shows that the ratio ˆ H/W ˆ P decreases when the Pigouvian abatement level rises. The aspect that is W new concerns the impact on welfare gain of the time period over which abatement costs are reduced. It is seen from the table that the longer it takes to

A selective review

75

Table 2.2 Ratio of welfare gains from innovation to welfare gains from pollution control, WH/WP Time lag until abatement costs halve Pigouvian abatement level (%) 10 40 60

10 years 2.98 1.07 0.79

20 years 0.88 0.46 0.41

40 years 0.16 0.16 0.17

reduce these costs, the smaller is the welfare gain from innovation. Thus, ˆ H and W ˆ P are almost the same when the Pigouvian according to the table, W abatement level is 40 per cent and abatement cost are reduced by 50 per cent in ten years. But if it takes 20 years or more to achieve this cost reduction ˆ H is signifi(because it is more expensive to develop new technologies), then W ˆ P. The key message here is that the conditions for W ˆ H to cantly smaller than W ˆ P are very stringent: the initial abatement level must be be large relative to W fairly modest and innovation must have the potential to reduce abatement costs rapidly. Parry et al. show that this result is robust against different assumptions for research cost function and choice of planning horizon. The effect of an ˆ H/W ˆ P as then the benefits from innoincrease in the discount rate is to lower W vation are more heavily discounted. A practical demonstration for the application of the results in Table 2.2 is furnished by the Kyoto treaty on climate change. The treaty commits the European Union to cut emissions of CO2 and other greenhouse gases 8 per cent below the 1990 baseline level during 2008–12. Assuming for the sake of the argument that the target of the envisioned emission reduction is set optiˆ H will be significantly mally, then one would expect in this example that W ˆ P. This is because of the difficulty of achieving cost reductions smaller than W of 50 per cent or more within a decade or so, without significant deployment of abatement technology. In an analogous case, the cost of technologies to reduce sulphur emissions from power plants decreased by 20 per cent over the last decade (Boward and Brinkman, 1998). The results of Parry et al. appear to imply that pollution control – rather than technological innovation – should be the dominant factor in the design of environmental policies. However, it is important to realize that this conclusion is based on the analysis of policy-induced welfare gains in only a single market, namely the market for novel abatement technology. In a following section, we shall examine whether Parry et al.’s policy recommendation keeps its validity if one allows for market interactions and, in particular, accounts for

76

Environmental and resource economics 2004/2005

inter-market technology spillovers. To set the context for this discussion, we next take a look at the various instruments available to a policymaker to address environmental externalities and technology spillovers. 4.2

The Role of the Policy Instrument

The discussion so far has focused only on a single instrument of environmental policy, namely taxes. As policymakers have more options to achieve environmental objectives, we consider briefly in this section the incentives of alternative policy instruments for inducing technological change. A more comprehensive review of this topic can be found in Jaffe et al. (2002) and Kemp (1997); a short discussion in the context of climate change policy is given by Fischer (2000). Environmental policy instruments are usually grouped into market-based instruments, such as taxes, subsidies and tradable permits, and the so-called ‘command-and-control’ instruments, which include emission standards and other forms of direct regulation. While the former type is the favorite of economists because of its cost effectiveness for achieving emission reductions, both types of control are likely to spur some technological progress. Thus a discussion of the consequences of the various policy instruments for inducing technological change has to start with the motives of firms to invest in novel abatement technologies. These include cost-savings that the new technology promises, reduction in emission tax payments, the prospects of gains from patent royalties, subsidy payments, or perhaps public relation motives, that is, the wish to be seen as ‘green’. Early papers have concentrated on elucidating how in a competitive setting the different policy instruments affect a firm’s decision to invest in abatement technology on the premise that a firm wants to reduce its marginal abatement costs (Downing and White, 1986; Magat, 1978, 1979; Zerbe, 1970). This is also the starting point for the analysis by Milliman and Prince, who studied the issue first for the case of a single representative firm (1989), and later extended their analysis to firms with heterogeneous abatement cost (1992). In their 1989 study Milliman and Prince used essentially a graphical argument of the kind depicted in Figure 2.4. Shown on the axes of the graph are the levels of abatement, and the marginal costs and benefits associated with it. These quantities are assumed to vary linearly with the degree of abatement. Technological progress is pictured as a rotation of the marginal cost curve in the direction of decreasing costs. The incentives the different policy instruments give to a firm to use novel abatement technology result from the costsavings they imply. In the graphical representation these incentives correspond to certain areas, which are indicated in Figure 2.4. By comparing the size of these areas Milliman and Prince established in their 1989 paper the following

A selective review

77

Standards (0ab) Free permits (0ab) Emission taxes (0ac) Auctioned permits (0afhb) Abatement subsidies (0ac)

Marginal costs/benefits

MC

MB MC' a

p0

p*

d

0

c

h

b

A0

f

A' Abatement

¯ A

Note: A firm facing an emission tax of p0 realizes savings corresponding to area 0ac if it adopts novel technology that lowers its marginal abatement cost curve from MC to MC′; with the novel technology the firm will abate to level A′ instead of A0, provided that the policymaker does not adjust the level of control. The source of the savings is the reduction in abatement costs (area 0ab) plus the net reduction in tax payments (area acb). The savings given by the other policy instruments are as indicated in the figure. Source: Milliman and Prince (1989).

Figure 2.4

Firm-level incentives to adopt under pollution control

ranking for the potential of different policy instruments to promote the adoption of novel emission-reducing technology: (i) auctioned permits; (ii) emission taxes and subsidies; and (iii) standards and free permits. Later work by Milliman and Prince (1992) and Jung et al. (1996) extended the analysis to the situation of heterogenous firms, but did not yield results that change this ranking order. Some authors have investigated if and how the ranking would change if some of the rather ideal assumptions underlying the work by Milliman and Prince were relaxed. A rather obvious shortcoming of the model is that it regards the permit price as exogenous and as independent of the adoption decisions of the

78

Environmental and resource economics 2004/2005

firms. Since the adoption of novel abatement technology lowers the permit price, it reduces the anticipated cost-savings of firms over time, and thus their incentive to adopt the new technology.9 The issue was analyzed in papers by Keohane (1999) and Requate and Unold (2003), who concluded that taxes provide higher incentives for adoption than tradable permits, and that auctioned and freely allocated permits have equal incentives if permit price is endogenous. Another extension of Milliman and Prince’s work has been undertaken by Fischer et al. (2003), who considered the incentives of firms to develop and to adopt environmentally friendly innovations when other firms can imitate. These authors could not establish a unique ranking of the policy instruments. For example, with full imitation, auctioned permits tend to be superior to emission taxes, because innovators can effectively appropriate the gains from the fall in permit price. However, when barriers to imitation are high, emission taxes are the preferred policy instrument for inducing environmentally friendly technologies, because the rents from innovation under a tax outweigh innovators’ profits under a permit scheme. This evaluation changes when the policymaker has only limited information on the aggregate abatement cost function. Focusing on the special situation that environmental marginal damages do not vary substantially with emission levels, Biglaiser et al. (1995) argue that, for the purpose of inducing technical change, taxes are superior to tradable permits. The reason is that, unlike taxes, a permit scheme would allow firms to strategically delay their plans to invest in novel clean technology in order to oppose more stringent future emission targets.10 The theoretical studies reviewed above indicate that market-based instruments generally provide greater incentives for firms to invest in environmentally friendly technologies than direct regulations. The magnitude of these incentives depends on the model context and assumptions, meaning that no general ranking of the policy instruments on their ‘innovation-promoting’ potential is possible. Moreover, all of the results on the effects of the policy instruments obtained so far neglect innovation by firms in non-polluting markets and the associated economy-wide consequences. Moving beyond this limitation requires a general equilibrium approach, which is the subject of the following sections.

5.

THE GROWTH THEORY PERSPECTIVE

In this section we discuss some macroeconomic aspects of the interaction of environmental policy with technological change using the vehicle of endogenous growth theory. Models of endogenous growth differ from the ‘standard’

A selective review

79

neoclassical Solow theory in that they do not rely on exogenous technological advance falling like ‘manna from heaven’ to explain growth. Instead, dedicated profit-seeking investment in knowledge by firms is seen as the cause for long-run growth. Since knowledge is difficult to appropriate by firms, it spills over, thus creating positive externalities. Both R&D investment per se and knowledge spillover appear as the combined source for endogenous growth in these types of model. They endow the macro-scale production function with the properties that are sufficient to guarantee steady-state growth (Romer, 1990; Arrow 1962b). In-depth reviews of the burgeoning literature on endogenous growth can be found in work by Barro and Sala-i-Martin (1995) and Aghion and Howitt (1999). The inclusion of environmental aspects into the framework of endogenous growth models is usually accomplished through the assumption that society ‘cares’ about the environment, which is then formalized by including some measure of environmental quality as a variable in the utility function that a social planner optimizes. In fact, all the environmental growth models discussed in the monograph by Aghion and Howitt (1999, ch. 5) follow this construction. However, this approach falls short of capturing the productivity effects associated with environmental quality. These arise, for instance, because a cleaner environment improves the health of workers, thereby boosting labor productivity, or because enhanced biodiversity provides a larger pool of genetic information, thereby spurring productivity in R&D. A set of rather general growth models that also include these types of ecologic–economic interactions was constructed in a series of papers by Bovenberg and Smulders (1995, 1996) and Smulders (1998) that are the subject of this section. The simplest type of these models (Smulders, 1995) lumps together physical and knowledge capital and so maintains the one-sector production structure of the Solow-type neoclassical growth models. It offers insight, in particular, into the consequences of optimal environmental policy on long-run economic growth, but suffers from the underdeveloped representation of the process of knowledge accumulation, which is simply modeled as a by-product of production. Despite this shortcoming, we include the model in this review as it provides both a conceptual and formal stepping stone for the construction of more complex and more satisfying growth models. Of interest here is the model extension by Smulders (1998), which provides both for endogenous and exogenous knowledge accumulation and is capable of illuminating the interaction between the optimal setting of policy targets and technological change. 5.1

A One-sector Endogenous Growth Model

Models of endogenous growth and the environment have to account for a variety of interactions between economic activities, technology and the environment. To

80

Environmental and resource economics 2004/2005

lay bare the various interdependencies and to avoid getting confused by details we will first give an overview of the structure of these types of models (Smulders, 1999). After this preparation, we proceed to study specific model realizations in more detail. 5.1.1 Prototypical model structure As before, we assume the existence of a social planner whose objective is to optimize social welfare over an infinite planning horizon. This requires that one specifies a social utility function U, and the factors on which it depends. Taking aggregate consumption X and environmental quality, as measured by an indicator N, as the relevant variables (UX > 0, UN ≥ 0), and assuming furthermore that the social planner discounts the utility of future generations by q, the relevant welfare function reads ∞ W=

∫

U(X, N)e–qtdt.

(2.16)

0

For the sake of convenience we disregard population growth and normalize population size to 1, so that there is no distinction between per capita quantities and their aggregate counterparts. Focusing next on modeling the ecosystem, we denote the productive services from the environment by R, and stipulate that the change in environmental quality resulting from production can be described by N˙ = Q(N) – R,

(2.17)

where the function Q(N) describes the regenerative potential of the ecosystem. It is useful, but not necessary, to think of the function Q(N) as having the hump-shaped appearance familiar from the theory of renewable resources (Clark, 1990). Of special interest in the following is the case for which N˙ = 0, and so R = Q(N). It pertains to a situation in which environmental services are used exactly to the degree with which they are regenerated, and which with some justification could be labeled ‘sustainable’. Production Y is allocated to consumption X, and to investment in both manmade and human capital that we collectively denote by M. Thus, ˙ = Y (N, R, M) – X, M

(2.18)

where YN > 0, YR > 0, YM > 0, and all inputs are necessary. The crucial aspect here is that production Y depends, in addition to capital M, on environmental quality, N, and services R.

A selective review

81

Equations (2.16)–(2.18) form the basis of many environmental growth models (Verdier, 1993; Gradus and Smulders, 1993; Bovenberg and de Mooij, 1997). Here we follow Smulders (1995) in developing this prototypical model into a form that allows one to draw definite conclusions. 5.1.2 Optimal long-run growth Smulders (1995) splits man-made capital M into two components, namely physical capital K and knowledge H, M = K + H,

(2.19)

so that the production function Y can now be written as Y = Y (K, H, N, R).

(2.20)

It is assumed that Y (·) exhibits the standard neoclassical properties, including constant returns to scale in K and H. With a view to ensuring the existence of steady-state growth, Smulders follows the common practice (for example Barro and Sala-i-Martin, 1999, p. 64) and employs a utility function (XNf)1–1/s U(X, N) = –––––––––, 1 – 1/s

(2.21)

for which the elasticity of intertemporal substitution s, U′(X) s = – –––––––––, XU′′(X)

(2.22)

and the share of environmental amenities in utility f, UNN f = ––––––, UX X

(2.23)

are constant. As will be seen later, the parameter f can be also interpreted as the social preference for environmental amenities. A social planner opting to conduct optimal policy will aim to maximize utility over an infinite time horizon with respect to X, R, K and H, subject to the ecological constraint (2.17) and the goods market constraint

82

Environmental and resource economics 2004/2005

Y = X + K˙ + H˙

(2.24)

that is implied by (2.18) and (2.19). The corresponding optimality conditions read: YH = YK UXNN N˙ X˙ s[r – q + (––––––) ––] = ––– UX N X UN/UX + YN Y˙ R ––––––––––– + QN + ––– = r, YR YR

(2.25) (2.26)

(2.27)

where in the last two equations we denoted the return to capital by r, that is, YK ≡ r.

(2.28)

The first condition, equation (2.25), is a requirement for the statically optimal allocation of physical and knowledge capital and states that both types of capital must yield the same returns in an optimum. As mentioned earlier, there are strong reasons to believe that this condition cannot be satisfied in a perfectly competitive market without help of governmental intervention. The difficulties of firms in appropriating the knowledge they generate through environmental R&D translate into a lack of incentives to accumulate knowledge capital to the socially optimal extent, so that satisfaction of (2.25) is likely to require policy intervention. We will not pursue this issue further but rather focus on the two dynamic optimality conditions (2.26) and (2.27). We first note that if X˙/X > 0, then consumption is low today relative to tomorrow (that is, consumption grows). Equation (2.26) then says that households are more willing to postpone consumption the more they benefit from it, that is, the more the interest rate r exceeds their pure time preference q, and the more they value future improvements in environmental amenities, as expressed by the term involving UXN. Through the presence of this term equation (2.26) furnishes a generalization of the standard Ramsey equation (for example, Barro and Sala-i-Martin, 1995, p. 63). Equation (2.27) is a generalization of the Hotelling rule, and guarantees that in an optimum the exploitation of the environment yields a return rN, given by its left-hand side,

A selective review

UN /UX + YN Y˙R rN = ––––––––––– + QN + –––, YR YR

83

(2.29)

which equals the return r on alternatives capital goods, that is, rN = r.

(2.30)

If the environment were a non-renewable resource without amenity value and with only the capacity to provide services to production by resource depletion, we would recover from the last two equations the usual version of the Hotelling rule. The price of environmental service YR should then grow at the rate that equals the rate of interest to compensate for the loss of future revenue from current depletion. The QN term in equation (2.29) indicates that the price of a renewable resource has to rise at a rate faster than r as long as QN is negative. The reason is that the extracting use of the environment not only depletes the resource but also reduces its future capability of regenerating. This reduces the revenue from the exploitation of the environment, which has to be compensated by an additional price increase. Next we focus on the implication of equations (2.26) and (2.27) for the characterization of steady-state growth. The model allows this type of growth provided environmental quality is at the sustainable level N* for which Q(N*) = R and N˙ * = 0. Smulders shows that to N* corresponds a constant value r, which when used in the Ramsey equation (2.26) ensures a steady-state growth rate g = X˙/X for consumption that is non-zero at the fixed value g = s(r – q) > 0,

(2.31)

and that physical capital, knowledge and production also grow in the long run with g, meaning that the characterization of the steady state is completed through the equations Y˙ K˙ H˙ X˙ –– = –– = –– = –– = g. Y K H X

(2.32)

These results show that sustainable growth is feasible provided that there are constant returns in production with respect to knowledge factors. 5.1.3 Policy consequences Further insight into the long-run prediction of the Smulders model can be gained through a graphical analysis. We first note that from (2.30) and the

84

Environmental and resource economics 2004/2005

assumption of constant returns to scale in production the return on capital r depends only on N and R. In the steady state, environmental use R equals the absorption capacity of the environment Q(N), so that r becomes simply a function r of N*, r = r(N,Q(N)),

(2.33)

where we have dropped the asterisks to simplify notation. The upper part of Figure 2.5a depicts the variation of r with N in a schematic fashion. Obviously, the rate of return to capital is zero for the value of N at which Q(N) = 0, as then the environment does not provide the services R that are essential φ large φ small

r rN

r rN

B B'

A

ru p (N, Q(N))

A'

N R

N R Q(N))

Q(N)) N

N g

g g = σ[ρ (N, Q(N)) – θ]

g¯ N*

N

(a) endogenous growth, γ < 1

N*

N

(b) exogenous growth, γ = 1

Note: The upper panels depict schematically the variation of rN with N together with those of r (endogenous growth) and ru (exogenous growth). The optimal state is achieved at the point of intersection of the corresponding curves. The lower panels show the levels of services provided by the environment (R) and the growth rate (g or g¯) in the social optimum. Source: Smulders, 1995, 1998

Figure 2.5 Steady-state solution for the endogenous and exogenous growth case

A selective review

85

for production. It increases for low levels of environmental quality N, since production benefits from improved environmental services, but decreases for high values of N as less productive use is made of the environment. The growth rate g, which according to (2.31) differs from r only by constant factors, follows the same pattern and is depicted in the lower part of Figure 2.5a. Next we turn attention to the rate of return to environmental quality rN, equation (2.29). By examining the dependence on N of the various terms comprising rN, it can be shown that this rate will tend to fall with improvement in environmental quality, so that drN/dN < 0, as indicated in the upper part of Figure 2.5a. Changes in the parameter f, which characterizes the social preference for environmental amenities, cause the rN curve to shift. If a higher amenity value is attached to the environment, investment in the environment will pay a higher return, and the rN curve moves outward. At the value N*, at which the rN and the r curves intersect, the rates of return of environmental and man-made capital are equal, and environmental quality is socially optimal according to (2.30). All the elements are now in place to analyze graphically some of the consequences of environmental policy. We consider first the case that society attaches a large value to environmental amenity, as quantified by the parameter f, and that accordingly the level of environmental quality, as expressed by N, is also large. Nonetheless we assume that it is still suboptimal, so that the return on natural capital exceeds that from physical capital (point B in Figure 2.5a). As a result of governmental policy, which aims at full internalization of all the benefits of the environment, the optimal growth equilibrium at the intersection of the r-curve and rN-curve is realized (point B′). As the corresponding growth rate g (lower part of Figure 2.5a) is smaller than the one in the initial state, optimal policy will in this case hurt long-run growth. Thus there is a tradeoff between economic growth and environmental quality. This conclusion changes, however, when initial environmental quality is suboptimal at low levels of N and f (point A in Figure 2.5a). Governmental policy correcting this situation will then strive to reach state A′. As the growth rate in the optimum is now larger than the one in the initial state, optimal environmental policy will now stimulate long-run growth. The model can be used to illuminate the economic consequences of environmental policy for a range of circumstances. Nonetheless, it does not provide much insight into the specific role played by technological change that is induced by environmental policy. The reason for this failure lies ultimately in the constraints implied by the one-sector model structure, which requires knowledge to grow endogenously as a by-product of production, and which cannot capture the process of knowledge accumulation with the required detail. Much more suited for our purpose are growth models whose representation of the knowledge accumulation process is flexible enough to account for

86

Environmental and resource economics 2004/2005

both endogenous and exogenous knowledge accumulation. Being able to vary the degree of ‘endogeneity’ in a model allows one to directly assess the economic consequences of environmental policy for different assumptions concerning the strength of the interaction between environmental policy and technological change. A model for this purpose has been developed by Smulders (1998). It builds in its formal structure on the presentation given earlier in this section and relies on previous work by Bovenberg and Smulders (1995, 1996). 5.2

Technological Change and the Cost of Environmental Policy

5.2.1 Extending the basic model The Smulders model (1998) regards knowledge H as a composite good comprising a component hen that is accumulated endogenously through investment in R&D, and a second component hex that increases as result of exogenous technological progress. No explicit stipulation is made concerning the relative weight with which hen and hex contribute to H. It is rather assumed that H is determined by some function H of hen and hex, so that the accumulation of composite knowledge can be described by H ≡ H(hen, hex), Hen > 0, Hex > 0.

(2.34)

H is taken to be homogeneous of degree 1 in its arguments, implying that H = Hhenhen + Hhexhex. The share of endogenous and exogenous knowledge is then quantified through the respective elasticities Hhen g = ––––– hen, H

Hhex 1 – g = ––––– hex. H

(2.35)

Knowledge H is regarded as ‘resource augmenting’, meaning that it enhances the efficiency with which environmental services R are employed in production. This is captured by specifying the production technology as Y = G(N)F(Z, K),

(2.36)

Z=H·R

(2.37)

where

measures the efficiency with which environmental services are employed in production, and F exhibits the usual neoclassical properties and, in addition,

A selective review

87

constant returns to scale. The effect of environmental quality on productivity is accounted for through the term G(N). The model is completed through the goods market equation, Y = X + K˙ + qh˙en,

(2.38)

which states that production is used for consumption, accumulation of (physical) capital goods, and knowledge creation. The symbol q quantifies the price of knowledge in terms of the final good.11 It is easy to see that when g = 1 and all technological change is endogenous, the model reduces to the one-sector endogenous growth model of the previous section. The similarities go even further. It can be shown that the optimal intertemporal allocation conditions are formally identical to the ones for the one-sector endogenous model so that (2.25)–(2.27) retain their validity. The only difference is that now Yhen/q replaces YH. As before, long-run growth is characterized by a constant value of environmental quality and by a common growth rate for the economic variables according to equation (2.32). It is here that the similarity stops. Unlike in the one-sector model, where the growth rate g was endogenously determined from equation (2.31) and (2.32), the growth rate is now given by the rate of exogenous technological progress and equals h˙ex/hex. The intuition is here that when g = 0, the model reduces essentially to the standard neoclassical Solow growth model, in which diminishing return to capital in the production function is compensated by exogenous technological progress to yield non-zero long-run growth. This dynamics remains intact even for the more general case: as long as g < 1, diminishing returns to investment in man-made capital mean that growth rates exceeding g¯ ≡ h˙ex/hex can only be maintained by devoting an increasing fraction of production to investment. This finally reduces consumption to zero, which is clearly suboptimal. Consequently, the long-term growth rate is g¯ for all gs smaller than 1.12 It is important here that g¯ is independent of environmental quality N. The same applies then to the long-run rate of return that is calculated from g¯ through the Ramsey equation (2.36) as g¯ ru = q + ––, s

(2.39)

where all terms are independent of N. In the graphical representation, Figure 2.5b, ru appears as a parallel to the N-axis. Its intersection with the rN-curve determines the socially optimal environmental quality N*, as the optimality condition (2.30) now takes the form rN = ru.

88

Environmental and resource economics 2004/2005

5.2.2 Policy implications Figure 2.6 compares the consequences of environmental policy when technological progress is exogenous to the situation reached when progress is endogenous. Focusing first on the exogenous case, we consider an economy that is initially in a suboptimal state with environmental quality N0. The government puts policy in place aimed at more fully accounting for consumption externalities. Smulders models this ‘policy shock’ by an increase in the value of the parameter f, equation (2.23), which measures the weight of amenities relative to produced consumption. Figure 2.6a shows that the increase in f will increase the long-run optimal value of environmental quality N, but the improvement is less pronounced when g is relative large, and technological progress is to a larger extent endogenous. The intuition here is that a larger role of endogenous technology development requires more investment to acquire the knowledge that stimulates long-run growth. This absorbs output that would otherwise go into consumption, and there is less room to invest in amenities. Smulders refers to this as the ‘burdenof-investment effect’ of endogenous growth. It follows that governmental policy that does not account for the burden of investment resulting from induced r

r r N, φ > 0 r N, φ = 0

p (N, Q(N))

r* ru

ru

N

N

N0 (a) exogenous case, γ < 1 r N, φ = 0

N0

N*

N'

(b) endogenous case, γ = 1 r N, φ > 0, γ high

r N, φ > 0, γ low

Note: The effect of environmental policy is modeled by shifting the rN-curve to the left. This results (a) when growth is exogenous in an increase of the socially optimal environmental quality; (b) for endogenous growth, however, the optimal improvement in environmental quality is less pronounced and is accompanied by a larger interest rate. Source:

Smulders (1998).

Figure 2.6 Impact of environmental policy on long-term environmental quality and interest rate

A selective review

89

technological change may overestimate the efficacy of policy for improving the environment. Technological change makes tight environmental policy less desirable as the burden of investment effect impedes environmental improvement. Figure 2.6b illustrates the different outcomes of environmental policy for exogenous and endogenous technological advance. When technological advance is exogenous (g < 1) and environmental quality is initially at state N0, optimal policy will improve the level of environmental quality to N′ without affecting interest rate ru. When technical progress is endogenous (g = 1), however, the optimum state is given by the intersection of the rN with the rcurve at which both environmental quality N* and interest rate r* are increased relative to the situation without policy intervention.13 The induced change in the interest rate r*, which is absent in the exogenous growth model, occurs in addition to the ‘burden-of-investment effect’. The direction of this effect depends on the initial value of N and the shape of the r-curve. If the rate of return to man-made capital increases in the long run, as is shown in Figure 2.6b, optimal environmental quality (N*) is smaller than in the exogenous growth model (N′), because investment in economic growth is rewarded by a higher return than investment in the environment. This mechanism counteracts policy aimed at improving environmental quality in the long run. Together with the burden-of-investment effect it reduces the level of optimal long-run environmental quality relative to the one that is socially optimal when technological progress is exogenous. To gain insight into the impact of endogenous technological change on the optimal timing of abatement and environmental policy, Smulders (1998) employed numerical simulations to study the transitional dynamics of both the endogenous and exogenous growth models. Contrary to the result obtained by Goulder and Mathai (2000), the outcome here is that endogenous technology change justifies higher near-term abatement and disfavors a ‘wait-and-see’ strategy. The reason is that in the presence of induced technological change environmental policy temporarily lowers the rate of return to man-made capital and it does so for a longer period of time than is the case for exogenous technical advance. In consequence, the incentives to invest in a clean environment increase. While this ‘interest rate effect’ justifies higher initial abatement, it diminishes over time as the rate of return to man-made capital recovers. Hence it is optimal for abatement and the stringency of environmental policy to decrease with time. In summary, then, the main insights provided by the Smulders model are: 1. Governmental policy to improve environmental quality can both stimulate and hurt long-term optimal growth depending on the initial level of environmental quality and on the value society attaches to environmental amenities. 2. The presence of induced technological change provides a reason for a

90

Environmental and resource economics 2004/2005

more ambitious environmental policy. It will typically act to stabilize long-term environmental quality at a level that is below that which could be optimally achieved when technology progress is exogenous.

6.

TECHNOLOGICAL CHANGE IN GREENHOUSE GAS ABATEMENT MODELS

All the models reviewed so far have focused only on one type of market failure, namely the one associated with environmental externalities, but have neglected R&D inefficiencies resulting from technology spillovers. To complete our survey we next review models that are capable of accounting for both deviations from the efficient market paradigm. Models of this type have been developed and used to simulate the consequences of policies aimed at mitigating climate change resulting from industrial atmospheric emissions of greenhouse gases. Because of their extended scope, these models are fairly complex, and thus no longer analytically tractable. A prominent representative for this class of models is the DICE-99 model (Dynamic Integrated model of Climate and the Economy) developed by Nordhaus (see Nordhaus and Boyer, 2000, for an extensive review), which is a highly aggregated global model linking together global production, energy consumption and a climate-related sector. DICE-99 builds upon the more detailed RICE-99 (Regional dynamic Integrated model of Climate and the Economy) model by the same author, which represents a geographically disaggregated eight-region view. While both DICE-99 and RICE-99 treat technological progress as exogenous, they have spawned a series of models in which technological change is endogenized. In his R&DICE version of DICE-99 Nordhaus (2002) uses a fixed proportion production function to separate the effect of induced innovation from that stemming from input factor substitution. Technology change is taken to manifest itself through the development of greenhouse abatement technology that lowers the carbon emissions required to produce a fixed output. RICE is also the parent of a model developed by Buonanno et al. (2003) for which the authors coined the acronym ETC-RICE. It borrows from RICE the neoclassical production and climate change sector, and implements induced technology change in the same way as R&DICE, namely through its effect on the carbon intensity of output. From the perspective of this review perhaps the most interesting variant of DICE-99 is the model developed by Popp (2002). A feature of this author’s ENTICE (for Endogenous Technological Change) model is that it can account for a multitude of channels through which technological progress can affect greenhouse gas emissions, including not only progress in abatement technology but also improvements in energy efficiency and fuel switching.

A selective review

91

In ENTICE the production sector is modeled by a Cobb–Douglas type production function, in which, next to capital and labor, energy services appear as an input factor. This input comprises both fossil fuel use and ‘energy-related human capital’, and is related to these quantities through a constant elasticityof-substitution (CES) functional form, in which the substitution parameter provides an indication of the ease with which input switching is feasible. ENTICE assumes that carbon-saving ‘energy knowledge’ increases with the accumulation of research through costly R&D in the energy sector, and depreciates as new knowledge supersedes old. The economic damage caused by industrial greenhouse gas emissions is modeled in the same way as in DICE: the damage function is taken to depend quadratically on the mean global temperature increase resulting from industrial CO2 emissions, which in turn is calculated on the basis of a set of geophysical relations. This damage is included in the production function to cause a fractional reduction in output. As discussed earlier, markets tend to invest in R&D at a socially suboptimal level because of technology spillovers and the difficulties associated with the appropriability of technology innovations. ENTICE models this type of market failure in an approximate and heuristic way. Taking up a suggestion by Nordhaus (2002), ENTICE implements a constraint in its objective function that forces the returns on energy R&D to be a multiple of the returns on physical capital. This inflates innovation costs and forces the market to underinvest relative to the social optimum at which investments would give equal returns to all forms of capital, both physical and non-physical. Specifically, the multiplier is given the value 4 in broad alignment with statistical data for the USA. Investment in R&D in the energy sector will crowd out research in other sectors. Since the return on energy R&D is assumed to be four times greater than that of other investments, full crowding out would imply that an increase in energy R&D leads to a fourfold decrease in the value of investments in the non-energy sectors. Popp argues that such a high crowding-out factor is not supported by statistics, and accordingly assumes in his ENTICE base case simulations only a 50 per cent crowding out. After the model has been calibrated against empirical estimates of energysaving R&D on induced innovation in the energy sector, it furnishes, in particular, a quantification of the welfare gains resulting from a socially optimal carbon tax. Popp differentiates the ‘direct welfare effect’ of new knowledge, which is associated with immediate environmental benefits, from a ‘productivity effect’, which refers to the fact that new knowledge makes future R&D more productive. In his ENTICE simulation Popp is able to differentiate between the two effects and to show that it is the latter that makes the dominant contribution to welfare over longer time periods, even when knowledge depreciation is accounted for.

92

Environmental and resource economics 2004/2005

Popp finds that the effect of induced technological change on key economic variables is small and is outweighed by that of factor substitution. Specifically, carbon-saving technological innovation induced by an optimal carbon tax imposed in 1995 will increase output, but only by 0.5 per cent in 2105. Another model that explores the connection between greenhouse gas abatement policy and technology change is the one developed by Goulder and Schneider (1999). Its scope is different from ENTICE and in fact from the other models considered so far as it does not adopt a social planning framework and refrains from ascertaining optimal policy. It rather takes a carbon tax as given, and simulates how profit-maximizing economic actors are responding to it. The model considers a conventional, carbon-based energy industry, and an alternative emissions-free industry. An R&D service sector is introduced whose function within the model is twofold: 1.

2.

It supplies both industries with innovations that allow them to reduce the cost of producing the particular sort of energy. This means specifically that technology advance in the conventional energy sector is modeled as productivity improving, a feature that is in clear contrast to ENTICE, where technology progress is carbon-saving. As the services the R&D sector provides are not free, its inclusion in the model provides a way to capture the opportunity cost resulting from a policy-induced reallocation of a supply of R&D services.

Intrasectoral innovation spillovers are captured in the Goulder and Schneider model in a way that is somewhat more systematic than the ENTICE treatment of this externality while still remaining heuristic: it is assumed that the magnitude of knowledge spillover within a sector is proportional to the industry-wide level of expenditure on R&D. Because of its structure, the Goulder and Schneider model offers an expanded view on the mechanisms through which carbon taxes affect technology change when knowledge spillover and policy-induced reallocation of R&D services are present. As far as the latter is concerned, the model demonstrates that it is the magnitude of the R&D opportunity cost that determines whether induced technology change retards or promotes economic growth when the carbon tax is exogenously fixed. While in the base case simulation induced technological change suppresses output, the situation changes significantly when R&D is ‘free’. In this case induced technological change is output enhancing. The opportunity costs associated with R&D reallocation are dependent on the level of knowledge spillovers. High spillovers within a sector mean that the social returns to R&D in this sector are larger than those in a sector with

A selective review

93

fewer spillovers and higher level of technology appropriability. A tax-induced reallocation of R&D away from the high-spillover sector will imply significant opportunity cost and a corresponding loss of output, as is indeed observed in the simulations. To put the effect of R&D opportunity costs into perspective Goulder and Schneider remark that irrespective of their value, the presence of induced technological change always increases the net benefit from a given tax in the sense that the output loss associated with achieving any level of greenhouse gas abatement is lowered by induced technological change.

7.

CONSEQUENCES FOR ENVIRONMENTAL POLICY

It is not straightforward to draw robust conclusions from the models that we reviewed on the interactions between environmental policy and technological change. This is partly due to differences in the way the concept of ‘technological progress’ is formalized. In the models by Goulder and Mathai (2000) and Parry et al. (2002) technological advance manifests itself through its impact on abatement costs – it lowers them. This is in contrast to the way Goulder and Schneider (1999) and Nordhaus’s R&DICE treat technological progress, namely as output enhancing. Smulders (1998) champions yet a different perspective: his concept of technological progress is ‘resource augmenting’ meaning that technological progress enhances the services that the environment provides. A ‘realistic’ model of technological change in the environmental field would have to encompass all of these different concepts. The ENTICE model (Popp, 2002) is probably the one furthest developed in this respect. Keeping in mind the diverse meaning of ‘technology progress’, we use the insight that the models provide to answer the earlier questions concerning the relationship between environmental policy and technological innovation. The Costs and Benefits of Environmental Policy All studies indicate that induced technical change has positive effects on the costs and benefits of environmental policy. Goulder and Mathai (2000) showed that in the presence of induced technological change any environmental policy target, for example a certain CO2 concentration in the atmosphere, can be achieved at lower costs. However, if the policymaker adheres to a benefit–cost criterion, an induced reduction in marginal abatement costs implies that a higher amount of abatement becomes optimal, which results in greater total abatement expenditures. Since this translates into larger net welfare gains, the presence of induced innovation strengthens the case for pollution control. The model by Parry et al. (2002) quantifies the welfare gains

94

Environmental and resource economics 2004/2005

stemming from induced innovation and shows that they are typically small relative to the benefits associated with pollution control. This result is consistent with the general equilibrium models by Goulder and Schneider (1999) and Popp (2002), which account also for the opportunity costs of investing in environmental R&D. The reallocation of resources to research and to develop ‘green’ technologies will crowd out R&D in other sectors of the economy, with detrimental effects on growth. However, these results need to be qualified as the models consider only one channel of knowledge accumulation, namely investment in R&D. The Optimal Timing of Abatement The source of knowledge accumulation plays an important role for conclusions about the optimal degree and timing of environmental policy. While a ‘wait-and-see’ strategy is appropriate when technological progress is autonomous, this is, in general, not the case when technological progress is endogenous. Specifically, when knowledge is generated via R&D, Goulder and Mathai (2000) as well as Popp (2002) showed that it is optimal to postpone some abatement to the future until the accumulation of knowledge lowers marginal abatement costs. But when abatement itself accumulates the knowledge via learning-by-doing, there is no general result concerning the optimal timing of abatement activities. Nonetheless, it is expected that in this case the potential of future cost reduction typically requires more near-term abatement. Smulders (1998) found that endogenous ‘resource-augmenting’ technical change provides a rationale for higher near-term emission reductions. The reason is that pollution control reduces the productivity of the economy for a longer period of time, so that investment in a clean environment becomes more attractive than investment in production. However, this effect is reversed when improvements in environmental quality outweigh production losses due to emission reductions. The models reviewed do not provide general results on the impact induced technological innovation has for altering the optimal time profiles of abatement and emission taxes. While the details of these changes are related to the mode of technology accumulation adopted by the model, most authors seem to agree that their magnitude will be rather modest. Policy Instruments In the presence of induced innovation the function of environmental policy is twofold: Setting constraints on environmentally damaging activities and stimulating environmentally friendly innovation. There appears to be general

A selective review

95

agreement on the superiority of market-based instruments over command-andcontrol regulations not only to correct environmental externalities costeffciently but also to stimulate private investments in ‘green’ technologies. Which type of market-based instrument should be used in a concrete situation depends on a variety of factors, including present and future marginal abatement costs and damages, and the information the regulator has about these quantities. Based on the models reviewed here, it can be said that endogenizing technological advance does not substantially alter the policy recommendations obtained from environmental economic models without technology change.

8.

CONCLUSIONS

While relatively simple models of the type discussed in this review elucidate important aspects of induced technological progress, they fall short of accounting for the complexity of the real world. Some comments regarding potential modeling extensions seem therefore to be in order. Technology advances through a variety of economic interactions. As we have seen, two idealized representations of this interaction have been to date the main subject of economic theorizing, namely technology advance driven by R&D and by learning-by-doing. Technology advance involving R&D proceeds through the costly allocation of resources to the task of innovation. Accordingly, if society allocates all its resources to production and nothing to R&D, then technology progress comes to a standstill. Conversely, when progress is the result of learning-by-doing, it is the sole consequence of the production and use of technology. Technological progress is then ‘free’ in the sense that there is no need for investment into R&D, although opportunity costs do arise. While both the R&D and the learning-by-doing approach tie well to realworld phenomena, they do not give a complete picture of reality, so that models that are based exclusively on one or the other are bound to miss important determinants of technological change, a fact that has been specifically emphasized by Clarke and Weyant (2002). Thus the learning-by-doing approach disregards historic public and private expenditures, and therefore tends to underestimate the cost of technological advance. There are problems with the R&D approach, too. For one, the R&D approach misses learning-by-doing effects. Moreover, investment in research should not be interpreted as emanating exclusively from R&D labs. Kline and Rosenberg (1986) argue that the notion that innovation is initiated by research of the scientific sort is wrong most of the time. Innovation evolves through cycles of design, testing, production and marketing, all of which draw on state-of-the-art knowledge and interact with public policies.

96

Environmental and resource economics 2004/2005

A second aspect that is not included in state-of-the art models of induced technological change relates to the discontinuous nature of technological progress. It arises when previously uncompetitive technologies reach market acceptance. While before market acceptance R&D investment has typically limited impact on aggregate production, this changes dramatically at the point of market acceptance where aggregate production undergoes a discontinuous shift. Discontinuities of this type can be simulated with bottom-up models that explicitly consider hundreds of technologies and that can therefore give detailed representations of changes in the aggregate production characteristics (Seebregts et al., 1999). However, analytical modeling of the discontinuous characteristics of technology progress is still in its infancy. Finally, the usefulness of models of induced technological change could be improved by explicit inclusion of uncertainties related to technological progress. Uncertainty about how far technology might advance, how fast it will do so, and how costly it will be translates into uncertainty about how the production function responds to policy stimuli. Given that it is unknown when technological change will occur and how abrupt it will be, the best approach is to employ an uncertainty framework that embraces a range of possibilities about the nature and extent of future technological change. The point here is that there is still a gap between how economic models depict the process of technological change and what happens in reality. Closing this gap is a challenge for the next generation of models of induced technological change and a requisite for an improved evaluation of the consequences of environmental policies.

NOTES 1.

2.

We exclude from consideration bottom-up energy system models (Seebregts et al., 1999; Edmonds et al., 2000). At the heart of these models is a detailed, typically computer-based, representation of the cost and performance characteristics of technologies in the energy sector, which is taken as the basis for modeling the pattern of technology penetration. They are thus less suited for shedding light on the macroeconomic consequences of environmental policy. While the description we just gave roughly summarizes key elements of the commonly accepted explanation of technological advance, there is a variant perspective. The ‘evolutionary’ approach to technological change, as pioneered by Nelson and Winter (1982), abandons the idea that firms can optimize R&D investment decisions. Because of the large uncertainties surrounding the outcome of R&D investments, firms will face significant difficulties in rationalizing their R&D expenditures. It is rather argued that firms embark on a purposive search for profit that is guided by their ‘routines’ and ‘rules of thumb’ to decide on how much to invest in R&D. The most successful firm is the one which adopts in the most efficient way to the competitive environment. The similarity of this view to the theory of biological evolution is obvious. From the evolutionary perspective, firms may miss opportunities for increased profits simply because they do not look hard enough when they face a competitive business climate. An external policy shock such as a new environmental constraint can therefore provide a stimulus to new search, possibly leading to discovery of

A selective review

3.

4. 5.

6.

7. 8. 9.

10. 11. 12.

13.

97

previously undetected profit opportunities. These new ways of doing may actually be more profitable, leading to the ‘win–win’ outcome that was asserted by Porter and Van der Linde (1995), and whose existence cannot be easily explained within the framework of profit-optimizing firms. While the Goulder and Mathai model is particularly tailored to deal with CO2 emissions, their conclusions apply to a wider context. However, as CO2 and climate change issues provide politically relevant and concrete examples, we will adhere to Goulder and Mathai’s terminology. First (second) derivatives are denoted by single (double) subscripts to function symbols. Dots over function symbols refer to time derivatives. This gives dAt dHt d(–pt) CAA(·) –––– + CAH(·) –––– = ––––––. dk dk dk This picture is of course completely consistent with the static view as espoused in the textbook literature (for example, Endres, 2000). If technological change lowers marginal abatement costs, then the optimal tax required to achieve a target level of abatement is reduced correspondingly. It is assumed that damages are a convex function of changes in the atmospheric CO2 concentration. As obviously WP + WH = (½)A*p, one finds after division by WP = (½)pAP that 1 + WH/WP = A*/AP. Expressing the ratio A*/AP in terms of b by taking advantage of AP = (1 – b)A* then yields equation (2.15). This argument can be illustrated with the help of Figure 2.4. Consider again a competitive industry under a tradable permit system. Assume further that the permit price will fall from p0 to p′ when all firms adopt the new technology. Jung et al. calculated that in this situation the cost-savings of the industry under auctioned permits equal area 0afhb in Figure 2.4, which consists of the reduction in abatement costs (area 0ab) plus the reduction in payments for permits due to the lower permit price (area afhb). However, these aggregate cost-savings are not identical with the incentives for firms to adopt the innovation. To see this, assume that all firms but one have adopted the cleaner technology, so that the permit price is approximately p´. For the last firm, being a price-taker in the permit market, the cost reduction from adopting the new technology is given by area 0db, which is smaller than the aggregate costsavings 0afhb. Thus, under a tradable permit system the incentives for firms to invest in cleaner technology decrease over time when the new technology spreads. In addition, this result holds whether tradable permits are freely allocated or auctioned, because the distribution itself does not affect firms’ decisions. This result echoes Weitzman’s (1974) rule that a price instrument is more efficient than a quantity instrument, if marginal damages are constant and information about marginal abatement costs is asymmetric. In the one-sector model, the relative price of knowledge equals unity because there is only one production technology. It is worth pointing out that in the Smulders model the qualitative difference between completely endogenous growth (g = 1) and only partly endogenous growth (g < 1) is maintained irrespective of how closely g approaches unity. As long as g < 1, environmental quality N does not influence long-term growth, which is determined by exogenous factors. However, environmental quality does have permanent effects on growth when g = 1. In the figure, we have drawn the rN-curve intersecting the r-curve at the upward-sloping part. This is appropriate, as discussed before, for low preference for environmental quality.

REFERENCES Aghion, P. and P. Howitt (1999), Endogenous Growth Theory, 3rd printing, Cambridge, MA, MIT Press.

98

Environmental and resource economics 2004/2005

Arrow, K.J. (1962a), ‘Economic welfare and the allocation of resources for inventions’, in R.R. Nelson (ed.), The Rate and Direction of Inventive Activity, Princeton: Princeton, University Press. Arrow, K.J. (1962b), ‘The economic implications of learning by doing’, Review of Economic Studies, 29, 153–73. Barro, R.J., and X. Sala-i-Martin (1995), Economic Growth, New York: McGraw-Hill. Biglaiser, G., J. Horowitz and J. Quiggin (1995), ‘Dynamic pollution regulation’, Journal of Regulatory Economics, 8, 33–44. Bovenberg, A.L. and R.A. De Mooij (1997), ‘Environmental tax reform and endogenous growth’, Journal of Public Economics, 63, 207–37. Bovenberg, A.L. and S. Smulders (1995), ‘Environmental quality and pollutionaugmenting technological change in a two-sector endogenous growth model’, Journal of Public Economics, 57, 369–91. Bovenberg, A.L. and S. Smulders (1996), ‘Transitional impacts of environmental policy in an endogenous growth model’, International Economic Review, 37, 861–93. Boward, W.L. and A.M.S. Brinkman (1998), ‘Retrofit FGD system price trends and influence factors’, in A.E. McBride and R.W. Porter (eds), Proceedings of the 60th America Power Conference, Chicago: Illinois Institute of Technology, pp. 326–30. Buonanno, P., C. Carraro and M. Galeotti (2003), ‘Endogenous induced technical change and the costs of Kyoto’, Resource and Energy Economics, 25, 11–34. Chiang, A. (1992), Elements of Dynamic Optimization, New York: McGraw-Hill. Clark, C.W. (1990), Mathematical Bioeconomics: The Optimal Management of Renewable Resources, 2nd edn, New York, Wiley. Clarke, L.E. and J.P. Weyant (2002), ‘Modeling induced technological change: an overview’, in: A. Grübler, N. Nakic´enovic´ and W.D. Nordhaus (eds), Technological Change and the Environment, Laxenburg, Austria: International Institute for Applied Systems Analysis. Downing, P.B. and L.J. White (1986), ‘Innovation in pollution control’, Environmental Economics and Management, 13, 18–29. Edmonds, J., J. Roop and M. Scott (2000), Technology and the Economics of Climate Chance Policy, Washington, D.C: Pew Center of Global Climate Change. Endres, A. (2000), Umweltökonomie; Darmstadt: Wissenschaftliche Buchgesellschaft. Fischer, C. (2000), ‘Climate change policy choices and technical innovation’, climate issue brief 20, Washington, D.C: Resources for the Future. Fischer, C., I.W.H. Parry and W.A. Pizer (2003), ‘Instrument choice for environmental protection when technological innovation is endogenous’, Journal of Environmental Economics and Management, 45, 523–45. Goulder, L.H. and K. Mathai (2000), ‘Optimal CO2 abatement in the presence of induced technological change’, Journal of Environmental Economics and Management, 39, 1–38. Goulder, L.H. and S.H. Schneider (1999), ‘Induced technological change and the attractiveness of CO2 abatement policies’, Resource and Energy Economics, 21, 211–53. Gradus, R., and S. Smulders (1993), ‘The trade-off between environmental care and long-term growth – pollution in three prototype growth models’, Journal of Economics, 58, 22–51. Grossman, G.M. and E. Helpman (1997), Innovation and Growth in the Global Economy, Cambridge, MA: MIT Press. Grubb, M., T. Chaphuis and M. Ha-Duong (1996), ‘Technologies, energy systems and

A selective review

99

the timing of CO2 abatement: an overview of economic issues’, in N. Nakic´ enovic´, W. Nordhaus, R. Richels and F. Toth (eds), Climate Change: Integrating Science, Economics, and Policy, IIASA workshop proceedings, Laxenburg, Austria: International Institute for Applied Systems Analysis. Grubb, M., J. Köhler and D. Anderson (2002), ‘Induced technical change in energy and environmental modeling: analytic approaches and policy implications’, Annual Review of Energy and the Environment, 27, 271–308. Jaffe, A.B., R.G. Newell and R.N. Stavins (2002), ‘Environmental policy and technological change’, Environmental and Resource Economics, 22, 41–69. Jung, C., K. Krutilla and R. Boyd (1996), ‘Incentives for advanced pollution abatement technology at the industry level: an evaluation of policy alternatives’, Journal of Environmental Economics and Management, 30, 95–111. Kemp, R. (1997), Environmental Policy and Technical Change, Cheltenham, UK and Lyme, NH: Edward Elgar. Keohane, N.O. (1999), ‘Policy instruments and the diffusion of pollution abatement technology’, discussion paper, Harvard University. Kline, S. and N. Rosenberg (1986), ‘An overview of innovation’, in R. Landua and N. Rosenberg (eds), The Positive Sum Strategy: Harnessing Technology for Economic Growth, Washington, D.C: National Academy Press, Magat, W.A. (1978), ‘Pollution control and technological advance: a dynamic model of the firm’, Journal of Environmental Economics and Management, 5, 1–25. Magat, W.A. (1979), ‘The effects of environmental regulation on innovation’, Law and Contemporary Problems, 43, 3–25. Milliman, S.R. and R. Prince (1989), ‘Firm incentives to promote technological change in pollution control’, Journal of Environmental Economics and Management, 17, 247–65. Milliman, S.R. and R. Prince (1992), ‘Firm incentives to promote technological change in pollution control: reply’, Journal of Environmental Economics and Management, 22, 292–6. Nelson, R.R. and S.G. Winter (1982), An Evolutionary Theory of Economic Change, Cambridge, MA: Harvard University Press. Nordhaus, W.D. and J. Boyer (2000), Warming the World – Economic Model of Global Warming, Cambridge, MA: MIT Press. Nordhaus, W.D. (2002), ‘Modeling induced innovation in climate-change policy’, in A. Grübler, N. Nakic´ enovic´ and W.D. Nordhaus (eds), Technological Change and the Environment, Washington, DC: Resources for the Future Press. Orr, L. (1976), ‘Incentive for innovation as the basis for effluent charge strategy’, American Economic Review, 60, 441–7. Parry, I.W.H (1998), ‘Pollution regulation and the efficiency gains from technological innovation’, Journal of Regulatory Economics, 14, 229–54. Parry, I.W.H., W.A. Pizer and C. Fischer (2002), ‘How large are the welfare gains from technological innovation induced by environmental policies?’, discussion paper 02-57, Washington, DC: Resources for the Future. Forthcoming in Journal of Regulatory Economics. Popp, D. (2002), ‘ENTICE: endogenous technological change in the DICE model of global warming’, discussion paper, Syracuse University, New York. Porter, M.E. and C. Van der Linde (1995), ‘Towards a new conception of the environment–competitiveness relationship’, Journal of Economic Perspectives, 9, 97–118. Requate, T. and W. Unold (2003), ‘Environmental policy incentives to adopt advanced

100

Environmental and resource economics 2004/2005

abatement technology: will the true standing please stand up?’, European Economic Review, 47, 125–46. Romer, D. (1996), Advanced Macroeconomics, New York: McGraw-Hill. Romer, P.M. (1990), ‘Endogenous technological change’, Journal of Political Economy, 98, 71–102. Seebregts, A.J., et al. Policy report ECN BS: ECN-C-99-025 prepared for the Energy Research Centre of the Netherlands. Seebregts, A., T. Kram, G. Schaeffer, A. Stoffer, S. Kypreos, L. Barreto, S. Messner and L. Schrattenholzer (1999), ‘Endogenous technological change in energy systems models: synthesis of experience with ERIS, MARKAL, and MESSAGE’, policy report ECN BS: ECN-C-99-025 prepared for the Energy Research Centre of the Netherlands, Pelton, Netherlands. Smulders, S. (1995), ‘Environmental policy and sustainable economic growth’, De Economist, 143, 163–95. Smulders, S. (1998) ‘Technological change, economic growth, and sustainability’, in J. van den Bergh and M. Hofkes (eds), Theory and Implementation of Economic Models for Sustainable Development, Dordrecht: Kluwer. Smulders, S. (1999), ‘Endogenous growth theory and the environment’, in J.C.J.M. van den Berg (ed.), Handbook of Environmental and Resource Economics, Cheltenham, UK and Northampton, MA: Edward Elgar. Verdier, T. (1993), ‘Environmental pollution and endogenous growth: a comparison between emission standards and technological standards’, working paper 57.93 Fondazione Eni Enrico Mattei, Milan. Weitzman, M.L. (1974), ‘Prices vs. quantities’, Review of Economic Studies, 41, 477–91. Wigley, M.L., R. Richels and J. Edmonds (1996), ‘Economic and environmental choices in the stabilization of atmospheric CO2 concentrations’, Nature, 379, 43 240–43. Wright, B.D. (1983), ‘The economics of invention incentives: patents, prizes, and research contracts’, American Economic Review, 73, 691–707. Zerbe, R.O. (1970), ‘Theoretical efficiency in pollution control’, Western Economic Journal, 8, 364–76.

3. Land use decisions and policy at the intensive and extensive margins Ian W. Hardie, Peter J. Parks and G. Cornelis van Kooten1 1.

BACKGROUND AND INTRODUCTION

There are few economic activities that do not involve land resources in some way. Land is a location for residential, commercial and industrial activities, an input in the production of private and public goods, and a factor in household production of leisure and recreational goods. While many economically relevant market and non-market outputs can be produced by natural landscapes, economic effort is frequently required to convert or ‘develop’ natural land into a condition that produces a desired mix of outputs. Because the quantity of land is finite, the opportunity cost of land development becomes apparent when undeveloped land becomes scarce. Public goods such as habitat for biologically diverse species or terrestrial stores of carbon (that would otherwise be released to the atmosphere to contribute to global warming) can be lost when land use or its cover is changed. The growing scarcity of natural landscapes throughout much of the world, combined with increased awareness of environmental and resource consequences of land use and land cover change, has renewed interest in, and increased research on, the economics of land use. This is partly because influencing land use or land cover is necessary to help solve the resource and environmental problems that face policymakers today. Private landowners’ decisions to change land use can create large spillover effects and environmental externalities at two ‘margins’ in the spectrum of land uses. One of these, to which we refer as the ‘extensive’ margin, occurs when land in a natural landscape (wilderness or nature) is first converted into agriculture or forests that are managed to produce market outputs (rural land). The other, referred to as the ‘intensive’ margin, occurs when land in agricultural or forestry use is converted to residential, commercial or industrial use (urban land). Large environmental and social costs can result in either case, and the market can fail in both situations to internalize these spillover and environmental costs. Thus these margins become ‘hot spots’ of land use 101

102

Environmental and resource economics 2004/2005

policy, as governments and non-governmental organizations seek to limit the divergence of social and private values.2 Although land economics has diminished as a separate field in applied economics, urban economists have remained and resource economists are becoming active in the analysis of land use at the intensive margin. Forest economists have remained active at the extensive margin. Urban economists typically model private land use decisions using variants of an asset valuation formulation attributed to Wicksell, while forest economists typically model land use decisions employing generalizations of Faustmann’s soil rent model. Both formulations are perfect foresight, net present value models, but they treat land use change differently. While a wide spectrum of other models has been used to analyze allocation of land among different uses, the Wicksell and Faustmann formulations are predominantly land use decision models. One of the objectives in this chapter is to review the Wicksell and Faustmann models to highlight their differences and similarities. Since the models serve the same basic purpose, we believe there may be synergies in reviewing both of them. A second objective is to survey and underscore differences and similarities in the policies used to solve resource and environmental problems resulting from land use change at the two margins. A diverse set of policy instruments is used to affect land use or cover (both referred to henceforth as land use) at these two interfaces. Zoning, central planning, regulations and market-based measures are all employed to influence and control land use decisions at the rural–urban interface (intensive margin). At the rural–nature interface (extensive margin), policymakers concerned with land use change have been more willing to let the market determine land use and then to rely on a variety of instruments, but most often regulations or subsidies, to correct for undesired consequences. However, policy prescriptions are evolving at the extensive margin toward the use of zoning as a means to increase social well-being, and privately based codes of forest practice and forest certification are now being implemented to address environmental spillovers and to ensure sustainable forestry. Thus a wider set of policy prescriptions is now being employed at this margin. As a consequence, policy choices are exhibiting a convergence at the two interfaces, with policy instruments showing greater similarity over time. A review of both helps clarify this convergence of land use policy instruments. The choice to focus on land development at the rural–nature and rural–urban interfaces limits the scope of the chapter. In particular, policies and models developed to analyze and protect nature in farming localities are omitted. Governments concerned with these conservation issues have increasingly employed long-term contracts as a way to protect the environment: examples include the Conservation Reserve Program in the United States and the Province of Saskatchewan’s Permanent Cover Program in Canada. Participation is voluntary in these programs, and financial incentives are

Land use decisions and policy

103

provided to get landowners to agree to idle parcels of land or to restrict farming practices. These contracts often call for cropland to be taken out of production and planted to trees or forage. In the United States, farm subsidies have been linked to conservation practices and cross-compliance has made farmers ineligible for agricultural program subsidies if they destroy wetlands or cultivate land that has not previously been producing annual crops. Thus our choice to omit agricultural land use models and associated policies means we do not cover some extensive-margin transfers of land from nature to agricultural use. To this point, these programs are exclusive to the United States, and to avoid substantially increasing the scope of this chapter, we have chosen not to include them.3 The chapter is organized into six sections. In sections 2 and 3, we are concerned with the intensive margin; the von Thünen location rent and Wicksell land use decision models are presented in section 2, while policies used to preserve rural land and to manage growth at the intensive margin are reviewed in section 3. Parallel presentations for the extensive margin are provided in sections 4 and 5. Variants and generalizations of the Faustmann model are developed in section 4, while policies used to protect nature and to ensure sustainable and environmentally sound forestry practices are reviewed in section 5. In the final section (section 6), we synthesize the main points of the chapter and draw some conclusions.

2.

THE URBAN–RURAL INTERFACE: MODELING LAND USE DECISIONS AT THE INTENSIVE MARGIN

The major land use issues at the intensive margin are generated by population increase and the associated formation of new households, which stimulate the conversion of land to residential, industrial and commercial use. Quantity of land converted is sometimes an issue, as jurisdictions seek to slow or limit population growth by regulating land use. But more often concern is focused on the spatial pattern of development. Government programs are implemented to eliminate sprawl, to preserve farmland and open space, to alleviate spillovers between nearby parcels with different uses (farmland next to residential, residential next to industrial), and to lower the costs to government of providing the infrastructure (roads, schools, water and sewers) necessary to support urban land uses. Because spatial organization of land use is of primary concern, policy is often formulated in conjunction with plans defining desired geographical distributions of permitted land uses. These desired land use plans are formulated by agencies within government jurisdictions that are often granted the power to withhold approval of land development projects, to validate compliance with existing laws, to administer fees and to negotiate

104

Environmental and resource economics 2004/2005

changes in development plans before approval. Because of this, spatial planning plays an important role in the implementation of government policy at the intensive margin. The conceptual framework for the economic analysis of land use at the intensive margin has developed primarily from Alonso’s adaptation of the von Thünen location rent model (Fujita, 1996). The model that emerges from this theoretical development is a competitive market model with an equilibrium condition that locates urban land use in a compact set around a ‘central business district’ (CBD). In its simplest form, the equilibrium condition that establishes this location can be expressed as (see Capozza and Helsley, 1989, p. 298): N(t)s¯ = fmb(t)2,

(3.1)

where N(t) is the number of households in the market at time t, ¯s is the (fixed) average lot size or land area consumed by a household, mb(t) is the distance from the CBD to the boundary of the area of urban land use, and f is a parameter. When the left- and right-hand sides of equation (3.1) are in the same units of measurement, f can be interpreted as the constant p (=22/7), and the right-hand side is the area of a circle of radius mb(t) that is centered on the CBD. Thus (3.1) states that, in equilibrium, the area within the circle of urban land use must be just large enough so that everyone is housed. Use of this simple spatial equilibrium condition results in the urban land area growing in annular rings as population and the number of households in the market increase with time t.4 Capozza and Helsley (1989) summarize the primary results concerning land value that come from this simple von Thünen model. They derive a land price gradient that originates from the CBD, decreases monotonically to the rural–urban boundary, drops precipitously at this boundary, and then continues to decrease as land is located even further from the urban center. This theory formalizes the idea of active land conversion at the edges of cities when urban population is increasing (the intensive margin), and it identifies the basic value components of this process. In Capozza and Helsley’s (1989) model, equilibrium land prices for the urban land use within the intensive margin take the form: ∞

A 1 T 1 Pu(t,m) = –– + D + –– –– [mb (t) – m] + –– Rt (t,m)e–r(t–t)dt, r r s¯ rt

∫

m ≤ mb (t). (3.2)

The equilibrium price for land outside of the boundary is:

Land use decisions and policy

105

∞

A 1 pr(t,m) = –– + –– Rt (t,m)e–r(t–t)dt, r r t

∫

m > mb (t).

(3.3)

These equations indicate that the equilibrium price (Pu) of urban land at any time t and distance m from the urban center is the equilibrium price of the rural land (Pr) plus two additional components – the cost D of developing the land (treated in (3.2) as a constant) and a location rent created by the distance of the site inside of the urban boundary mb. The location rent depends on the per-unit distance commuting cost T, the average size of a developed lot s¯, and the discount rate r. Commuting costs increase and location rent decreases as m increases since the site is then located farther from the CBD. Location rent increases over time for any given site if the city grows and the urban boundary moves farther from the CBD. The equilibrium price of rural land is the capitalized value of its agricultural rent A/r, plus a development premium. The development premium is the capitalized value of the future increases in rent R of urban land that will occur as the city grows (the subscript indicates a partial derivative with respect to time t). Thus the effect of urbanization extends beyond the rural–urban boundary if population increases and the urban area grows, for then the land market capitalizes the value of future development into the current price of rural land. This premium has sometimes been omitted in the analysis of farmland prices in the agricultural economics literature, which has traditionally defined the price of farmland as the capitalized rent that accrues from farming (Clark et al., 1993; Just and Miranowski, 1993). But it has also been made a central component of the analysis of rural land values (Shi et al., 1997; Hardie et al., 2001). Two observations of immediate interest emanate from this urban growth model. One is the dominance of urban land use. According to this model, quantity of urban land depends upon the number of households, and population growth is exogenous to the land market. If additional demanders of urban land services enter the market, their willingness to pay for land is high enough that they can claim however much rural land they desire. Thus land is guaranteed to remain in rural or natural uses only if government sees fit to ensure that use. While much of government policy affecting land use at the intensive margin is concerned with the preservation of rural land, this policy emphasis does not always dominate. Maintaining rural land within the urban boundary will expand the size of the city and increase the costs of providing schools, roads, sewers and other infrastructure. Policy may be concerned with this aspect of urban growth as well as with the social benefits provided by open space within an urban area. A second immediate implication of the model is the consequences of the land development premium for farming. As noted by Lopez et al. (1988), land

106

Environmental and resource economics 2004/2005

next to the rural–urban boundary takes on the characteristics of an appreciating financial asset that can be held for speculative purposes. Farmers’ planning horizons can be shortened by the prospect of selling their land and this can lead to a reluctance to maintain and replace farm machinery, drainage systems and other farm infrastructure. In addition to this ‘impermanence syndrome,’ higher land values also result in a change in the character of farming. ‘Traditional’ farms, meaning enterprises constructed similarly to farms in non-metropolitan regions, incur higher costs without increased revenues, while ‘adaptive’ farms, including pick-your-own, Christmas tree, nursery and other specialty agricultural enterprises, obtain per-acre returns of as much as 7.5 times that of traditional farms (Heimlich and Barnard, 1992). ‘Recreational’ farms also emerge, with smaller acreages and very little market output per dollar of expenditure on each acre. Rural landowners become a mixture of farmers, land speculators and lifestyle consumers, with different desires for policies that protect the right to farm, regulate farming, and preserve farmland through subsidies. The amenity value of having open space in an urban area and the spillovers from farming (noise, smell, chemicals) also encourage urban land consumers to enter these policy debates. As a consequence, policy development becomes a complex endeavor, with desires for the preservation of farmland conflicting with desires to limit the negative externalities from farming, and desires for more open space and larger land units conflicting with desires for more efficient infrastructure and lower property taxes (Gardner, 1994). Forcing all land use change into annular rings imposes obvious limitations on the analysis of land use decisions. So does the assumption that all rural land is homogeneous in quality and has the same value in agricultural use. This suggests modification of the basic von Thünen urban growth model, both to remove the spatial equilibrium condition that confines all land use change to the urban boundary and to add land quality attributes. When this is done, we are left with a supply relationship that Anderson (1993) traces back to Wicksell, a demand function, and a competitive market equilibrium condition. The supply relationship can be expressed as: t

∫

P(m,s,t,h;z) = e–rt A(t,m;z)dt + ∞

e–rt ––– s

[∫ t

0

]

e–r(t–t) R(s,t,m,h;z)dt – D(s,h;z) .

(3.4)

This is Fujita’s (1982) specification, augmented by an exogenous vector z of immutable land characteristics (floodplain, wetland, topography) and a vector h of ‘improvements’ (house, landscaping, parking areas).5

Land use decisions and policy

107

Equation (3.4) gives the net present value for an acre (or hectare) of land, obtained from the premise that land located at distance m from the CBD will earn agricultural land rents until time t, be instantly developed at cost D, and thereafter yield rents R from the urban land use. But (3.4) can also be thought of as a land developer’s profit maximization problem in which the per-acre land value P can be maximized by choosing the optimum time t to develop, the optimum lot size s to provide, and the optimum level of improvements h (Hardie and Nickerson, 2003). This supply interpretation assumes that all owners are potential producers of urban land (though they may rent the land to themselves after development), there is perfect foresight, and housing is durable (since the model does not inlcude replacement costs). This model applies to land at any date, since the choice of t is endogenous, and at any distance from the urban center, since t may be far in the future. It also allows for the possibility that land of different quality may develop at different distances at the same date. If increases in population cause bid rents for housing to increase faster than agricultural rents over time, and if landowners maximize land values, then optimization implies that an urban lot with location m and characteristics z will be produced whenever returns just cover the costs of development: R(s,t,h;m,z) = sA(t;m,z) + rD(s,h;z).

(3.5)

This first-order condition, obtained from maximizing P(m,s,t,h;z) with respect to t, is a consequence of the eventual dominance of the rent for urban land. If, in addition, land developers correctly assess consumers’ maximum willingness to pay for urban lots, so that R in equation (3.4) is the homebuyers’ bid rent function, then the Wicksell formulation contains the price of an urban lot (PL) at time t,6 ∞

∫

pL(t,h;m,z) = e–r(t–t) R(s,t,h;m,z)dt.

(3.6)

t

Substitution of (3.5) into (3.6) and evaluation of the integral yields sA(t;m,z) pL(t,h;m,z) – ––––––––———= D(s,h;z). r

(3.7)

The supply relationship expressed in equation (3.4), when combined with the ‘rational expectations’ assumption that land developers can correctly assess consumers’ willingness to pay for urban lots, embodies the competitive

108

Environmental and resource economics 2004/2005

market equilibrium condition that the difference in the prices of rural and urban land is just equal to the cost of development.7 This equilibrium condition is independent of the spatial equilibrium condition (3.1) and it describes a well-functioning competitive land market at the intensive margin.

3.

POLICIES TO ADDRESS MARKET FAILURES AT THE INTENSIVE MARGIN

If a well-functioning land market exists, the question underlying policy intervention becomes: ‘Where is the market failure?’ Brueckner’s (2000) answer consists of three parts: (1) the social value of open space is not accounted for in development costs; (2) congestion externalities are not internalized and people live farther away than they would if these costs were incorporated; and (3) the impact of new development on infrastructure is not fully incorporated into the developed land price. Brueckner omits the possible failure of developers to internalize the effect of their developments on the environment, and he takes the assumptions of the urban land market model as given. In particular, the model’s perfect foresight assumption eliminates the NIMBY (‘not in my back yard’) response to land use change (Dear, 1992). Since homeowners are assumed to have full knowledge of future spillovers from nearby land uses at the time of purchase, willingness to pay (cf. equation 3.6) for an urban lot would adjust for (known) future spillovers, and, when those expected spillovers occur, the homeowner would not react. This model does not allow for the possibility that a person might purchase a home site with the intention to lobby for removal of a spillover. Nor does it incorporate positive costs of moving between houses. One may certainly question a model that treats these cases as one and the same and, further, assumes that households are able to move between houses without cost at any time. However, these assumptions are needed to construct a well-defined bid rent function. One might also question the assumption that all landowners have the same discount rate, for if that were true, farmers nearing retirement age would be no more likely to sell land for development than young farmers just beginning their enterprise. 3.1

Preserving Rural Land

Although not part of the model, the NIMBY response of suburban homeowners to noise, odor and other spillovers from farming has led to right-to-farm legislation in most states within the United States and in some provinces in Canada. This legislation seeks to supersede the common law of nuisance by granting farmers who employ acceptable farming practices exemptions from litigation that seeks to restrict their farming operations (Lapping and

Land use decisions and policy

109

Leutwiler, 1987). Its goal is the protection of farming rather than the preservation of farmland, although right-to-farm laws do sometimes provide incentives to keep land in agriculture. The original New Jersey law, for example, states that farmers in an official farmland preservation program are to benefit from an ‘irrebuttable’ presumption that their operation does not constitute a nuisance, while farmers not in a preservation program are to benefit from a ‘rebuttable’ presumption (Lisansky and Clark, 1987). Right-to-farm laws are sometimes integrated with laws providing for the formation of agricultural districts. New York’s agricultural district law, for example, packages right-to-farm provisions, protection from public regulations that might restrict farming practices, stringent review of eminent domain takings and restrictions on the provision of infrastructure for urban uses with the granting of relief from property tax assessments (Boisvert and Bills, 1984; Bills and Boisvert, 1987). The objective of this legislation is to keep commercial farming viable in areas near urban centers, maintain a critical mass of farm supply and farm market facilities, and indirectly conserve and protect agricultural lands. Protection is indirect because these laws do not regulate the sale or conversion of rural land, and conversion can take place whenever the premium for doing so becomes sufficient. In the context of the market model, the rightto-farm and agricultural district laws seek to delay development by increasing the agricultural rents, A(t,m;z). But these rents depend on time and development pressure, and whether the laws have affected them enough to successfully delay development is still an open question. Tax incentives designed to give farmers a break from having to pay high property taxes are widespread policies developed to maintain farming in areas where farmland has a large development premium. These policies can extend to land speculators and recreational farm owners if they are classified as farmers, even though these owners may not contribute much to farming activity in the region. Tax policies are not, by themselves, capable of compensating rural landowners for providing a public good (open space or farmland protection) at private expense, and they generally do not deter landowners from selling out for development when the opportunity arises (Heimlich and Anderson, 2001). Indeed, the various tax policies that have been designed to preserve farming may actually have the opposite effect: they may increase the area affected by lack of investment in agriculture and, by reducing the costs of land speculation, may provide incentives that encourage urban sprawl (Barichello et al., 1995; Corbett, 1990; Ervin et al., 1977).8 As evidence has accumulated that preferential tax assessments do more to subsidize farmland owners than to conserve farmland, governments have increasingly initiated programs to purchase development rights and conservation easements (Wiebe et al., 1996). These programs involve separating and purchasing some but not all of an owner’s rights to a property: separated rights

110

Environmental and resource economics 2004/2005

might include, for example, the right to build residential or commercial buildings, to drain sloughs, to burn associated uplands, or to remove endangered species of trees. In the United States, most purchases have been in the form of agricultural conservation easements that restrict residential, commercial or industrial uses, but that allow active farming. These are targeted at prime farmland and have the goal of permanent preservation of farmland in urbanizing areas (Heimlich and Anderson, 2001).9 Significant impetus for these programs has come from ballot referendums that show taxpayers are willing to pay for the provision of open space with bond issues. In November 2001, for example, voters passed 85 of 115 state and local open space spending measures, which provided for more than $1.2 billion in public funds for open space protection efforts (Hollis and Fulton, 2002). This suggests that Brueckner’s identification is correct and that the market for land at the intensive margin is failing to incorporate completely the social value of open space. In Europe, where price support and other agricultural production subsidies have maintained land in agriculture and where supranational policies are now dictating reduced farmland cultivation, emphasis has changed from preserving farmland to preserving countryside (Alterman, 1997). This has implications for open space benefits when farmers are concerned about trespass. Actively farmed land is not likely to provide the same environmental or amenity benefits as, say, a regional park providing recreational access or a greenbelt providing natural habitat. Such limitation on access is more of a concern in the United States than, say, in Britain or Sweden, where rights to traverse farmland are well established. Perhaps because of this limitation, non-profit private land trusts, such as the Nature Conservancy, the Conservation Fund and the Trust for Public Land, have become active in the preservation of open space within the United States. These organizations purchase properties or easements on lands that provide environmental benefits and seek to protect land slated for urban development. Purchased land is often turned over to state and local governments. This establishes a means for the governments to provide some of the social values that are not obtained from farmland with limited access. 3.2

Managing Growth

An important difference between preferential tax assessments and purchase of development rights is the potential role of planning. Preferential tax assessments typically are extended to all eligible landowners regardless of the location of their property. But purchases can be targeted to sites where social or environmental benefits are deemed to be particularly high, such as along a wildlife corridor or within a densely populated area. While the potential for targeting exists, it generally is not realized. Hollis and Fulton (2002) find few cases where agencies and organizations concerned with open space coordinate

Land use decisions and policy

111

with agencies developing zoning and land use plans. The agencies concerned with land use planning are more likely to rely on zoning regulations, impact fees and infrastructure development to obtain their desired landscapes.10 Transfer of development rights Transferable development rights and wetlands mitigation banking are exceptions to this dichotomy. These constitute cases where separation of development rights have been integrated with land use planning. Wetlands mitigation banking (WMB) allows landowners to develop wetland sites on their property if they have sufficient credits from investment in the completed rehabilitation of a WMB site. Land use planning enters this program through the designation of the WMB sites (Fernandez and Karp, 1998). Sites can be chosen that provide large high-quality habitats with superior potential to sustain desired ecosystems. Given good choices, the investments in the WMB can provide greater community-wide environmental benefits than equivalent investments in the maintenance of wetlands on sites that are being developed. Good planning is crucial to obtain higher benefits, because WMB is a ‘no net loss’ program that links area restored to wetland area removed by development. In contrast, a program such as the US Farm Bill’s Wetland Reserve Program allows only for on-site mitigation and removes the planning element from the policy (Parks and Kramer, 1995). Zoning-based transferable development right (TDR) programs are initiated with a partial taking of property rights by government in source areas where development is to be ‘down-zoned’ (Johnston and Madison, 1997). Acreage that initially might have been zoned to allow one dwelling unit per acre, for example, may be down-zoned to allow only one dwelling unit per 20 or 50 acres. When the down-zoning occurs, landowners in the affected source areas are granted the option to sell the separated development rights to land developers in designated development areas (sinks). Owners of property in these sinks or target areas can purchase the transferable development rights and use them to gain variances from lot size and per-acre dwelling unit zoning restrictions. Thus a developer in a target area where development is restricted to one house per acre may, for example, obtain a variance that allows production of two dwelling units per acre. Landowners who lose property rights consequently are compensated in a development rights market, but at rates driven by the opportunity costs created by zoning instead of by consumer willingness to pay for urban land. Prices of the restricted land also can increase above agricultural use values if the demand for recreational farming sites is sufficient (Nickerson and Lynch, 1999). Governments incur the costs of planning and administration of these TDR programs, but can reduce infrastructure costs by providing fewer services in the source areas and more efficient services in the sinks. This appears to be a primary motivation for these programs.

112

Environmental and resource economics 2004/2005

Zoning and planning Zoning regulations can be developed to reduce negative externalities that some land uses impose on others, to meet fiscal goals such as increasing a local tax base or lowering infrastructure costs, or to ‘maintain the homogeneity of exclusive residential districts’ (Rolleston, 1987). Large lot zoning (typically minimum lot sizes of three or more acres) has been enacted to control growth, but has been found to be ineffective for this purpose: urban land use remains the highest-valued alternative and the number of new homes built does not decrease enough to compensate for the greater amount of land devoted to each dwelling unit (Coughlin and Keene, 1981; Heimlich and Anderson, 2001). Exclusive-use zoning and urban growth boundaries have been successful at restricting farmland conversion in Oregon, Hawaii, Saskatchewan and Quebec, although they have recently been overwhelmed in Israel by increased population and housing pressures (Nelson, 1992; Alterman 1997). Such strict controls can affect income distribution and lower social welfare by constraining the demand for low-density development. Fuguitt and Brown (1990) provide some evidence of this demand: they have found that 70 per cent of Americans would prefer to live in a rural setting within 30 miles of a city of at least 50 000 people. Despite this stated preference, citizens in Oregon and Hawaii have continued to support exclusive-use zoning. Wallace (1988) and McMillan and McDonald (1991) find that variances and revisions can cause zoning other than exclusive-use zoning to ‘follow the market.’11 If so, these regulations may serve more to guide development as it takes place (‘growth management’) than to control the amount of development located in a particular jurisdiction. The initial precept of non-exclusive-use zoning was to separate incompatible land uses and to abate negative externalities. This was expected to raise land values, but this effect has been difficult to verify empirically (Fishel, 1990; McMillan and McDonald, 1993). The idea that zoning should be used to establish homogeneous use is now under challenge by planners who espouse a ‘new urbanism’ of mixed-use zoning with intermingled residential, commercial and industrial land uses (Grant, 2002). Zoning as a tool for growth management has seen less application in European countries such as Britain, The Netherlands and France. There growth is typically managed by direct review of proposed development of specific sites for adherence to government plans. Compensation is not paid when development of rural land is refused, and ‘takings’ are not the issue that they are in the United States.12 A review by Alterman (1997) finds that the citizens of these countries are more committed to urban containment, higherdensity living and the preservation of countryside. This makes planning-based control easier to implement and less likely to lower social welfare. Zoning for fiscal reasons includes the development of industrial and business parks to increase the tax base and the management of residential growth

Land use decisions and policy

113

to reduce infrastructure costs (Cameron and Stephenson, 2002). Management to reduce infrastructure costs almost always involves the containment of sprawl: in a review of studies of the cost of sprawl, Burchell et al. (1998) found that compact development reduced infrastructure costs by 5 to 25 per cent and operating costs of municipalities and schools by 2 to 5 per cent. Provision of infrastructure also can be used as an instrument to affect the location of development. Maryland’s ‘Smart Growth’ program designates priorityfunding areas for which infrastructure development is targeted. The objective is to influence the location of urban land development by increasing the value of the site within the priority area.13 Infrastructure policy can also be used to discourage development at chosen locations: local jurisdictions can place moratoriums on the development of urban land in designated areas until adequate school or sewer capacity is provided, and developers of land in areas where infrastructure is lacking can be charged impact fees to pay for additional infrastructure that would be required to service new development. Thus zoning is only one of several policy instruments that can be used to manage growth in a rural–urban area. Instruments other than zoning, particularly the provision of public transportation and the development of roads, may be the most important tools for growth management, particularly if zoning does follow the market. Brueckner’s congestion externalities depend on a failure of commuters to recognize the social costs created by their use of existing roads. But the provision of new roads or light-rail transit can direct the location of new development, and the development of the transportation network can have an important influence on individuals’ housing decisions (Smith et al., 2002). Citizens experiencing congestion effects can also demand new and improved roads (Heimlich and Anderson, 2001). Public transportation in the form of trains and buses can serve as a good or poor substitute for automobile travel – a key difference between some of the European countries and the United States. Because several instruments are involved, growth management may incur greater costs of coordination and implementation across government jurisdictions and agencies than will more specific policies aimed at the correction of particular market failures. The question of whether planning and infrastructure policy will manage growth better than market forces has not been resolved. Unfettered land use change at the intensive margin is a complicated dynamic process. In addition to changes in the character of farming, spatially discontinuous development will naturally occur as households with different preferences and incomes trade off the cost of commuting against the benefits of living in the country. Large lot housing dependent on wells and septic systems will develop at the outer edges of the metropolitan area. Then, as land values increase, more dense development will occur on nearby parcels. Demands for infrastructure

114

Environmental and resource economics 2004/2005

will increase and the eventual provision of infrastructure will lead to further development. Businesses will develop to be near customers and labor sources when urban densities become sufficient. Peiser (1989) finds that discontinuous development in a freely functioning land market can lead to higher final housing densities as parcels left undeveloped in the short run increase in value and eventually convert to more intensive use. The rub, however, is in the intermediate period when government can intervene in the market, alter where and how development takes place, lower its costs of providing infrastructure and provide social benefits by preserving farming and open space. Because of this, changes in land use at the intensive margin can be expected to depend both on the functioning of the land market and on public policy.

4.

MODELING LAND USE CONVERSION AT THE EXTENSIVE MARGIN

Change at the extensive margin is mainly between land used in forestry or agriculture versus its use in nature. At a specific point in time, there may exist marginal agricultural land that, from the perspective of society, would be better left as nature (for example wetlands, natural range), or forest that would be better left unharvested. In practice, the social allocation of land use at the extensive margin relies to a much greater extent on market policies than is the case at the intensive margin. While it is straightforward to model policies for dealing with the divergence between privately optimal and socially optimal land use decisions when land use activities are spatially separable, difficulties arise when activities on one parcel of land can affect the optimal decisions to be taken on other parcels, whether these are adjacent or not. Although spatial dependence is also present at the intensive margin, this dependence is frequently taken into consideration through zoning ordinances and land use planning. While zoning may adjust spatial flows of land benefits within the intensive margin, zoning cannot guarantee that a land allocation is economically efficient. Thinking about land use changes at the extensive margin can be facilitated by developing an economic model of the land use choices for forest and agricultural land available at this margin. The model presented here is similar to the Faustmann model (Faustmann, 1849), expanded to consider partial harvests, carbon and other non-timber benefits, the influence of adjacent or separate stands on the production of benefits other than timber, and potential conversion to non-forest use. The Faustmann model was developed to determine optimum harvest cycles (rotations) for a single stand that is permanently dedicated to timber production. Among the many changes to the basic model, relevant extensions that have been considered include: (1) the influence of

Land use decisions and policy

115

non-timber forest amenities whose production increases monotonically with age – convex joint production with respect to stand age (Hartman, 1976); (2) the influence of non-timber forest amenities whose production does not increase monotonically with age – non-convex joint production (Swallow et al., 1990); (3) the influence of neighboring stands (Swallow and Wear, 1993); and (4) the possible conversion of land to non-forest use (McConnell et al., 1983; Parks et al., 1998). Carbon stored in forestland is treated as an amenity that increases monotonically with stand age, and is addressed using a method developed by van Kooten et al. (1995). The influence of neighboring stands is addressed through a non-timber environmental benefits function. A forest is presumed to comprise several stands, and aggregate environmental benefits are a non-separable function of activities chosen on each of the component stands. Stands can be adjacent (Krcmar and van Kooten, 2002; Koskela and Ollikainen, 1999, 2002; Swallow and Wear, 1993; Swallow et al., 1997), or separated by distance (Bogdanski et al., 2002). When economic benefits are a linear function of the amount harvested, clear felling at an optimum finite rotation length maximizes net benefits. Introducing non-separable forest-level environmental amenity benefits constitutes an opportunity cost of stand-level harvest that can make partial harvests part of an optimum solution. Conversion to non-forest use (for example, agriculture) occurs when the present value of continued forest benefits no longer exceeds the opportunity cost of agricultural use (McConnell et al., 1983). Corner solutions lead to zero or infinite length optimum rotations, respectively implying immediate harvest by clear felling or permanent preservation for non-timber amenity production. Addition of distance and quality considerations allow for a rural landscape that includes non-contiguous areas of different land uses (Parks et al., 1998). The objective of the modified Faustmann model is to choose harvest dates and harvest amounts (including thereby the amount of inventory left standing) that maximize the net discounted social benefits from two forest stands with different owners. The stands may or may not be adjacent to one another. Benefits from forest use are derived from timber harvest, carbon storage and other environmental benefits (which may be related to timber left standing). To maintain focus on forest use and provision of nature, forest and nature benefits are assumed to exceed those of other uses such as agriculture. This is a simplification of convenience that emphasizes a forest solution to the rural land allocation problem. For each parcel, the economic decision is how much volume, if any, to harvest from the stand. Harvest choices are constrained to the interval between no harvest (benefits only from nature) and clear felling of the stand; any timber left standing contributes in various ways to environmental amenities. Time is also an implicit or explicit decision variable because, whenever the harvest amount is optimally greater than

116

Environmental and resource economics 2004/2005

zero, the timber must be of a positive age (usually a minimum of 15 years old). Choice of the level and timing of timber harvest on a stand is also a choice of the level of inventory (volume) to be left on the stand. Unless the stand is mature and additional growth is effectively offset by decay, delaying harvest allows the remaining timber to grow and permits a larger future harvest. More importantly, standing timber has non-market (nature) value, while growth provides benefits by removing CO2 from the atmosphere. When non-market and carbon values are appropriately taken into account, it still may not pay to grow trees for timber production. It is possible that the land has little or no value in agriculture or in commercial timber production, and is best suited to produce only non-market (for example biodiversity, water storage) and carbon storage amenities. The modified Faustmann model captures all these possibilities. Assume that there are two stands, denoted 1 and 2, and suppose that choice is unaffected by other consumption and investment decisions; this makes income and benefits derived from forest activities separable from non-forest related activities. We express annual total net benefits at any given time as: B(h1(t), h2(t), v1(t), v2(t), v˙ (t)) = F1(h1(t)) + F2(h2(t)) + C(v˙1(t), h1(t)) + C(v˙2(t), h2(t)) + E(v1(t), v2(t)).

(3.8)

In expression (3.8), vi(t) refers to the volume of timber in stand i growing at time t, with v˙ (t) = ∂v(t)/∂t = v˙1(t) + v˙2(t) ≥ 0, v˙ i(t) ≥ 0, i = 1,2; h1(t) and h2(t) are quantities harvested from each stand, which are also functions of time. (To minimize notation, time arguments for volume and harvest functions will henceforth be suppressed, unless required for purposes of clarity.) Total value is the sum of timber, carbon and environmental values. Timber value, Fi(hi), i = 1,2, is a function of harvest, and may vary with the quality of the timber grown on the stand. Carbon value, C(v˙i, hi), is the change in carbon stored in stand type i = 1,2. Increases in standing timber volume increase stored carbon (∂C/∂ v˙i > 0 since v˙i ≥ 0), while harvests of timber release carbon (∂C/∂hi < 0). Environmental value, E(v1, v2), includes non-carbon environmental amenities, and these depend on timber volume in each of the two stands. Under this specification, spatial interaction occurs through the environmental benefit function, where it is assumed that ∂E/∂vi > 0, ∂2E/∂vi2 < 0, and ∂2E/∂v1∂v2 ≠ 0. Commercial timber benefits are given by Fi(hi) = (p – c)hi – R, where p is the price received for logs (in $ per m3), c is the harvest cost (in $ per m3), and R is a fixed cost associated with harvest. Fixed costs R may include a legislative requirement to regenerate harvested sites. For convenience, p and c are assumed not to vary with log size and fixed costs R will be incorporated into the harvest cost c. Note that, under these conditions, timber

Land use decisions and policy

117

benefit is linear in harvest, so that optimal harvest at any time hi*(t) would be either to clear-fell the available volume of timber at that time, vi(t), or not harvest at all. Optimal partial harvests, 0 < hi*(t) < vi(t), require that opportunity costs such as ∂2E/∂v1∂v2 be considered. Assume that the growth function, gi (vi), is known and has the properties that ∂gi/∂vi ≥ 0 and ∂2gi/∂vi2 < 0, ∀vi, and that growth differs between stands. This enables us to consider the difference between, say, growing hybrid poplar on marginal agricultural or denuded forestland versus growing native species. It also enables us to model different non-carbon environmental benefits on each stand, and different interaction effects between the stands. This is discussed further below. The objectives of the landowners and society can diverge. Each landowner’s objective is to Maximize

T

∫ F (h )e i

hi

i

–rt

dt,

(3.9)

0

where T is the length of the harvest cycle, with T possibly being infinite. Each owner solves what amounts to a Faustmann rotation problem, given the linearity in Fi(hi) = (p – c – R)hi (Samuelson, 1976). In contrast, the objective of an authority responsible for all benefits from both stands is to Maximize

T

∫ {F (h ) + F (h ) + C(v˙ , h ) + C(v˙ , h ) + E(v , v )}e 1

h1h2

1

2

2

1

1

2

2

0

1

2

–rtdt,

(3.10)

perhaps subject to a constraint on the public expenditure required to bring about convergence between private and social objectives. The rate r used to discount benefits need not be the same for the private landowners and the authority, but, for simplicity, we assume they are the same. Adding non-carbon environmental benefits E(v1, v2) that are not separable – E(v1, v2) ≠ E(v1) + E(v2) – creates spatial interdependence among stands, and provides an analytical basis for optimal partial harvests not present in Faustmann- and Hartmantype problems. Physical relationships require that constraints be placed on problems (3.9) and (3.10). These constraints are: hi(t) ≤ vi(t), i = 1,2

(3.11)

v˙i(t) = gi(vi(t)) – hi, i = 1,2.

(3.12)

118

Environmental and resource economics 2004/2005

In addition, the following conditions ensure that optimal solutions to problems (3.9) and (3.10) have land in forest and not in its next best alternative (agriculture): T

∫ F (h **)e i

–rtdt

i

0

Ai ≥ ––– (1 – e–rT) – Mi, i = 1,2 r

(3.13)

T

∫ {F (h 1

0

*)

1

+ F2(h2*) + C(v˙1, h1*) + C(v˙2, h2*) + E(v1*, v2*)}e–rtdt (3.14) A1 + A2 ≥ —————(1 – e–rT) – M1 – M2 r

where Ai refers to the (constant-over-time) annual returns to agriculture for stand i, Mi denotes the cost of converting stand i to agriculture, and superscripts ** and * denote optimal private and social choices, respectively.14 It is reasonable to expect that E(v1, v2) > 0 and that C(v˙i, hi) > 0, especially if storage of carbon in wood product sinks or biomass burning in place of fossil fuels are taken into account. This means that society might optimally prefer to keep land in forestry rather than agriculture, even though the landowner might wish optimally to convert forestland to agriculture. One means to prevent conversion of forest into agriculture is to reduce the net benefits of conversion, T

Ai ––– (1 – e–rt) – Mi – r

∫

Fi(hi*)e–rtdt.

0

This can be accomplished, for example, by providing a subsidy to landowners who retain land in forestry, by taxing those who convert land, or by removing incentives, such as subsidized loans or special tax breaks, related to agricultural land ‘improvements.’ Alternatively, the authority can lower Ai by discontinuing output or input subsidies (including farm credit subsidies that reduce r), or by taxing agricultural commodities. We discuss some programs that aim to do this in the next section. Under the circumstances specified by objective functions (3.9) and (3.10), the problem facing the authority is to design incentives that motivate landowners to take into account the external effects of their harvesting decisions. These incentives should motivate landowners to make optimal land use choices

Land use decisions and policy

119

consistent with (3.10) rather than (3.9). An intermediate step between (3.9) and (3.10) is to consider carbon only, ignoring non-carbon environmental amenities and possible stand interdependencies. 4.1

Internalizing Carbon Fluxes from Land Use

The problem of carbon fluxes can be addressed by incorporating a price for carbon into the private landowners’ decision calculus. That is, landowners can be compensated for carbon uptake and taxed for carbon released at time of harvest (see van Kooten et al., 1999). If a market for carbon credits from sinks can be established, and this is not without problems (see Marland et al., 2001), such a market could take the place of carbon taxes and subsidies. Whether taxes or credits are used, the decision maker would face a price of pc ($ per tonne carbon, or $ per tC) for any carbon sequestered or a cost of –pc for any carbon released as a result of harvest. Suppose that the forestland owner chooses to harvest trees at the time that leads to the maximum financial benefit. Suppose also that the time horizon is infinite (T→∞) and that the optimal harvest strategy is to clear-fell a site at equal periodic intervals corresponding to t*, so that hi* = vi(t*). If carbon uptake benefits are taken into account, the net benefit at time of harvest equals [(p – c) – pca(1 – b)]vi (t*), where (as discussed above) p is the price of timber and c harvesting cost, a is a parameter that converts timber into carbon, and b is the proportion of carbon stored in products or landfills after harvest. Then, assuming stands are independent (so they can be treated as a single stand) and excluding E(v1, v2), problem (3.10) can be redefined as one of finding t* that maximizes the present value of the timber plus carbon sequestration benefits over all future rotations. The net present value is (van Kooten et al., 1995): t

pca(bv(t)e–rt + r ∫ v(s)e–rsds) 0 PV = ––––––––––––––––––––––––—— —————— 1 – e–rt

[

]

Carbon

(p – c)v(t)e–rt + ———————— 1 – e–rt

[

]

, Timber

(3.15) where the first and second terms on the right-hand side of the expression refer to the net discounted values of the carbon and commercial timber benefits, respectively. The optimum rotation length t* can be found by setting the first derivative of expression (3.15) with respect to t equal to zero and solving for t*. The resulting expression is quite complex, but, upon setting pc = 0, one gets the usual Faustmann (financial) rotation age. The important point here is that the authority can delay private harvesting decisions by increasing the carbon subsidy/tax, pc. Indeed, harvests might even be delayed infinitely if carbon

120

Environmental and resource economics 2004/2005

prices are sufficiently high and b sufficiently low. Notice also that, if the trees have no commercial timber value (p = 0) but only value in sequestering and storing carbon, it may still be optimal to harvest the stand periodically, but only if not all of the stored carbon is released at the time of harvest (b ≠ 0) and costs of harvesting are not prohibitive. 4.2

Including Environmental Amenities: Land Use Interdependencies

Until now we have not considered non-carbon environmental amenities and the possibility that harvest activities on one site will affect environmental amenities, and thus the socially optimal harvest strategy, on other sites. But such effects can matter. For example, wetlands are likely to have much more value in a region characterized by fence-row-to-fence-row cultivation (a monotonous and visually unappealing landscape) than in an area already containing similar habitat and landscape variety. Likewise, tree plantations with native species can have greater amenity value in areas planted to hybrid poplar than in regions where native species are abundant.

γ (t) IV

III

II

I 0

t Stand age

Figure 3.1 Relationship between stand age and amenity value, various amenities

Land use decisions and policy

121

The spatial problem in forestry is that many non-timber values (for example wilderness preservation, provision of forage for wild ungulates, wildlife habitat) vary in different ways with forest age. Some are even unrelated to forest age (see Calish et al., 1978; Bowes and Krutilla, 1989). Examples of the relationship between a forest stand’s age and non-timber amenities are provided in Figure 3.1. Benefit stream I, for example, might represent the value of wildlife species (for example herbivores) adapted to young forests with plentiful forage; wildlife values that are independent of forest age would be represented by II; and III would represent the value of species reliant on more mature forests, such as trout and spotted owls. Benefit stream IV represents the sum of these amenity flows (Swallow et al., 1990). When non-timber values are combined with commercial timber benefits, solving for the rotation length that maximizes total net (timber plus nontimber) benefits is more complicated. Rather than finding a global maximum, one often discovers local minima or local maxima upon solving the first-order conditions for an optimal solution. For total benefit stream IV in Figure 3.1, for example, the same marginal conditions that define a local maximum also apply to a local minimum. Neither of these is a global maximum. Thus, when total benefits are not convex with respect to time, policies based on first-order conditions could lead to poor policy prescriptions. This policy-relevant non-convexity can prevent a tax or subsidy policy from achieving the socially desirable harvest levels or rotation age. The reason can also be illustrated using marginal benefits and costs associated with rotation length (Figure 3.2). First-order optimization conditions would indicate that a socially optimal (internal) solution for a finite timber harvest would occur when the marginal opportunity cost of delaying harvest (MOC) equals the discounted marginal benefit of delay (MBD). Suppose that the private, financial rotation age is given by t*. Points 1 and 2, with accompanying rotation ages t1 and t2, represent cases where MBD = MOC is satisfied. Because MOC intersects MBD from above, a non-convexity occurs at 1: second-order optimization conditions are not satisfied and the rotation age is not optimal. At 2, the second-order conditions are satisfied and the rotation is socially optimal. Providing the forest manager with a myopic subsidy equal to the value of the non-timber benefits will result in a rotation age t1 that is shorter than the financial rotation age t*. This subsidy to get the manager to take into account environmental spillovers results in a too-soon harvest, and reduces non-market values compared to a later harvest at t*. However, a policy that rewards the forest owner with an appropriate subsidy only over the period t1≤ t ≤ t2 will achieve the desired social optimum. For some standing forests the external amenity benefits might be so great that it would not be socially optimal to harvest the forest at any time in the future. This would be the case if equilibrium points existed at ages beyond

122

Environmental and resource economics 2004/2005

$

MOC

1

0

MBD

2

t1

t*

t2

t

Rotation Age Figure 3.2

Non-convexities and optimal Hartman–Faustmann rotation age

those indicated in Figure 3.2 (for example MOC and MBD may intersect again at some future time in such a way that MOC is upward and MBD downward sloping). If this is the case and society inherits ‘ancient’ forests, it may be worthwhile delaying harvests indefinitely. Finally, consider our problem (3.10), where environmental amenities at one site are affected by harvest decisions at another site. For convenience, we ignore carbon flux benefits. Assume that trees on a site grow according to the following function: vi(t) = aivi – bivi2, i = 1,2. Non-carbon environmental amenities are assumed to be a function of growth on both stands and assumed to take a Cobb–Douglas form: E(v1, v2) = gv1a v2b, with g, a, b ≥ 0 and a + b < 1. Then problem (3.10) can be written as: Maximize

T

∫ {(p – c) (h + h ) + gv 1

h1h2 with constraints

0

2

a

1

v2b}e–rt dt, (3.16)

Land use decisions and policy

123

hi(t) ≤ vi(t), i = 1,2, and

(3.17)

v˙i(t) = aivi – bivi2 – hi, i = 1,2.

(3.18)

For the problem defined by expressions (3.16), (3.17) and (3.18), the current-value Hamiltonian is given as (with time notation on control and state variables suppressed): H(v1, v2, h1, h2, q1, q2) = (p – c) (h1 + h2) + gv1a v2b + q1[a1v1 – b1v12 – h1] + q2[a2v2 – b2v22 – h2]

(3.19)

The set of admissible controls must satisfy conditions (3.17), which require that the complementary slackness conditions must also be satisfied (Leonard and Van Long, 1992, p. 199): li[vi – hi] = 0, [vi – hi] ≥ 0, li ≥ 0, i = 1,2,

(3.20)

where li are the Lagrangian multipliers. This problem involves two state variables and two controls, and the need to satisfy inequality constraints. This makes it much more difficult to solve analytically than the one-state, onecontrol problem (Leonard and Van Long, 1992). Hence numerical mathematical programming techniques are generally employed (see Hof, 1993; Hazell and Norton, 1986). Since our objective here is only to illustrate the problem created when activities on one site affect those on other sites, we only report theoretical solutions obtained by other researchers. Swallow and Wear (1993) and Swallow et al. (1997) consider multiple use across forest stands. Consider two sites, one publicly owned and the other private. Suppose that harvest of the private forest stand affects the flow of amenity benefits from the public stand, thus shifting both the MOC and MBD functions in Figure 3.2 at the public site. While it may be financially optimal to harvest the public stand, the public manager may wish to delay harvest in anticipation of felling of the private site, thus extending the public rotation age beyond that which would be socially optimal in the single-stand case. When both sites are managed for their joint commercial timber and amenity values, the sequence of harvest schedules can take rather odd forms. For example, even though two forest stands may be nearly similar in all respects, it might be socially optimal to permit one site to mature to beyond 100 years before harvesting it, and to harvest the other site several times during this period. Given the information needs to implement a subsidy in this case, Vincent and Binkley (1993) recommend the use of zoning as a public policy for increasing social well-being.

124

Environmental and resource economics 2004/2005

When both sites are privately owned, the authority needs to design incentives that cause each of the owners (or the single owner of the two stands) to take into account the effect on the amenity values of one stand of decisions taken on the other. Truly spatial problems occur when more than two stands and two amenities are considered, with socially optimal solutions diverging from those derived from two-stand analyses. Bogdanski et al. (2002) consider the case of three stands, two of which are adjacent, with values derived from timber harvesting, domestic forage use, wildlife production and passive use. When there are only two stands, they obtain results similar to those of Swallow et al. (1997). Depending on parameter values, non-timber outputs, and the distribution of values across the forest landscape, introducing a third stand can affect the degree of specialization in outputs for each stand. As long as harvests on one stand affect the benefits from standing timber on other stands (joint environmental benefits are not separable with respect to stand volumes), it may be socially optimal for harvests and inventories to differ across stands, even if the stands are identical and not adjacent. When there are one or two stands, non-convexities can lead to specialization as an optimal solution. This solution can be implemented using regulation (zoning). However, non-convexities need not be the main policy determinants, because, as Bogdanski et al. (2002) demonstrate, many factors other than nonconvexities can determine how closely or differently stands are managed. Management that ignores the explicit location of stands, spatial scale of habitats and/or interdependence among stands can result in lower social wellbeing. Indeed, research indicates that different management strategies can arise in different forest regions even when the technical production relationships and objectives are identical. But the converse may also be true: it is possible that similar management strategies should optimally be followed in different jurisdictions even if the physical forest, prices and discount rates are different. Forest regulations and taxes used to promote or protect non-timber amenities need to be sensitive to the spatial location of stands, the natural physical differences of stands, and stand interdependencies. Therefore, forest policy tools and initiatives that promote or support the principle of socially optimal use of forest resources, such as codes of forest practices and land use zoning, need to vary from one forest region to another to account for spatial considerations. The conclusion of theoretical economics research using a model such as that described above is that there is no one dominant land management paradigm. In many jurisdictions, governments have introduced various laws to regulate timber-harvesting practices on public and private land as a means to promote multiple-use forest management and achieve efficient forest resource use. In some cases, strict laws apply on all forest sites (Cook, 1998), but such

Land use decisions and policy

125

an approach fails to account for linkages and interdependencies between forest areas, regional differences in marginal values, or differences in forest ecosystems. Other jurisdictions (for example Sweden) recognize the need for flexible forest management (Wilson et al., 1998). Initiatives that do not account for spatial differences and that are inflexible over time will fail to achieve efficient resource use. Brown et al. (1993) argue that policy rigidities must be avoided; laws of forest practices and management programs need to adapt over time as more information becomes available so as to better account for the spatial and temporal aspects of forestry, and to achieve better social outcomes. If forest practice laws fail to do so, they will fail to achieve economically efficient resource use. What policies have been implemented to address spillovers at the extensive margin? In the next section, we examine these in greater detail. In some cases, we describe policies in place; in others, we point to policies that are being considered or in the process of being implemented.

5.

POLICIES TO ADDRESS LAND USE SPILLOVERS AT THE EXTENSIVE MARGIN

While we have already suggested some policy instruments for dealing with the divergence between privately and socially optimal land use activities at the extensive margin, reality is often more complex. A challenging task facing policymakers is to develop appropriate policy instruments for addressing environmental spillovers. Different instruments that governments can use include command-and-control regulations (such as planning and growth management), and market-based incentives that promote flexibility in achieving environmental objectives (Stavins, 2002). Market incentives include tax-subsidy and cap-and-trade schemes (where a quota or cap is set on emissions, say, and firms are free to buy and sell quotas). However, state involvement is generally required for cap-and-trade schemes, if only to determine the cap level and enforce and monitor the subsequent trading mechanism. Reliance on private transactions to resolve environmental spillovers has been generally eschewed because empirical evidence of its success is lacking. The usual conclusion is that transaction costs of reaching agreements are too onerous, so some form of state involvement is required. Even where firms (landowners in our case) have voluntarily agreed to address environmental spillovers, the explicit threat of state intervention is generally a prerequisite (Segerson and Miceli, 1998). There is now increasing evidence of the emergence of non-state, marketdriven governance structures for addressing environmental spillovers (Kolk et al., 1999; Khanna, 2001). Private unilateral action may be undertaken because it leads to a reduction in production costs (for example energy savings).

126

Environmental and resource economics 2004/2005

Examples are 3M Company’s ‘Pollution Prevention Pays’ and Dow Corporation’s ‘Waste Reduction Always Pays’ programs, which have reduced production or operating costs (Stavins, 2002). Of such governance structures, private certification of sustainable forest management practices is possibly one of the more comprehensive examples (Kiker and Putz, 1997; Swallow and Sedjo, 2002), and it is examined further below. Before doing that, however, we consider forest-sector policies and more general policies affecting land use at the extensive margin. 5.1

Forest-sector and General Policies to Protect Nature

In the past decade or so, there has been a spate of public policies designed to protect nature in forest ecosystems. The main characteristic of these programs is their reliance on regulation, or command and control, as opposed to market instruments. We discuss these policies in the remainder of this section. Zoning of protected areas Zoning is increasingly used in countries’ conservation programs to protect representative wildlife habitats. In 1991, some 5 per cent of the earth’s land surface was protected; in 1997, some 9 per cent of the globe’s land surface was protected (Green and Paine, 1997). The Brundtland Commission (WCED, 1987) suggests that 20 per cent of tropical forestlands be protected (p. 152), and that there be a threefold increase in the amount of land set aside for species and ecosystem preservation (pp. 165–6). As a result, there appears to be international consensus that, as a rule of thumb, nations should commit 12 per cent of their land base to protect biological resources. Many countries have either formulated or are in the course of designing plans to move towards this target. For example, British Columbia’s Protected Areas Strategy was designed to keep more than 12 per cent of the Province’s land area as wilderness reserves (van Kooten, 1999). The creation of conservation reserves raises at least two questions. Is 12 per cent an adequate level of protection? Can such a level of protection be sustained? Protected areas constitute in situ protection of biodiversity and other environmental values. Along with ex situ facilities, they help sustain many species, including some in serious danger of extinction. But the protected areas were not originally established to conserve biological diversity, and many parks and wilderness areas are considered to be too small effectively to conserve intact ecosystems or to provide for their inhabitants. At the end of the twentieth Century, there were 30 350 protected areas covering 13.2 million km2 in the world, but 59 per cent of these were less than 1 000 ha in size and six large protected areas accounted for 18.5 per cent of the total (Green and Paine,

Land use decisions and policy

127

1997). The largest protected areas are thinly populated and contain very little in the way of biodiversity. Further, rates of protection vary greatly by biomes. Protected areas account for 16.3 per cent of mixed island systems, 10.3 per cent of subtropical and temperate rain forests, and 9.1 per cent of mixed mountain systems, but less than 1 per cent of temperate grassland and 1.1 per cent of lake systems (Green and Paine, 1997). It is little wonder that biologists argue that greater areas need to be protected, although they do recognize that this needs to be done in an economically wise or efficient manner (Pressey, 2000; Sinclair, 2000). The more relevant question might be whether significant conservation reserves can be sustained. First, creation of protected areas has not gone unopposed, particularly in developing countries where local people were often evicted from newly established reserves without compensation. Benefits from protected areas frequently extend beyond the sites to society at large in the form of existence values, or hunting benefits for an international elite, while costs are borne locally. Unless local citizens have a stake in a protected area, they are likely to undermine efforts to preserve the ecosystem. Ineffective management and insufficient funding to maintain protected areas, in both developing and developed countries, have led to the conclusion that ‘all reserves in the world are presently in a state of decline as a result of attrition from human interference’ (Sinclair, 2000, p. 43). Ironically, in many countries where parks are major sources of tourist revenue, little is reinvested in conservation. Poaching is a particular problem in many developing countries because the values of products from some endangered and threatened species, such as rhinoceros, elephant, tiger and bear, are high compared to the incomes of people living in the regions where such animals are found. Poaching is also a problem in developed countries, where poaching of bear and other wildlife, as well as timber (for example to make cedar shakes for roofs), also appears difficult to stop. As in the rural–urban interface, zoning may not, by itself, be adequate. Triad zoning Zoning has also been recommended in Canada, which has perhaps the highest degree of public ownership of forestland in the world (Wilson et al., 1998). In order to address environmental concerns surrounding use of public forestlands, some Canadian provinces are looking to zone forestland into three components – non-commercial use (no logging whatsoever), multiple use (a mix of forestry operations and other activities) and intensive forestry. The acronym TRIAD derives from the need to create three zones of use. The guiding idea is that, by investing heavily in silviculture in the intensively managed zone, enough timber will be produced to cover loss of harvests from the other zones. As such, this policy is based on (albeit incomplete) research results from models similar to those presented in the previous section.

128

Environmental and resource economics 2004/2005

There are problems with this zoning approach to protect nature. First, TRIAD zoning does not necessarily lead to an economically efficient outcome from society’s point of view. As we have seen, when there are interdependencies among spatially separated stands, whatever is done on one stand will affect how other stands are managed. It does not necessarily follow that permanently setting aside one particular area never to be harvested is efficient. Permanent set-asides may or may not be economically efficient. Indeed, periodic harvests, perhaps even more than 100 years apart, may increase social welfare. Second, with risks of fire, disease and climate change, it is possible that amenity values on the permanently set-aside area can be lost for a long period of time. Salvage logging on the no-logging area along with reduced logging in intensively managed areas may be appropriate. Of course, some of these observations apply to conservation reserves as well, although TRIAD zoning is supplementary to biological reservation. TRIAD zoning on a regional scale involving perhaps 20 or more million ha. of working forest may simply be infeasible, because such zoning will reduce overall flexibility. It seems to us that adaptive management with appropriate economic incentives may better protect non-timber amenities. For example, wood engineering has made it possible to use fiber from hybrid poplar plantations to produce solid wood products that are comparable in terms of their properties (strength, flexibility and so on) with products from traditional plantations and original forests. However, in areas such as temperate British Columbia, where hybrid poplar grows faster than in other regions of Canada, there is a lack of private land for plantations, and Provincial regulations on public land often preclude the planting of hybrid poplar on a large scale, even on land zoned for intensive forest management. Unless environmental amenity constraints are sufficiently relaxed in the intensively managed, timber-only zone, it may not be possible to offset losses in timber output from increased set asides by designating special zones for single use (see Krcmar et al., 2003). Endangered species legislation Endangered species legislation is another example of regulation of land use in order to protect nature at the extensive margin. To our knowledge, the United States is the only country to have enacted endangered species legislation that affects how private landowners can use their land. While Canada also has endangered species legislation, it only applies to federal lands and not to private lands or land under the ownership of a Provincial government (the most common form of public ownership). In the United States, Congress passed the Endangered Species Act (ESA) in 1973 and re-authorized it in 1988. ESA was interpreted by the Courts to mean that species were to be saved regardless of costs, which established an impractical duty to preserve all species. Therefore, shortly after ESA was enacted, the Congress established

Land use decisions and policy

129

the Endangered Species Committee, comprised of relevant agency heads and other representatives, to resolve conflicts between federal government projects and ESA. Environmentalists saw this as an attempt to get around ESA, and dubbed the Committee the ‘God Squad.’ But it has rarely met since it was established and has seldom taken action to overturn Court decisions. Endangered species legislation affects land use because endangered species’ habitat is found on land, or, in the case of fish, it affects land use (for example removal of water for irrigation or clearing trees too close to streams affects fish habitat). ESA can lead to less than socially optimal land use for several reasons. First, because private landowners usually incur the costs of protecting the habitat of endangered species, they have an incentive to destroy wildlife habitat on their land before it is identified as harboring an endangered species or before an identified species is listed under ESA. Thus more habitat may potentially be destroyed than is socially optimal. Second, the legislation could prevent certain land use activities that generate high net social benefits that would exceed the benefits from protecting a specific species. Perhaps the species has little value to society or better habitat exists elsewhere, or preventing the land use activity will not succeed in preventing the species from becoming extinct. Most land use regulations are not designed to achieve a socially optimal allocation of land. Protection of species can be modeled in ways described in the previous section, but only if species numbers are related to land use activities. If that is not the case, then other approaches to analyzing optimal land use will need to be applied. 5.2

Codes of Forest Practice and Forest Certification

Beginning in the early 1990s, many forest jurisdictions began to introduce codes of forest practice in order to address environmental spillovers associated with forest operations and to ensure sustainable forest management. Sweden’s code was embodied in the 1993 Forestry Act. The Act itself consisted of only a few pages, with on-the-ground application left to local state foresters. In contradistinction, the Province of British Columbia’s 1995 Forest Practices Code consists of thousands of pages of detailed requirements that place an onerous burden not only on firms but also on the state foresters responsible for enforcing it (van Kooten, 1999). In British Columbia, there were delays in getting forest management plans approved (necessary because logging companies harvest timber from public lands), harvesting costs increased dramatically, and flexibility in regeneration was reduced. Forest practices codes are, perhaps, an unneeded regulatory burden if zoning regulations are implemented, because of the development of non-state, market-driven schemes for the certification of forest management practices.

130

Environmental and resource economics 2004/2005

Failure to sign a global convention on forestry at the Earth Summit in Rio de Janeiro in 1992 led environmental, non-governmental organizations to develop a private, voluntary regulatory scheme for sustainable forest management. In 1993, a coalition of environmental groups (led by the World Wide Fund for Nature, or WWF), foresters and timber companies formed the Forest Stewardship Council (FSC) to develop standards for sustainable forest management (SFM) and to certify companies practicing sustainable forestry. In response, competing domestic certification schemes have developed, primarily in Canada, the United States and Europe. These were implemented by the forest industry in North America and by forest landowners in Europe. As of June 2002, 122 million ha. of forests have been certified globally, but this constitutes only about 3 per cent of the world’s forests. Nearly 9 per cent of forests in North America are certified and nearly 6 per cent of those in Europe, but only about 0.5 per cent of forests elsewhere. Only 8 per cent of all certified forests are in Asia, Africa and Latin America. Most of the certified forestland is certified under a domestic scheme as opposed to that of the FSC. While the intention of legislated codes of forest practice is to foster joint production of commercial timber and nature, forest certification has perhaps eliminated the need to have state regulations. Indeed, inflexible local codes governing forest practices that must be retained by a certification scheme (which is the case for FSC certification) could become an obstacle to firms seeking to become certified. Forest certification is a non-state, market-driven governance structure that addresses environmental spillovers related to land use in forestry. Whether such an approach could be extended to other land uses as well is unclear, but certification could conceivably be extended to terrestrial carbon sinks, where it could be used to facilitate a CO2 emissions trading market that also includes carbon offset credits. The potential for other land uses to become certified is something that governments need to consider in the development of future land use policy.

6.

CONCLUSIONS

Scarcity of undeveloped rural or natural land has increased the attention being paid to trends between urban land uses, rural land uses, and land in its natural state. A range of analytical techniques has been developed to study land use at the intensive margin (between urban and rural uses) and the extensive margin (between rural uses and nature). The opportunity cost of developing either rural or natural land may include lost benefits that are not valued in markets. Economic policy issues frequently involve the external opportunity cost of developing rural land for urban purposes, or developing natural land for agriculture or forestry purposes. Policy approaches such as regulations and market-

Land use decisions and policy

131

based incentives have been developed to incorporate these external opportunity costs into the land development process, and to reconcile private land use decisions with public preferences for multiple benefits from the landscape. One common perspective used to analyze land use at the intensive margin is the bid rent framework of von Thünen. In this basic model, land is optimally allocated in a competitive land market to the use that provides greatest net present value. Net benefit from a particular land use option depends on returns and costs associated with that option, and on a land parcel’s location relative to a central business district. Innovations to the urban bid rent model gave rise to the widely used urban growth model, which includes a premium for potential development in the value of rural land. Introducing land quality in addition to location results in a more realistic landscape that includes non-contiguous land uses. When external benefits are associated with rural use, then policies are needed to influence the allocation of land between urban and rural uses. Specific policy issues at the intensive margin include the opportunity cost of lost open space, congestion externalities of development, and the external costs of development infrastructure (for example schools, sewers, roads, public services). Policy approaches to help sustain agriculture in a changing landscape include right-to-farm legislation and preferential property tax valuation. When development premiums for rural land are substantial, incentives for continued rural use may be provided through conservation easements or purchase of development rights programs. Regulatory approaches such as land use zoning and large minimum developed lot sizes have also been used to influence the intensive margin and the developed landscape. These tend to be an alternative to planning or growth management policies in which land use conversion is considered on a parcel-by-parcel basis. The optimum timber harvest model developed by Faustmann provides a common perspective for analyzing land use at the extensive margin. The Faustmann model has been extended in numerous directions to include the influence of (1) non-timber benefits, (2) neighboring stands, and (3) non-forest land use. Including non-timber benefits can lead to an important optimum outcome in which a forest is permanently preserved for non-timber production. Whether the joint production of timber and non-timber benefits is convex with respect to time is crucial in determining optimum forest management policies. Including neighboring stands (or other non-linear harvest effects) is crucial in distinguishing optimum partial harvests from optimum clear-felling. When this is done, it becomes necessary to distinguish between private landowner and social optimums, since the optimum partial harvest results from spillovers between forest stands. Considering non-forest rural land use is essential to represent accurately any landscape that consists of more than a single forest use.

132

Environmental and resource economics 2004/2005

Specific policy issues at the extensive margin include the opportunity cost of developing natural lands for agriculture, and the competing joint production of timber and non-timber benefits in forested landscapes. Policy approaches used to sustain environmental benefits in agricultural landscapes include conservation payment programs, and (in the USA) linking agricultural subsidies to environmental performance standards. Regulatory approaches used to obtain optimum joint timber and non-timber production include forest zoning, TRIAD zoning (in Canada), and endangered species habitat protection (in the USA). Forest practice-related performance standards are also used in forest landscapes throughout the world, frequently taking the form of forest practice certification. New research opportunities are emerging throughout land economics, particularly in the development of models, methods and policy analyses. For example, new conceptual models are beginning to integrate urban, rural and natural land uses. There is increasing recognition of rural use in urban models, non-agricultural use in agricultural models, and non-forest (non-timber) use in forest (timber) models. The relevance of land to current policy problems is the engine that drives recent innovation in both the conceptual and analytical areas. The role of the landscape in environmental public goods such as carbon storage, species habitat and water quality is receiving increased attention. Most of the policy analyst’s traditional instruments have been applied or proposed for landrelated policy problems (command and control, charges and subsidies, and transferable property rights). There appears to be a growing interest in the synthesis of market-based, regulatory and planning approaches at both the intensive and extensive margins. At both the intensive and extensive margins, there is a movement toward combining zoning and planning to manage development of land in complex landscape situations. At the intensive margin, planning on the part of the landowner seeking approval for development plays a big role, and government increasingly uses zoning or regional plans (Europe) to guide this landowner planning. At the extensive margin, zoning is being recommended as a ‘better’ policy and planning of forestland management is playing a bigger role for landowners seeking approval for forest stewardship certification. This planning–zoning alternative seems to complement a market-based economic policy (tax or subsidize), which may be too imprecise to obtain some desired spatial configurations of land uses. Economists have yet fully to compare the advantages and disadvantages of the two approaches to controlling land development, making this a potentially fruitful area for further research.

Land use decisions and policy

133

NOTES 1. 2.

3.

4. 5. 6. 7.

8.

9. 10.

11.

12.

13. 14.

Authors are cited alphabetically. Senior authorship is not assigned. Use of the terms ‘intensive’ and ‘extensive’ margin is based on von Thünen’s isolated state, where the use of land as an input into agriculture and forestry stops at the edge of the central city (the intensive margin) and at the edge of the hinterlands (the extensive margin) beyond which land is left in its natural state (van Kooten and Bulte, 2000, ch. 4). The other major topics of land use change at the wilderness–rural land interface that are omitted are tropical deforestation and desertification (primarily in Africa). Two of the three authors have published recent survey articles on deforestation (van Kooten et al., 1999; Parks et al., 1998) and choose not to revisit that topic here. Desertification involves issues of land ownership (use and expropriation of commons, use without ownership) that are placed outside of this chapter by our maintained assumption that land ownership is either fully private or fully public. Meyer et al. (2003) also provides a current review of this topic. More general forms of this equilibrium condition can, of course, be employed in this type of model. See, for example, Cheshire and Sheppard (1995) or Henderson and Mitra (1996). Note that agricultural land value is now treated as a function of time, distance from a CBD and environmental land characteristics. Capozza and Helsley simplified their algebra by treating this value as an exogenous constant. We use PL to denote price in the Wicksell model (equation 3.6) and PU to denote price in the von Thünen case (equation 3.2). Theoretical results can change if the perfect foresight assumption that developers know the consumers’ bid rent function at all time periods is relaxed. Titman (1985), Capozza and Helsley (1990), and Batabyal (1996) have shown that uncertainty and irreversibility can combine to delay the timing of development. Capozza and Li (1994) also show that replacement of capital can affect the timing of development. Thus the durable capital assumption also may be limiting. Smith et al. (2002) raise the associated question: can consumers of urban land successfully forecast future land use patterns, particularly those related to open space? The literature has several competing definitions of urban sprawl; see, for example, Galster et al. (2001) and Ewing et al. (2002). Common elements seem to be unnecessary automobile use to perform routine household chores and excessive infrastructure costs. In general, the term can refer to any inefficiently (however defined) low density of developed land use. By 2002, state and local governments within the USA had purchased conservation easements for 1.1 million acres of farmland at a cost averaging approximately $1 746 per acre (American Farm Trust, Fact Sheet: Status of State PACE Programs, November 2002). Impact fees are charged to cover costs of supplying additional educational facilities, sewerage facilities, upgraded roads or other public infrastructure deemed to be needed by the added households. These fees normally must be paid to local governments to obtain development approval. The primary counter-argument is presented by Hamilton (1978), who argues that zoning applications by competing jurisdictions within a local metropolitan area can lead to situations in which the supply of housing will be below and the price above purely competitive levels. Hamilton’s focus is on the ability of local jurisdictions to use zoning to restrict the movement of households within a given market (cf. Tiebout, 1956). The issue in the United States stems from the Fifth Amendment to the Constitution, which states that property cannot be taken for public purpose without just compensation. Several court cases have dealt with the degree to which this amendment extends to regulations that take some but not all property rights (Lant, 1994). Irwin and Bockstael (2002) find that this program significantly affects the probability that land will be developed. Since the planning horizon is T, the infinite stream of agricultural benefits on the RHS of equations (3.13) and (3.14) is converted to a finite horizon of T via the term (1–e–rT).

134

Environmental and resource economics 2004/2005

REFERENCES Alterman, R. (1997), ‘The challenge of farmland preservation: lessons from a sixnation comparison’, Journal of the American Planning Association, 63(2), 220–43. Anderson, J.E. (1993), ‘Use-value property tax assessment: effects on land development’, Land Economics, 69 (August), 263–9. Barichello, R.R., R.M. Porter and G.C. van Kooten (1995), ‘Institutions, economic incentives and sustainable rural land use in British Columbia’, Chapter 2 in A. Scott, J. Robinson and D. Cohen (eds), Managing Natural Resources in British Columbia. Markets, Regulations, and Sustainable Development, Vancouver: UBC Press, pp. 7–53. Batabyal, A.A. (1996), ‘The timing of land development: an invariance result’, American Journal of Agricultural Economics, 78 (November), 1092–7. Bills, N.L. and R.N. Boisvert (1987), ‘New York’s experience in farmland retention through agricultural districts and use-value assessment’, Chapter 19 in W. Lockeretz (ed.), Sustaining Agriculture Near Cities, Ankeny, IA: Soil and Water Conservation Society. Bogdanski, B.E.C., G.C. van Kooten and C.S. Binkley (2002), ‘Spatial richness in multiple-use forest management’, FEPA working paper, mimeograph, Vancouver: Forest Economics and Policy Analysis Research Unit, University of British Columbia. Boisvert, R.N. and N.L. Bills (1984), ‘Variability of New York’s agricultural use values and its implications for policy’, Northeastern Journal of Agricultural and Resource Economics, 13 (October), 254–63. Bowes, M.D. and J.V. Krutilla (1989), Multiple-Use Management: The Economics of Public Forestlands, Washington: Resources for the Future. Brown, T.C., D. Brown and D. Binkley (1993), ‘Laws and programs for controlling nonpoint source pollution in forest areas’, Water Resources Bulletin, 29(1), 1–13. Brueckner, J.K. (2000), ‘Urban sprawl: diagnosis and remedies’, International Regional Science Review, 23, 160–71. Burchell, R.W., N.A. Shad, D. Listokin, H. Phillips, A. Downs, S. Seskin, J.S. Davis, T. Moore, D. Helton and M. Gail (1998), The Costs of Sprawl – Revisited, Transit Cooperative Research Program Report 39, 1073–4872, Washington, DC: National Academy Press. Calish, S., R.D. Fight and D.E. Teeguarden (1978), ‘How do nontimber values affect douglas-fir rotations?’, Journal of Forestry, 76 (April), 217–21. Cameron, S. and K. Stephenson (2002), ‘Does sprawl cost us all? Isolating the effects of housing patterns on public water and sewer costs’, Journal of the American Planning Association, 68(1), 56–70. Capozza, D.R. and R.W. Helsley (1989), ‘The fundamentals of land prices and urban growth’, Journal of Urban Economics, 26, 295–306. Capozza, D.R. and R.W. Helsley (1990), ‘The stochastic city’, Journal of Urban Economics, 28, 187–203. Capozza, D.R. and Y. Li (1994), ‘The intensity and timing of development: the case of land’, American Economic Review, 84 (September), 889–904. Cheshire, P. and S. Sheppard (1995), ‘On the price of land and the value of amenities’, Economica, 62, 247–67. Clark, J.S., M. Fulton and J.T. Scott Jr (1993), ‘The inconsistency of land values, land rents, and capitalization formulas’, American Journal of Agricultural Economics, 74 (February), 147–55.

Land use decisions and policy

135

Cook, T. (1998), ‘Sustainable practices? An analysis of B.C.’s forest practices code’, Chapter 9 in C. Tollefson (ed.), The Wealth of Forests Vancouver: UBC Press, pp. 204–31. Corbett, R. (ed.) (1990), Protecting Our Common Future: Conflict Resolution Within the Farming Community, Ottawa: Canada Mortgage and Housing Corporation. Coughlin, Robert E. and John E. Keene (1981), ‘The protection of farmland: a reference guidebook for state and local governments’, in National Agricultural Lands Study, Washington, DC: US Government Printing Office, 284pp. Dear, Michael (1992), ‘Understanding and overcoming the NIMBY syndrome’, Journal of the American Planning Association, 58 (3), 288–300. Ervin, D.E., J.B. Fitch, R.K. Godwin, W.B. Shepard and H.H. Stoevener (1977), Land Use Control: Evaluating Economic and Political Effects, Cambridge, MA: Ballinger. Ewing, R., R. Pendall and D. Chen (2002), Measuring Sprawl and Its Impact, Washington, DC: Smart Growth America, accessed at: www.smartgrowthamerica.org. Faustmann, M. (1849), ‘On the determination of the value which forest land and immature stands possess for forestry’, reprinted in Journal of Forest Economics, 1(1995), 7–44. Fernandez, Linda and Larry Karp (1998), ‘Restoring wetlands through wetlands mitigation banks’, Environmental and Resource Economics, 12 (October), 323–44. Fishel, William A. (1990), ‘Introduction: four maxims for research on land use controls’, Land Economics, 66 (August), 229–36. Fuguitt, Glenn V. and David L. Brown (1990), ‘Residential preferences and population redistribution, 1972–1988’, Demography, 27 (4), 589–600. Fujita, Masahisa (1982), ‘Spatial patterns of residential development’, Journal of Urban Economics, 12, 22–52. Fujita, Masahisa (1996), ‘Urban land use theory,’ in Regional and Urban Economics, Parts 1 and 2, Encyclopedia of Economics, Richard Arnott (ed.), Amsterdam: Harwood Academic Publishers. Galster, G., R. Hanson, M.R. Ratcliffe, H. Wolman, S. Coleman and J. Freihage (2001), ‘Wrestling sprawl to the ground: defining and measuring an elusive concept’, Housing Policy Debate, 12 (4), 681–717. Gardner, Bruce L. (1994), ‘Commercial agriculture in metropolitan areas: economics and regulatory issues’, Agricultural and Resource Economics Review, 23 (April), 100–109. Grant, Jill (2002), ‘Mixed use in theory and practice’, Journal of the American Planning Association, 68 (1), 71–84. Green, M.J.B. and M. Paine (1997), ‘State of the world’s protected areas at the end of the twentieth century’, paper presented at the World Conservation Union (IUCN) World Commission on Protected Areas Symposium on ‘Protected Areas in the 21st Century: From Islands to Networks, held in Albany, Australia, 24–29 November, accessed 23 October 2001 at: http://www.wcmc.org.uk/protected_areas/albany.pdf. Hamilton, Bruce W. (1978), ‘Zoning and the exercise of market power’, Journal of Urban Economics, 5, 116–30. Hardie, Ian and Cynthia Nickerson (2003), ‘The effect of a forest conservation regulation on the value of subdivisions in Maryland’, working paper WP 03-10, Department of Agricultural and Resource Economics, University of Maryland, College Park. Hardie, Ian W., Tulika A. Narayan and Bruce L. Gardner (2001), ‘The joint influence

136

Environmental and resource economics 2004/2005

of agricultural and nonfarm factors on real estate values: an application to the MidAtlantic region’, American Journal of Agricultural Economics, 83 (February), 120–32. Hartman, R. (1976), ‘The harvesting decision when the standing forest has value’, Economic Inquiry, 14, 52–8. Hazell, P.B.R. and R.D. Norton (1986), Mathematical Programming for Economic Analysis in Agriculture, New York: Macmillan. Heimlich, Ralph E. and William D. Anderson (2001), ‘Development at the urban fringe and beyond’, agricultural economic report no. 803, Economic Research Service, United States Department of Agriculture, Washington, DC. Heimlich, Ralph E. and Charles H. Barnard (1992), ‘Agricultural adaptation to urbanization: farm types in the Northeast metropolitan areas’, Northeastern Journal of Agricultural and Resource Economics, 21 (April), 50–60. Henderson, Vernon and Arindam Mitra (1996), ‘The new urban landscape: developers and edge cities’, Regional Science and Urban Economics, 26, 613–43. Hof, John (1993), Coactive Forest Management, New York: Wiley. Hollis, Linda E. and William Fulton (2002), ‘Open space protection: conservation meets growth management’, discussion paper, Brookings Institutional Center on Urban and Metropolitan Policy, April, www.solimar.org. Irwin, Elena G. and Nancy E. Bockstael (2002), ‘Urban sprawl as a spatial economic process’, Conference paper CP02A14, Lincoln Institute of Land Policy, Cambridge, MA. Johnston, Robert A. and Mary E. Madison (1997), ‘From landmarks to landscapes: a review of current practices in transfer of development rights’, Journal of the American Planning Association, 63 (3), 365–78. Just, R.E. and J.A. Miranowski (1993), ‘Understanding farmland price changes’, American Journal of Agricultural Economics, 74 (February), 156–68. Khanna, M. (2001), ‘Non-mandatory approaches to environmental protection’, Journal of Economic Surveys, 15 (July), 291–324. Kiker, C.F. and F.E. Putz (1997), ‘Ecological certification of forest products: economic challenges’, Ecological Economics, 20 (January), 37–51. Kolk, A., R. van Tulder and C. Welters (1999), ‘International codes of conduct and corporate social responsibility: can transnational corporations regulate themselves?’, Transnational Corporations, 8 (1), 143–80. Koskela, E. and M. Ollikainen (1999), ‘Optimal public harvesting under the interdependence of public and private forests’, Forest Science, 45 (2), 259–71. Koskela, E. and M. Ollikainen (2002), ‘Optimal private and public harvesting under spatial and temporal interdependence’, Forest Science, forthcoming. Krcmar, E. and G.C. van Kooten (2002), ‘Timber, carbon uptake and structural diversity tradeoffs in forest nanagement’, FEPA working paper, mimeograph, Forest Economics and Policy Analysis Research Unit, University of British Columbia, Vancouver. Krcmar, E., I. Vertinsky and G.C. van Kooten (2003), ‘Modeling alterntiave zoning strategies in forest management’, International Transactions in Operations Research, in press. Lant, C.R. (1994), ‘The role of property rights in economic research on U.S. wetlands policy’, Ecological Economics, 11, 27–33. Lapping, M.B. and N.R. Leutwiler (1987), ‘Agriculture in conflict: right-to-farm laws and the peri-urban milieu for farming’, Chapter 17 in William Lockeretz (ed.), Sustaining Agriculture Near Cities, Ankeny, IA: Soil and Water Conservation Society.

Land use decisions and policy

137

Leonard, D. and N. Van Long (1992), Optimal Control Theory and Static Optimization in Economics. Cambridge, UK: Cambridge University Press. Lisansky, J. and G. Clark (1987), ‘Farmer–nonfarmer conflicts in the urban fringe: will right-to-farm help?’, Chapter 18 in William Lockeretz (ed.), Sustaining Agriculture Near Cities, Ankeny, IA: Soil and Water Conservation Society. Lopez, R.A., A.O. Adelaja and M.S. Andrews (1988), ‘The effects of suburbanization on agriculture’, American Journal of Agricultural Economics, 70 (May), 346–58. Marland, G., K. Fruit and R. Sedjo (2001), ‘Accounting for sequestered carbon: the question of permanence’, Environmental Science & Policy, 4 (6), 259–68. McConnell, K.E., J.N. Daberkow and I.W. Hardie (1983), ‘Planning timber production with evolving prices and costs’, Land Economics, 59 (3), 292. McMillan, D.P. and J.F. McDonald (1991), ‘A Markoff chain model of zoning change’, Journal of Urban Economics, 30, 257–70. McMillan, D.P. and J.F. McDonald (1993), ‘Could zoning have increased land values in Chicago?’, Journal of Urban Economics, 33, 167–88. Meyer, A.L., G.C. van Kooten and S. Wang (2003), ‘Institutional, social and economic roots of deforestation: further evidence of an environmental Kuznets relation?’, International Forestry Review, in press. Nelson, A.C. (1992), ‘Preserving prime farmland in the face of urbanization: lessons from Oregon’, Journal of the American Planning Association, 58 (4), 467–88. Nickerson, C.J. and L. Lynch (1999), ‘The effect of farmland preservation programs on farmland prices’, working paper WP-99-08. Department of Agricultural and Resource Economics, University of Maryland, College Park, MD. Parks, P.J. and R.A. Kramer (1995), ‘A policy simulation of the Wetland Reserve Program’, Journal of Environmental Economics and Management, 28 (2), 223–40. Parks, P.J., E.B. Barbier and J.C. Burgess (1998), ‘The economics of forest land use in temperate and tropical areas’, Environmental and Resource Economics, 11 (3–4), 473–87. Peiser, R.B. (1989), ‘Density and urban sprawl’, Land Economics, 65 (August), 193–204. Pressey, R.L. (2000), ‘The end of conservation on the cheap, revisited’, Chapter 5 in G.C. van Kooten, E.H. Bulte and A.R.E. Sinclair, Conserving Nature’s Diversity, Aldershot, UK: Ashgate, pp. 45–67. Rolleston, B.S. (1987), ‘Determinants of restrict suburban zoning: an empirical analysis’, Journal of Urban Economics, 21, 1–21. Samuelson, P.A. (1976), ‘Economics of forestry in an evolving society’, Economic Inquiry, 14, 466–92. Segerson, K. and T.J. Miceli (1998), ‘Voluntary environmental agreements: good or bad news for environmental protection?’, Journal of Environmental Economics and Management, 36, 109–30. Shi, Y.J., T.T. Phipps and D. Colyer (1997), ‘Agricultural land values under urbanizing influences’, Land Economics, 73 (1), 90–100. Sinclair, A.R.E. (2000), ‘Is conservation achieving its ends?’, Chapter 4 in G.C. van Kooten, E.H. Bulte and A.R.E. Sinclair, Conserving Nature’s Diversity, Aldershot, UK: Ashgate, pp. 30–44. Smith, V.K., C. Poulos and H. Kim (2002), ‘Treating open space as an urban amenity’, Resource and Energy Economics, 24, 107–29. Stavins, R.N. (2002), ‘Lessons from the American experience with market-based environmental policies’, in J.D. Donahue and J.S. Nye Jr. (eds), Harnessing the Huricane: The Challenge of Market-Based Governance, New York: Brookings Institution Press.

138

Environmental and resource economics 2004/2005

Swallow, S.K. and R.A. Sedjo (2002), ‘Voluntary eco-labeling and the price premium’, Land Economics, 78 (May), 272–84. Swallow, S.K. and D.N. Wear (1993), ‘Spatial interactions in multiple-use forestry and substitution and wealth effects for the single stand’, Journal of Environmental Economics and Management, 25, 103–20. Swallow, S.K., P.J. Parks and D.N. Wear (1990), ‘Policy-relevant nonconvexities in the production of multiple forest benefits’, Journal of Environmental Economics and Management, 19, 264–80. Swallow, S.K., P. Talukdar and D.N. Wear (1997), ‘Spatial and temporal specialization in forest ecosystem management under sole ownership’, American Journal of Agricultural Economics, 79, 311–26. Tiebout, C.M. (1956), ‘A pure theory of local expenditures’, Journal of Political Economy, 64, 416–24. Titman, S. (1985), ‘Urban land prices under uncertainty’, American Economic Review, 75 (June), 505–14. van Kooten, G.C. (1999), ‘Preserving species without an endangered species act: British Columbia’s forest practices code’, Chapter 4 in M. Boman, R. Brännlund and B. Kristrom (eds), Topics in Environmental Economics, Dordrecht: Kluwer Academic Publishers, pp. 63–82. van Kooten, G.C. and E.H. Bulte (2000), The Economics of Nature, Oxford: Blackwell Publishers. van Kooten, G.G., C.S. Binkley and G. Delacourt (1995), ‘Effect of carbon taxes and subsidies on optimal forest rotation age and supply of carbon services’, American Journal of Agricultural Economics, 77 (May), 365–74. van Kooten, G.C., E. Krcmar-Nozic, B. Stennes and R. van Gorkom (1999), ‘Economics of fossil fuel substitution and wood product sinks when trees are planted to sequester carbon on agricultural lands in Western Canada’, Canadian Journal of Forest Research, 29 (11), 1669–78. Vincent, J.R. and C.S. Binkley (1993), ‘Efficient multiple-use forestry may require land use specialization’, Land Economics, 69 (November), 370–76. Wallace, N.E. (1988), ‘The market effects of zoning undeveloped land: does zoning follow the market?’, Journal of Urban Economics, 23, 307–26. WCED (World Commission on Environment and Development) (1987), Our Common Future, Oxford: Oxford University Press. Wiebe, K., A. Tegene and B. Kuhn (1996), ‘Partial interests in land: policy tools for resource use and conservation’, agricultural economic report no. 744, Economic Research Service, United States Department of Agriculture, Washington, DC. Wilson, B., G.C. van Kooten, I. Vertinsky and L.M. Arthur (eds) (1998), Forest Policy. International Case Studies, Wallingford, UK: CABI Publishing.

4. Indicators of sustainability Eric Neumayer* 1. INTRODUCTION If a political or geographical entity such as a nation-state, a region or a city is committed to sustainable development, then measuring sustainability becomes very important. Only with the help of such measurement will it be able to assess whether and if so which policy measures are necessary to achieve sustainability. I will mainly concentrate on the nation-state level here and inquire how one can measure whether the economy of a country is sustainable and what such measurement tells us. As will be seen, there are many indicators aspiring to answer this question and they come to starkly differing conclusions. For example, some indicators tell us that most countries, particularly the developed ones, have no apparent problem with sustainability, whereas others suggest that the economies of many countries, and the developed ones in particular, are clearly unsustainable. Part of the reason why existing indicators come to such differing conclusions is that they differ in their understanding of sustainability and thus differ in what they measure. To start with, on a very fundamental level, most indicators focus exclusively on intergenerational equity, that is equity between generations. However, at least one of the indicators I look at, namely the index of sustainable economic welfare (ISEW), explicitly tries to combine intergenerational equity with intragenerational equity, that is, equity within the current generation. Furthermore, sustainability, even where it refers exclusively to intergenerational equity, comes in two main forms, namely weak sustainability (WS) and strong sustainability (SS). I define WS as the requirement to keep per capita utility non-declining at any moment in time, that is over the whole future development path, which represents a common economic definition.1 WS is based on the often implicit assumption of unlimited substitutability of natural capital in the production of consumption goods and the generation of utility through other forms of capital such as man-made (produced) capital and human capital.2 SS holds that natural capital is non-substitutable either in its entirety or at least with reference to certain so-called critical functions of natural capital. It is sometimes defined in value terms as the requirement to keep the value of natural capital intact. Such a definition is somewhat problematic as it does not 139

140

Environmental and resource economics 2004/2005

constrain substitutability amongst various forms of natural capital itself. A different definition therefore sees SS as the requirement to maintain all or at least some of the critical functions of natural capital intact. This would call for setting standards in physical terms and for specific forms of natural capital, which would have to be obeyed. This clearly constrains substitutability within natural capital. It typically implies keeping human impact within the natural regenerative capacity of the environment – see Neumayer (1999a, 2003) for a more extensive discussion. SS thus often focuses on environmental sustainability. One must not forget, however, that, like WS, SS is also driven by considerations of intergenerational equity as the assumption is that keeping the value of natural capital at least constant or maintaining specific forms of natural capital is necessary to protect the welfare of future generations (Daly, 1992). Some of the indicators I look at try to measure WS whereas others are indicators of SS. The indicators can of course be criticized partly not for what they are and how reliable they are in their measurement, but for what they are supposed to indicate, namely WS or SS. I will not engage in such a discussion and have made my view clear in Neumayer (1999a, 2003). Instead, I take the indicators for what they are supposed to measure and provide a critical analysis of how well they achieve this task. I will distinguish between monetary indicators, physical indicators and hybrid approaches, which try to combine physical standards with monetary valuation. All indicators of WS are monetary indicators. This is not surprising as the assumption of unlimited substitutability allows commensurability such that all values can be expressed in monetary terms. SS, on the other hand, with its focus on environmental sustainability, is measured either with physical indicators or with hybrid ones. Inevitably, not all indicators can be covered. For example, I do not include the so-called Green Accounting Research Project (GARP) (Markandya et al., 2000) since its main objective is not to construct a sustainability indicator, but is confined to producing damage cost estimates mainly for air pollution. Similarly, I do not discuss any environmental indicator systems, whose main objective is environmental monitoring without a clear sustainability rule attached to it, such as, for example, the pressure-state-response indicator set of the Organisation for Economic Co-operation and Development (OECD, 2001). The same is true for accounting tools such as physical input–output tables used in, for example, the national accounting matrix including environmental accounts (NAMEA) (see, for example, Steenge, 1999). I also do not cover the concept of environmental space (Hille, 1997). Some of its basic ideas are taken on board in ecological footprints and material flows, but it is less well known than these two other indicators, which are discussed. Vitousek et al.’s (1986) famous indicator of the human appropriation of net primary productivity is similarly not included. The concept of

Indicators of sustainability

141

ecological footprints is inspired by and partly built upon this indicator. The same is true for energy-based indicators called emergy/exergy based on thermodynamics (see Odum,1996; Herendeen, 1999; Ferrari et al., 2001; Dincer, 2002). No doubt many others could be listed here as being omitted from the analysis, but I hope to cover the best known and most popular indicators of both weak and strong sustainability. Pezzey and Toman (2002a, p. 213) in their assessment of progress and problems in the economics of sustainability come to the conclusion that ‘theoretical work has vastly outstripped empirical work’. I agree with this assessment. Partly, the gap between theoretical and empirical work is to be explained by data problems – both in terms of missing data and in terms of poor quality of existing data. Often, heroic assumptions and crude simplifications need to be made for empirical indicators to resemble even faintly the theoretical ideal. But theory is most useful if it can be applied in reality and substantial effort has already been undertaken on developing empirical indicators of sustainability. It is the objective of this chapter to review and critically assess these efforts, always conscious of the limits and problems such indicators encounter in putting a theoretical ideal into practice. Having noted the data problem, I will mainly concentrate on methodological aspects in the critical analysis. It will be seen that all empirical sustainability indicators encounter substantial methodological and other criticism. This chapter thus extends the more theoretically oriented contributions published before in this series, such as Aronsson and Löfgren (1998) and Pezzey and Toman (2002a), with a distinctively empirically oriented analysis. It is organized as follows: section 2 starts with genuine savings and the index of sustainable economic welfare as monetary indicators of weak sustainability. Section 3 discusses ecological footprints and material flows as physical indicators in the spirit of SS. As hybrid approaches, I deal with the concept of sustainability gaps, the Greened National Statistical and Modelling Procedures project, better known under its acronym GREENSTAMP, and the ‘sustainable national income according to Hueting’ (SNI) in section 4. All these are indicators of SS. Section 5 concludes.

2.

MONETARY INDICATORS

2.1

Genuine Savings

2.1.1 Justification and basic idea Genuine savings (GS) is an indicator of WS within the welfare-theoretic approach to the economics of sustainability, building upon Solow’s (1974) and Hartwick’s (1977) path-breaking work. The term ‘genuine’ was introduced by

142

Environmental and resource economics 2004/2005

Hamilton (1994) to distinguish genuine savings, which refers to changes in all utility-relevant stocks of capital including natural capital, human capital as well as (in principle at least) social capital,3 from traditional net savings, which refers only to man-made or produced capital.4 Assume a framework in which population is constant and the social welfare function is a discounted utilitarian function with a constant rate of discount. Also assume that dynamic welfare is maximized such that the competitive economy develops along the intertemporally efficient path with all externalities optimally internalized – see, for example, the dynamic optimization or optimal economic growth models in Hartwick (1990), Hamilton (1994, 1996) and Neumayer (1999a, 2003). Further, assume that the productivity of the economy is fully captured by all capital stocks, a condition formally known as stationary technology (Asheim, 2003). For example, all technological progress is captured in manmade capital and human capital. Finally, the assumptions of weak sustainability need to hold, namely that either other forms of capital can substitute for the depletion of natural capital without limit or that natural capital is superabundant or that technical progress can always overcome any apparent resource constraint (Neumayer, 1999a, 2003).5 Within such a framework, it can be shown that the economy of a country cannot be weakly sustainable if its GS rate is below zero (Pezzey, 2002; Pezzey and Toman, 2002b). The policy recommendation following from this result would be to keep GS above zero: invest into all forms of capital at least as much as there is depreciation of all forms of capitals. 2.1.2 Empirical studies Early crude attempts to estimate GS figures have been undertaken by Pearce and Atkinson (1993). Comprehensive GS accounting is, however, confined to estimates undertaken by the World Bank (1997, 2002). Given Kirk Hamilton’s affiliation with the World Bank’s Environment Department, it is perhaps not surprising that the Bank has been the main proponent of GS as an indicator of WS and has estimated GS figures for most countries in the world from 1970 onwards.6 GS data are now included in the annual updates of the Bank’s World Development Indicators.7 In the most comprehensive formulation of GS, it is made operational by the Bank as follows: GS = investment in man-made capital – net foreign borrowing + net official transfers – depreciation of man-made capital – net depreciation of natural capital + current education expenditures Two observations are immediately apparent from this definition: first, (net) investment in human capital is simply approximated by current education

Indicators of sustainability

143

expenditures.8 This is certainly rather crude, but it is difficult to see how investment in human capital could be estimated otherwise for so many countries over such a long time horizon. Dasgupta (2001a, p. C9f.) argues that it is an overestimate since human capital is lost when people die. Against this one might object that part of the human capital might have been passed on so that the human capital is not really lost once individuals die or, to be precise, leave the workforce. In any case, such correction would be difficult to undertake and I will not bother further with the calculation of net investment in human capital here. Second, social capital is not included due to the insurmountable difficulties in measurement. Gross investment in man-made capital is already included in the national accounts as it forms part of gross national product (GNP) or gross domestic product (GDP). Depreciation of man-made capital is less commonly estimated since net national product (NNP) or net domestic product (NDP) play a less prominent role in macroeconomics. Still, some estimations exist, on which one can build. What is much more difficult and contested is measuring the depreciation of natural capital. Natural capital can be depreciated mainly due to two human activities: resource extraction or harvesting and environmental pollution. To start with the latter, the Bank only includes the estimated damage of carbon dioxide emissions, where each ton of carbon emitted is valued at US$20 per metric tonne of carbon. The value is taken from Fankhauser (1995) and is often regarded as a consensus estimate. As concerns depreciation of natural capital due to resource extraction, the Bank includes the following resources: oil, natural gas, hard coal, brown coal, bauxite, copper, iron, lead, nickel, zinc, phosphate, tin, gold and silver. Forests are the only renewable resource taken into account. How should depreciation of natural capital due to resource extraction be measured? This question is controversially debated in the relevant literature (Hartwick, 1977, 1990; Hartwick and Hageman, 1993; El Serafy, 1981, 1989, 1991; Vincent, 1997; Santopietro, 1998; Davis and Moore, 2000). It does have a very simple answer, however, as long as one stays in the framework of a competitive intertemporally efficient economy (Hartwick, 1990; Hamilton, 1994, 1996; Neumayer, 1999a, 2003). In this framework natural capital depreciation is equal to total Hotelling (1931) rent: (P – MC) . R

(4.1)

where P is the resource price, MC is marginal cost and R is resource extraction. In the case of a renewable resource, R would be resource harvesting beyond natural regeneration. One of the major difficulties of applying this theoretically correct method in reality is that data on marginal cost are frequently unavailable. Average costs are more available. The World Bank

144

Environmental and resource economics 2004/2005

therefore replaces marginal cost with the more readily available average costs and calculates depreciation according to the following formula: (P – AC) . R9

(4.2)

What are the main results of GS computations? Figure 4.1 shows the development of GS relative to GNP for regions over the period 1976 to 2000. The OECD countries as well as East and South Asia never have negative GS rates. The same is true for the world as a whole. The Latin American and Caribbean region touches zero and slightly below only some time in the early 1980s. The problematic regions are Sub-Saharan Africa, whose GS rates go negative in the early 1980s, and North Africa and the Middle East with GS rates that are negative throughout the whole period, with few exceptions. At this level of regional aggregation, it already becomes clear that the regions with the greatest natural resource extraction are the most problematic ones. Indeed, if one looks at individual countries, those with major dependence on natural resource extraction are also the ones that often have negative GS rates. The main message from the Bank’s GS computations is therefore that many developing countries that are dependent on resource exploitation are weakly unsustainable (see Hamilton and Clemens, 1999 for more detail). In the case of Sub-Saharan Africa, it is noteworthy that their net savings before natural capital depreciation are often already negative such that their economies are on a weakly

30 20 10 %

0 –10 –20

Source:

World Bank (2002).

Figure 4.1

Genuine savings rates in per cent of GNP

2000

1998

1996

1994

North Africa and Middle East World OECD

1992

1990

1988

1986

1982

1980

1978

1976

–40

1984

Carribean and Latin America East Asia South Asia Sub-Saharan Africa

–30

Indicators of sustainability

145

unsustainable path quite independently of depreciation due to natural resource exploitation. 2.1.3 Critical assessment The concept of GS itself as well as the way it is measured in practice by the World Bank has encountered much criticism. I can only briefly analyse some important aspects here. A more comprehensive discussion is provided in Neumayer (1999a, pp. 154–77). To start with some conceptual problems, the framework in which GS is derived as an indicator of WS is very restrictive. The assumption of intertemporal efficiency is hard to defend as actual economies are highly unlikely to develop along the optimal path. If environmental and other externalities are not internalized, then existing prices and quantities differ from the optimal ones. In this case, positive GS rates can go hand in hand with unsustainable resource exploitation and environmental degradation. This is one of the reasons why positive GS cannot be taken as indicating achievement of WS (Asheim, 1994; Pezzey and Withagen, 1998). Furthermore, even if efficient, an economy need not be sustainable. If it is not, then shifting towards sustainability will change prices. Pezzey and Toman (2002b, p. 17) therefore suggest that ‘sustainability prices and sustainability itself are thus related in a circular fashion: Without sustainability prices, we cannot know whether the economy is currently sustainable; but without knowing whether the economy is currently sustainable, currently observed prices tell us nothing definite about sustainability.’ Even if the economy developed along an optimal path as assumed in the derivation of GS, then positive GS at any moment of time does not indicate WS since WS is defined in keeping per capita utility at least constant over all time, whereas GS at any moment of time merely presents a point measure of sustainability. The upshot of all this is that GS is at best a one-sided indicator of sustainability: negative GS rates signal unsustainability, but positive GS rates cannot be interpreted as an indication that WS has been achieved. Whilst this point is acknowledged by Hamilton and the Bank, it is likely to be ignored by policymakers and others. The danger is that positive GS rates are taken as proof for the achievement of WS. The assumption of a stationary technology is similarly unrealistic. This assumption breaks down if technological progress is partly exogenous in the sense that it is not fully captured by total capital (Weitzman, 1997; Aronsson and Löfgren, 1998). The same is true for changing terms of trade in open resource-trading economies (Hartwick, 1994; Asheim, 1986, 1996; Sefton and Weale, 1996). Developing natural-resource-exporting countries with negative GS might still be weakly sustainable if exogenous technical progress or future terms-of-trade improvements raise future consumption possibilities. The GS rate becomes similarly ambiguous if the discount rate is not assumed to be

146

Environmental and resource economics 2004/2005

constant throughout. For this case Asheim et al. (2003) show that even negative GS rates at any moment of time need not imply weak unsustainability. The GS rule also becomes much more complex if the unrealistic assumption of constant population is abandoned. Results depend on whether population growth is assumed to be exponential and whether social welfare only depends on per capita utility or also on population size (see Hamilton, 2003; Asheim, 2002; Arrow et al., 2003). Not surprisingly, the many developing countries with strong population growth appear to be even less weakly sustainable once population growth is accounted for. On the more practical side, the weak unsustainability of many naturalresource-dependent developing countries crucially depends on the way the Bank calculates natural capital depreciation. Neumayer (1999a, pp. 164–77) and Neumayer (2000a) argue in favour of what has become known as the El Serafy (1989, 1991) method instead of equation (4.2). The formula for the El Serafy method is: 1 (P – AC) . R . —————— , (1 + r)n + 1

[

]

(4.3)

where r is the discount rate and n is the number of remaining years of the resource stock. For simplicity, this is often set equal to the static reserves-toproduction ratio, which is the number of years the reserve stock would last if production were the same in the future as in the base year. If r > 0 and n > 0, then (4.3) will produce a smaller depreciation term for resource extraction than (4.2). If either n or r is large, the depreciation will be rather small. Equation (4.3) is also called the ‘user cost’ of resource extraction since it indicates the share of resource receipts that should be considered as capital depreciation. The formula for the El Serafy method is derived from the following reasoning: receipts from non-renewable resource extraction should not fully count as what El Serafy calls ‘sustainable income’ because resource extraction leads to a lowering of the resource stock and thus brings with it an element of depreciation of the resource capital stock.10 Whilst the receipts from the resource stock will end at some finite time, ‘sustainable income’ by definition must last forever. Hence, ‘sustainable income’ is defined as that part of resource receipts which if received infinitely would have a present value just equal to the present value of the finite stream of resource receipts over the lifetime of the resource. Natural capital depreciation is then the difference between resource rents and ‘sustainable income’. Appendix 1 shows why this reasoning leads to equation (4.3). There exists a range of arguments why it might be better to use the El Serafy method than the World Bank method in real-world computations of

Indicators of sustainability

147

natural capital depreciation. First, Hartwick and Hageman (1993) show that the El Serafy method can be understood as an approximation to equation (4.1), which, to repeat, is the theoretically correct depreciation in a framework of a competitive intertemporally efficient economy. Its main advantage over the World Bank method in equation (4.2) is that the El Serafy method can use average cost without apology as it does not depend on marginal cost. The World Bank method, on the other hand, needs to replace marginal cost with average cost as marginal cost is not readily available. Due to the replacement of marginal with average cost it can also merely represent an approximation to the theoretically correct method. Which of the two methods creates the greater bias is not clear in general. Under certain assumptions about the resource extraction cost function, the two methods can be shown to be two polar cases of the true depreciation value and the bias depends on the elasticity of the marginal cost curve with respect to the quantity extracted (Vincent, 1997; Serôa da Motta and Ferraz do Amaral, 2000). The upshot of this is that even within a framework of a competitive intertemporally efficient economy the El Serafy method can provide a better approximation to natural capital depreciation than the World Bank method. Second, the true appeal of the El Serafy method stems from the fact that it does not depend on the assumption of efficient resource pricing – resource rent growing at the rate of interest according to Hotelling’s (1931) rule – because it is not dependent on an optimization model. In other words, it does not depend on the assumption of a competitive economy developing along the intertemporally efficient path. It is an ‘ex post approach, capable of accounting for any entrepreneurial decisions regarding extraction’ (El Serafy, 1997, p. 222). The World Bank’s method, on the other hand, depends on efficient resource pricing. Now, none of the data the Bank uses are guaranteed to be the ones that would be generated if a country’s economy developed along an optimal path and the Bank does not estimate any shadow values. Hence, for consistency reasons it might be better to use a method such as the El Serafy method, which does not depend on efficient resource pricing either. Interestingly, in Atkinson et al. (1997, p. 60f.) the same authors on whose work the World Bank is built admit that since ‘there is little evidence for efficient pricing of resources in the ground’, it ‘may be advisable to value resource depletion as a user cost with a non-zero discount rate’ according to the El Serafy method. Third, the World Bank method leads to implausible results for many resource-dependent countries. Take Saudi Arabia as an example. Figure 4.2 plots the GS rate of Saudi Arabia over the period 1970 to 1999, once as computed with the World Bank method and once with the El Serafy method using a discount rate of 4 per cent. The message could not be more different: Saudi Arabia is hugely unsustainable if one applies the World Bank method,

148

Environmental and resource economics 2004/2005

80.00

GS (El Serafy method) GS (World Bank method)

60.00 40.00 %

20.00 0.00 –20.00

1998

1996

1994

1992

1990

1988

1986

1984

1982

1980

1978

1976

1974

1972

–60.00

1970

–40.00

Source: Own computations from data in World Bank (2002) and British Petroleum (various years).

Figure 4.2

Genuine savings rates for Saudi Arabia in per cent of GNP

but has no apparent problems with WS if the El Serafy method is employed. Application of the World Bank method leads to results that are hard to believe. The World Bank (2003) estimates the total natural capital wealth of Saudi Arabia in 1994 at about US$ 807 billion as a low estimate and US$ 2.3 trillion as a high estimate. This contrasts with a total value of natural capital depreciation of Saudi Arabia over the period 1970–99 estimated at US$ 1.5 trillion To be fair, it has to be said that the way the World Bank estimates a country’s natural capital stock itself is methodologically different from the way it estimates the depreciation of it (see World Bank, 1997, pp. 18f. and 37f.). The natural capital stock is underestimated relative to its depreciation. Still, the World Bank method paints a picture of Saudi Arabia that is hard to believe. The El Serafy method, on the other hand, takes into account the country’s enormous oil and natural gas reserves and thereby comes to an entirely different conclusion about the WS of the Saudi Arabian economy. These enormous reserves are simply ignored in the World Bank method due to its assumption of efficient resource pricing. But, as El Serafy (2001, p. 205) stresses, ‘there should clearly be a fundamental difference for national accounting between extracting a given volume out of a large or a small stock. Extracting the same volume amounting to 5 per cent of the stock has different implications for income and sustainability from extracting the same quantity if it amounts to 50 per cent of the stock.’

Indicators of sustainability

149

The change in results with regard to GS once the El Serafy method is employed is not restricted to Saudi Arabia. Instead, many of the countries which appear unsustainable according to the World Bank no longer are once the El Serafy method is used. This is not true for all countries, but true for most of them and in particular those with substantial reserves (Neumayer, 1999a, 2000a). Hamilton (2000, p. 5) admits that the World Bank method ‘arguably over-estimates the value of resource depletion, particularly for countries having large reserves to production ratios’. But the GS figures are published by the Bank without any such qualification. Of course, the El Serafy method is not without problems either. To start with, one needs to choose a discount rate and it is far from clear what the right discount rate should be. A prudent rule would be to choose a rate of discount that approximates the real rate of return to investing the receipts from resource extraction into other forms of capital. Also, one needs to estimate n, the remaining lifetime of the resource stock. Given uncertainty about the future, it is not clear what n should be. A simple rule of thumb is to set n equal to the static reserves-to-production ratio, which becomes updated in every period of accounting as information about the level of extraction and the stock of remaining reserves changes. Since estimates of reserve stocks are sometimes difficult to establish for some resources in some countries, the exact value for n can be contentious, and a prudent accountant should use a lower-bound estimate. For example, only reserves that can be economically exploited at current technology should be included. As a further criticism of the way resource depletion is accounted for, one could argue that the resources depleted in poor developing countries with high resource exploitation go to rich developed countries for their benefit and that therefore the rich countries are responsible if resource exploitation is unsustainable. Proops et al. (1999) have shown that if natural capital depreciation is attributed to the country of resource consumption, then not surprisingly the GS position of resource-exporting developing countries improves, whereas that of resource-importing developed countries is not so safely positive any longer. There are good arguments why resource exploitation should be attributed to the extracting country itself and not to the consuming country, however. This is because the purpose of resource accounting is to try to measure whether and by how much the natural capital stock of a country is depreciating. It simply does not matter who is ‘responsible’ or to ‘blame’ for its depreciation. Furthermore, even in terms of justice one could argue that a resource-rich country is endowed with an extra capital asset by nature. It is up to this country to make sustainable use of it. If it burns all resource rents in consumption instead of investing them into the human capital of its people and in manufactured capital of its economy, this is its own fault. It is the responsibility of each country to keep its own capital stock intact.

150

Environmental and resource economics 2004/2005

A different criticism of the World Bank estimates of GS relates to the way environmental pollution is accounted for. Obviously, carbon dioxide is by far not the only and, as some would argue, not the most relevant of all pollutants. Similar concerns can be raised with respect to renewable resources, as currently forestry is the only one included. Such vital renewable resources as water, soil, fish and, more generally, biodiversity are missing due to lack of adequate data. The Bank is not ignorant in this respect, but sees itself unable to compute environmental damage costs for other environmental pollutants and other renewable resource degradation due to lack of data. One of the dangers is that the WS position of major polluters, particularly the developed countries, is overestimated. That the more developed countries by and large do not become detected as weakly unsustainable is mostly to be explained by their usually quite high net saving rates. Nevertheless, these countries would no longer have such outstandingly good WS performance if more pollutants were taken into account. Indeed, Atkinson et al. (1997, p. 93) show that the WS position of developed countries becomes less comfortable and in a few instances there are even signs of unsustainability if nitrogen oxide, sulphur dioxide and particulate matter emissions are taken into account in addition to carbon dioxide. As a final criticism, it is not entirely clear what specific policies should be undertaken following the detection of negative GS rates. There are many ways to increase GS, from reducing resource depletion and environmental pollution to increasing investment in man made capital. Surely, bringing pollution or resource exploitation down to zero, for example, cannot be a sensible policy as it would be a highly inefficient way of raising GS. The World Bank seems to favour raising investment in man-made capital as the way out of negative GS rates, as becomes clear in the following quotation from World Bank (1997, p. 35): ‘The depressed rates of genuine saving . . . represent an opportunity not seized. . . . [I]t is often the gross saving effort that is insufficient in these countries, which points the finger squarely at broader macroeconomic policies’. 2.2

Index of Sustainable Economic Welfare (ISEW) and Genuine Progress Indicator (GPI)

2.2.1 Justification and basic idea The basic idea of the Index of Sustainable Economic Welfare (ISEW), also known under the name Genuine Progress Indicator (GPI), is to provide a substitute indicator for GNP or GDP, which are regarded as misleading indicators of both current welfare or instantaneous utility and sustainability. It stands in the tradition of and indeed partly builds upon earlier attempts to provide a more comprehensive indicator of welfare and to incorporate environmental and/or sustainability aspects into such an indicator – see, for example, Nordhaus and

Indicators of sustainability

151

Tobin’s (1972) Measure of Economic Welfare (MEW), Zolotas’s (1981) Economic Aspects of Welfare (EAW) and Eisner’s (1990) Total Incomes System of Accounts (TISA).11 The MEW and the EAW take some environmental aspects into account. The MEW adjusts the welfare measure for ‘disamenities of urban life’ such as ‘pollution, litter, congestion, noise’ based on hedonic valuation studies.12 The EAW subtracts air pollution damage costs together with half of the estimated control costs for air and water pollution and the full control costs for solid wastes from the welfare measure. It also deducts the costs of resource depletion. The TISA, on the other hand, does not include any environmental aspects in its measurement, but like the MEW and the EAW seeks to broaden the concept of capital and investment accounted for. Daly et al. (1989) and Cobb and Cobb (1994) were the first to propose and develop an ISEW for the USA. Building upon their work, computation of an ISEW usually starts from personal consumption expenditures. These expenditures are weighted with an index of income inequality. Then, certain welfarerelevant contributions are added, whereas certain welfare-relevant losses are subtracted. As an example, take the US study of Cobb and Cobb (1994): after having weighted personal consumption expenditures by a modified Gini coefficient of pre-tax income distribution data, they add the estimates of the value of the services from household labour, consumer durables and streets and highways. They also add net private investment into man-made capital and changes in the net international investment position of the USA. They subtract most expenditures on health and education because these are regarded as mostly defensive expenditures. They also subtract expenditures on consumer durables, estimates of the costs of commuting, car accidents, and the costs of environmental degradation such as water, air and noise pollution, loss of wetlands and farmlands, the depletion of non-renewable resources and longterm environmental damages due to CO2 emissions. The ISEW is simply the sum of the weighted personal consumption expenditures and all the mentioned corrections. The guiding idea of these adjustments is to subtract items regarded as contributing to either welfare or sustainability, and to deduct items that reduce either welfare or sustainability. Like GS, the ISEW is an indicator of WS since in adding up environmental and non-environmental values substitutability of natural capital is implicitly assumed. This is somewhat ironic given that most proponents of the ISEW, and Herman Daly in particular, are also proponents of SS. The policy recommendation is to ensure that the ISEW is not decreasing. One can interpret the ISEW loosely as a kind of extended or greened net national product (gNNP), which is defined as ‘comprehensive consumption’ plus GS, where comprehensive means that all utility-relevant flows are included in the consumption vector, not just the consumption of material

152

Environmental and resource economics 2004/2005

goods. The theoretical sustainability foundation of the ISEW then follows from the fact that under certain assumptions preventing GS from becoming negative is equivalent to preventing gNNP from falling (Pezzey and Toman, 2002a, p. 184; Asheim, 2003). 2.2.2 Empirical studies An ISEW has been constructed for Austria (Stockhammer et al., 1997), Chile (Castañeda, 1999), Germany (Diefenbacher, 1994), Italy (Guenno and Tiezzi, 1998), the Netherlands (Rosenberg et al., 1995), Scotland (Moffatt and Wilson, 1994), Sweden (Jackson and Stymne, 1996), Thailand (Clarke and Islam, 2003) and the UK (Jackson et al., 1997). Sometimes these studies come, with only slightly changed methodology, under the name of Genuine Progress Indicator (GPI), as, for example, in the case of Australia (Hamilton, 1999) and the US (Redefining Progress, 1999, 2001). For Australia there also exists a related measure, which comes under the name of sustainable net benefit index (SNBI) (Lawn and Sanders, 1999). The specific methodology employed changes a bit from country to country depending on data availability and the preferences of the authors, but all studies come to the same basic conclusion: starting from around the 1970s or the early 1980s, depending on the country, the ISEW or GPI no longer rises very much or even falls, whereas GNP or GDP continues to rise. As an explanation for this widening gap between ISEW or GPI on the one hand and GNP or GDP on the other, Max-Neef (1995, p. 117) has put forward the so-called ‘threshold hypothesis’: ‘for every society there seems to be a period in which economic growth (as conventionally measured) brings about an improvement in the quality of life, but only up to a point – the threshold point – beyond which, if there is more economic growth, quality of life may begin to deteriorate’. This ‘threshold hypothesis’ is referred to in almost every study of ISEW or GPI and Max-Neef (ibid.) himself regarded the evidence from these studies ‘a fine illustration of the Threshold Hypothesis’. Figure 4.3 provides a nice illustration of the so-called threshold effect looking at the US GPI, which is basically the updated version of the US ISEW. It shows the development of the United States GPI per capita in comparison to GDP per capita. Whilst the two graphs roughly move in parallel with each other, from around the 1970s increasing divergence can be observed. Whereas GDP is still increasing, the GPI is no longer increasing or even slightly falling. This picture is typical for practically all ISEW or GPI studies. 2.2.3 Critical assessment As pointed out above, the ISEW can be interpreted as a kind of gNNP and under certain assumptions preventing GS from becoming negative is equivalent to preventing gNNP from falling. Whilst this provides the sustainability

Indicators of sustainability

30 000

153

GDP p.c. GPI p.c.

25 000

$US

20 000 15 000 10 000

Source:

2000

1995

1990

1985

1980

1975

1970

1965

1960

1955

0

1950

5 000

Redefining Progress (various years).

Figure 4.3

United States GDP versus GPI per capita (US$ of 1996)

foundation for the measure, it also means that all the critical points raised in our discussion of GS similarly apply to the ISEW. I will not discuss them again here, but concentrate on the problematic aspects specific to this measure. Many authors have criticized various aspects of the ISEW (see Nordhaus, 1992; several authors in Cobb and Cobb, 1994; Atkinson, 1995; Neumayer, 1999a, 1999b, 2000b). The criticism ranges from general conceptual points to detailed methodological aspects. Before coming to the detailed methodological criticism, let us note two fundamental points. First, contrary to GS, the ISEW is not rigorously derived from a theoretical model. Thus, whilst it has a basic theoretical foundation as pointed out above, the specific adjustments undertaken and their justification are often somewhat ad hoc, as will be seen in our discussion of some specific components of the ISEW below. Indeed, with the notable exception of Lawn (2003), proponents of the ISEW and related indicators have devoted comparatively little effort to theoretically justifying their measure. Second, similar to GS, it is not quite clear what specific policies are to be undertaken if the ISEW or GPI is falling as many different policies could raise it. Ironically, one might even suggest stronger growth in consumption expenditures as these usually form the starting point for the corrections undertaken.13 Let us now turn towards specific components of the ISEW and the method employed for their valuation. I will mainly concentrate here on the two aspects, which are directly and significantly responsible for the ‘threshold

154

Environmental and resource economics 2004/2005

effect’, namely the way in which energy resource depletion and long-term environmental damage are computed. We will see that in both cases the computations depend on arbitrary or even flawed assumptions. Before that, let us briefly discuss two other contested components, however, namely intragenerational income inequality and defensive expenditures. 2.2.3.1 Income inequality As mentioned in the introduction, the ISEW and GPI stand out from other indicators of sustainability in that they try to include intragenerational equity considerations as well. Most studies use the Gini coefficient, where usually one year is set as the base year for the index. In the US GPI, for example, 1968 is set as 100, because ‘it represented the lowest Gini coefficient over the 1950–1998 period, thus the least income inequality. All other years are then compared to this benchmark’ and an index is created (Redefining Progress, 1999, p. 14). The inequality-adjusted consumption expenditures are reached via dividing unadjusted expenditures by this index and multiplying by 100. However, the inclusion and valuation of intragenerational income inequality has encountered many criticisms. Critics such as Mishan (1994, p. 172) argue that ‘all efforts to adjust the welfare index to accommodate changes in distribution . . . must be regarded with misgivings. They are either arbitrary or politically biased and are, therefore, invariably a focus of attack.’ At the least, there is no consensus that less inequality is always better than more inequality. Even if there was a consensus, doubts remain whether such a value judgement can be easily monetized. 2.2.3.2 Defensive expenditures With regard to defensive expenditures, the very concept is elusive and what should count as a defensive expenditure is rather arbitrary. Cobb and Cobb (1994, p. 53) exclude most expenditures for education because they believe that education ‘contributes little to productivity’. In their perspective it should therefore not count as investment. This is clearly at odds with the importance attached by most economists to human capital and education. However, more relevant to this section here is that Cobb and Cobb do not want to count education as consumption either since ‘most schooling appears to be defensive. In other words, people attend school because others are in school and the failure to attend would mean falling behind in the competition for diplomas or degrees that confer higher incomes on their recipients’ (ibid., p. 54). Following this line of argument, one could classify many if not most expenditure items as defensive in character. For example, if health expenditures are defensive expenditures against illness, why should food and drinking expenditures not count as defensive expenditures against hunger and thirst? Are holiday and entertainment expenditures to be considered defensive expenditures against boredom? Daly et al. (1989, p. 78) defend their concept of

Indicators of sustainability

155

subtracting defensive costs by saying that ‘ “defensive” means a defense against the unwanted side effects of other production, not a defense against normal baseline environmental conditions of cold, rain and so on’. But even accepting Daly et al.’s definition, one could argue that at least part of food, drink, entertainment and holiday expenditures are caused by the stressful, exhausting and boring modes of modern production that make these expenditures necessary as a defence against their unwanted side-effects. As the revised System of National Accounts points out: ‘Pushed to its logical conclusion, scarcely any consumption improves welfare in this line of argument’ (Commission of the European Communities et al., 1993, p. 14). 2.2.3.3 Resource depletion With regard to resource depletion, studies differ in that some use the resource rent method, whereas others use the replacement cost method. Those using the resource rent method, such as Daly et al. (1989), Stockhammer et al. (1997), Diefenbacher (1994) and Guenno and Tiezzi (1998) deduct total resource rents from consumption expenditures according to equation (4.2); that is, in the same way that the World Bank computes depreciation of the natural capital stock for its GS rates. As we have seen in the discussion of GS, there might be good reasons to employ the El Serafy method instead, which leads to lower estimates of the true depreciation of natural capital due to resource exploitation. However, much more problematic is the replacement cost method, which has been used by the majority of studies and I will concentrate the discussion on this. Cobb and Cobb (1994) were the first to use the replacement cost method. Each barrel of oil equivalent was valued at a replacement cost which was assumed to escalate by 3 per cent per annum in real terms between 1950 and 1990 and was anchored around an assumed cost of $75 in 1988. Castañeda (1999), Rosenberg et al. (1995), Moffatt and Wilson (1994), Jackson and Stymne (1996), Jackson et al. (1997), Hamilton (1999) and Redefining Progress (1999) have all followed Cobb and Cobb’s example with slight modifications. It is already questionable to assume that all non-renewable energy resource consumption needs to be replaced fully by renewable resources given that there are still huge reserves of non-renewable resources available for many years to come (British Petroleum, 2002). I will concentrate on the 3 per cent escalation factor, however, which clearly gives rise to a threshold effect if resource consumption is not falling and GNP/GDP is not rising too much. As a rationale for this assumption of constantly increasing replacement costs, Cobb and Cobb (1994, p. 267) refer to the costs per foot of oil drilling which they report to have increased by about 6 per cent per annum in real terms during the period of high oil prices in the 1970s, which triggered the exploration and drilling of more difficult to exploit oil fields. They reason that ‘when the limits of a resource are being reached, the cost of extracting the next

156

Environmental and resource economics 2004/2005

unit is more costly than the previous unit’ and that ‘this principle presumably applies also to renewable fuels, though not as dramatically as to oil and gas’, which is why the escalation factor is assumed to be 3 per cent in real terms instead of 6 per cent. Especially with respect to renewable energy resources, such reasoning might be erroneous, however. The most likely candidate for replacing non-renewable fuels is renewable solar energy, with solar energy influx many times exceeding current world energy demand. Also, costs for solar energy use are currently high because the technology is still in the early stages of development, but costs will fall over time as technology improves. Instead of assuming replacement costs to escalate by 3 per cent per year, it might therefore be more appropriate to assume that replacement costs are falling over time. Neumayer (2000b) has shown that if replacement costs are not assumed to escalate, then depletion of energy resources no longer contributes to a ‘threshold effect’. So far, I have not made clear whether resource use refers to the domestic extraction or the domestic consumption of resources, which could to a large extent be imported. The reason is that studies differ on this respect. On the one hand, all studies using the resource rent method value the extraction of resources. This is correct, as the resource rent method attempts to determine the sustainable parts of an income stream derived from resource extraction. Only rents from resource extraction, not from consumption, enter the national income accounts. Therefore, to deduct the value of consumption instead would mean to deduct something that has never been added in the first place. On the other hand, the studies using the replacement cost method are not consistent in their reference point. Whereas the revised US ISEW in Cobb and Cobb (1994) and the US GPI estimate the cost for replacing national extraction, the Australian GPI and the Chilean, Dutch, Scottish, Swedish and UK ISEW estimate the cost for replacing national consumption of non-renewable resources. Methodologically correct is the valuation of consumption, not extraction. This is because the rationale behind the method is to replace non-renewable resource use. Where these resources come from, whether they are imported or domestically extracted, simply does not matter. The idea behind the replacement cost method is not to cancel out non-sustainable income streams, as it is with the resource rent method. Instead, the idea is to estimate the costs of replacing all non-renewable resources in use for the production of goods and services. 2.2.3.4 Long-term environmental damage Even more problematic is another component of the ISEW/GPI. Almost all studies compute annual values of long-term environmental damage, or the costs of climate change, as this item is sometimes also called. Strangely, however, all studies but the Australian GPI let this annual damage accumulate over time. Most studies follow the approach taken by Daly et al. (1989) and Cobb and Cobb (1994) in

Indicators of sustainability

157

valuing each barrel of oil equivalent of annual non-renewable energy resource consumption at $0.50 in 1972 dollars. This value is deducted from the ISEW in this year, but also in all following years. Similarly, in any given year not only the value for current resource consumption is deducted, but the values from all past years as well. Cobb and Cobb (1994, p. 74) provide as justification for this accumulation approach that they imagined that a tax or rent of $0.50 per barrel-equivalent had been levied on all nonrenewable energy consumed during that period and set aside to accumulate in a noninterest-bearing account . . . . That account might be thought of as a fund available to compensate future generations for the long-term damage caused by the use of fossil fuels and atomic energy.

Jackson et al. (1997, p. 23) realize in their computation of the UK ISEW that ‘the major problem with this approach . . . is the arbitrary way in which a charge is calculated’. Instead they purport to value each tonne of greenhouse gas emissions with its marginal social cost, which they correctly define as reflecting ‘the total (discounted) value of all the future damage arising from that tonne of emissions’. Strangely, however, they follow Cobb and Cobb’s (1994) lead in letting this damage accumulate over time. Stockhammer et al. (1997) similarly compute marginal social damage costs for the Austrian ISEW – and let the estimated damage accumulate over time. This is simply wrong, however, as the marginal social cost per tonne of carbon emitted already derives from the total present discounted value of all future damage caused by this tonne of carbon. Accumulation of the annual damage therefore leads to multiple counting of the same effect. Accumulation leads to an almost exponentially increasing term to be deducted from consumption expenditures, and Neumayer (2000b) demonstrates that if long-term environmental damage is not accumulated, then it no longer gives rise to a ‘threshold effect’. 2.2.3.5 An example: the UK ISEW As an example of the extreme sensitivity of results with respect to changing the methodology for computing resource depletion and long-term environmental damage, let us have a look at the case of the UK. Figure 4.4 shows the development of GDP and the original ISEW, indexed with base year 1970 (data taken from Jackson et al., 1997). Whilst the ISEW grows faster than GDP at the beginning, the ISEW starts falling in the early 1980s and there is a rising gap apparent between GDP and the ISEW – a fine example of the alleged ‘threshold effect’. If, however, the costs of replacing resource depletion are no longer escalated and long-term environmental damage is not accumulated, then one can see that the new, corrected ISEW is rising in line with GDP, and indeed faster. In particular, there is no longer any evidence for the ‘threshold effect’, which has completely disappeared.

158

Environmental and resource economics 2004/2005

180 170 160 150 140 130 120 110 100 90 80

1996

1994

1992

1990

1988

1986

1984

1982

1980

1978

1976

1974

1972

1970

Index of UK GDP Index of UK ISEW Index of corrected UK ISEW

Note: ‘Corrected’ means ISEW without 3 per cent escalation factor for resource depletion and without accumulation of carbon dioxide damage. Source:

Jackson et al. (1997) and own computations from World Bank (2002).

Figure 4.4 Index of GDP, ISEW and corrected ISEW for the United Kingdom (1970 = 100)

3.

PHYSICAL INDICATORS

3.1

Ecological Footprints: Measuring Sustainability by Land Area

3.1.1 Justification and basic idea The concept of ecological footprint (EF) focuses on environmental sustainability rather than intergenerational equity more generally. Its objective is to translate all the ecological impact of human economic activity into the ‘area required to produce the resources consumed and to assimilate the wastes generated . . . under the predominant management and production practices in any given year’ (Wackernagel et al., 2002, p. 9266). Since the focus is on consumption, the required land area is attributed to the consumer rather than the producer since the consumer rather than the producer is deemed responsible for the impact. That is, for example, resources extracted in a developing country, but exported to a developed country, count towards the ecological footprint of the developed country. Land rather than money is taken as the unit of accounting since, according to its proponents, ‘monetary analysis is misleading as it suggests substitutability, allows for the discounting of the

Indicators of sustainability

159

future and focuses on marginal rather than absolute values’ (Wackernagel et al., 1999, p. 376f.). EF is regarded by its proponents as an indicator in the spirit of SS.14 If the EF exceeds the bioproductive land area available, then the carrying capacity of the land area is exceeded. This is called an ecological deficit and the economic activity causing the EF is judged to be unsustainable. The EF is linked to the somewhat older concept of a sustainable population size. If the EF of a country, for example, exceeds the bioproductive land area available, then this can also be interpreted to the effect that the area’s population is bigger than its sustainable size at given levels of per capita ecological footprints. Ironically, from this perspective there would be unsustainable ‘overpopulation’ in the developed countries, which as will be shown below typically have ecological deficits, rather than in the developing countries. Proponents of EF usually do not emphasize the link to sustainable population size, possibly because they want to stress that consumption levels causing the high EF are unsustainable rather than blaming ‘over-population’. The concept of EF builds on earlier measures of the impact of humans on ecosystems such as Vitousek et al.’s (1986) measure of human appropriation of so-called net primary productivity (NPP) and Odum’s (1996) accounting of energy flows. The following impacts are included (Wackernagel et al., 2002): (1) crop growing for food, animal feed, fibre, oil and rubber; (2) animal grazing for meat, hides, wool and milk; (3) harvesting of timber for wood, fibre and fuel; (4) fishing in oceans and freshwater; (5) infrastructure for housing, transportation, industrial production and hydro-electric power; and, finally, (6) fossil fuel burning. In addition, 12 per cent of the bioproductive land area is reserved for biodiversity conservation. Due to data problems, neither waste production nor freshwater withdrawal are included. Also note that the extraction of non-renewable mineral resources is not included at all and non-renewable energy resources are taken into account only with respect to the land area required to cope with the environmental damage of fossil fuel burning (see below). The reason is probably that it is difficult to convert non-renewable resource extraction into a required land area. Obviously, not all bioproductive land is the same. All land area is therefore standardized into one common global measurement unit using yield and equivalence factors. Equivalence factors make different categories of land use roughly comparable with each other, whereas yield factors make land of the same land use category, but with differing productivity, comparable. Proponents of EF emphasize that, wherever possible, they use publicly available governmentally approved data and that their calculations are conservative in the sense of under- rather than overestimating the EF (Wackernagel and Silverstein, 2000; Wackernagel and Yount, 2000; Wackernagel et al., 2002). Of all the human impacts, accounting for fossil fuel is the most important

160

Environmental and resource economics 2004/2005

one, responsible for slightly less than half of the global EF in 1999. This socalled energy footprint is the one that has grown fastest over the last decades and in which the disparity between the developed and developing countries is largest. It is also the most contested component of EF, however. It is calculated as the forest land area required to hypothetically sequester enough carbon from the atmosphere to avoid any increase in the atmospheric concentration of carbon. This is done under the assumption that about 35 per cent of carbon emissions are absorbed by the world’s oceans15 and, somewhat strangely, that nuclear energy and fossil energy are treated equally ‘because of inconclusive data about the long-term area demand of nuclear power’ (Wackernagel et al., 2002, p. 9267).16 It is perhaps counter-intuitive that the estimated land area can exceed the actually existing ecologically productive land area on earth. For example, a forest logged down at twice its regeneration rate is accounted for at twice its area (Wackernagel et al., 2002). This is taken as a sign of unsustainability: ‘humans are consuming resources at a rate that would require more land than actually exists’ (Wackernagel and Yount, 2000, p. 26). 3.1.2 Empirical studies The global bioproductive land area is estimated at about one quarter of the earth’s surface (WWF, 2002). The most recent comprehensive study tracking the ecological impact of human activity on a worldwide scale comes to the conclusion that ‘humanity’s load corresponded to 70 per cent of the capacity of the global biosphere in 1961, and grew to 120 per cent in 1999’ (Wackernagel et al., 2002, p. 9266). In per capita terms, an EF of 2.33 ha per capita stands in contrast to existing global biocapacity of only 1.91 ha per capita. An EF has also been calculated for nations, regions and even cities (see Chambers et al., 2000, pp. 133–44). Some nations and regions, particularly the developed ones, have a much larger EF than bioproductive land area available and therefore an ecological deficit. Not surprisingly, all cities examined also run such a deficit. Table 4.1 lists the ecological footprint and the ecological deficit of a selection of countries in per capita terms. 3.1.3 Critical assessment A whole range of methodological and other aspects of EF has encountered criticism – see, for example, van den Bergh and Verbruggen (1999), Ayres (2000) and IVM (2002). On a very fundamental level, one could argue that EF adds up apples and oranges in adding such diverse items as actual land use for agricultural products and purely hypothetical land use for the absorption of carbon dioxide emissions. I will concentrate here on more specific points, however. The first criticism addresses the way in which the EF for fossil fuel use is computed. Ayres

Indicators of sustainability

161

Table 4.1 Ecological footprint and deficit of selected countries and the world (ha per person).

United Arab Emirates USA Canada New Zealand Australia Belgium/Luxembourg UK Ireland Netherlands Japan Germany Russia Brazil World Gabon China Central African Republic India Note: Source:

Ecological footprint

Ecological deficit

10.13 9.70 8.84 8.68 7.58 6.72 5.35 5.33 4.81 4.77 4.71 4.49 2.38 2.28 2.12 1.54 0.92 0.77

8.88 4.43 –5.40 –14.28 –7.03 5.59 3.70 –0.81 4.02 4.06 2.96 –0.35 –3.65 0.38 –26.57 0.51 –8.13 0.09

A negative ecological deficit repesents an ecological surplus. WWF (2002).

(2000) argues that there are many more technical possibilities to sequester carbon from the atmosphere than land-intensive forestry. He mentions pumping compressed carbon dioxide into empty oil and gas wells or liquefied carbon dioxide into the deep oceans. One might dismiss these as merely potential future possibilities, which might or might not be possible and might have undesirable environmental side-effects. More importantly, however, the fossil fuel could be replaced with renewable energy, particularly wind and solar energy.17 Whilst this would be prohibitively costly, the required land area would be much lower than under the forestry option as renewable resources are far more landefficient. Wind turbines and photovoltaic generators could even be placed on land that is not bioproductive or already in use, such as at sea or in deserts or on top of buildings. Such land use would not subtract from the bioproductive land available at all. IVM (2002) has calculated that the energy footprint becomes negligible with little impact on the overall EF if, hypothetically, 50 per cent of world energy demand were satisfied with renewable energy, which

162

Environmental and resource economics 2004/2005

IVM (2002) claims to be technically possible, and for the remaining energy demand mainly low carbon fuels such as natural gas are used. To repeat, the economic cost at the current state of technology would be enormous, but given that EF is blind towards monetary valuation and therefore costs, its proponents cannot argue against considering renewable energy as a hypothetical solution to the carbon dioxide emission problem. It is also no argument against this alternative computation that the use of renewable energy on such a large scale is purely hypothetical at the moment. The same argument would apply to the forestry option employed by the proponents of EF, which is equally purely hypothetical. If the energy footprint becomes negligible, then the global EF is well within the limit of bioproductive land area available. Similarly, many developed countries no longer exhibit an ecological deficit. The second criticism is not directly targeted at EF itself, but at a certain interpretation following from the concept of an ecological deficit, which is derived from an EF. Whereas such a deficit would commonly be regarded by economists as a normal exchange of goods, in which trading partners have differing comparative advantage, to the mutual benefit of both partners (van den Bergh and Verbruggen, 1999), proponents of EF see ecological deficits as inherently dangerous and undesirable, particularly at the level of nation-states. The main reason for this anti-trade bias is that ‘trade reduces the most effective incentive for resource conservation in any import region, the regional population’s otherwise dependence on local natural capital’ (Rees and Wackernagel, 1996, pp. 238f.). As a result, a ‘restoration of balance away from the present emphasis on global economic integration and interregional dependency toward enhanced ecological independence and greater intraregional self-reliance’ is recommended (ibid., p. 241). Willey and Ferguson (1999, p. 2) are even more explicit in proclaiming that ‘all nations should live within their own ecological capacity’. Against this, van den Bergh and Verbruggen (1999, p. 66) maintain that national boundaries are geopolitical and cultural artefacts and therefore have no environmental meaning. A third criticism is that it is not quite clear what the policy recommendation of EF analysis is. Certainly, from the perspective of its proponents, an EF greater than the bioproductive land area available is unsustainable in the spirit of SS. But is an EF just smaller than this area enough? Wackernagel and Yount (2000, p. 38) insist that ‘footprint assessments are not intended to advocate maximum reduction of human load’, but what exactly is advocated then? As a last criticism, it is doubtful whether EF really represents an indicator of SS. EF does not constrain substitutability within natural capital. This does not conflict with the first definition of SS, which refers to the value of total natural capital, but it does conflict with the second definition of SS, which constrains substitutability within natural capital as well and requires maintaining critical functions of natural capital intact. Furthermore, in making total

Indicators of sustainability

163

available bioproductive land area the yardstick against which hypothetical land use is measured, human activities, which are clearly not strongly sustainable, need not be indicated as such by the EF measure. We have seen above that the land area required to hypothetically absorb carbon dioxide emissions can be much reduced if renewable energy production is taken as the hypothetical option rather than reforestation. Doing so would then suggest that the global EF is well within the global bioproductive land area available, even though carbon dioxide emissions would still be clearly beyond the natural regenerative capacity of the global atmosphere, which violates the SS requirement. Even with the reforestation option, the fact that a global ecological deficit exists and EF therefore indicates a violation of SS is purely coincidental. This is because if only there were more bioproductive land area available globally, then according to current EF methodology the world as a whole need not have an ecological deficit. Global human impact would still be in violation of strong sustainability, however, but it would not be indicated as such by EF. 3.2

Material Flows: Measuring Sustainability by Weight

3.2.1 Justification and basic idea The concept of material flows (MF) is inspired by early work by Ayres and Kneese (1969) on industrial metabolism.18 Its starting point is a deep dissatisfaction with environmental policies that focus mainly on emissions and waste products. Its proponents maintain that many environmental problems are caused long before pollutants are emitted and waste is produced because MF need to be moved in order to produce products. It is the sheer size of MF that creates environmental problems, according to its proponents, and this size needs to be reduced substantially in order to lower the pressure on the environment. Reduction of MF is propagated as a good candidate for a ‘one single long-term goal in environmental policy’ (Hinterberger and Wegner, 1996, p. 7). Similar to EF, the concept of MF is regarded by its proponents as an indicator in the spirit of SS (Hinterberger et al., 1997, p. 12). From their perspective, ‘a core environmental condition of sustainability is a physical steady-state system, with the smallest-feasible flows of resources at the . . . input and output boundaries between the technosphere and the ecosphere’ (Spangenberg et al., 1999, p. 492). The concept of MF, first developed by Schmidt-Bleek (1993a, 1993b), is inspired by Herman Daly (1977) and his emphasis on the growing scale or material throughput of the economy as the main cause of environmental degradation. It therefore shares Daly’s emphasis on optimal scale and a limit to or reduction of throughput in a ‘steady-state’ economy rather than efficient allocation as the priority for environmental policymaking. Daly likes to illustrate the importance of scale by invoking the metaphor of a ship that sinks if it is too heavily loaded even if the cargo is efficiently allocated on board. The

164

Environmental and resource economics 2004/2005

emphasis on scale rather than efficiency also partly explains why weight is used as the unit of accounting rather than money. The other reason has to do with the perceived difficulties of monetary valuation of environmental degradation, on which more below. From the perspective of MF, the focus needs to shift from the ‘sink’ side of the economy to the ‘source’ side. This is so for a number of reasons. First, the preoccupation with emissions and waste tends to ignore that all consumption goods come with a hidden ‘ecological rucksack’, which is defined as ‘the sum of all the materials that are not physically included in the economic output under consideration, but have been necessary for production, use, recycling and disposal’ (Spangenberg et al., 1999, p. 498). These typically occur at the resource extraction or harvesting stage. Examples would be earth and rock displaced during non-renewable resource extraction and soil erosion in agriculture (Matthews et al., 2000, p. 1). Second, the precautionary principle is invoked to justify giving priority to a reduction of MF (Hinterberger et al., 1997). Given that uncertainty and ignorance render a precise assessment of the ecological impact of pollutants difficult and imply that many forms of environmental damage cannot be known in advance, reducing MF is seen as a promising alternative as it will reduce the pressure on the environment across the board. Third, Spangenberg et al. (1999) also argue that no environmental policy will ever be able to efficiently control the thousands of substances emitted into the environmental sinks. Their monetary valuation, which is necessary for finding the efficient level of pollution, is regarded as an impossible task. In comparison, it would be much easier to control mineral and energy materials entering the economic system, the number of which are estimated between 50 and 100 in the case of Germany. The aim and policy recommendation is to reduce MF by a factor of four (von Weizsäcker et al., 1997) or, more ambitiously, by a factor of ten (Factor 10 Club, 1994), at least in developed countries, over the next 40 to 50 years, which is regarded as technically feasible (Hinterberger et al., 1997).19 All material inputs are classified into five main categories: abiotic raw materials (mineral and energy resources), biotic raw materials, moved soil (agriculture and forestry), water and air. MF reduction should be achieved in all categories (Hinterberger et al., 1997). However, in most empirical studies such as Adriaanse et al. (1997) and Matthews et al. (2000) water flows are excluded because they ‘are so large that they would completely dominate all other material flows and would obscure the meaning and, thus, the usefulness of the indicators’ (Matthews et al., 2000, p. 8). 3.2.2 Empirical studies The most prominent of empirical studies is that of Adriaanse et al. (1997), which computed MF for Germany, Japan, the Netherlands and the USA over

Indicators of sustainability

165

the period 1975 to 1994. This study has been updated to 1996 in Matthews et al. (2000) and extended to cover Austria as well. Matthews et al. (2000) also distinguishes between MF from different economic sectors as well as MF into different environmental media, namely air, land and water. It also allows us to establish which MF remain in the economy longer than one year and which are dissipative and therefore difficult to recover and recycle. In Adriaanse et al. (1997, p. 7) the hidden flows of imported primary natural resources and semimanufactures are attributed both to the country of final consumption of materials and to the country exporting the materials for further use in another country. In Matthews et al. (2000, p. 43) it is recognized, however, that this would lead to double-counting among trading countries. Hidden flows are therefore attributed solely to the importing country (ibid., pp. 14 and 43). In all five countries under study in Matthews et al. (2000), MF have increased by between 16 (Netherlands) and 28 (USA) per cent between 1975 and 1996. These absolute increases in weight have occurred despite substantial reductions in the material flow intensity of GDP, which has fallen between 26 (USA) and 42 (Japan) per cent. In other words, the decoupling of MF from GDP has not been strong enough to bring about absolute reductions in MF. In cross-country comparison, one is not surprised to find that the Japanese economy has the lowest per capita MF (11.2 million tonnes per capita in 1996), whereas the USA has the highest (25.1 million tonnes per capita in 1996). Note that for these figures MF refers to what Matthews et al. (ibid., p. 7) call domestic process output, that is, materials extracted from the domestic environment plus imported materials used in the domestic economy and flowing to the domestic environment. It does not include domestic hidden flows, which do not themselves enter the domestic economy. A similar picture emerges if the analysis is extended to 15 European Union member states over the period 1980 to 2000, for which Table 4.2 provides information. The decoupling of MF from GDP has not been strong enough to prevent absolute increases in MF in most countries as well as the EU as a whole. Note that in this table the definition of MF differs slightly from those presented above. Other studies have been undertaken for other countries as well as for certain regions, sectors and even infrastructure projects by the Wuppertal Institute for Climate, Environment and Energy and by the Sustainable European Research Institute (SERI), Vienna. Both institutes are major proponents of the concept. The basic idea of MF has also sparked interest in national statistical agencies, particularly in Germany and the Netherlands (see, for example, de Haan, 1999 and Tjahjadi et al., 1999). Dahme et al. (1998) have proposed ranking countries according to their MF and to use these data to construct an extension to the United Nations Development Programme’s Human Development Index (HDI), called ‘Sustainable Human Development Index’ (SHDI).

166

Environmental and resource economics 2004/2005

Table 4.2 Relative change (%) in MF and MF intensity in 15 EU member states, 1980–2000 Country

Austria Belgium, Luxembourg Denmark Finland France Germany Greece Ireland Italy Netherlands Portugal Spain Sweden United Kingdom EU 15

DMI

DMC

DMC DMI per capita per capita

17

2

–5

41 28 14 3 –1 52 31 9 16 45 54 8 11 5

1 5 8 –4 –9 49 25 1 –6 39 48 –4 –1 3

–4 1 –1 –12 –13 35 12 –2 –17 32 39 –10 –7 –3

DMC/euro

DMI/euro

9

–36

–27

35 23 5 –6 –6 38 18 6 3 38 44 2 5 –1

–35 29 –36 –38 –40 11 –58 –31 –43 –23 –14 –35 –40 –34

–9 13 –32 –33 –35 13 –56 –26 –30 –20 –11 –27 –32 –33

Note: DMI: all materials extracted for use in a country plus imported materials; DMC: DMI less exported materials. Source:

Eurostat (2002, p. 16).

3.2.3 Critical assessment The most important criticism of the concept of MF is that it adds up apples and oranges (Gawel, 1998). From an ecological point of view, two forms of material throughput with differing environmental damage impacts cannot be meaningfully added together just because one can express both in weight terms. Without further analysis of what the material throughput consists of and what are its environmental implications, there is no reason to presume that, say, Japan’s MF of 11.2 tonnes per capita is any better than the MF of the USA at 25.1 tonnes per capita. Indeed, one could argue that the very statement that the MF of the US is 25.1 tonnes per capita in 1996 is entirely void of meaning. Similarly, it is pointless simply to rank countries according to the size of their MF. In its prescription to reduce general MF across the board, the concept seems to draw the erroneous conclusion from the difficulties of valuing environmental damage that one cannot distinguish successfully according to differences in environmental damage at all. It is simply not true that, as Hinterberger and

Indicators of sustainability

167

Luks (1998, p. 7) suggest, ‘in most cases it is impossible to distinguish between “good” and “bad” throughput’. In its call for general MF reduction across the board, the concept goes from a rejection of one extreme belief, namely the possibility of comprehensive environmental valuation, to the other extreme, which is seemingly blind towards admittedly incomplete attempts at valuation. The call for general reductions in MF is not guaranteed to be ecologically effective, but is guaranteed to be highly economically inefficient with respect to whatever reduction in environmental damage might be achieved (Gawel, 2000). The failure to appreciate the importance of valuing benefits and opportunity costs unnecessarily renders the concept largely unattractive. Because general reductions in MF are not guaranteed to be ecologically effective, it is also doubtful whether MF can really function as an indicator of SS. Following the policy recommendation of reducing general MF by a certain factor need not reduce the stress on critical functions of natural capital if the specific MF, which are threatening these functions, are not directly addressed. Having presented our criticism of general MF reductions, there is much more potential in the concept once one abandons the idea of such across-theboard reductions. For example, it is true that an environmental policy that merely focuses on the ‘sink’ side of the economy will tend to neglect the many environmental problems that are caused during the entire production process and MF is to be credited with redrawing our attention to this. Furthermore, once one starts distinguishing between more and less dangerous materials, then reductions in those material flows that tend to threaten critical functions of natural capital moves us towards SS. The proponents of MF have started to take these criticisms more seriously. For example, Matthews et al. (2000, p. 3) states that ‘we recognize that it is at the level of sub-accounts – the examination of specific material flows, and categories of like flows – that materials flow analysis will have most relevance to detailed policy-making’. The same document also developed a pilot study for the United States, in which material flows are distinguished according to their physical and chemical properties. Similarly, Hinterberger et al. (1999, pp. 364f.) recognize a need for differentiating material flows and suggest that MF reductions need to be regarded as complementary to fine-tuned environmental policies tackling problems at the ‘sink’ side of the economy rather than substituting for them. Gawel (2000, pp. 165–7) is right, though, in arguing that proponents of MF need to be clear whether they see general MF reductions as the panacea for most if not all environmental problems, or regard differentiated MF reductions as a policy tool complementary to environmental policies targeting specific pollutants at the ‘sink’ side of the economy.

168

4.

Environmental and resource economics 2004/2005

HYBRID APPROACHES

Hybrid approaches are those that combine physical indicators with monetary valuation. Typically, no monetary values are put upon items of natural capital. Rather, only the monetary costs of achieving the standards computed. Roefie Hueting’s (1980, 1991) work is the starting point for a number of hybrid approaches. I will look at the three most important ones: the so-called sustainability gaps, the Greened National Statistical and Modelling Procedures (GREENSTAMP) and the ‘sustainable national income according to Hueting’ (SNI). 4.1

The Starting Point: Hueting’s Pioneering Work

Hueting’s point of departure is the suggestion that human impact has reached a level that threatens the integrity of environmental functions, which represents a ‘new scarcity’ unknown before (Hueting, 1980). His proposal was to define standards that maintain vital environmental functions intact, to estimate the costs of achieving these sustainability standards and to subtract these costs from national income. Also subtracted should be all those expenditures that are defensive and, according to Hueting, are wrongly counted as value added in the national accounts: compensatory, restoratory and preventive environmental expenditures. The resulting ‘sustainable national income’ (SNI) is defined as ‘the maximum attainable level of production and consumption, using the technology of the year under review, whereby the vital functions, that is possible uses, of the physical surroundings remain available forever’ (Hueting and de Boer, 2001, p. 24). Hueting is well aware that his is a ‘partial equilibrium and static approach’ since effects on other sectors of the economy are not taken into account (Hueting, 1991, p. 205). As Hueting (ibid., p. 204) points out, his proposal was provoked by the need for a practical indicator in the face of insurmountable problems of creating a theoretically correct indicator: In the course of a working visit to Indonesia in 1986, I was provoked by the following remark made by the Indonesian minister for Population and Environment: ‘In my policy making I need an indicator in money terms for losses in environment and resources, as a counterweight to the indicator for production, namely national income. If a theoretically sound indicator is not possible, then think up one that is rather less theoretically sound.’

Hueting therefore regards his proposal as a workable, if second-best, alternative to the theoretically correct, but in his view practically impossible, valuation of environmental functions with the help of shadow prices.20 As we shall see, all the three hybrid approaches dealt with here share this basic conviction.

Indicators of sustainability

4.2 4.2.1

169

Indicators Based on Hueting’s Work Sustainability gaps

4.2.1.1 Justification and basic idea The concept of ‘sustainability gaps’, developed by Ekins and Simon (1999, 2001), is based on Hueting’s work even though his work is strangely never cited by Ekins and Simon. Similar to Hueting, the proponents of the concept of sustainability gaps reject the welfare-theoretic approach to the economics of sustainability due to insurmountable problems of measuring natural capital depreciation. Its basic idea is to measure the gap between pre-specified environmentally sustainable standards and current violation of these standards in physical terms and to translate this gap into monetary terms via valuation techniques. Environmental sustainability is defined as ‘the maintenance of important environmental functions’ (Ekins and Simon, 1999, p. 39). The concept thus provides a measure of SS. The suggested standards are as follows (ibid., p. 47): • Stable climate. • Undepleted ozone layer. • Biodiversity at current levels. • No loss of function for non-renewable resources. • Sustainable harvest at desired level for renewable resources. • Limiting emissions to critical loads in order to protect human health. • Maintenance of an unspoilt countryside. • Maintenance of environmental security in restricting environmental risks to low levels. Once the standards are defined, the resulting gap between the standards and current practice can be calculated. One can also calculate the years it would take at current trends to achieve the standards. Going one step further, one can, in principle at least, transform the sustainability gaps into monetary values by estimating the opportunity costs necessary to achieve the sustainability standards. First, one needs to establish the necessary measures to achieve the standards. These measures can be either in the form of reducing the output of certain goods and services whose production causes environmental degradation, or in the form of input substitution and pollution abatement in production processes, or finally in the form of direct restoration and preservation. Next, cost curves have to be established for the implementation of each measure. Then all measures are sorted with respect to their marginal cost in order to arrive at an overall cost curve for achieving the sustainability standard. Hypothetically, the measure with the least cost is undertaken first, then the measure with the next highest cost and so on. In so far as there might

170

Environmental and resource economics 2004/2005

be practical obstacles for following this sequence of least-cost measures, the estimate for the sustainability gap is too low. Ekins and Simon (2001, p. 20) warn very explicitly against the idea of subtracting the monetized sustainability gap from GNP or GDP and against an interpretation of the gap as the actual amount of money that would need to be spent to achieve sustainability: the calculation of sustainability gaps ‘is very much a static, partial equilibrium calculation, representing at a moment in time the aggregation of expenditures that would need to be made to reduce the various dimensions of the physical sustainability gap to zero’. If these expenditures were actually undertaken, however, then prices would change, which contradicts the partial equilibrium assumption. 4.2.1.2 Empirical studies Ekins and Simon (2001) provide empirical estimates of the sustainability gap for the UK and the Netherlands. The one for the UK refers to carbon dioxide and some other air pollutants only, whereas the one for the Netherlands covers more environmental areas due to better data availability. Not surprisingly, their study finds substantial gaps between current practice and the pre-specified environmental standards. In many cases it would take decades to achieve the standards at current trends and for some environmental aspects sustainability standards could never be achieved without reversing current trends. No monetary valuation is attempted due to lack of comprehensive data. Ekins and Simon (2001, p. 21) point out that ‘considerable statistical effort is still needed’ to derive a monetary value of sustainability gaps ‘across all relevant environmental themes’. 4.2.2

Greened National Statistical and Modelling Procedures GREENSTAMP)

4.2.2.1 Justification and basic idea The Greened National Statistical and Modelling Procedures (GREENSTAMP) are the result of a research project financed by the European Community over the period 1994 to 1996. The starting point of GREENSTAMP is also a rejection on mainly practical grounds of the welfare-theoretic approach to the economics of sustainability as represented, for example, by GS (Brouwer et al., 1999). Its proponents believe that it is practically impossible to value depreciation of natural capital reliably and comprehensively, as required by the welfare-theoretic approach. To follow such an approach would be ‘largely illusory for providing a meaningful indicator of sustainability’ (ibid., p. 14). Instead, proponents of GREENSTAMP want to estimate with the help of multi-sector national economic models what the feasible economic output would be if pre-specified environmental standards were to be achieved. In specifying environmental standards, which have to be obeyed, GREENSTAMP is also an

Indicators of sustainability

171

indicator of SS. In estimating the opportunity costs of obeying these standards, the approach is inspired by Hueting. However, its proponents deviate from Hueting’s original proposal to deduct the costs of achieving the environmental standards from actual national income. They believe that such an approach would estimate a ‘sustainable’ income that ‘is probably lower than the national income that could be obtained, and maintained durably, while respecting the norms’ (Brouwer et al., 1999, pp. 15f.). Since achieving the pre-specified environmental standards would imply non-marginal changes, for which the partial equilibrium framework becomes untenable, general equilibrium modelling is the preferred alternative. According to GREENSTAMP proponents, the hypothetical national income that could be obtained while obeying the norms can therefore only be estimated if the feasible economic output itself is subject to modelling (O’Connor and Ryan, 1999). 4.2.2.2 Empirical estimations The GREENSTAMP methodology has been tested with the help of the so-called M3ED (Modèle Economie Energie Environnement Développement) multi-sectoral dynamic simulation model. Model runs have been undertaken, among others, for France (O’Connor and Ryan, 1999) and the Czech Republic (Kolar and O’Connor, 2000). One of the advantages of the modelling approach is that the model can be run with different assumptions about the environmental standards. The modelling is explicitly dynamic and future oriented (ex ante approach). Modelling the transition to the specified environmental standards forms an important part of the analysis. The feasible economic output is therefore estimated over a period of time and projected into the future, which can be done with appropriate assumptions made future values. Accepting that any environmental standards set or assumptions taken about the future are always subjective, GREENSTAMP is defended by its proponents as a valuable exercise to better understand the conditions of achieving sustainability, however defined: ‘The information of most value is not found in the aggregate figures themselves – which are always open to alteration through changing assumptions – but in the richness of information and understanding obtained through construction and comparison of the different model outputs and scenarios’ (O’Connor and Ryan, 1999, p. 130). The model runs for France, for example, have been undertaken for four distinct scenarios ranging from very pessimistic to very optimistic assumptions about technological advances and from very lenient to very stringent environmental standards. 4.2.3

‘Sustainable national income according to Hueting’

4.2.3.1 Justification and basic idea The calculations of a ‘sustainable national income according to Hueting’ (SNI) for the Netherlands, undertaken

172

Environmental and resource economics 2004/2005

by a group of researchers at the Free University Amsterdam and the Wageningen University, also build on Hueting’s work, which this time is very explicitly acknowledged. Like GREENSTAMP, the proponents of SNI realize that the adjustments to national income following the observance of externally imposed environmental standards can only be undertaken in a general equilibrium framework. Contrary to GREENSTAMP’s dynamic and future- as well as transitionoriented modelling approach, the SNI explicitly follows a static comparative or ex post approach. It is defined as ‘the situation of the economy after an instantaneous change towards sustainable resource use’ (Gerlagh et al., 2001, p. 3). The aim is to establish what the income for a given year would have been if the economy had had to obey the environmental standards. Transition dynamics do not matter as two static situations are compared with each other: once before and once after the sustainability standards are imposed upon economic activity. This follows from a desire that ‘the SNI calculations should not be burdened with transition costs’ (Gerlagh et al., 2001, p. 3). In the process of calculation, a range of simplifying assumptions is made (Gerlagh et al., 2001, 2002). For example: •

• •

•

•

• •

As already mentioned, all transition or adaptation costs are ignored as ‘in a way of speaking, it is assumed that the change to a sustainable economy is foreseen in advance, long enough that economic agents are able to integrate this transition in the planning of their investment decisions’ (Gerlagh et al., 2002, p. 164). Abatement costs are assumed to be the same for all sectors as no sector-specific data are available. ‘Defensive’ expenditures, that is expenditures whose aim is environmental restoration, prevention of environmental degradation or compensation for such degradation are subtracted from national income if they enter the national accounts as value added. This follows from the consideration that actual and potential expenditures to reach the specified sustainability standards are essentially substitutes. Costs for remedying environmental problems, which have accumulated over a long time, are also distributed over a long time period instead of attributed to one year only. The labour supply is supposed to be inelastic and the labour market clears through an adjusting wage rate, thus ensuring employment neutrality. The income and price elasticities of various goods need to be specified. The trade balance is assumed to be equal to the national savings

Indicators of sustainability

•

•

173

balance, which is in turn assumed to constitute a constant share of national income. With respect to price changes in world markets, two variants are calculated: one in which prices on the world market do not change, whereas in the other price changes on the world market are presumed to be proportional to price changes in the Netherlands. Similarly, because prices will change following the imposition of environmental standards, the SNI can be compared to national income based either on the initial prices or on the new prices. Together with the two scenarios about price changes on the world market, this creates a total of four variants for the general equilibrium model.

4.2.3.2 Empirical studies Gerlagh et al. (2002) calculate different variants of a Dutch SNI for the year 1990 in an applied general equilibrium model with 27 production sectors. Nine environmental themes are covered: climate change, ozone depletion, acidification, eutrophication, particulate matter and volatile organic compound emissions, heavy metal dispersion into water, dehydration of land and soil contamination. For all these themes, environmental sustainability standards are set such that emissions stay within the natural regenerative capacity of the environment. For the last two themes, this rule translates into a standard of zero dehydration and zero soil contamination. Then abatement cost data based on currently available technologies are collected to estimate the costs of reaching the specified standards. Abatement costs consist of operation and maintenance costs for technical abatement measures in the first place and value added from output losses otherwise where these technical measures have been exhausted and output reductions are the only way left to reduce emissions. In their calculations, Gerlagh et al. (2002) find that the costs of reducing greenhouse gas emissions represent the highest share of the costs of achieving the sustainability standards. They estimate that to reach less than 70 per cent of the sustainability standards is relatively cheap, reducing national income only by about 10 per cent. Further improvements quickly become very expensive, however. Whereas the conventional net national income is estimated at about 450 billion guilders, the SNI, that is, the income where 100 per cent of the sustainability standards are obeyed, is calculated at about 250 billion guilders. In Hofkes et al. (2002) the calculations are repeated for the year 1995 and a comparison is drawn with the calculations for 1990. They find that ‘SNI improves substantially from 1990 to 1995. Growth rates in sustainable income levels exceed growth rates in national income. . . . Over the period 1990–1995 an absolute delinking of economic growth and environmental pressure has taken place’ (Hofkes et al., 2002, p. 21).

174

4.3

Environmental and resource economics 2004/2005

Critical Assessment

Similar to Hueting’s original proposal, the monetary valuation of the sustainability gaps suffers from its partial equilibrium approach for establishing the cost curves. The costs for the implementation of each measure are estimated under the ceteris paribus assumption. However, if all those measures that are necessary to achieve the sustainability standards were effectively undertaken, then the ceteris paribus assumption would become fictitious. The relative prices of consumption goods and input factors would change, as would the extent and structure of environmental degradation. Economic restructuring, feedbacks and interlinkages would have to be considered in a total equilibrium analysis of the economy. This task can only be achieved with comprehensive modelling as undertaken by the other two hybrid approaches. It is also unclear what the appropriate timeframe is for achieving certain sustainability standards. This holds especially true with respect to standards for non-renewable resources. Simon and Ekins postulate that the use of nonrenewable resources must not diminish their ‘function’, which can be achieved via more efficient use, repair, reuse, recycling and substitution with renewable substitutes. However, it is unclear whether the maintenance of functionality must be achieved instantaneously or over a long time period. The latter would be more sensible as there is no immediate danger of running out of non-renewable resources. With respect to GREENSTAMP and SNI, the modelling approach is their chief advantage as it avoids the implausible partial equilibrium assumptions. The hypothetical character of the estimated feasible economic output as the result of a modelling exercise also represents the greatest weakness of these indicators, however. The results and indeed the whole modelling exercise are difficult for laymen to understand. Moreover, the model dependency of the estimates means that the results crucially depend on the underlying assumptions made. The section on the SNI has illustrated this point in listing a number of assumptions needed, all of which are contestable, of course. GREENSTAMP and SNI are therefore more valuable with respect to the research they generate on how to construct abatement cost curves, how to deal with environmental defensive expenditures and what is needed in terms of environmental statistics and reporting both at the firm and at the macroeconomic level. Another important property is that they help to focus the discussion on which environmental standards are considered sustainable. This is explicitly acknowledged by the proponents of GREENSTAMP, which do not regard their calculations as providing one single, all-encompassing indicator of sustainability, but rather as a way to improve the many building stones needed for a better informed policymaking towards sustainability. The SNI, on the other hand, is more ambitious. Whilst it is not supposed to replace national

Indicators of sustainability

175

income, certainly in the eyes of Roefie Hueting it is meant to provide a real alternative to it. With all three hybrid approaches, one needs to be careful in interpreting the estimated monetary value of the sustainability gap, the estimated feasible economic output in the case of GREENSTAMP and the estimated SNI, respectively. A high value for the sustainability gap, a great difference between actual and estimated feasible output or between national income and the estimated SNI can mean either of two things. It can either mean that the actual economy is far away from the sustainability norms or that the economy is close to fulfilling the norms, but doing so would be very costly. The environmental implications can therefore be quite different for the same monetary value. Similarly, a given monetary value for the sustainability gap, a given difference between actual and estimated feasible output or between national income and SNI does not tell us anything about the relative achievement of sustainability with respect to different norms. It could be that certain norms are drastically violated while others are almost achieved or it could be that the economy is equally far away from achieving all norms. Also, a constant or falling value of the sustainability gap, a closing of the gap between actual and feasible economic output or between national income and the SNI tells us nothing about the state of the environment in itself. This is because this could be either the consequence of the economy moving closer to fulfilling the sustainability standards or the consequence of a lowering of costs for achieving the standards due to, for example, technical progress. Detailed knowledge of the sustainability norms and the economy’s distance from these norms is therefore essential and one should never rely on the aggregate monetary calculations alone. The reader might wonder why I have not said anything critical on the concept and treatment of defensive expenditures in hybrid approaches. The reason is that our main concern that I raised in the discussion of the ISEW or GPI, namely the essential arbitrariness of what constitutes a defensive expenditure, is much less pronounced here. This is because such expenditures refer only to environmental expenditures and have a clear reference point in the form of the environmental standards set.

5.

CONCLUSION

This chapter has reviewed the most prominent indicators of WS and SS. It has presented the basic idea and justification for each indicator, has reviewed empirical studies and has discussed the criticisms they encounter. Much emphasis has been placed on the critical and problematic aspects. This is not to say that these indicators do not also have strengths and point to aspects of

176

Environmental and resource economics 2004/2005

sustainability that are worth paying attention to. The basic point of GS is that resource-dependent developing countries need to ensure they invest enough in alternative forms of capital to be weakly sustainable, and it is certainly a valid one. The ISEW rightly claims that GNP/GDP are poor indicators of either current welfare or sustainability. In their defence, it should be noted that they were never constructed as indicators of welfare or sustainability. The revised system of national accounts states this with unambiguous clarity: ‘Neither gross nor net domestic product is a measure of welfare. Domestic product is an indicator of overall production activity’ (Commission of the European Communities et al., 1993, p. 41). Still, the proponents of ISEW are correct in arguing that policy makers, the media and some economists in their simplified models literally equate GNP/GDP with welfare. Ecological footprints and material flows remind us that the environmental impact of the goods and services we consume go far beyond what is contained in them or directly observable from their use. Both caution us against such fallacies as, for example, believing that a car fuelled by liquid hydrogen is a zero emissions car, as the German car manufacturer BMW claims in one of its advertisements. Sustainability gaps, GREENSTAMP and SNI induce us to think about which environmental standards one would want to impose on economic activity and to calculate a first-approximation estimate of their opportunity costs. Having said that all indicators make some valid basic point, the devil lies in the detail. Problems start when these indicators try to come up with a concrete measurement of sustainability. With respect to WS, it is true that resource-dependent developing countries might have problems with achieving sustainability, as GS claims. But do most of them have massive problems, as the World Bank’s method for computing natural capital depreciation would suggest, or do only some of them have less drastic problems, as the El Serafy method would suggest? The World Bank method certainly overestimates natural capital depreciation, but the El Serafy method might underestimate it. An indicator that comes to radically different conclusions for quite a few countries depending on which method is used, with none of them obviously superior, is equally obviously problematic. Indeed, I have dealt with two popular methods, but there are others with different merits and problems. A panel on ‘Integrated Environmental and Economic Accounting’ in its report to the US Congress came to the conclusion that ‘no single valuation method has been shown to be free of problems’ (Nordhaus and Kokkelenberg, 2000, p. 102). This represents a huge setback to efforts at providing reliable and relatively non-contentious estimates of the monetary value of natural capital depreciation. It is a sure recipe for confusion amongst policymakers. The ISEW studies claim that even developed countries have massive problems in achieving WS. However, this claim was shown to be the consequence

Indicators of sustainability

177

of a dubious, perhaps even flawed, methodology for computing natural resource depletion and long-term environmental damage. Unless these methodological shortcomings are corrected, the ISEW cannot be taken seriously. Even then, questions remain with respect to the treatment of income inequality and so-called defensive expenditures. The exclusion of welfareenhancing factors such as increasing life expectancy, leisure time and consumer-good quality and the like are other aspects which I could not deal with in detail and which are discussed in, for example, Neumayer (1999a, 1999b, 2003). One cannot exclude the possibility that (some) developed countries are weakly unsustainable, but we know for certain that they are not strongly sustainable. Indeed, we know this with such certainty that one wonders why anyone would bother to construct an indicator such as EF, whose very objective seems to be demonstrating just this. I have explained above how the overshoot of the EF beyond the bioproductive land area available crucially depends on its method of translating carbon dioxide emissions into land area. As a matter of fact, carbon dioxide emissions are well beyond the natural absorptive capacity of the atmosphere and are on the rise. This cannot be strongly sustainable as it will disturb and in many cases destroy the natural functions of the global atmosphere. But we knew this a long time ago and we did not need EF to tell us what we knew all the time with what is arguably the result of a complex methodological artefact. Outspoken critics of the concept maintain that due to methodological flaws, EF does not have ‘any value for policy evaluation or planning purposes’ (Ayres, 2000, p. 349) and is ‘unsuitable as a tool for informing policy-making’ (van den Bergh and Verbruggen, 1999, p. 71). As long as the methodology for computing the land area necessary to bring carbon dioxide emissions within the natural absorptive capacity is unchanged, I have to agree with this judgement. Even if the methodology became changed on these aspects, doubts regarding the validity and reliability of the EF computations remain for reasons I have no space to go into here (see the special issue of the journal Ecological Economics, Volume 32, Issue 3, 2000). Proponents of the concept of MF are correct in pointing out the misery of an environmental policy that is obsessed with emissions and waste and ignores the environmental damage created along the whole production and consumption process of goods and services. Also, there is some fundamental truth in the statement that ‘unless economic growth can be dramatically decoupled from resource use and waste generation, environmental pressures will increase rapidly’ (Matthews et al., 2000, p. v). However, there are certainly more effective and more efficient ways to achieve SS than to reduce MF across the board by a factor of ten (or four or 20 or whatever, for that matter). When material flows are differentiated according to their potential to threaten critical functions

178

Environmental and resource economics 2004/2005

of natural capital, then the comprehensive coverage of potential environmental impacts ‘from cradle to grave’ has much to offer. I fully agree with Hinterberger et al. (1999, p. 371) that the concept of MF is still in its infancy and might well have great potential if future research pushes it in the right direction. The general importance of material flows and its potential promise is therefore not contested here. What is contested is whether material flows can be aggregated by weight and whether the recommendation to reduce material flows across the board makes either economic or ecological sense. All hybrid approaches provide interesting information on how far away an economy is from reaching pre-specified environmental standards. Problems start where monetary valuation begins. The concept of sustainability gaps suffers from the untenable ceteris paribus clause. With large-scale abatement undertaken, quantities and prices change, which defeats the partial equilibrium assumption. Only general equilibrium modelling can overcome this problem, and both GREENSTAMP and SNI provide interesting exercises in modelling the costs of reaching pre-specified environmental standards. However, because general equilibrium modelling is required, many assumptions need to be made, which by necessity are contentious. As some of their proponents readily admit, hybrid approaches, ‘whatever concept they engage, are highly sensitive to model calibration, specification of environmental standards, technological change and other assumptions used’ (O’Connor et al., 2001, p. 16). The SNI calculations roughly point out that to achieve SS would cost about 50 per cent of national income in the case of the Netherlands. This is quite a substantial cost, which renders it very doubtful whether any country would be willing to incur such a cost. Fortunately, the SNI provides an upperbound estimate. This is because the comparative static SNI approach necessarily overestimates the true costs of achieving SS as it is based on current technology. SS could only be achieved over a long period of time, however, in which technology would change, which would make the move towards SS much cheaper. All in all, our critical review of sustainability indicators paints a rather bleak picture. One can live with the fact that there are two opposing paradigms of sustainability that need their own indicators. What is more disturbing is that there is no entirely convincing and reliable way of measuring to what extent WS or SS is achieved. As concerns WS, I believe that the GS approach with the El Serafy method employed holds the greatest promise. As concerns SS, the hybrid approaches are also promising, in spite of their various difficulties and problematic aspects. The quest for developing better sustainability indicators is far from over. An important step, which I have not dealt with in detail, would be to provide more comprehensive and better-

Indicators of sustainability

179

quality resource and environmental data. As mentioned in the introduction, empirical indicators of sustainability face the problem of missing as well as poor-quality data. National and international statistical agencies are still far from providing the necessary data at sufficient quality. Better data will not cure methodological flaws, however, and I hope that the analysis here has helped to focus the minds of interested researchers on where most work still needs to be done to develop more valid and reliable indicators of sustainability.

APPENDIX 1: DERIVATION OF USER COSTS ACCORDING TO THE EL SERAFY METHOD The formula for computing user costs according to the El Serafy method can be derived as follows: let P be the resource price, AC average extraction cost, R the amount of resource extracted, r the discount rate and n the number of remaining years of the resource stock if extraction was the same in the future as in the base year; that is, n is the static reserves-to-extraction ratio. Then the present value of total resource rents RR ≡ (P – AC) . R is equal to:

[

]

1 RR 1 – —————— n RR (1 + r)n+1 ∑ ————= ————————————. i = 0 (1 + r)i 1 1 – ——— 1+r

(4A.1)

The present value of an infinite stream of ‘sustainable income’ SI is SI SI(1 + r) SI ∑ ———— = ————— = ——————. i = 0 (1 + r)i r 1 1 – —— 1+r ∞

(4A.2)

Setting (4A.1) and (4A.2) equal and rearranging expresses SI as a fraction of RR:

[

]

1 SI = RR 1 –——————— (1 + r)n + 1

The user costs, representing the depreciation of the resource stock, would thus be

180

Environmental and resource economics 2004/2005

[

]

[

]

1 1 (RR – SI) = RR ——————— = (P – AC) . R —————— . n + 1 (1 + r) (1 + r)n + 1

NOTES * 1.

2. 3. 4.

5.

6. 7. 8. 9.

10. 11. 12. 13.

I would like to thank Tom Tietenberg, Henk Folmer, Jeroen C.J.M. van den Bergh, Mathis Wackernagel, Salah El Serafy and two anonymous referees for many helpful suggestions and comments. All errors are mine. Pezzy and Toman (2002a) prefer to define sustainability in terms of current utility not exceeding its maximum sustainable level for all time, but note that their definition is implied by the one we use here. For simplicity, we will also not distinguish between non-declining utility and non-declining capacity or opportunities to provide utility. Natural capital means everything in nature providing utility to human beings. Human capital is skills and knowledge. Social capital is difficult to define. It refers to things like the amount of trust, the extent of social networks, the willingness of individuals to cooperate with each other and their ‘civic engagement’ in social groups such as churches and unions (Putnam, 1993). In our view, which is shared by Dasgupta (2001a, 2001b), genuine investment would be a better term to use than genuine savings. The reason is that in macroeconomics, savings is often defined as private savings. For example, in a closed economy, savings is equal to investment plus government expenditures minus taxes. Savings in the usage of genuine savings instead refers to the sum of private plus public savings (taxes minus government expenditures), which generates the equality between total savings and investment. Asheim et al. (2003) argue that the Hartwick rule and therefore GS as an indicator of WS does not depend on the assumption of substitutability of natural capital. However, their counter-example depends on the assumption that production is based on the extraction of a renewable resource within the limits of the maximum rate of natural regeneration. No actual economy fulfils this criterion, hence one is justified in arguing that the relevance of the Hartwick rule depends on the assumptions of WS. Missing are mainly some small countries. It is highly ironic, however, that at the same time that GS has become an established term in the academic literature, the Bank now calls its published data ‘adjusted net saving’ rather than GS. We will stick to the term GS. Note that in the traditional national accounts capital expenditures on education are already counted towards investment in man-made capital. In the case of forestry, the depreciation term is equal to the stumpage value (market price minus logging costs) times the quantity of timber and fuelwood commercially harvested beyond regeneration rates. Dasgupta (2001a, p. C10) criticizes this, stating that if the logged land were converted into farmland, then the ‘social worth of the land as farm should be included as an addition to the economy’s capital base’. This would be difficult to do in practice and since the value of forest depletion is minor compared to the other resources, we will not further pursue this criticism here. The same reasoning applies to renewable resources if harvesting exceeds natural regeneration. See Eisner (1990) for an overview. Such studies derive the value from environmental disamenities in comparing, for example, house prices from real estate, which is similar in all respects but the environmental disamenity. Admittedly, the proponents of ISEW and GPI would counter-argue that such an option would be self-defeating as raising consumption expenditures would also raise many of the deduction items, which would then more than compensate any expected increase in the ISEW and GPI.

Indicators of sustainability 14. 15. 16. 17.

18. 19. 20.

181

Even from the perspective of their proponents, EF is not fully compatible with SS, however, as it does not directly require compensating future generations for past and current fossil fuel use with an alternative energy resource (Wackernagel and Silverstein, 2000, p. 392). Initially, the absorptive capacity of the oceans was not included, which sparked a lot of criticism (for example Ayres, 2000). Taking out nuclear energy lowers the global EF by between 4 per cent (Wackernagel et al., p. 9268) and 7 per cent (IMV, 2002, p. 18). These two different groups of authors have a controversy about the exact amount. It is unclear why some proponents of EF who recognize that ‘a more sound basis for the “energy footprint” is the required area of land needed to produce the specified energy renewably’, as Ferguson (2002, p. 310) does, want to restrict the alternative computation to biologically grown resources (sugar cane/ethanol), which are much more land-intensive than wind and solar energy. See Fischer-Kowalski (1998) and Fischer-Kowalski and Hüttler (1999) for an intellectual history of material flows covering the period 1860 to 1998. Hinterberger et al. (1999, p. 368) suggest that various countries as well as over 100 companies, which are members of the World Business Council on Sustainable Development (WBCSD), have stated their commitment to reduce material flows. In some sense, the hybrid indicator approach is reminiscent of Baumol and Oates’s (1971) standards–prices approach in the economics of pollution control, where standards are set somewhat arbitrarily given that the efficient level of pollution is often difficult if not impossible to establish.

REFERENCES Adriaanse, Albert, Stefan Bringezu, Allen Hammond, Yuichi Moriguchi, Eric Rodenburg, Donald Rogich and Helmut Schütz (1997), Resource Flows: The Material Basis of Industrial Economies, Washington, DC, Wuppertal, The Hague and Tsukuba: World Resources Institute, Wuppertal Institute, Netherlands Ministry of Housing, Spatial Planning and Environment and National Institute for Environmental Studies. Aronsson, Thomas and Karl-Gustaf Löfgren (1998), ‘Progress and problems in the economics of sustainability’, in Tom Tietenberg and Henk Folmer (eds), International Yearbook of Environmental and Resource Economics 1998/1999, Cheltenham, UK and Lyme, NH: Edward Elgar. Arrow, Kenneth J., Partha Dasgupta and Karl-Göran Mäler (2003), ‘The genuine savings criterion and the value of population’, Economic Theory, 21, 217–25. Asheim, Geir B. (1986), ‘Hartwick’s Rule in open economies’, Canadian Journal of Economics, 19 (3), 395–402. Asheim, Geir B. (1994), ‘Net national product as an indicator of sustainability’, Scandinavian Journal of Economics, 96, 257–65. Asheim, Geir B. (1996), ‘Capital gains and net national product in open economies’, Journal of Public Economics, 59, 419–34. Asheim, Geir B. (2002), Green National Accounting with a Changing Population, working paper, University of Oslo, Oslo. Asheim, Geir B. (2003), ‘Green national accounting for welfare and sustainability: a taxonomy of assumptions and results’, Scottish Journal of Political Economy, 50 (2), 113–30. Asheim, Geir B., Wolfgang Buchholz and Cees Withagen (2003), ‘The Hartwick Rule: myths and facts’, Environmental and Resource Economics, 25 (2), 129–50.

182

Environmental and resource economics 2004/2005

Atkinson, Giles (1995), ‘Measuring sustainable economic welfare: a critique of the UK ISEW’, working paper GEC 95-08, Centre for Social and Economic Research on the Global Environment, Norwich and London. Atkinson, Giles, Richard Dubourg, Kirk Hamilton, Mohan Munasinghe, David Pearce and Carlos Young (1997), Measuring Sustainable Development – Macroeconomics and the Environment, Cheltenham, UK and Lyme, NH: Edward Elgar. Ayres, Robert U. (2000), ‘Commentary on the utility of the ecological footprint concept’, Ecological Economics, 32, 347–9. Ayres, Robert U. and Allen V. Kneese (1969), ‘Production, consumption, and externalities’, American Economic Review, 59 (3), 282–97. Baumol, William J. and Wallace E. Oates (1971), ‘The use of standards and prices for protection of the environment’, Swedish Journal of Economics, 73, 42–54. British Petroleum (various years), BP Statistical Review of World Energy, London: British Petroleum. Brouwer, Roy, Martin O’Connor and Walter Radermacher (1999), ‘GREEned National STAtistical and Modelling Procedures: the GREENSTAMP approach to the calculation of environmentally adjusted national income figures’, International Journal of Sustainable Development, 2, 7–31. Castañeda, B.E. (1999), ‘An index of sustainable economic welfare (ISEW) for Chile’, Ecological Economics 28, 231–44. Chambers, Nicky, Craig Simmons and Mathis Wackernagel (2000), Sharing Nature’s Interest – Ecological Footprints as an Indicator of Sustainability, London: Earthscan. Clarke, Matthew and Sardar M.N. Islam (2003), ‘Diminishing and negative returns of economic growth: an index of sustainable economic welfare (ISEW) for Thailand’, mimeo, Melbourne: Victoria University. Cobb, C.W. and Cobb, J.B. (eds) (1994), The Green National Product: A Proposed Index of Sustainable Economic Welfare, Lanham: University Press of America. Commission of the European Communities, Eurostat, International Monetary Fund, Organisation of Economic Co-operation and Development, United Nations and World Bank (1993), System of National Accounts 1993, Brussels, Luxembourg, New York, Paris, Washington, DC. Dahme, Kai, Friedrich Hinterberger, Helmut Schütz and Eberhard K. Seifert (1998), ‘Sustainable human development index: a suggestion for “greening” the UN’s indicator’, mimeo, Wuppertal: Wuppertal Institute for Climate, Environment and Energy. Daly, Herman E. (1977), Steady-State Economics, San Francisco: W.H. Freeman. Daly, Herman E. (1992), Steady-State Economics – Second Edition with New Essays, London: Earthscan. Daly, H., Cobb, H.E. and Cobb, J.B. (1989), For the Commond Good, Boston: Beacon Press. Dasgupta, Partha (2001a), ‘Valuing objects and evaluating policies in imperfect economies’, Economic Journal, 111, C1-C29. Dasgupta, Partha (2001b), Human Well-being and the Natural Environment, Oxford: Oxford University Press. Davis, Graham A. and David J. Moore (2000), ‘Valuing mineral stocks and depletion in green national income accounts’, Environment and Development Economics, 5, 109–27. De Haan, Mark (1999), ‘On the international harmonisation of environmental accounting: comparing the national accounting matrix including environmental accounts of

Indicators of sustainability

183

Sweden, Germany, the UK, Japan and the Netherlands’, Structural Change and Economic Dynamics, 10, 151–60. Diefenbacher, H. (1994), ‘The index of sustainable economic welfare: a case study of the Federal Republic of Germany’, in C.W. Cobb and J.B. Cobb (eds), The Green National Product: A Proposed Index of Sustainable Economic Welfare, Lanham: University Press of America, pp. 215–45. Dincer, Ibrahim (2002), ‘The role of exergy in energy policy making’, Energy Policy, 30, 137–49. Eisner, Robert (1990), The Total Incomes System of Accounts, Chicago: Chicago University Press. Ekins, Paul and Sandrine Simon (1999), ‘The sustainability gap: a practical indicator of sustainability in the framework of the national accounts’, International Journal of Sustainable Development, 2, 32–58. Ekins, Paul and Sandrine Simon (2001), ‘Estimating sustainability gaps: methods and preliminary applications for the UK and the Netherlands’, Ecological Economics, 37, 5–22. El Serafy, Salah (1981), ‘Absorptive capacity, the demand for revenue, and the supply of petroleum’, Journal of Energy and Development, 7, 73–88. El Serafy, Salah (1989), ‘The proper calculation of income from depletable natural resources’. in Y.J. Ahmad, S.E. Serafy and E. Lutz (eds), Environmental Accounting for Sustainable Development: A UNDP–World Bank Symposium, Washington, DC: World Bank, pp. 10–18. El Serafy, Salah (1991), ‘The environment as capital’, in Robert Costanza (ed.), Ecological Economics: The Science and Management of Sustainability, New York: Columbia University Press, pp. 168–75. El Serafy, Salah (1997), ‘Green accounting and economic policy’, Ecological Economics, 21, 217–29. El Serafy, Salah (2001), ‘Steering by the right compass: the quest for a better assessment of the national product’, in Ekko C. van Ierland, Jan van der Straaten and Herman R.J. Vollebergh (eds), Economic Growth and Valuation of the Environment: A Debate, Cheltenham, UK and Northampton, MA: Edward Elgar, pp. 189–210. Eurostat (2002), Material Use in the European Union 1980–2000: Indicators and Analysis, Brussels: Statistical Office of the European Commission. Factor 10 Club (1997), Statement to Government and Business Leaders, Wuppertal: Wuppertal Institute for Climate, Environment and Energy. Fankhauser, Samuel (1995), Valuing Climate Change: The Economics of the Greenhouse, London: Earthscan. Ferguson, Andrew (2002), ‘The assumptions underlying eco-footprinting’, Population and Environment, 23, 303–13. Ferrari, Sylvie, Stephane Genoud and Jean-Baptiste Lesourd (2001), ‘Thermodynamics and economics: towards exergy-based indicators of sustainable development’, Schweizerische Zeitschrift für Volkswirtschaft und Statistik, 137, 319–36. Fischer-Kowalski, Marina (1998), ‘Society’s metabolism – the intellectual history of materials flow analysis, part I, 1860–1970’, Journal of Industrial Ecology, 2, 61–78. Fischer-Kowalski, Marina and Walter Hüttler (1999), ‘Society’s metabolism – the intellectual history of materials flow analysis, part II, 1970–1998’, Journal of Industrial Ecology, 2, 107–36. Gawel, Erik (1998), ‘Das Elend der Stoffstromökonomie – Eine Kritik’, Konjunkturpolitik, 44, pp. 173–206.

184

Environmental and resource economics 2004/2005

Gawel, Erik (2000), ‘Probleme einer Stoffstromökonomik’, Konjunkturpolitik, 46 (1–2), 164–89. Gerlagh, Reyer, Rob Dellink, Marjan Hofkes and Harmen Verbruggen (2001), ‘An applied general equilibrium model to calculate a sustainable national income for the Netherlands’, Report W-01/16, Amsterdam: Free University, Institute for Environmental Studies. Gerlagh, Reyer, Rob Dellink, Marjan Hofkes and Harmen Verbruggen (2002), ‘A measure of sustainable national income for the Netherlands’, Ecological Economics, 41, 157–74. Guenno, G. and S. Tiezzi (1998), ‘An index of sustainable economic welfare for Italy’, working paper 5/98, Fondazione Eni Enrico Mattei, Milano. Hamilton, Clive (1999), ‘The genuine progress indicator: methodological developments and results from Australia’, Ecological Economics, 30, 13–28. Hamilton, Kirk (1994), ‘Green adjustments to GDP’, Resources Policy, 20, 155–68. Hamilton, Kirk (1996), ‘Pollution and pollution abatement in the national accounts’, Review of Income and Wealth, 42, 13–33. Hamilton, Kirk (2000), ‘Genuine saving as a sustainability indicator’, Environment Department paper no. 77, Washington, DC: World Bank. Hamilton, Kirk (2003), ‘Sustaining economic welfare – estimating changes in per capita wealth’, policy research working paper 2498, Washington, DC: World Bank. Hamilton, Kirk and Michael Clemens (1999), ‘Genuine savings rates in developing countries’, World Bank Economic Review, 13, 75–98. Hartwick, John M. (1977), ‘Intergenerational equity and the investing of rents from exhaustible resources’, American Economic Review, 67, 972–4. Hartwick, John M. (1990), ‘Natural resources, national accounting and economic depreciation, Journal of Public Economics, 43, 291–304. Hartwick, John M. (1996), ‘Sustainability and constant consumption paths in open economies with exhaustible resources’, Review of International Economics, 3, 275–83. Hartwick, John M. and Anja Hageman (1993), ‘Economic depreciation of mineral stocks and the contribution of El Serafy’, in Ernst Lutz (ed.), Toward Improved Accounting for the Environment, Washington, DC: World Bank, pp. 211–35. Herendeen, Robert A. (1999), ‘EMERGY, value, ecology and economics’, in Jeroen C.J.M. van den Bergh (ed.), Handbook of Environmental and Resource Economics, Cheltenham, UK and Northampton, MA: Edward Elgar, pp. 954–64. Hille, John (1997), ‘The concept of environmental space’, expert corner no. 1997/2, Copenhagen: European Environment Agency. Hinterberger, Friedrich and Fred Luks (1998), ‘Dematerialization, employment and competitiveness in a globalized economy’, plenary session paper, Fifth Biennial Conference of the International Society for Ecological Economics, 15–19 November, Santiago de Chile. Hinterberger, Friedrich and Gerhard Wegner (1996), ‘Limited knowledge and the precautionary principle: on the feasibility of environmental policies’, mimeo, Wuppertal: Wuppertal Institute for Climate, Environment and Energy. Hinterberger, Friedrich, Fred Luks and Friedrich Schmidt-Bleek (1997), ‘Material flows vs. natural capital – what makes an economy sustainable?’, Ecological Economics, 23, 1–14. Hinterberger, Friedrich, Fred Luks and Marcus Stewen (1999), ‘Wie ökonomisch ist die Stoffstromökonomik? Eine Gegenkritik’, Konjunkturpolitik, 45 (4), 358–75. Hofkes, Marjan, Reyer Gerlagh, Wietze Lise and Harmen Verbruggen (2002),

Indicators of sustainability

185

‘Sustainable national income: a trend analysis for the Netherlands for 1990–1995’, report R-02/02, Free University, Institute for Environmental Studies, Amsterdam. Hotelling, Harold (1931), ‘The economics of exhaustible resources’, Journal of Political Economy, 39, 137–75. Hueting, Roefie (1980), New Scarcity and Economic Growth: More Welfare Through Less Production?, Amsterdam, New York and Oxford: New Holland Publishing. Hueting, Roefie (1991), ‘Correcting national income for environmental losses: a practical solution for a theoretical dilemma’, in Robert Costanza (ed.), Ecological Economics: The Science and Management of Sustainability, New York: Columbia University Press, pp. 194–213. Hueting, Roefie and Bart de Boer (2001), ‘Environmental valuation and sustainable national income according to Hueting’, in Ekko C. van Ierland, Jan van der Straaten and Herman R.J. Vollebergh (eds), Economic Growth and Valuation of the Environment: A Debate, Cheltenham, UK and Northampton, MA: Edward Elgar, pp. 17–77. IMV (2002), Assessing the Ecological Footprint, Copenhagen: Danish Environmental Institute. Jackson, T. and S. Stymne (1996), Sustainable Economic Welfare in Sweden: A Pilot Index 1950–1992, Stockholm: Stockholm Environment Institute. Jackson, T., F. Laing, A. MacGillivray, N. Marks, J. Ralls and S. Stymne (1997), An Index of Sustainable Economic Welfare for the UK 1950–1996, Guildford: University of Surrey Centre for Environmental Strategy. Kolar, Jan and Martin O’Connor (2000), ‘Natural resources and environmental accounting in the Czech Republic – an overview of methodology and results’, paper prepared for the EVE ‘Green Accounting’ Workshop, 6–7 March, Milan. Lawn, P.A. (2003), ‘A theoretical foundation to support the index of sustainable economic welfare (ISEW), genuine progress indicator (GPI), and other related indexes’, Ecological Economics 44 (1), 105–18. Lawn, P.A. and R.D. Sanders (1999), ‘Has Australia surpassed its optimal macroeconomic scale? Finding out with the aid of “benefit” and “cost” accounts and a sustainable net benefit index’, Ecological Economics, 28, 213–29. Markandya, Anil, Alistair Hunt and Pamela Mason (2000), ‘Valuing damages for green accounting purposes: the GARP II approach’, working paper 93/2000, Fondazione Eni Enrico Mattei, Milan. Matthews, Emily et al. (2000), The Weight of Nations – Material Outflows from Industrial Economies, Washington, DC: World Resources Institute. Max-Neef, M. (1995), ‘Economic growth and quality of life: a threshold hypothesis’, Ecological Economics, 15, 115–18. Mishan, Ezra J. (1994), ‘Is a welfare index possible?’, in Clifford W. Cobb and John B. Cobb (eds), The Green National Product: A Proposed Index of Sustainable Economic Welfare, Lanham: University Press of America, pp. 169–92. Moffatt, I. and M.C. Wilson (1994), ‘An index of sustainable economic welfare for Scotland, 1980–1991’, International Journal of Sustainable Development and World Ecology, 1, 264–91. Neumayer, Eric (1999a), Weak versus Strong Sustainability: Exploring the Limits of Two Opposing Paradigms, Cheltenham, UK and Northampton, MA: Edward Elgar.. Neumayer, Eric (1999b), ‘The ISEW: not an index of sustainable economic welfare’, Social Indicators Research, 48, 77–101.

186

Environmental and resource economics 2004/2005

Neumayer, Eric (2000a), ‘Resource accounting in measures of unsustainability: challenging the World Bank’s conclusions’, Environmental and Resource Economics, 15, 257–78. Neumayer, Eric (2000b), ‘On the methodology of ISEW, GPI and related measures: some constructive suggestions and some doubt on the “Threshold” hypothesis’, Ecological Economics, 34, 347–61. Neumayer, Eric (2003), Weak Versus Strong Sustainability: Exploring the Limits of Two Opposing Paradigms, 2nd revised edn, Cheltenham, UK and Northampton, MA: Edward Elgar. Nordhaus, William D. (1992), ‘Is growth sustainable? Reflections on the concept of sustainable economic growth’, paper prepared for the International Economic Association Conference, October, Varenna. Nordhaus, William D. and Edward C. Kokkelenberg (eds) (2000), Nature’s Numbers – Expanding the National Economic Accounts to Include the Environment, Washington, DC: National Academy Press. Nordhaus, William D. and James Tobin (1972), ‘Is growth obsolete?’, in National Bureau of Economic Research, Economic Growth, research general series no. 96F, New York: Columbia University Press. O’Connor, Martin and Grant Ryan (1999), ‘Macroeconomic cost-effectiveness and the use of multi-sectoral dynamic modelling as an environmental valuation tool’, International Journal of Sustainable Development, 2, 127–63. O’Connor, Martin, Anton Steurer and Marialuisa Tamborra (2001), ‘Greening national accounts’, environmental valuation in Europe policy research brief number 9, Cambridge Research for the Environment, Cambridge. Odum, H.T. (1996), Environmental Accounting: EMERGY and Environmental Decisionmaking, New York: Wiley. OECD (2001), OECD Environmental Indicators 2001, Paris: Organisation for Economic Co-operation and Development. Pearce, David W. and Giles Atkinson (1993), ‘Capital theory and the measurement of sustainable development: an indicator of “weak” sustainability’, Ecological Economics, 8 (2), 103–8. Pezzey, John and Cees Withagen (1998),’The rise, fall and sustainability of capitalresource economies’, Scandinavian Journal of Economics, 100, 513–27. Pezzey, John C.V. and Michael A. Toman (2002a), ‘Progress and problems in the economics of sustainability’, in Tom Tietenberg and Henk Folmer (eds), International Yearbook of Environmental and Resource Economics 2002/2003, Cheltenham, UK and Northampton, MA: Edward Elgar. Pezzy, John C.V. and Michael A. Toman (2002b), ‘The economics of sustainability: a review of journal articles’, discussion paper 02-03, Washington, DC: Resources for the Future. Proops, John L.R., Giles Atkinson, Burkhard Frhr. v. Schlotheim and Sandrine Simon (1999), ‘International trade and the sustainability footprint: a practical criterion for its assessment’, Ecological Economics, 28, 75–97. Putnam, Robert D. (1993), Making Democracy Work – Civic Traditions in Modern Italy, Princeton: Princeton University Press. Redefining Progress (1999), The 1998 U.S. Genuine Progress Indicator: Methodology Handbook. San Francisco: Redefining Progress. Redefining Progress (2001), The 2000 Genuine Progress Indicator, San Francisco: Redefining Progress. Rees, William and Mathis Wackernagel (1996), ‘Urban ecological footprints: why

Indicators of sustainability

187

cities cannot be sustainable – and why they are a key to sustainability’, Environmental Impact Assessment Review, 16, 223–48. Rosenberg, D., P. Oegema and M. Bovy (1995), ISEW for the Netherlands: Preliminary Results and Some Proposals for Further Research, Amsterdam: IMSA. Santopietro, George D. (1998), ‘Alternative methods for estimating resource rent and depletion cost: the case of Argentina’s YPF’, Resources Policy, 24, 39–48. Schmidt-Bleek, Friedrich (1993a), ‘MIPS – a universal ecological measure’?, Fresenius Environmental Bulletin, 2, 306–11. Schmidt-Bleek, Friedrich (1993b), ‘MIPS re-visited’, Fresenius Environmental Bulletin, 2, 407–12. Sefton, J.A. and M.R. Weale (1996), ‘The net national product and exhaustible resources: the effects of foreign trade’, Journal of Public Economics, 61 (1), 21–47. Serôa da Motta, Ronaldo and Ferraz do Amaral, Claudio A. (2000), ‘Estimating timber depreciation in the Brazilian Amazon’, Environment and Development Economics, 5, 129–42. Solow, Robert M. (1974), ‘Intergenerational equity and exhaustible resources’, in Review of Economic Studies for the Symposium on the Economics of Exhaustible Resources, pp. 29–46. Spangenberg, Joachim, Friedrich Hinterberger, Stephan Moll and Helmut Schütz (1999), ‘Material flow analysis, TMR and the MIPS concept: a contribution to the development of indicators for measuring changes in consumption and production patterns’, International Journal of Sustainable Development, 2, 491–505. Steenge, Albert E. (1999), ‘Input–output theory and institutional aspects of environmental policy’, Structural Change and Economic Dynamics, 10, 161–76. Stockhammer, E., H. Hochreiter, B. Obermayr and K. Steiner (1997), ‘The index of sustainable economic welfare (ISEW) as an alternative to GDP in measuring GDP in measuring economic welfare. The results of the Austrian (revised) ISEW calculation 1955–1992’, Ecological Economics, 21, 19–34. Tjahjadi, B., D. Schäfer, W. Radermacher and H. Höh (1999), ‘Material and energy flow accounting in Germany – data base for applying the national accounting matrix including environmental accounts concept’, Structural Change and Economic Dynamics, 10, 73–97. Van den Bergh, C.J.M. Jeroen and Harmen Verbruggen (1999), ‘Spatial sustainability, trade and indicators: an evaluation of the “ecological footprint” ’, Ecological Economics, 29, 61–72. Vincent, Jeffrey R. (1997), ‘Resource depletion and economic sustainability in Malaysia’, Environment and Development Economics, 2, 19–37. Vitousek, Peter M., Paul R. Ehrlich, Anne H. Ehrlich and Pamela A. Mason (1986), ‘Human appropriation of the products of photosynthesis’, BioScience, 36, 368–73. von Weizsäcker, Ernst, Amory B. Lovins and L. Hunter Lovins (1997), Factor Four: Doubling Wealth – Halving Resource Use, The New Report to the Club of Rome, London: Earthscan. Wackernagel, Mathis and J. David Yount (2000), ‘Footprints for sustainability: the next steps’, Environment, Development and Sustainability, 2, 21–42. Wackernagel, Mathis and Judith Silverstein (2000), ‘Big things first: focusing on the scale imperative with the ecological footprint’, Ecological Economics, 32, 391–4. Wackernagel, Mathis, Larry Onisto, Patricia Bello, Alejandro Callejas Linares, Ina Susana López Falfán, Jesus Méndez García, Ana Isabel Suárez Guerrero, Mathis Guadalupe Suárez Guerrero (1999), ‘National natural capital accounting with the ecological footprint analysis’, Ecological Economics, 29, 375–90.

188

Environmental and resource economics 2004/2005

Wackernagel, Mathis, Niels B. Schulz, Diana Deumling, Alejandro Callejas Linares, Martin Jenkins, Valerie Kapos, Chad Monfreda, Jonathan Loh, Norman Myers, Richard Norgaard and Jørgen Randers (2002), ‘Tracking the ecological overshoot of the human economy’, Proceedings of the National Academy of Sciences, 99, 9266–71. Weitzman, Martin L. (1997), ‘Sustainability and technical progress’, Scandinavian Journal of Economics, 99, 1–13. Willey, David and Andrew Ferguson (1999), Carrying Capacity Ethics, London: Optimum Population Trust. World Bank (1997), Expanding the Measure of Wealth: Indicators of Environmentally Sustainable Development, Washington, DC: World Bank. World Bank (2002), Adjusted Net Savings Data, Washington, DC: World Bank, http://lnweb18.worldbank.org/ESSD/essdext.nsf/44ByDocName/GreenAccounting AdjustedNetSavings. World Bank (2003), Wealth Estimates, Washington, DC: World Bank, http://lnweb18.worldbank.org/ESSD/essdext.nsf/44ByDocName/GreenAccounting WealthEstimates. WWF (2002), Living Planet Report 2002, Gland: World Wide Fund for Nature. Zolotas, Xenophon (1981), Economic Growth and Declining Social Welfare, New York: New York University Press.

5. Value transfer and environmental policy Ståle Navrud WHY VALUE TRANSFER? Increased use of economic analyses in the environment, transport, energy, health and cultural sectors has increased the demand for information on the economic value of environmental and other non-market goods by decisionmakers. Due to limited time and resources when decisions have to be made, new environmental valuation studies often cannot be performed, and decision makers must rely on transfer of economic estimates from previous studies (often termed ‘study sites’) of similar changes in environmental quality to value the environmental change at the ‘policy site’. This procedure is most often termed ‘benefit transfer’, but since damage estimates can also be transferred, I will use the more general term ‘value transfer’. Value transfer increases the uncertainty in the estimated environmental value, and a crucial question becomes: what level of accuracy is acceptable, and how does the need for accuracy vary with the policy use of the value? Results from validity tests of value transfer procedures have shown that the uncertainty in spatial and temporal benefit transfer could be quite large. Thus one should be careful in using value transfer in policy uses where the demand for accuracy is high. The practice of value transfer can be traced back to be the calculation of lost recreational value from the Hell’s Canyon hydroelectric project 30 years ago (see Krutilla and Fisher, 1975, chs 5 and 6). The first large-scale user of value transfer, however, was the USDA Forest Service. In preparation for the 1980 Resource Planning Assessment (RPA) the Forest Service launched a large-scale effort to collect data on the economic values associated with recreational use of forestlands, in order to balance these against timber production and other uses. These early examples of value transfer were, however, conducted in an uncritical manner, often lacking sound theoretical, statistical and empirical basis, and did not question the validity and reliability of the transferred values. The validity of value transfer was placed firmly on the agenda about ten years ago in a set of papers in a special section of a 1992 189

190

Environmental and resource economics 2004/2005

issue of Water Resources Research (vol. 28, no. 3). Since then there has been a steady growth in the literature on testing validity of benefit transfer, the development of transfer methods and statistical techniques, and applications of these to environment and health. Reviews of these developments can be found in Desvousges et al. (1998) and Navrud and Ready (forthcoming). This chapter is organized as follows. The next section reviews the different types of policy use where value transfer has been used. Then I review environmental valuation techniques, and describe the methods used to transfer value estimates from these techniques. Finally, I discuss tests of validity of value transfers, the main challenges to value transfer and offer my suggestions for future research to address these challenges.

POLICY USE OF ENVIRONMENTAL VALUATION ESTIMATES Environmental valuation studies have four main types of policy use, where value transfers have also been performed: 1.

Cost–benefit analysis (CBA) of investment projects and policies (both ex ante and ex post analyses). 2. Environmental costing in order to map the marginal environmental and health damages of, for example, air, water and soil pollution from energy production, waste treatment and other production and consumption activities. These marginal external costs can be used in investment decisions and operation (for example as the basis for ‘green taxes’). 3. Environmental accounting at the national level (green national accounts) and at the firm level (environmental reporting and accounting). 4. Natural Resource Damage Assessment (NRDA)/liability for environmental damages, that is, compensation payments for natural resource injuries from e.g. oil spills and other pollution incidents. Environmental valuation techniques are mostly used in CBAs, but have also been used in NDRAs in the USA, for example Carson et al. (2003); and environmental costing of electricity production from different energy sources in both the USA and Europe, for example Rowe et al. (1995), Desvousges et al. (1998) and European Commission – DG XII (1995, 1999). New York State and a few other US states have used valuation for environmental costing to construct ‘adders’ to their electricity prices. Adders are increments added to the private marginal costs that allow you to get closer to full marginal social cost of electricity production. In this case they were used to make more rational dispatch decisions for electricity generation by using marginal social rather than marginal private costs

Value transfer and environmental policy

191

(Brennan et al., 1996; Tietenberg, 2003, p. 175). Finally, environmental valuation has been used in green national accounting exercises, for example the Green Accounting Research Project (GARP) of the European Commission (Tamborra, 1999, GARP II, 1999). The UN’s statistical division UNSTAT has actively supported the development of resource accounting systems (for example the Handbook on Integrated Environmental Economic Accounts). The need for accuracy in the economic estimates increases, and thus the applicability of value transfer techniques decreases, as we move down the above list of potential policy uses. For NRDAs, and partly also for environmental accounting and costing, there seems to be a more direct link between the outcome of the analysis and policy impact, and the group affected is better defined, than in a CBA (Navrud and Pruckner, 1997). However, even in CBAs the need for accuracy can be large, for example if the estimated costs and transferred benefit estimates of a new environmental policy are very close. CBA has a long tradition in the USA as a project evaluation tool, and has also been used extensively as an input in decision-making ever since President Reagan issued Executive Order (EO) 12292 in 1981, necessitating a formal analysis of costs and benefits for federal environmental regulations that impose significant costs or economic impacts (that is, Regulatory Impact Analysis). However, some statutes, for example the Clean Air Act, specifically rule out CBA as a basis for setting ambient standards, and resorts to cost-effectiveness analysis (CEA). In Europe, CBA has a long tradition in evaluation of transportation investment projects in many countries, but environmental valuation techniques were in most cases not applied. There seems to be no legal basis for CBA in any European country, with the exception of the law of interior transport (1982) in France requiring the use of CBA to evaluate public transportation investments, and the UK Environment Act (1995; section 39) requiring a comparison of costs and benefits. Some countries have administrative CBA guidelines for project and policy evaluation, and in a few cases these include a section on environmental valuation techniques. Paragraph 130r of the Maastricht Treaty, which focuses on the EU’s environmental goals, environmental protection measures and international cooperation in general, says that the EU will consider the burden and advantage of environmental action or non-action. Furthermore, the ‘Fifth Activity Programme for Environmental Protection Towards Sustainability’ (1993–2000) says: In accordance with the Treaty, an analysis of the potential costs and benefits of action and non-action will be undertaken in developing specific formal proposals within the Commission. In developing such proposals every care will be taken as far as possible to avoid the imposition of disproportionate costs and to ensure that the benefits will outweigh the costs over time. (European Community 1993, p. 142)

192

Environmental and resource economics 2004/2005

The 1994 Communication from the Commission to the Council of the European Parliament, entitled ‘Directions for the EU on Environmental Indicators and Green National Accounting – The Integration of Environmental and Economic Information Systems’ (COM (94)670, final 21.12.94), states a specific action for ‘improving the methodology and enlarging the scope for monetary valuation of environmental damage’. More recently, the European Commission (EC)’s Green Paper, entitled ‘For a European Union Energy Policy’, states that ‘internalisation of external costs is central to energy and environmental policy’. During the last few years the Directorate General (DG) Environment of the European Commission has performed several CBAs of new air quality targets; see, for example, European Commission–DG XI Environment (1997, 1998). These analyses rely heavily on the work done within the ExternE research project (European Commission – DG XII Research, 1995, 1999). EC DG Environment is now in the process of preparing guidelines on benefit assessments for all DG Environment policy and project assessments. The new EC Water Directive explicitly mentions CBA. The EC adopted the White Paper on Environmental Liability on 9 February 2000 (COM(2000) 66 final), and on 30 March 2000 the Environmental Council meeting supported the construction of a community framework directive on environmental liability that covers environmental damage – both contamination of sites (where liability exists in all member states) and damages to biodiversity as well as traditional damage (health and property). EC DG Environment has now started work to assess the applicability and adequacy of environmental valuation and value transfer methods to value biodiversity damages for the purpose of environmental liability. International organizations such as the OECD, the World Bank and regional development banks and UNEP (United Nations Environment Program) have produced guidelines on environmental and health valuation and value transfer techniques; for example OECD (1989, 1994, 1995); Asian Development Bank (1996), and UNEP (1995, ch. 12). In many cases they have used valuation techniques as an integral part of CBA of investment projects, for example the World Bank’s evaluation of water and sanitation projects (Whittington, 1998).

TRANSFER OF INFORMATION CBAs of new environmental policies or projects with environmental impacts are often based on some kind of damage function approach (DFA) (see Box 5.1). In a DFA, there is not only uncertainty in the final step of economic valuation of environmental and health impacts. Uncertainty in transfer of

Value transfer and environmental policy

193

information is aggregated over all four steps. Thus there is uncertainty in atmospheric and marine dispersion models, dose-response and exposureresponse functions, or expert assessments of environmental and health impact where we lack these models and functions. Therefore, several disciplines can provide insights into environmental value transfer, for example environmental economics, statistics/econometrics, decision theory, ecology, geography, sociology, cognitive psychology and philosophy.

BOX 5.1 DAMAGE FUNCTION APPROACH (DFA) APPLIED TO EMISSIONS TO AIR AND WATER Step 1: Step 2: Step 3:

Step 4:

*Emissions and other residuals* → \Dispersion model\ → *Changed concentrations and other conditions* → \Dose-response functions (environment) and exposure-response functions (health)\ →*Physical impacts* → \New environmental valuation study, or existing studies combined with Benefit transfer techniques\ → *Damages or benefits*

\ . . . \ = model * . . . * = output (or input)

Cropper et al. (1997) clearly illustrates the uncertainty in the transfer of information in other parts of the DFA than valuation. They showed that the transfers of exposure-response functions from Philadelphia in the USA to Delhi in India are likely to be misleading. Alberini and Krupnick (1997) reach the same type of conclusion in their study of transfer of exposure-response functions from the USA to Taiwan. Uncertainty also originates of course from the environmental valuation techniques used in the original valuation studies, and techniques to transfer these monetary values spatially and temporally. To conduct step 4 in the DFA by benefit transfer, detailed reviews and/or databases of valuation studies would ease the job; see Navrud (1992, 1999) for a review and database of all types of valuation studies, respectively, and Carson (forthcoming) for a review of contingent valuation studies worldwide.

194

Environmental and resource economics 2004/2005

REVIEW OF ENVIRONMENTAL VALUATION TECHNIQUES The economic estimates of environmental, cultural heritage and health impacts used in a benefit transfer exercise would typically be based upon individual preferences, either observed behaviour (revealed preferences; RP) towards some marketed good with a connection to the non-market good of interest; or stated preferences (SP) in surveys with respect to the non-marketed good. Table 5.1 provides an overview of the different types of RP and SP valuation techniques. Revealed preference techniques can be divided into direct and indirect methods. Direct methods include simulated market exercises, that is, constructing a real market for a public good. This is most often impossible, and, when possible, it is usually very time-consuming and costly. In, for example, Switzerland and California, referenda are held regularly over public goods programmes; for example a proposed increase in taxes to pay for a programme to reduce water pollution. Presuming an informed voter, the decision on how to vote is based on the voter’s assessment of whether the marginal utility of the programme is greater than the marginal utility of the amount he/she would have to pay. In order to use the results of actual referenda to value a public good, we need data on voting behaviour for different levels of the good at a fixed tax price or for a fixed level of the good at different tax prices. However, in most actual referenda the voters only vote for or against one specified tax rate to reach one specified water quality level. Contingent valuation (CV) Table 5.1 Classification of environmental valuation techniques

Revealed preferences (RP)

Stated preferences (SP)

Indirect

Direct

Household production function (HPF) approach

Simulated markets

• Travel cost (TC) method • Averting costs (AC)

Actual referenda

Hedonic price (HP) analysis

Market prices Replacement costs (RC)

Choice experiments (CE)

Contingent valuation (CV)

• Conjoint analysis • Contingent ranking • Contingent rating • Pairwise comparisons

Value transfer and environmental policy

195

surveys using discrete choice elicitation techniques seek to overcome this by conducting hypothetical referenda. Another advantage of CV surveys over actual referenda is that they secure a more representative sample of the population than actual referenda, which often have low participation rates and are dominated by better-educated and better-off citizens. In CV surveys the respondents are better informed, since information provided to the voters in the actual referenda is often complicated, difficult to understand, and incomplete as it does not address aspects of the programme of concern to the public. Some environmental impacts can be valued using dose-response functions and market prices, for example impacts on crops, forests and building materials (corrosion and soiling) from air pollution. This approach uses only the physical or biological dose-response relationship to estimate the response to a change in some environmental parameter. The observed market price of the activity or entity is then multiplied by the magnitude of the physical or biological response to obtain a monetary measure of damage. Thus neither behavioural adaptations nor price responses are taken into account. Simple multiplication provides an accurate estimate of economic behaviour and value – in this case changes in gross revenue – only if economic agents are limited in the ways in which they can adapt to the environmental effect, and if the effect is small enough to have little or no impact on relative prices. This combination of circumstances is very unlikely. If, for example, crop damages from air pollution are large enough to change prices, changes in consumer and producer surpluses have to be calculated. If farmers undertake preventive measures, like switching to crops that are less sensitive to air pollution, the simple multiplication approach will overestimate damage costs. Thus other approaches should be used; see Adams and Crocker (1991). The replacement cost method (also termed restoration cost method) has been used to estimate economic damages from soil erosion by using market prices for soil and fertilizers to calculate what it would cost to replace the lost soils. This approach has also been used to calculate loss of ecosystem functions. Restoration costs are, however, just arbitrary values that might bear little relationship to true social values. Individuals’ willingness to pay for the restoration of environmental and cultural amenities may be more or less than the cost of replacement. The greatest advantage of these direct RP methods is that they are relatively simple to use. But as noted earlier, the methods ignore the behavioural responses of individuals to changes in the environmental amenities. They also obscure the distinction between benefits and costs – there is no guarantee that people are actually willing to pay the estimated cost. The indirect RP methods entail two main groups of methods: the household production function approach (including the popular travel cost method and averting cost method) and hedonic price analysis.

196

Environmental and resource economics 2004/2005

The household production function (HPF) approach involves investigating changes in consumption of commodities that are substitutes or complements to the environmental attribute. The travel cost (TC) method, used widely to measure the demand for recreation, is a prominent example. The costs of travelling to a recreational site together with participation rates, visitor attributes and information about substitute sites are used to derive a measure for the use value of the recreational activity at the site. Travel can be used to infer the demand for recreation only if it is a necessary part of the visit, or in economic terms is a weak complement. TC models build on a set of strict assumptions, which are seldom fulfilled, and the results are sensitive to the specification of the TC model, the choice of functional forms, treatment of travel time and substitute sites and so on. However, they can be relatively cheap to perform (compared to SP methods), and give reasonably reliable estimates for use values of natural resources (for example recreational fishing, hunting and hiking) for the current quality of a site. Another example of the use of the household production function approach is averting costs (AC) (also known as defensive or preventive expenditures) to infer value. Averting inputs include air filters, water purifiers, noise insulation, and other means of mitigating personal impacts of pollution. Such inputs substitute for changes in environmental attributes; in effect the quality of a consumer’s personal environment is a function of the quality of the collective environment and the use of averting inputs. We measure the value of changes in the collective environment by examining costs incurred in using averting inputs to make the personal environment different from the collective environment. A rational consumer will buy averting inputs to the point where the marginal rate of substitution between purchased inputs and the collective environment equals the price ratio. By characterizing the rate of substitution and knowing the price paid for the substitute, we can infer the price that consumers would be willing to pay for a change in the environment. The common element in household production methods is the use of changes in the quantities of complements to estimate the value of a change in quality. HPF uses actual behaviour as the basis for valuation, but is limited to use value. Non-use values, which do not entail direct consumption, cannot be estimated by looking at complements or substitutes. HPF approaches have mostly been used to value recreational activities, health and material damages. Hedonic price (HP) analysis refers to the estimation of implicit prices for individual attributes of a market commodity. Some environmental goods and services can be viewed as attributes of a market commodity, such as real property. For example, proximity to noisy streets, noisy airports and polluted waterways; smell from hog operations, factories, sewage treatment plants and waste disposal sites; exposure to polluted air, and access to parks or scenic vistas are purchased along with residential property. Part of the variation in

Value transfer and environmental policy

197

property prices is due to differences in these amenities. Other applications have been to wages for jobs that entail different levels of mortality risks (termed hedonic wage models) to estimate the ‘value of a statistical life’ (VOSL). HP data can be quite costly to obtain, as there is often no database of residential properties that have data on environmental amenities and other attributes determining the property price. In addition, the second stage of the HP analysis is often impossible since we lack socioeconomic data about the buyers of residential properties. The HP function is very sensitive to the specification and functional form, and it is often difficult to find a measure for the environmental amenity where data exist, and in which the bidders for residential properties can recognize marginal changes and has complete information about at the time they bid for the property. Two examples: (i) there are often no data on traffic noise levels, and using the annual average number of vehicles on the nearest road or distance to this road as a proxy variable for noise levels could easily value all road-traffic-related externalities (including accident risks, health impacts from air pollution, barrier effects and soiling); (ii) properties are shown to potential buyers on Sundays when there is little traffic on the nearby road, and thus they place their bid on the property with incomplete information about the road traffic noise level. While (indirect) RP methods are based on actual behaviour in a market for goods related to the environmental good in question (and hence the value for the environmental goods is elicited based on sets of strict assumptions about this relationship), SP methods measure the value of the environmental good in question by constructing a hypothetical market for the good. This hypothetical nature is the main argument against SP methods. However, no strict assumptions about the relationship between marketed complements or substitutes, or attributes of a marketed good and the environmental good, have to be made. SP methods also have the advantages of being able to (i) measure the total economic value (TEV), (ii) include both use and non-use value (also termed passive use value), (iii) derive the ‘correct’ Hicksian welfare measure, and (iv) measure future changes in environmental quality. The SP methods can be divided into direct and indirect approaches. The direct contingent valuation (CV) method is by far the most used method, but over the past few years the indirect approaches of choice experiments (CE) have gained popularity. The main difference between these two approaches is that while the CV method is typically a two-alternative (referendum) approach, CE employs a series of questions with more than two alternatives that are designed to elicit responses allowing for estimation of preferences over attributes of an environmental state. A contingent valuation (CV) survey constructs scenarios that suggest different possible future government actions. Under the simplest and most commonly used CV question format, the respondent is offered a binary choice

198

Environmental and resource economics 2004/2005

between two alternatives, one being the status quo policy, and the other having a cost greater than maintaining the status quo. The respondent is told that the government will impose the stated cost (for example increased taxes, higher prices associated with regulation, or user fees) if the non-status quo alternative is provided. The key elements here are that the respondent provides a ‘favour/not favour answer’ with respect to the alternative policy (versus the status quo), based on information about what the alternative policy will provide, how it will be provided, how much it will cost, and how it will be charged for (that is, payment vehicle). This way of eliciting willingness to pay (WTP) is termed binary discrete choice (DC). In such a closed-ended version of CV, respondents can also be asked to make multiple discrete choices in double- and multiple-DC WTP questions. An alternative elicitation method is open-ended questions, where respondents are asked directly about the most they would be willing to pay to get the alternative policy. A payment card with amounts ranging from zero to some expected upper amount could be used as a visual aid. Then the data could be treated statistically as interval data; that is, if you say yes to pay 50 euro as the highest amount, but say no to 100 euro, we know that the respondent has a WTP within this interval. One of the main challenges in a CV study is to describe the change in the environmental amenity that the alternative policy will provide, in a way that is understandable to the respondent and is at the same time scientifically correct. Concerns raised by CV critics over the reliability of the CV approach led the US National Oceanic and Atmospheric Administration (NOAA) to convene a panel of eminent experts co-chaired by Nobel Prize winners Kenneth Arrow and Robert Solow to examine the issue. In January 1993, the Panel, after lengthy public hearings and reviewing many written submissions issued a report which concluded that ‘CV studies can produce estimates reliable enough to be the starting point for a judicial or administrative determination of natural resources damages – including lost passive use value’ (Arrow et al., 1993). The Panel suggested guidelines for use in Natural Resource Damage Assessment (NRDA) legal cases to help ensure the reliability of CV surveys on passive use values including the use of in-person interviews, a binary discrete choice question, a careful description of the good and its substitutes, and several different tests should be included in the report on survey results. Since the Panel has issued the report, many empirical tests have been conducted and several key theoretical issues have been clarified. The simplest test corresponds to a well-known economic maxim, the higher the cost the lower the demand. This price sensitivity test can easily be tested in the binary discrete choice format, by observing whether the percentage favouring the project falls as the randomly assigned cost of the project increases, which rarely fails in empirical applications. The test that has attracted the most attention in recent years is whether WTP estimates from CV studies increase in a

Value transfer and environmental policy

199

plausible manner with the quantity or scope of the good being provided. CV critics often argue that insensitivity to scope results from what they term ‘warm-glow’, by which they mean getting moral satisfaction from the act of paying for the good independent of the characteristics of the actual environmental good. There have now been a considerable number of tests of the scope insensitivity hypothesis (also termed ‘embedding’), and a review of the empirical evidence suggests that the hypothesis is rejected in a large majority of the tests performed (Carson 1997). Producing a good CV survey instrument requires substantial development work, typically including focus groups, in-depth interviews, and pre-test and pilot studies to help determine whether people find the good and scenario presented plausible and understandable. The task of translating technical material into a form understood by the general public is often a difficult one. Adding to the high costs of CV surveys is the recommended mode of survey administration – in-person interviews (Arrow et al., 1993). Mail and telephone surveys are dramatically cheaper, but mail surveys suffer from sample selection bias (that is, those returning the survey are typically more interested in the issue than those who do not) and phone surveys have severe drawbacks if the good is complicated or visual aids are needed. CV results can be quite sensitive to the treatment of potential outliers. Open-ended survey questions typically elicit a large number of so-called protest zeros and a small number of extremely high responses. In discrete choice CV questions, econometric modelling assumptions can often have a substantial influence on the estimated mean and median WTP. Any careful analysis will involve a series of judgemental decisions about how to handle specific issues involving the data, and these decisions should be clearly noted. According to Carson (2000), the recent debate surrounding the use of CV is, to some degree, simply a reflection of the large sums at stake in major environmental decisions involving passive use and the general distrust that some economists have for information collected from surveys. The spotlight placed upon CV has matured it; its theoretical foundations and limits to its users are now better understood. The CV method has still not reached the routine application stage, and all CV surveys should include new research/tests. Carson (2000) concludes that perhaps the most pressing need is to establish how to reduce the costs of CV surveys while still maintaining a high degree of reliability, and suggests combined telephone–mail–telephone surveys to reduce survey administration costs, and implementation of research programmes designed to solve some of the more generic representation issues such as lowlevel risk and large-scale ecosystems. Choice experiments (CEs) have been employed in the marketing, transportation and psychology literature for some time, and arose from conjoint analysis, which is commonly used in marketing and transportation research.

200

Environmental and resource economics 2004/2005

CEs differ from typical conjoint methods in that individuals are asked to choose from alternative bundles of attributes instead of ranking or rating them. Under the CE approach respondents are asked to pick their most favoured out of a set of three or more alternatives, and are typically given multiple sets of choice questions. Because CEs are based on attributes, they allow the researcher to value attributes as well as situational changes. Furthermore, in the case of damage to a particular attribute, compensating amounts of other goods (rather than compensation based on money) can be calculated. This is one of the approaches that can be used in Natural Resource Damage Assessments (NRDAs). An attribute-based approach is necessary to measure the type or amount of other ‘goods’ that are required for compensation (Adamowicz et al., 1998). This approach can provide substantially more information about a range of possible alternative policies as well as reduce the sample size needed compared to contingent valuation (CV). However, survey design issues with the CE approach are often much more complex due to the number of goods that must be described and the statistical methods that must be employed. Although results from original studies using all these valuation techniques could be used for value transfer, studies using CV, TC and HP methods dominate. More recently, CE studies have been used for value transfer (Hanley and Wright, 2003, Scarpa et al., 2003). Since CE has the potential to decompose the economic value into values for each component/attribute of an environmental good (and value more changes in environmental quality in the same survey), it could be better suited for value transfer than CV studies. However, more research is needed on the role and validity of CE in value transfer. Value transfer based on any of these four methods can be performed in several ways.

TYPOLOGY OF VALUE TRANSFER METHODS There are two main approaches to benefit transfer: 1.

2.

Unit value transfer • Simple unit transfer • Unit transfer with income adjustments Function transfer • Benefit function transfer • Meta analysis

Unit Value Transfer Simple unit transfer is the easiest approach to transferring benefit estimates

Value transfer and environmental policy

201

from one site to another. This approach assumes that the well-being experienced by an average individual at the study site is the same as will be experienced by the average individual at the policy site. Thus we can directly transfer the benefit estimate, often expressed as mean WTP/household/year, from the study site to the policy site. For the past few decades this procedure has routinely been used in the United States to estimate the recreational benefits associated with multipurpose reservoir developments and forest management. The selection of these unit values could be based on estimates from only one or a few valuation studies considered to be close to the policy site (both geographically and in terms of the good valued), or based on an average WTP estimate from literature reviews of many studies. Walsh et al. (1992, table 1) presents a summary of unit values of days spent in various recreational activities, obtained from 287 CV and TC studies. The US Oil Spill Act recommends transfer of unit values for assessing the damages resulting from small ‘Type A’ spills or accidents using the National Resource Damage Assessment Model for Coastal and Marine Environment. This model transfers benefit estimates from various sources to produce damage assessments based on limited physical information about the spill site. The obvious problem with this transfer of unit values for recreational activities is that individuals at the policy site may not value recreational activities the same as the average individual at the study sites. There are two principal reasons for this difference. First, people at the policy site might be different from individuals at the study sites in terms of income, education, religion, ethnic group or other socioeconomic characteristics that affect their demand for recreation. Second, even if individuals’ preferences for recreation at the policy and study sites were the same, the recreational opportunities might not be. Unit values for non-use values of, for example, ecosystems from CV studies might be even more difficult to transfer than recreational (use) values for at least two reasons. First, the unit of transfer is more difficult to define. While the obvious choice of unit for use values is consumer surplus (CS) per activity day, there is greater variability in reporting non-use values from CV surveys, in terms of both WTP for whom, and for what time period. WTP is reported both per household or per individual and as a one-time payment, annually for a limited time period, annually for an indefinite time, or even monthly payments. Second, the WTP is reported for one or more specified discrete changes in environmental quality, and not on a marginal basis. Therefore, the magnitude of the change should be close, in order to get valid transfers of estimates of mean, annual WTP per household. Also the initial levels of environmental quality should be close if one is to expect non-linearity in the benefit estimate or underlying physical impacts.

202

Environmental and resource economics 2004/2005

For health impacts the question of which units to transfer seems somewhat simpler. With regard to mortality the unit could be the value of a statistical life (VOSL). For morbidity, it is more complicated since several units of value are used. For light respiratory symptoms such as coughing, headaches and itchy eyes, symptom days (defined as a specified symptom experienced one day by one individual) are often used (for example Navrud, 2001). Values for more serious illnesses are reported in terms of value per case. However, the description of these different symptoms and illnesses varies in terms of, for example, severity. A better alternative would therefore be to construct values for episodes of illness defined as type of symptoms, duration and severity (described in terms of restrictions in activity levels, whether one would have to go to the hospital and so on); see, for example, Ready et al. (forthcoming). On the issue of units to transfer, one should also keep in mind that the valuation step is often part of a damage function approach (see Box 5.1). Therefore a linkage has to be developed between the units of the endpoints of doseresponse functions, and the unit of the economic estimates. This has been done successfully, for example, for visibility changes at national parks (measured as percentage change in miles of visibility); see Smith and Osborne (1996) and health impacts (for example RSD – respiratory symptoms day, VOSL – value of statistical life, VOLYL – value of life years lost; see Alberini and Krupnick, 2002), but is much more difficult for complex changes in environmental quality and natural resources. The simple unit value transfer approach should not be used for transfer between countries with different income levels and costs of living. Therefore, unit transfer with income adjustments has been applied. Since most of the environmental valuation studies have been conducted in developed countries, this has become the general practice when conducting CBAs of infrastructure projects in developing countries, for example in the World Bank. The adjusted benefit estimate Bp′ at the policy site can be calculated as Bp′ = Bs (Yp/Ys)ß, where Bs is the original benefit estimate from the study site, Ys and Yp are the income levels at the study and policy site, respectively, and ß is the income elasticity of demand for the environmental good in question. There is, however, little empirical evidence on how the income elasticity of demand ß for different environmental goods and health impacts varies with income. The primary assumption in adjusting WTP values in proportion to some measure of income is that the income elasticity (of demand for environmental quality) is 1.0. Krupnick et al. (1996, p. 320) note in their benefit transfer exercise for impacts of air pollution in Central and Eastern Europe (CEE) that there is no reason to think that WTP for environmental quality varies proportionally

Value transfer and environmental policy

203

with income. They note that empirical evidence from the USA shows that the premature mortality risk (from, for example, air pollution) is an inferior good. The relative income approach (that is, assuming an income elasticity of 1.0) will then understate the WTP of lower-income countries. Thus Krupnick et al. use an income elasticity of 0.35 (with 1.0 as a sensitivity analysis) when transferring US mortality values to CEE. Alberini and Krupnick (2002) conclude from their value transfer comparison that assuming an income elasticity of WTP of 1.0, or even making other adjustments, does not appear to be reliable for valuing morbidity and mortality risks in developing countries. When we lack data on the income levels of the affected populations at the policy and study sites, gross domestic product (GDP) per capita figures have been used as proxies for income in international benefit transfers. However, Barton (1999) clearly shows how this approach could give wrong results in international value transfers when income levels at the study and/or policy site deviate from the average income level in the individual countries. Using the official exchange rates to convert transferred estimates in US dollars to the national currencies does not reflect the true purchasing power of currencies, since the official exchange rates reflect political and macroeconomic risk factors. If a currency is weak on the international market (partly because it is not fully convertible), people tend to buy domestically produced goods and services that are readily available locally. This enhances the purchasing powers of such currencies on local markets. To reflect the true underlying purchasing power of international currencies, the US International Comparison Program (ICP) has developed measures of real GDP on an internationally comparable scale. The transformation factors are called purchasing power parities (PPPs). The Asian Development Bank manual on economic valuation of environmental impacts (ADB, 1996) provides monetary values for health and environmental impacts that are adjusted in proportion to per capita gross domestic product (GDP). They note that it would be more appropriate to use PPP estimates of per capita GDP because these estimates have been adjusted to reflect a comparable amount of goods and services that could be purchased with the per capita national income in each country. Even if PPP-adjusted GDP figures and exchange rates can be used to adjust for differences in income and cost of living in different countries, it will not be able to correct for differences in individual preferences, initial environmental quality, and cultural and institutional conditions between countries (or even within different parts of a country). Function Transfer Transferring the entire benefit function is conceptually more appealing than

204

Environmental and resource economics 2004/2005

just transferring unit values because more information is effectively taken into account in the transfer. The benefit relationship to be transferred from the study site(s) to the policy site could be estimated using either revealed preference (RP) approaches like TC and HP methods or stated preference (SP) approaches like the CV method and choice experiments (CE). For a CV study, the benefit function can be written as: WTPij = b0 + b1Gj + b2 Hij + e,

(5.1)

where WTPij = the willingness to pay of household i at site j, Gj = the set of characteristics of the environmental good at site j, and Hij = the set of characteristics of household i at site j, b0 , b1 and b2 are sets of parameters, and e is the random error. To implement this approach the analyst would have to find a study in the existing literature with estimates of the constant b0 and the sets of parameters, b1 and b2. Then the analyst would have to collect data on the two groups of independent variables, G and H, at the policy site, insert them into equation (5.1), and calculate households’ willingness to pay at the policy site. The main problem with the benefit function approach is due to the exclusion of relevant variables in the WTP (or bid) function estimated in a single study. When the estimation is based on observations from a single study of one or a small number of recreational sites or a particular change in environmental quality, a lack of variation in some of the independent variables usually prohibits inclusion of these variables. For domestic benefit transfers researchers tackle this problem by choosing the study site to be as similar as possible to the policy site. Instead of transferring the benefit function from one selected valuation study, results from several valuation studies could be combined in a metaanalysis to estimate one common benefit function. Meta-analysis has been used to synthesize research findings and improve the quality of literature reviews of valuation studies in order to come up with adjusted unit values. In a meta-analysis, several original studies are analysed as a group, where the result from each study is treated as a single observation in a regression analysis. If multiple results from each study are used, various meta-regression specifications can be used to account for such panel effects. The meta-analysis allows us to evaluate the influence of a wider range in characteristics of the environmental good, the features of the samples used in each analysis (including characteristics of the population affected by the change in environmental quality), and the modelling assumptions. The resulting regression equations explaining variations in unit values can then be used together with data collected on the independent variables in the model that describes the policy site to construct an adjusted unit value. The regression

Value transfer and environmental policy

205

from a meta-analysis would look similar to equation (5.1), but with one added independent variable; Cs = characteristics of the study s (and the dependent variable would be WTPs = mean willingness to pay from study s). Smith and Kaoru’s (1990) and Walsh et al.’s (1990, 1992) meta-analyses of TC recreation demand models using both TC and CV studies for the USDA Forest Service’s resource planning programme were the first attempts to apply meta-analysis to environmental valuation. Later there have been applications to HP models valuing air quality (Smith and Huang, 1993), CV studies of both use and non-use values of water quality improvements (Magnussen, 1993), CV studies of groundwater protection (Boyle et al., 1994), TC studies of freshwater fishing (Sturtevant et al., 1995), CV studies of visibility changes at national parks (Smith and Osborne, 1996), CV studies of morbidity using quality of life years (QUALY) indexes (Johnson et al., 1996), CV studies of endangered species (Loomis and White, 1996), CV studies of environmental functions of wetlands (Brouwer et al., 1997), HP studies of aircraft noise (Schipper et al., 1998), CV studies of landscape changes (Santos, 1998), CV studies of WTP for wastewater treatment in coastal areas (Barton, 1999), and outdoor recreation (Shrestha and Loomis, 2001). The last five studies are international meta-analyses, including both European and North American studies. All the others, except Magnussen (1993), analyse US studies only. Many of these meta-analyses of relatively homogeneous environmental goods and health effects are not particularly useful for benefit transfer even within the USA, where most of these analyses have been conducted, because they focus mostly on methodological differences.1 Methodological variables such as ‘payment vehicle’, ‘elicitation format’ and ‘response rates’ (as a general indicator of quality of mail surveys) in CV studies, and model assumptions, specifications and estimators in TC and HP studies, are not particularly useful in predicting values for specified change in environmental quality at the policy site. This focus on variables describing the methodological choices made is partly due to the fact that some of these analyses were not constructed for benefit transfer (for example Smith and Kaoru, 1990; Smith and Huang, 1993, Smith and Osborne, 1996). Another reason is that inadequate information was reported in the published studies with regard to characteristics of the study site, the change in environmental quality valued, and income and other socio-economic characteristics of the sampled population. In particular, the last class of variables would be necessary in international benefit transfer, assuming cross-country heterogeneity in preferences for environmental goods and health effects. In most of the meta-analyses secondary information was collected on at least some of these initially omitted site and population characteristics variables, or for some proxy for them. These variables make it possible to value impacts outside the domain of a single valuation study, which is a main advantage of

206

Environmental and resource economics 2004/2005

meta-analysis over the benefit function transfer approach. However, often the use of secondary data and/or proxy variables introduces added uncertainty, for example using income data for the population in the region instead of income data for the fishermen at the study site. On the other hand, these secondary data are more readily available at the policy site without having to do a new survey. Most meta-analyses caution against using them for adjusting unit values due to potential biases from omitted variables and specification/measurement error of included variables. To increase the applicability of meta-analysis for value transfer, one could select studies that are as similar as possible with regard to methodology, and thus be able to single out the effects of site and population characteristics on the value estimates. However, the problem is that there are usually so few valuation studies of a specific environmental good or health impact that one cannot do a statistically sound analysis.

VALIDITY AND RELIABILITY OF BENEFIT TRANSFER While there are very detailed guidelines, although disputed, on how to carry out high-quality original valuation studies, for example Arrow et al. (1993) for contingent valuation (CV) surveys, no such (universally accepted) guidelines exist for benefit transfer. Although Smith (1992) called for the development of a standard protocol or guidelines for conducting benefit transfer studies more than ten years ago, no such generally accepted protocol exists. More recent studies, however, comparing benefit transfers with new CV studies of the same site to test the validity of benefit transfer, provide valuable input in the development of such guidelines. Loomis (1992) argues that cross-state benefit transfer in the USA (even for identically defined activities) is likely to be inaccurate, after rejecting the hypotheses that the demand equations and average benefits per trips are equal for ocean sport salmon fishing in Oregon versus Washington, and for freshwater steelhead fishing in Oregon versus Idaho. Bergland et al. (1995) and Downing and Ozuno (1996) used the benefit function transfer and unit value approaches. Downing and Ozuno only looked at use value, while Bergland et al. also cover transfer of non-use value. Bergland et al. (1995) conducted the same CV study of increased use and non-use values for water quality improvements at two Norwegian lakes (let us call them A and B for simplicity), constructed benefit functions for A and B, and then transferred the benefit function of lake A to value the water quality improvement in lake B, and vice versa. The mean values were also transferred and compared with the original CV estimate, since the two lakes are rather similar with regard to size and type of pollution problem. When selecting the

Value transfer and environmental policy

207

independent variables for the demand function two different approaches were used: (i) selecting variables which give the largest explanatory power, and (ii) selecting variables for which it is possible to obtain data at the policy site without having to do a costly survey. The last approach would ease future transfers, but could give less reliable estimates. Several tests for transferability were conducted, but all indicate lack of transferability statistically speaking (that is, transferred and original values are significantly different at the 5 per cent level). However, the transfer error, defined as the difference between predicted (transferred) mean WTP and observed mean WTP (in the original study), as a percentage of the observed mean WTP, is ‘only’ 20–40 per cent. In one lake the transferred values were higher, in the other they were lower than the estimate from the original study. Thus, from this study one cannot conclude which procedure would produce the highest values. While Bergland et al. (1995) test benefit transfers spatially by conducting two CV studies at the same time, Downing and Ozuno (1996) test benefit transfer both spatially and intertemporally through CV and TC models of recreational angling at eight bays along the Texas coast. Using a 5 per cent significance level, they found that 91–100 per cent of the estimates were not transferable across bays, but 50–63 per cent of within-bay estimates were transferable across time. Like Bergland et al. (1995), they conclude that geographical benefit transfer is generally not statistically reliable. Transfer errors are, however, ‘only’ 1–34 per cent. Brouwer and Spaninks (1999) reached the same conclusion in their CV studies of use and non-use values of biodiversity (meadow birds and bankside flowers) of two Dutch peat meadow sites. The original CV study gave significantly higher estimates than transferred CV estimate from the other peat meadow area, but the transfer error is again relatively low: 22–40 per cent. Scarpa et al. (2003) estimated the use value of 26 recreational forests in Ireland using a discrete choice CV survey. As opposed to Downing and Ozuno (1996), they find that benefit function transfer (spatially) is reliable. Fifty-one and 62 per cent of the median and mean WTP estimates, respectively, are transferable; that is, the original and transferred estimates are not significantly different at the 10 per cent level. Kirchhoff et al. (1997) found in a comparative survey of recreationists in the states of Arizona and New Mexico that out of 24 comparisons of benefit measures, only two involved errors exceeding 100 per cent, and 16 out of 24 indicated errors of less than 50 per cent, which is a magnitude of error that still might be acceptable in a preliminary assessment of a proposed policy’s impacts. All the validity tests of benefit transfer mentioned above look at different sites within the same country. Ready et al. (forthcoming) conducted the same CV study in five European countries (the Netherlands, Norway, Portugal,

208

Environmental and resource economics 2004/2005

Spain and the United Kingdom). They found that the transfer error in valuing respiratory symptoms (that could be caused by air pollution) was ± 37–39 per cent in terms of predicting mean willingness to pay (WTP) to avoid the symptom in one country from the data of the other countries. The observed transfer error should be compared with the variability in the original estimate within a country of ± 16 per cent (estimated using Monte Carlo simulations). Unit value transfer, unit value transfer with income adjustment (using PPP indexes for the cities in which the studies were conducted, since national PPP indexes were not representative for these specific cities) and benefit function transfer performed equally well (or poorly). The remaining differences in valuation between countries are due to other factors than income/purchase power, education level, age, sex, number of children in the household and health status variables. Thus cultural and attitudinal factors seem to be important in explaining differences in valuation across countries. In another benefit transfer validity test across countries, Rozan (1999) found transfer errors of 15–30 per cent. She conducted the same CV survey in Strasbourg, France and the neighbouring city of Kehl, Germany; and asked respondents to state their WTP for a specified improvement in air quality. Barton and Mourato (2003) and Chestnut et al. (1997) are examples of benefit transfer validity tests between developed and developing countries. These studies show transfer errors often larger than between developed countries. To conclude, results from validity tests show that the uncertainty in value transfers both spatially and temporally could be quite large. Thus benefit transfer should be applied to environmental valuation where the demand for accuracy is not too high. Kristofersson and Navrud (forthcoming) propose equivalency tests for validity of value transfer. Equivalence testing is better adapted to this task than traditional testing because it combines the concepts of statistical significance and policy significance into one test, by defining an acceptable transfer error before conducting the validity test. The level of acceptable transfer error will depend on the intended policy use.

CHALLENGES IN VALUE TRANSFER Value transfer is less than ideal, but so are most valuation efforts in the sense that better estimates could be obtained if more time and money were available. Analysts must constantly judge how to provide policy advice in a timely manner, subject to the resource constraints they face. Analysts should compare the cost of doing a new, original valuation study with the potential loss of making the wrong decision when using the transferred estimate. Decision theory and Bayesian analysis could be used to assess the need for further information about both monetary values and other steps in the damage function

Value transfer and environmental policy

209

approach (DFA); see Barton (1999) for an application. Studies that can reduce the uncertainty in transfer of information in other stages of the DFA (see Box 5.1) should also be conducted. Transfer methods may be particularly useful in policy contexts where rough or crude economic benefits may be sufficient to make a judgement regarding the advisability of a policy or project. Thus value transfer could be used in cost–benefit analyses of projects and policies, but one should be more careful in using transferred values in environmental costing and accounting exercises at the national and firm levels, and particularly in calculating compensation payments for natural resource injuries. Five main difficulties and, thus, challenges in value transfer were identified: 1. 2.

3.

4.

5.

Difficult access and low quality of many existing studies from which to transfer values. Valuation of new policies or projects are difficult in that: • expected change resulting from a policy is outside the range of previous experience; • most previous studies value a discrete change in environmental quality; how to convert these values into values for marginal changes resulting from a new policy or project; • most previous studies value a gain in environmental quality; how to convert these values to losses in order to value loss in environmental quality of, for example, energy and transportation projects. How to adjust for differences in the study site(s) and policy site that are not accounted for in the specification of the valuation model (function transfer) or in the procedure used to adjust the unit value (adjusted unit values). How to determine the ‘extent of the market’. To calculate aggregated benefits, the mean benefit estimate has to be multiplied by the total number of affected households (that is, households that find their wellbeing affected by the change in the quality of the environmental good). Guidelines on how to determine the size of the affected population are needed. While original valuation studies can be constructed to value many benefit (or cost) components simultaneously, benefit transfer studies often involve transfer and aggregation of individual components. Simply adding them assumes the independence in value between the components (that is, the independent valuation and summation (IVS) procedure). If components are substitutes or complements, this simple adding-up procedure would over- and underestimate the total benefits (or costs), respectively. Thus correction factors to take these interdependencies into

210

Environmental and resource economics 2004/2005

account have to be applied. It remains to see whether it is possible to construct general sets of correction factors for groups of environmental goods.

FUTHER RESEARCH The response to these main challenges in benefit transfer could be development of: 1.

Improved benefit transfer techniques, establishment of correction factors for interdependencies of transferred benefit and cost factors, and a protocol for benefit transfer including recommendations for how to determine the size of the ‘affected population’, and correct for interdependencies among components of the environmental good. 2. A database of environmental valuation studies. Recently, there have been great advances in both these issues, but they all need to be explored further: new approaches to value transfer have been suggested by Smith et al. (2002) (preference calibration approach), and Atkinson et al. (1992), León and Vásquez-Polo (1998) and Barton (1999) (Bayesian statistics and statistical decision theory). Smith et al. propose a preference calibration approach to value transfer to ensure that the transferred values are consistent with budget-constrained utility maximization behaviour. The first and foremost advantage of their approach is that the resulting benefit or damage estimate can never be inconsistent with household income, as they have been in some high-profile applications such as Constanza et al. (1998). Bayesian statistical techniques have been introduced both for the estimation of contingent valuation models in original valuation studies (León and Vásquez-Polo, 1998; McLeod and Bergland, 1999) and for explicit use in value transfer (Atkinson et al., 1992) The general idea is that we have some prior information about the benefits or costs from a project. If we decide to do a new valuation study specifically for that project, the statistical analysis should recognize the existence of prior information (Bayesian statistics). The decision whether or not to do a new survey should explicitly recognize the uncertain information available (statistical decision theory). On the issue of correction factors for interdependencies, Santos (1998) demonstrated that the frequently used independent valuation and summation (IVS) procedure for multi-attribute landscape changes would overestimate the benefits by 48 to 80 per cent in two case studies. This confirms that the error IVS introduces in policymaking cannot be ignored, as previously stressed by Hoehn and Randall (1989), Hoehn (1991) and Hoehn and Loomis (1993).

Value transfer and environmental policy

211

Ruijgrok (2001) tests the use of ecological classification in transferring values for nature reserves on the Dutch coast, and concludes that this approach offers new opportunities for transferring economic benefits within a limited geographical area. Thus more research is needed on the appropriate spatial scale of the classification. Based on a review of value transfer studies and validity tests of transfer, Brouwer (2000) propose a seven-step protocol as a first attempt towards good practice for using value transfer techniques in CBA. Step 1 involves the identification of the relevant ecological functions of the goods and services under consideration, and their importance for sustaining ecosystems and hence human systems. Step 2 focuses on identifying beneficiaries of the ecological functions/services preserved or foregone, and is interdependent with Step 3, which identifies values held by different stakeholder groups in order to be able to sketch out the reasons why they value the environmental good/service under consideration. Step 4 assesses the scope, acceptability and legitimacy of the valuation process(es): monetary and/or deliberative. In Step 5 appropriate studies are selected, and study quality assessed by looking at their internal and external validity. Step 6 looks at the research design of the selected studies, and tries to assess comparability between them, and what kind of adjustment may be chosen to account for differences in design/approach for each chosen study. In Step 7 values as obtained through the previous six steps are discussed with (representatives of) stakeholders, before they are extrapolated over the relevant population affected by the environmental change under consideration. Finally, the economic aggregate is included in a CBA. The web-based database EVRI (Environmental Valuation Reference Inventory, www.evri.ec.gc.ca/EVRI/) now contains about 1000 valuation studies. Since the database was initiated by Environment Canada, in cooperation with the US Environmental Protection Agency, the majority of these studies are from North America, and only 23 per cent are European valuation studies. Based on a European user survey, Navrud (1999) evaluated the EVRI database, and found it well suited for European conditions. Few studies outside North America and Europe are registered in the database. Thus there is a need to increase the number of existing valuation studies captured in this database for all countries. Since many valuation studies are old and use outdated methodology and there are few studies for many environmental goods, there is also a great need for new, original valuation studies using state-of-the-art methodology, and designed with value transfer in mind. There is also a need for comparative studies in terms of conducting the same valuation studies of environmental amenities, cultural assets and health impacts in many countries at the same time, and testing the validity of transfer to produce calibration factors which would improve value transfers between countries and especially between

212

Environmental and resource economics 2004/2005

developed and developing countries. Ready et al. (forthcoming), Barton and Mourato (2003) and Krupnick et al. (1996) are examples of studies testing transfer of values between countries. They also demonstrate the practice of using purchase power parity (PPP)-adjusted exchange rates when transferring between countries with different currencies (and preferably PPP indices from the study area within each country, instead of national averages). Studies testing the validity of transfer over time are also scarce. Downing and Ozuno (1996), McLeod and Bergland (1996), and Carson et al. (2003, p. 277) are some of the very few studies in this area. Carson et al. (2003) found that the results of conducting the same survey (that is, their Exxon Valdez oil spill CV study) at three different points in time over the course of a year yielded WTP distributions that were statistically indistinguishable (Carson, 1997). However, there is a great need for testing validity of transfers over more than one year as a basis for constructing a procedure to replace the current practice of assuming that the WTP for environmental goods and health effects increases at a rate equal to the consumer price index.

NOTES In addition to the individual papers cited in this chapter, I would like to recommend two books as sources for new ideas for value transfer research: Desvousges et al. (1998), and an edited collection of papers on recent advances in value transfer (Navrud and Ready, forthcoming). 1. Carson et al. (1996) is an example of a meta-analysis of different environmental goods and health effects, which was performed with the sole purpose of comparing results from valuation studies using both stated preference (CV) and revealed preference methods (TC, HP, defensive expenditures and actual market data).

REFERENCES Adamowicz, V, P. Boxall, M. Williams and J. Louviere (1998), ‘Stated preference approaches for measuring passive use values: choice experiments and contingent valuation’, American Journal of Agricultural Economics, 80 (February 1998), 64–75. Adams, R.M. and T.D. Crocker (1991), ‘Material damages’, Chapter IX in J.R. Braden and C.D. Kolstad (eds), Measuring the Demand for Environmental Quality, Amsterdam and New York: Elsevier Science Publishers B.V. (North-Holland). Alberini, A. and A. Krupnick (1997), ‘Air pollution and acute respiratory illness: evidence from Taiwan and Los Angeles’, American Journal of Agricultural Economics, 79 (5), 1620–24. Alberini, A. and A. Krupnick (2002), ‘Valuing the health effects of pollution’, in T. Tietenberg and H. Folmer (eds) (2002); The International Yearbook of Environmental and Resource Economics 2002/2003, Cheltenham, UK and Northampton, MA: Edward Elgar.

Value transfer and environmental policy

213

Arrow, K.J., R. Solow, E. Leamer, P. Portney, R. Radner and H. Schuman (1993), ‘Report of the NOAA Panel on Contingent Valuation’, Federal Register, 58 (January), 4601–14. Asian Development Bank (ADB) (1996), Economic Evaluation of Environmental Impacts. A Workbook, parts I and II, Environment Division, Asian Development Bank, Manila, Philippines. Atkinson, S.E, T.D. Crocker and J.F. Shogren (1992), ‘Bayesian exchangeability, benefit transfer and research efficiency’, Water Resources Research, 28 (3), 715–22. Barton, D.N. (1999), ‘The quick, the cheap and the dirty. Benefit transfer approaches to the non-market valuation of coastal water quality in Costa Rica’, Ph.D. thesis, Department of Economics and Social Sciences, Agricultural University of Norway. Barton, D. and S. Mourato (2003), ‘Transferring the benefits of avoided health effects from water pollution between Portugal and Costa Rica’, Environment and Development Economics, 8 (2), May, 351–72. Bergland, O., K. Magnussen and S. Navrud (1995), ‘Benefit transfer: testing for accuracy and reliability’, discussion paper D-03/95, Department of Economics, Agricultural University of Norway. Revised version in R.J.G.M. Florax, P. Nijkamp and K. Willis (eds) (2000), Comparative Environmental Economic Assessment: Meta Analysis and Benefit Transfer, Dordrecht: Kluwer Academic Publishers. Boyle, K.J., G.L. Poe and J.C. Bergstrom (1994), ‘What do we know about groundwater values? preliminary implications from a meta analysis of contingent valuation studies’, American Journal of Agricultural Economics, 76 (December), 1055–61. Brennan, T.J., K.L. Palmer, R.J. Kopp, A.J. Krupnick, V. Stagliano and D. Burtraw (1996), A Shock to the System: Restructuring America’s Electricity Industry, Washington, DC: Resources for the Future. Brouwer, R. (2000), ‘Environmental value transfer: state of the art and future prospects’, Ecological Economics, 32, 137–52. Brouwer, R. and F.A. Spaninks (1999), ‘The validity of environmental benefit transfer: further empirical testing’, Environmental and Resource Economics, 14 (1), 95–117. Brouwer, R., I.H. Langford, I.J. Bateman, T.C. Crowards and R.K. Turner (1997), ‘A meta-analysis of wetland contingent valuation studies’, CSERGE working paper GEC 97-20. Centre for Social and Economic Research on the Global Environment, University of East Anglia and Unviersity College London. Carson, R.T. (1997), ‘Contingent valuation and tests of insensitivity to scope’, in R. Kopp, W. Pommerhene and N. Schwartz (eds), Determining the Value of Nonmarketed Goods: Economic, Psychological and Policy Relevant Aspects of Contingent Valuation Methods, Dordrecht: Kluwer Academic Publishers. Carson, R.T. (2000), ‘Contingent valuation: a user’s guide’, Environmental Science and Technology, 34, 1413–18. Carson, R.T. (forthcoming), Contingent Valuation. A Comprehensive Bibliography and History, Cheltenham, UK and Northampton, MA: Edward Elgar. Carson, R.T., N.E. Flores, K.M. Martin and J.L. Wright (1996), ‘Contingent valuation and revealed preference methodologies: comparing the estimates for quasi-public goods’, Land Economics, 72, 80–99. Carson, R.T., R.C. Mitchell, M. Hanemann, R.J. Kopp, S. Presser and P.A. Ruud

214

Environmental and resource economics 2004/2005

(2003), ‘Contingent valuation and lost passive use: damages from the Exxon Valdez oil spill’, Environmental and Resource Economics, 25 (3), 257–86. Chestnut, L.G., B.D. Ostro et al. (1997), ‘Transferability of air pollution control health benefits estimates from the United States to developing countries: evidence from the Bangkok study’, American Journal of Agricultural Economics, 79 (5), 1630–35. Constanza, R., R. d’Arge, R. de Groot, S. Farber, M. Grasso, B. Hannon, K. Limburg, S. Naeem, R. O’Neill, J. Paruelo, R. Raskin, P. Sutton and M. van den Belt (1998), ‘The value of ecosystem services: putting the issue in perspective’, Ecological Economics, 25 (1), 67–72 (reprinted from Nature 1997, 387, 253–60. Cropper, M.L., N.B. Simon, A. Alberini, S. Arora and P.K. Sharma (1997), ‘The health benefits of air pollution control in Delhi’, American Journal of Agricultural Economics, 79 (5), 1625–9. Desvousges, W.H., F.R. Johnson and H.S. Banzhaf (1998), Environmental Policy Analysis with Limited Information Principles and Applications of the Transfer Method, Cheltenham, UK and Northampton, MA: Edward Elgar. Downing, M. and T. Ozuno (1996), ‘Testing the reliability of the benefit function transfer approach’, Journal of Environmental and Resource Economics and Management, 30 (3), 316–22. European Commission – DG XI Environment (1997), Economic Evaluation of Air Quality Targets for Sulphur Dioxide, Nitrogen Dioxide, Fine and Suspended Particulate Matter and Lead, Final Report, Luxembourg: European Commission, Office for Official Publications of the European Communities. European Commission – DG XI Environment (1998), Economic Evaluation of Air Quality Targets for Tropospheric Ozone. Parts A, B and C. Final Report, Luxembourg: European Commission, Office for Official Publications of the European Communities. European Commission – DG XII (1995), ‘ExternE – externalities of energy. vol. 2, methodology’, European Commission Directorate General XII Science Research and Development report EUR 16521, Brussels. European Commission – DG XII (1999), ‘ExternE – externalities of energy. vol. 7: methodology 1998 update’, European Commission Directorate General XII, report EUR 19083, Brussels. European Community (1993), Towards Sustainability: A European Community Programme of Policy and Action in Relation to the Environment and Sustainable Development, Brussels: Office for Official Publications of the European Communities. GARP II (1999), ‘Final report of the Green Accounting Research Project II to the European Commission – DG XII, RTD Programme “Environment and Climate” ’, contract ENV4-CT96-0285, November. Hanley, N. and R.E Wright (2003), ‘Testing for benefits transfer under the Water Framework Directive using choice experiment’, chapter 8 in S. Navrud and R. Ready (eds), Environmental Value Transfer: Issues and Methods, Dordrecht: Kluwer Academic Publishers. Hoehn, J.P. (1991), ‘Valuing the multidimensional impacts of environmental policy: theory and methods’, American Journal of Agricultural Economics, 73, 289–99. Hoehn, J.P. and J.B. Loomis (1993), ‘Substitution effects in the valuation of multiple environmental programs’, Journal of Environmental Economics and Management, 25, 56–75. Hoehn, J.P. and A. Randall (1989), ‘Too many proposals pass the benefit cost test’, American Economic Review, 79, 544–51. Johnson, F.R., E.E. Fries and H.S. Banzhaf (1996), Valuing Morbidity: An Integration

Value transfer and environmental policy

215

of the Willingness-to-Pay and Health Status Index Literatures, Durham, NC: Triangle Economic Research and Duke University. Kirchhoff, S., B.G. Colby and J.T. LaFrance (1997), ‘Evaluating the performance of benefit transfer: an empirical inquiry’, Journal of Environmental Economics and Management, 33 (1), 75–93. Kristofersson, D. and S. Navrud (2003), ‘Can use and non-use values be transferred across countries?’, chapter 13 in S. Navrud and R. Ready (eds), Environmental Value Transfer: Issues and Methods, Dordrecht: Kluwer Academic Publishers. Kristofersson, D. and S. Navrud (forthcoming), ‘Validity tests of benefit transfer – are we performing the wrong tests?’, forthcoming in Environmental and Resource Economics. Krupnick, A., K., Harrison, E. Nickell and M. Toman (1996), ‘The value of health benefits from ambient air quality improvements in Central and Eastern Europe: an exercise in benefits transfer’, Environmental and Resource Economics, 7, 307–32. Krutilla, J.V. and Anthony C. Fisher (1975), The Economics of Natural Environments Studies in the Valuation of Commodity and Amenity Resources, Baltimore, MD: John Hopkins Press. León, C.J. and F.J. Vásquez-Polo (1998), ‘A Bayesian approach to double bounded contingent valuation’, Environment and Resource Economics, 11, 197–215. Loomis, J.B. (1992), ‘The evolution of a more rigorous approach to benefit transfer: benefit function transfer’, Water Resources Research, 28 (3), 701–6. Loomis, J.B. and D.S. White (1996), ‘Economic benefits of rare and endangered species: summary and meta-analysis’, Ecological Economics, 18, 197–206. McLeod, D. and O. Bergland (1999), ‘Willingness-to-pay estimates using the doublebounded dichotomous-choice contingent valuation format: a test for validity and precision in a Bayesian framework’, Land Economics, 75 (1), 115–25. Magnussen, K. (1993), ‘Mini meta analysis of Norwegian water quality improvements valuation studies’, Norwegian Institute for Water Research, Oslo. Navrud, S. (1992), Pricing the European Environment, Oslo: Scandinavian University Press; Oxford and New York: Oxford University Press. Navrud, S. (1999), ‘Assessment of environmental valuation reference inventory (EVRI) and the expansion of its coverage to the EU, parts I, II and III’, European Commission, DG Environment, available at http://europa.eu.int/comm/environment/enveco/studies2.htm#24 Navrud, S. (2001), ‘Valuing health impacts from air pollution in Europe. New empirical evidence on morbidity’, Environmental and Resource Economics, 20 (4); 305–29. Navrud, S. and G.J. Pruckner (1997), ‘Environmental valuation – to use or not to use? A comparative study of the United States and Europe’, Environmental and Resource Economics, 10 (1), 1–26. Navrud, S. and R. Ready (eds) (forthcoming), Environmental Value Transfer: Issues and Methods, Dordrecht: Kluwer Academic Publishers. OECD (1989), Environmental Policy Benefits: Monetary Valuation, Paris: Organisation for Economic Co-operation and Development (OECD). OECD (1994), Project and Policy Appraisal: Integrating Economics and Environment, Paris: Organisation for Economic Co-operation and Development (OECD). OECD (1995), The Economic Appraisal of Environmental Projects and Policies. A Practical Guide, Paris: Organisation for Economic Co-operation and Development (OECD).

216

Environmental and resource economics 2004/2005

Ready, R., S. Navrud, B. Day, R. Doubourg, F. Machado, S. Mourato, F. Spanninks and M.X.V. Rodriquez (forthcoming), ‘Benefit transfer in Europe: how reliable are transfers between countries?’, forthcoming in Environment and Resource Economics. Rowe, R.D., C.M. Lang, L.G. Chestnut, D.A. Latimer, D.A. Rae, S.M. Bernow and D.E. White (1995), The New York Electricity Externality Study, volume I and II, New York: Hagler Bailley Consulting, Inc., Oceana Publications Inc. Rozan, A. (1999), ‘Benefit transfer: a comparison of WTP for air quality between France and Germany’, paper presented at the EU Concerted Action ‘Environmental Valuation in Europe’ (EVE) Workshop on Benefit Transfer, 14–16 October, Lillehammer, Norway. Ruijgrok, E.C.M. (2001), ‘Transferring economic values on the basis of an ecological classification of nature’, Ecological Economics, 39 (3), 399–408. Santos, J.M.L. (1998), The Economic Valuation of Landscape Change. Theory and Policies for Land Use Conservation, Cheltenham, UK and Northampton, MA: Edward Elgar. Scarpa, R., G.W. Hutchinson, S.M. Chilton and J. Buongiorno (2003), ‘Reliability of benefit value transfers from contingent valuation data with forest-specific attributes’, chapter 11 in S. Navrud and R. Ready (eds), Environmental Value Transfer: Issues and Methods, Dordrecht: Kluwer Academic Publishers. Schipper, Y., P. Nijkamp and P. Rietveld (1998), ‘Why do aircraft noise value estimates differ? A meta-analysis’, Journal of Air Transport Management, 4, 117–24. Shrestha, R.K. and J.B. Loomis (2001), ‘Testing a meta-analysis model for benefit transfer in international outdoor recreation’, Ecological Economics, 39 (1), 67–83. Smith, V.K. (1992), ‘On separating defensible benefit transfers from smoke and mirrors’, Water Resources Research, 28 (3), 685–94. Smith, V.K. and J.C. Huang (1993), ‘Hedonic models and air pollution. Twenty-five years and counting’, Environmental and Resource Economics, 3, 381–94. Smith, V.K. and Y. Kaoru (1990), ‘Signals or noise? Explaining the variation in recreation benefit estimates’, American Journal of Agricultural Economics, 72 (2), 419–33. Smith, V.K. and L. Osborne (1996), ‘Do contingent valuation estimates pass a “scope” test? A meta analysis’, Journal of Environmental Economics and Management, 31 (3), 287–301. Smith, V.K., G. Van Houtven and S.K. Pattanayak (2002), ‘Benefit transfer via preference calibration: “prudential algebra” for policy’, Land Economics, 78 (1), 132–52. Sturtevant, L.A., F.R. Johnson and W.H. Desvousges (1995), ‘A meta-analysis of recreational fishing’, working paper prepared by Triangle Economic Research, Durham, NC. Tamborra, M. (1999), ‘Towards a green accounting system for the European Union: the contribution of GARP II’, FEEM Newsletter, 2, November, 16–18, Fondazione Eni Enrico Mattei, Milan, Italy. Tietenberg, T.H. (2003), Environmental and Natural Resource Economics, Reading, MA: Addison-Wesley. UNEP (1995), chapter 12 in ‘Economic values of biodiversity’, Global Biodiversity Assessment, United Nations Environment Program (UNEP), Cambridge; Cambridge University Press. Walsh, R.G., D.M. Johnson and J.R. McKean (1990), ‘Nonmarket values from two

Value transfer and environmental policy

217

decades of research on recreation demand’ in Advances in Applied Micro-economics, vol. 5, Greenwich, CT: JAI Press Inc., pp. 167–93. Walsh, R.G., D.M. Johnson and J.R. McKean (1992), ‘Benefit transfer of outdoor recreation demand studies, 1968–1988’, Water Resources Research, 28 (3), 707–14. Whittington, D. (1998), ‘Administering contingent valuation surveys in developing countries’, World Development, 26 (1), 21–30.

6. Joint implementation in climate change policy Suzi Kerr and Catherine Leining* 1.

INTRODUCTION

The textbook economists’ model of a tradable permit system cannot usually be applied perfectly at either the domestic or international scale because of the difficulty and/or expense of defining allocations to and monitoring emissions of some groups, as well as for political reasons.1 It may be impossible to bring these groups fully into a tradable permit system but it is often possible to find compromise solutions to gain some benefits from trade: lower costs of achieving the environmental outcome, greater engagement of actors in the overall process, and greater equity by allowing all groups to gain some benefit. A variety of compromise trading models suit different circumstances. The Kyoto Protocol limits greenhouse gas emissions in the Annex B2 (developed) countries that ratify it. To reduce the costs of achieving the overall emission limits, three trading mechanisms are available: international emissions trading; joint implementation; and the clean development mechanism. They all ultimately transfer units that can be used for compliance but they are available to different groups in different time periods and have different rules.3 The two key distinctions are first, between Annex B (developed countries) and non-Annex B (developing countries), and second, based on the quality of domestic emissions monitoring. Only countries that have ratified Kyoto can participate. Under all mechanisms, trading can be carried out either by the government, or by legal entities (companies or individuals) where the government devolves this authority. The government remains ultimately responsible for compliance. International emissions trading (IET) is the simplest form of emissions trading where assigned amount units (AAUs) are simply moved from the registry of one Annex B country to another.4 This is available from 2008 onward. Countries that sell using this mechanism must have shown that they have adequate monitoring and enforcement of emissions (compliance with Articles 5 and 75) so that sales reflect real reductions in emissions. 218

Joint implementation in climate change policy

219

Joint implementation (JI) is available to all Annex B countries. It is a project-based mechanism where reductions are established by comparison with an estimated ‘baseline’ (what would have happened without the project) rather than relative to the national emission limits set in Kyoto. Emission reduction units (ERUs) can be created from 2008 onward.6 It does not require good national-level monitoring. It has high transaction costs because all projects must be internationally approved.7 Thus the main sellers should be countries that cannot engage in IET because their monitoring is inadequate. This is likely to be primarily economies in transition. The clean development mechanism (CDM) is available to sellers from developing countries only. The sources of emission reductions are somewhat limited (especially with regard to land use emissions). Like JI, it is project based and each trade requires international approval. In this case approval is needed because the seller countries do not face emissions limits so reductions can be measured only relative to agreed project baselines. Certified emission reductions (CERs) can be created from 2000 onward for use after 2008. Thus JI fills a specific niche in the international greenhouse gas trading market. It has a range of benefits for both buyers and sellers. JI allows parties that lag in developing their domestic climate institutions, and hence cannot participate in IET, to host projects and receive the benefits from trade. We focus most on these parties. They need to use a particular form of JI, called ‘Track II’.8 In contrast, Track I JI is very similar to IET so we cover it only briefly. Under JI, selling entities, those who reduce emissions, can make profit if they sell the ERUs for more than it cost to create them. These include receiving credit; technology transfer and learning if the buyer is actively involved in the project; and learning more broadly about how trading programs work, which will help them in future. Selling countries benefit economically if their companies engage in environmentally sound, profitable sales. In addition, some projects will help develop monitoring infrastructure, such as databases, that will facilitate the seller country’s move toward the more efficient trading mechanism, IET. For example, many countries may not have accurate data for national coal use, which is essential for estimating national emissions. The organizers of a JI project that covers the major coal users may collect data that help address this gap. The technology transfer and learning embedded in JI projects may help move the country from a high emissions path toward a less greenhouse gas (GHG) intensive path, which will make compliance with future targets easier. This is an issue of particular concern to some Annex B countries with transitional economies that may have difficulty satisfying the international monitoring requirements, Articles 5 and 7, by 2008. JI is not an ‘aid’ program. The buyers primarily participate because they gain benefits from the lower AAU prices. The option of JI also offers investors

220

Environmental and resource economics 2004/2005

security that they can acquire ERUs even if the seller or ‘host’ party fails to meet its Article 5 and 7 requirements. They also might gain ancillary commercial benefits. A JI project might give a company the chance to form a joint venture with a company in a country in transition by reducing undue regulatory barriers. This can provide learning opportunities and market access. Countries whose companies buy ERUs gain direct benefit because their companies comply at lower cost and expand their market opportunities. They might also gain in the longer term from better access to AAUs from those countries through established relationships, and from better monitoring systems, which will make those countries more effective partners in global cooperation on climate change. Countries that are neighbors of economies in transition with serious local air pollution problems could gain through reduced air pollution if the emissions reductions are associated with generally cleaner production. The European Union could benefit from JI because it could help the accession countries meet the European Union’s environmental standards. The situation with joint implementation (JI) is similar to that which arises with offsets from small sources within larger trading systems or voluntary optin procedures for extra sources. For example in the US Acid Rain program electric utilities were brought into the program in two phases but second-phase utilities were able to opt in to the first phase if they chose.9 JI probably won’t be a major mechanism for long (if Kyoto is a success) because it is costly and environmentally more risky than IET with good national monitoring. Studying it is still useful, however, as it will probably be important up until 2012 and even 2017, and it will offer lessons for the clean development mechanism and for domestic programs where coverage is not comprehensive. Box 6.1 explains some of the jargon used in the context of the Kyoto Protocol. 1.1

What Literature is Already Available on JI?

Many articles have been written on various aspects of JI. We do not attempt a complete literature survey but provide some key general references that will facilitate further exploration. Literature on specific issues that we cover later is discussed in that context. A series of excellent early papers present the basic issues in JI and how it links to other forms of international trading. These include Tietenberg et al. (1999), Mullins and Baron (1997), Nentjes (1994), Ridley (1998) and Hanafi (1998). Jepma (1995) is an edited volume with a wide range of perspectives on JI, which in 1995 still covered trade with both developed and developing countries. Zhang and Nentjes (1999), Bohm (1994) and Woerdman (2000) compare JI with a broader trading system such as IET. Kelly and Leining (2000) compares the institutions needed for JI relative to CDM. Parkinson et

Joint implementation in climate change policy

221

BOX 6.1 KEY ‘KYOTO’ JARGON FOR THE UNINITIATED Annex I (B) countries: Developed countries that have signed the Protocol and agreed to meet emission targets COP/MOP: Conference of the Parties/Meeting of the Parties – the international decisionmaking bodies Article 3: Specifies that Annex I countries must meet emission reduction targets Article 6: Creates joint implementation Articles 5 and 7: Specify the monitoring requirements that Annex I countries must satisfy before they can participate in international emissions trading IET: International emissions trading JI: Joint implementation Track II JI: The mechanism for joint implementation when the selling country has not satisfied Articles 5 and 7 AAUs: Assigned amount units – their initial allocation is defined by the Article 3 targets ERUs: Tradable emission reduction units created through joint implementation PDD: Project design document. This is required for Track II JI

al. (1999) consider the implications of ‘interim period banking’, or credit for reductions pre-2008 in both CDM and JI. Bailey and Jackson (1999) and OECD/IEA/IETA (2002) consider issues specific to dealing with economies in transition. In this chapter we do not touch on the issue of the potential contribution of JI in terms of emissions reductions or its costs. Most models of global trading can contribute some evidence on the potential reductions from JI simply by considering Eastern Europe as a region. See for example various results from the Energy Modeling Forum.10 Bohm and Carlén (1999) report laboratory experiments on JI trading that shed light on how actual gains might differ from potential gains. Institutions whose contributions are well worth exploring are

222

Environmental and resource economics 2004/2005

the OECD, the International Energy Agency and the Center for Clean Air Policy.11 1.2

What is the History of JI and How Does That Affect its Future?

JI began as a part of Activities Implemented Jointly (AIJ), a pilot program established in Berlin in 1995. AIJ covered both developed and developing countries (now covered by the CDM). The pilot phase provided no credit for emissions reductions but it had some real value. A lot of thinking was done about how to define the rules for trading. It introduced the idea of trading to many groups who had never considered it before, and thus built a community of interest. As a result of the projects we learned some lessons about monitoring emissions, defining baselines and setting up contracts between very different countries with different cultures and institutions. Many projects illustrated the risks involved in what are complex projects with sometimes unreliable or unskillful partners. On the other hand the pilot phase might have had some negative effects. There was a lot of hype about the potential benefits from involvement but many of the people who did get involved learned to their dismay that after considerable investment in project design there were few buyers and very low prices. Some proponents did not seem to fully appreciate that there would only be significant demand for credits when emissions limitations were enforced and the credits could be used to satisfy those limitations. Negative experience with low prices may reduce sellers’ future enthusiasm. Second, the nature of the projects and the players was probably not representative of the types of projects we might expect when real money is involved after 2008. The projects tended to be small, so methodologies were designed that were suitable for tackling small projects rather than larger sectoral projects. The methodological issues in these two cases are very different but larger projects may well be more effective and efficient in many countries. Buyers were involved for either public relations or altruistic reasons or because they were interested in learning. This meant that they were not as concerned about commercial aspects, so there were fewer incentives to cheat the system. Thus important lessons about strategic behavior could not be learned. Because there was no credit from these programs, there was also limited regulation. This led to widely varying methodologies and widely varying levels of environmental integrity; that is, some projects did not lead to as many real reductions as they generated emission credits. The projects with poor environmental outcomes might have increased the skepticism of some environmental groups that oppose trading. The United Nations Framework Convention on Climate Change

Joint implementation in climate change policy

223

(UNFCCC) website provides a great deal of information about the pilot phase of Activities Implemented Jointly and specific projects developed within it.12 None of these projects created credits and many were in developing countries, so not relevant to JI, but the list gives a flavor of the type and size of projects that people have experimented with so far. It also gives a clear, simple outline of the key issues in project definition; although this was written around 1997, these remain the key issues for JI. It provides a list of references that form an entry point into the extensive grey literature. Two major mechanisms have recently been set up to facilitate joint implementation post 2008: a buyer program, the Dutch ERUPT program,13 and an international agency program, the Prototype Carbon Fund, run as a public/private partnership by the World Bank.14 ERUPT is particularly focused on energy-related projects in Central and Eastern Europe. The Dutch government provides the funds and seeks emission reduction projects through a tendering process. The Prototype Carbon Fund has been established to demonstrate how JI (and CDM) projects can contribute to sustainable development. It is an active market creator, and is dedicated to capacity building and sharing information. No projects have been internationally approved to date. This chapter starts by outlining the current international rules governing joint implementation. We provide a summary of key jargon (see Box 6.1) for those who are unfamiliar with the complex Kyoto language. We then discuss two key international issues that are still unresolved: baseline development and monitoring. We then turn to domestic governance of joint implementation and how the private sector might engage in JI. At this point we consider how JI fits within the suite of flexibility mechanisms, why sellers and buyers might choose to engage in each, and how the different mechanisms might interact in the market for tradable units. We conclude with some thoughts about productive directions for future research.

2.

CURRENT INTERNATIONAL FRAMEWORK

2.1

International Governance of Joint Implementation

In the 1997 Kyoto Protocol, Article 6 creates the mechanism of joint implementation (JI). This Article enables Annex B parties to meet a portion of their Article 3 emission reduction or limitation commitments by transferring or acquiring emission reduction units (ERUs) generated in projects in Annex B countries that reduce anthropogenic emissions by sources or enhance anthropogenic removals by sinks. In the 2002 Marrakesh Accords, two decisions advanced the development

224

Environmental and resource economics 2004/2005

of JI. Decision 15/CP.7 reinforces three important principles governing JI. First, Annex I country purchases of JI credits must be in addition to their own efforts to reduce emissions rather than simply replacing them; this is known as ‘supplementarity’. Second, JI must maintain the environmental integrity of the Kyoto Protocol; that is, JI trades must not lead to higher global emissions. Third, trade in ERUs does not alter commitments in Annex B.15 Decision 16/CP.7 creates detailed draft guidelines for implementing JI. These guidelines are provided as recommendations to the COP/MOP. If they are accepted, through a decision at the first meeting of the COP/MOP following entry into force of the Kyoto Protocol, they will become final. The next section presents an overview of these recommendations: the governing institutions for JI; the project eligibility and participant requirements; and the ERU verification, issuance, and carry-over procedures. 2.1.1 Governing institutions Under Decision 16/CP.7 the ultimate governing authority for JI is the COP/MOP, which has the role of developing the rules, modalities and guidelines for implementation. The COP/MOP also appoints and oversees the Article 6 Supervisory Committee (see below). Both buyer and seller (host) parties must designate a focal point. The focal point of the host party must confirm that a JI project assists the host country in achieving sustainable development.16 When the host country is in full compliance with its Article 5 and 7 requirements as well as the recording of its assigned amount and its commitment period reserve, the host has the authority to verify and issue ERUs.17 When the host party is not in full compliance, it can issue ERUs only after the Article 6 Supervisory Committee has verified them.18 This committee manages the accreditation of independent entities (see below), and can recommend the format of the Article 6 ‘project design document’ (PDD) as well as revisions to the guidelines and criteria for baselines and monitoring. Accredited independent entities (AIEs) are legal entities that fulfill a list of operational and competency requirements laid out in Appendix A to the decision. AIEs are tasked with reviewing the PDD prepared by project participants, making it available for public comment, determining project eligibility, reviewing the monitoring report, and giving recommendations on verifying ERUs.19, 20 2.1.2 Eligibility of projects and participants JI project activities essentially are bound by the same limitations imposed on domestic activities under Article 3 and hence IET (for example, they are restricted to the gases, sectors and source categories listed in Annex A of the Protocol as well as the source and sink categories defined in Articles 3.3 and

Joint implementation in climate change policy

225

3.4).21 Decision 16/CP.7 of the Marrakesh Accords states that Annex I parties are to ‘refrain’ from using ERUs from nuclear facilities to meet their commitments. This is not really enforceable in IET because AAUs are fungible, but it does restrict JI. All JI projects must satisfy three criteria: they must have the approval of the parties involved; the reductions in emissions by sources or enhancement of removals by sinks must be ‘additional to any that would otherwise occur; the acquisition of ERUs must be supplemental to domestic actions by Annex I Parties for meeting the Article 3 commitments.’ Article 6 enables Annex I countries to authorize their legal entities to engage in generating, transferring or acquiring ERUs.22 The ultimate burden of compliance with the Protocol requirements stays with the national government of the Annex I party. 2.1.3 ERU verification, issuance, and carry-over procedures The Marrakesh Accords create a two-track procedure for verifying ERUs as follows.23 Track I The first track applies to host parties that are in compliance with their Article 5 and 7 requirements for national inventories, national registries, and reporting. These parties can verify and issue ERUs with no international oversight. This self-verification does not undermine the environmental integrity of the Protocol because any over-crediting of JI project activities by the host party will be reflected in the difference between the national emissions inventory and the party’s assigned amount. Thus, as long as the host party complies at the close of the first commitment period, then any over-crediting of JI projects will not represent a liability to the atmosphere, but simply a cost to the host party. If the party fails to comply with its commitment, it will face punitive measures on that basis and no additional punitive measures specific to JI will be needed. It clearly will be in the interest of the host party to develop domestic procedures for project approval, additionality assessment, baseline development, and monitoring that are transparent to project participants and that prevent accidental over-crediting of JI project activities. Essentially from an international perspective, Track I ERUs are the same as AAUs that a government has agreed to sell but with a different name; that is, track I JI is equal to IET.24 Track II The second track applies to host parties that do not comply with their Article 5 and 7 requirements. This track provides for independent review of the project and verification of the ERUs by AIEs under the oversight of the Article 6 Supervisory Committee.25 Project participants are required to complete a PDD that demonstrates party approval of the project, the additionality of the

226

Environmental and resource economics 2004/2005

project, and the inclusion of a baseline and monitoring plan in accordance with criteria defined in the Marrakesh Accords.26,27 No ERUs are created until reductions actually take place and are monitored and verified by the AIE. This system allows countries to undertake JI projects, but creates appropriate safeguards to prevent an unregulated flood of ERUs and AAUs from Annex I countries whose non-compliance cannot be detected because they are not properly reporting their national inventories. In addition to these formal mechanisms, there will be pressure from civil society to ensure environmental integrity. The public might punish firms that knowingly purchase unsound ERUs (or CERs). Thus it may be in the interest of project participants to support domestic and international procedures that build investor confidence in the environmental integrity of the ERUs from their projects. The Annex I host party issues ERUs once they have been verified under either track described above.28 Each ERU is labeled according to the specific JI project from which it was generated, the host party, and verification by the host party or the Supervisory Committee. Once an ERU has been issued, it can be cancelled or retired by the host country to meet its commitment, traded through IET, or banked.29,30 Under Decision 16/CP.7, any administrative costs arising from ERU verification relating to the function of the Article 6 Supervisory Committee shall be borne by both the parties included in Annex I and the project participants according to a decision to be made at COP/MOP1. In addition, Annex I parties are ‘invited’ to finance the administrative expenses for operating JI by making contributions to the UNFCCC Trust Fund for Supplementary Activities to facilitate preparatory work by the secretariat. There are some vague provisions also encouraging Annex I countries to facilitate the participation in JI by the ‘economies in transition’. Thus project participants will probably have to pay some amount toward general JI administration by the Supervisory Committee. Track II participants will need to pay fees for ERU verification by an AIE. In addition, it is possible that project participants will have to pay fees to their governments’ focal points for host country and buyer country project approval. If they (and even the parties they belong to) do not bear the full costs of verifying their project, JI will be more attractive to parties and project participants than is efficient.31 The largest outstanding international issue is how to set baselines for projects under Track II. The problem is very similar to the setting of CDM baselines. We discuss this in some detail in the next section. We also touch on outstanding concerns about optimal Track II monitoring.

3.

BASELINES FOR TRACK II JI PROJECTS

The draft JI agreement specifies that

Joint implementation in climate change policy

227

The baseline for an Article 6 project is the scenario that reasonably represents the anthropogenic emissions by sources or anthropogenic removals by sinks of GHGs that would occur in the absence of the proposed project.

And that A baseline shall be established: (a) On a project-specific basis and/or using a multi-project emission factor; (b) In a transparent manner with regard to the choice of approaches, assumptions, methodologies, parameters, data sources and key factors; (c) Taking into account relevant national and/or sectoral policies and circumstances, such as sectoral reform initiatives, local fuel availability, power sector expansion plans, and the economic situation in the project sector; (d) In such a way that ERUs cannot be earned for decreases in activity levels outside the project activity or due to force majeure; (e) Taking account of uncertainties and using conservative assumptions.

Conceptually this is fine but in reality accurate baselines are extremely difficult, if not impossible, to create. Essentially we are trying to predict the future behavior of humans in the context of economic changes they do not fully control. This is something we are invariably bad at, and we will never get a chance to observe how well we did because the project, by definition, will change behavior. If a baseline is set too high (emissions would have been lower than predicted), when emissions are actually measured, the difference between baseline and measured emissions will overstate the true effect of the project. This over-rewards participants and creates an environmental integrity problem; the excess ERUs will be used to allow higher actual emissions elsewhere. In contrast, if the baseline is set too low, participants will be under-rewarded. If participants are aware that the baseline is too low when the project begins, they might choose not to participate even though they could have reduced emissions in a cost-effective way. If the participants are not aware that the baseline is too low, they will find it hard to reduce monitored emissions far below the baseline. The risk of a baseline that is too low raises uncertainty and makes JI projects less attractive. Given this impossible task, we need to find a satisfactory compromise that achieves sufficient accuracy to assure reasonable environmental integrity, minimizes random variation, for equity reasons, and particularly avoids the potential for strategic bias that rewards less honest people. The tradeoff is that the baseline must also have relatively low cost: financially, in delays, and in terms of the skills required. This is important because otherwise potential participation in and gains from the program will be severely limited. OECD IEA (2000) offers an in-depth discussion of these issues and gives some suggestions for solutions in specific sectors: cement, iron and steel,

228

Environmental and resource economics 2004/2005

electricity generation and energy efficiency. Fischer (2002) offers an economics-focused assessment of three options for baseline setting: historical emissions, average industry emissions, and expected emissions. 3.1

Methodological Challenges

3.1.1 Prediction Predicting the future is hard. That said, it is less hard in some situations than others. It is easier in relatively stable sectors and countries. Predicting emissions in the USA is much easier than in Russia for both economic and political reasons. Predicting emissions from large utilities is probably (though not always) easier than from the renewables sector simply because utilities have been around a long time. The law of large numbers makes prediction more accurate as project scale increases. Many of the idiosyncratic factors that have large effects in small projects will balance each other out in a large project. In a large project, there are fewer avenues for ‘leakage’, where a project affects emissions outside the project, as many are captured within the project (see further discussion of leakage under ‘monitoring’), so estimating a baseline outside the project boundaries is less critical. Short-term predictions can be either less accurate because of transient shocks that would smooth out over several years, or more accurate because they do not need to deal with unpredictable technology shifts and major policy changes. Thus encouraging larger projects and requiring only medium-term predictions (see section below on adjusting baselines) could improve the quality of prediction. Baselines will never be perfectly accurate. Thus we need to trade off the benefits of engaging extra actors in the mitigation effort against the environmental risk that JI projects impose.32 3.1.2 Who should create the baseline? The draft Annex B decision implicitly assumes that the project participants choose their own baseline and then justify it. This might need to be challenged. The choice of best baseline designer involves two tradeoffs: between knowledge and incentives, and between skill and cost. Those directly involved in the project have the best knowledge about what would have happened without the project. This argues for having them involved in the baseline setting process. However, they also have the incentive to overstate baseline emissions because they directly profit. The challenge is to elicit their information but not allow it to create bias. Any non-transparent, non-replicable analysis provided by the participants (buyer or seller) is unreliable.33 If in the PDD they can be required to provide information that can be cross-checked (in random audit), their input will be useful. Creating a baseline (particularly using more aggregate approaches) requires

Joint implementation in climate change policy

229

skills that participants may not have. This problem can be addressed partly by creating standard methodologies and partly by using consultants and AIEs. There is a tradeoff between requiring high quality and controlling cost. Even consultants can be captured by project participants, so the methodologies need to be simple so others can understand them and the information used should be auditable. Complex models may be more accurate when used correctly but they are also more subject to manipulation.34 3.1.3 Adverse selection If all baselines were perfect, it would not matter which projects went ahead and which did not for environmental integrity. If, however, there are errors (even if they are unbiased over all potential projects) and the project developers can identify the errors because they have private information, the errors will lead to systematic bias that will harm the environment. Projects with overly generous baselines will be more likely to be carried out. The errors will no longer wash out over a large number of projects. If there are errors that no one can predict, project developer behavior will be unaffected and there will be no bias. 3.2

Possible Methodologies

3.2.1

Regional/sectoral top-down/bottom-up, or engineering-based, modeled baselines For the best long-term, unbiased predictions that minimize leakage, the larger the scale the better. Baselines estimated at a national or regional scale using historical data combined with bottom-up modeling offer the least biased and potentially most accurate predictions. These are the type of models that underlie integrated assessment models used in developed countries.35 They are general equilibrium models, so they control for leakage within (and potentially outside) the country. To provide detailed bottom-up information for all sources is, however, prohibitively expensive, so these models would not be accurate if applied at the source level. For small projects, these are often of limited value. The error on each project would be unacceptably high and, with adverse selection, the degree of bias could be considerable. They might best be used to create regional baselines or baselines for specific sectors, while controlling for interactions with other sectors and regions through the general equilibrium model. If a country can create and monitor a national ‘project’, which would make a national baseline useful, it can use IET.36 3.2.2 Sectoral benchmarking This is a simpler system that tries to balance the benefits of scale with sectorspecific detail without having prohibitively high transaction costs.

230

Environmental and resource economics 2004/2005

Multi-project baselines or ‘benchmarks’ project a business-as-usual emission rate per unit of output for example, tons of CO2/kWh for the power sector, or tons of CO2/ton of cement for the cement industry) based on aggregated data for a region, sector, project type or technology. Individual projects compare their emission rates to the multi-project baseline emission rate (OECD/IEA, 2000; Lawson et al., 2000). As with any more aggregated baseline, a benchmark may have high variance of errors for individual projects. With adverse selection this can create bias in project choice and loss of environmental integrity. Making the benchmark more stringent can reduce this problem (Lawson and Helme, 2000), but may come at the cost of loss of participation. On the other hand, the lower transaction costs, relative to project-specific baselines, may compensate developers for the greater stringency needed. A shortcoming of this approach is that it simply estimates emissions rates. It does not take account of changes in production levels. Thus it is vulnerable to leakage through changes in production levels. It is appropriate only for projects that focus on upgrading technology. It also leaves open the question of how best to generate the baseline from aggregated data. 3.2.3 Project-specific These use detailed information and assumptions about plant-specific fuel use, technology use and output levels over a period of time. In some ways this is the most accurate because of the specific detail it uses. However, it suffers from leakage because of the small scale, and from potential manipulation by project developers because the data are less replicable. It is also costly because it requires technical capacity and needs to be repeated for every project. Establishing simple standardized methodologies that use project specific data only where they can be audited can minimize costs and potential for manipulation. 3.2.4 Revelation mechanisms? An area that deserves more examination is the possibility of creating a mechanism that induces unbiased revelation of private information about baselines. Wirl et al. (1998) demonstrate the incentive to distort baselines in a principal–agent framework and show that an exogenous baseline avoids manipulation. They also illustrate how the efficiency loss required to ensure honest revelation in their model sharply reduces the predicted emissions reductions that could be achieved through JI and that those who already have relatively low emissions make most reductions. Baselines are set so low that many genuinely valuable projects are unprofitable. Kerr (1995) shows how private information about both costs of emissions reduction, and the risk associated with emissions reduction, can be partially

Joint implementation in climate change policy

231

revealed by requiring project organizers to choose between a number of contracts that offer a combination of a credit price and a total number of credits for the project. She models the situation where a fund tries to maximize reductions from a fixed budget. Reductions that would have happened anyway are simply zero cost reductions. The result is that those who have genuinely high baseline emissions will choose small reduction projects with a small number of credits but also a high credit price, while those with low baselines will choose to have a large number of credits with a low price per credit. This relies on a negative correlation between the average cost of reducing from a given emissions level (a higher baseline implies higher costs of reduction from any given level of emissions because there are fewer free or cheap ‘reductions’) and the optimal number of emission reductions offered. If a project developer lies about the cost of reducing to a certain level of emissions (or equivalently lies about their baseline/marginal abatement cost curve – an artificially high baseline lowers the cost of ‘reductions’), that is, says it is high, then they are ‘punished’ by only being able to have a small project. The appropriate schedule of average payment per credit awarded and total contract size induces accurate revelation. Different schedules might need to be developed for different countries and sectors. This approach might be applied to setting different rules for small and large projects (see section 3.4). Hagem (1996), Janssen (1999) and Fischer (2002) also consider the issues of incomplete information, adverse selection and strategic behavior in JI. Ausubel et al. (2003) develop a mechanism to reveal baselines in an auction context.37 They solicit baseline information at the same time as bids by making payments dependent on average industry bids as well as the individual’s information. It would be interesting to adapt both of these approaches to a trading situation and do empirical work to explore how they might be applicable in practice. If they could be made to work, they could produce unbiased baseline predictions but a great deal of uncertainty would remain. This uncertainty would be borne by the project participants. 3.3

Updating Baselines

Over anything other than the immediate future, the main source of error in unmanipulated baselines will be exogenous changes in technology, and global and local macroeconomic and political shocks. These could quickly lead baselines that are fixed at the beginning of the project to be hopelessly inaccurate. One way to address this problem is to update the baseline as information on these shocks is revealed.38 In some circumstances this reduces uncertainty for both the international community and the project developer (the shocks should affect the baseline and the actual measured emissions roughly equally, so the updating of the baseline offsets the exogenous shift in emissions and reduces

232

Environmental and resource economics 2004/2005

total risk).39 In others there is a tradeoff between imposing risk on investors and reducing environmental risk. Baselines should be updated only in response to changes that the project developers cannot possibly affect. Otherwise updating will invite strategic behavior. Benchmarks are a form of updatable baseline because the baseline automatically shifts with the level of output. However, output is not exogenous. If a project developer makes good profits from credits on each unit of output, they will have an incentive to inflate output inefficiently, thus increasing total emissions and raising the overall cost of emission reductions (Fischer, 2001). The benchmark emission rates could themselves be updated by using data on emissions rates in parts of industry that are not involved in climate change regulation. As more and more countries and companies are involved in regulation, however, this updating would become less accurate.40 The formula by which baselines will be updated needs to be defined clearly in advance to reduce uncertainty and avoid difficult expost negotiations. 3.4

Transaction Costs and Participation: Different Rules for Small Projects?

The CDM has a ‘fast track’ to facilitate participation by those who offer small projects. The rules are less stringent. Is this appropriate? Not in general. In expectation, there is at least as much environmental risk from a large number of small projects as from one large project. If all are equally biased, the total environmental integrity loss will be the same. If the rules can be made less stringent for small projects, they should also be able to be less stringent for large projects. One legitimate reason for a difference in rules is a desire to stimulate participation by groups who can offer only small projects. This could be either for equity reasons or to engage these groups so they can participate more fully in future. These projects could lead to learning and technology development. If this is the justification, the relaxed rules should not be for small projects per se but for specific types of projects offered by specific types of groups regardless of the project size. Otherwise large projects might be broken into small parts to fit the rules. This mistake should not be repeated in joint implementation. If ‘small’ can be defined in terms of the number of credits gained relative to the total level of emissions, and companies that would genuinely have high emissions in the baseline case also find it more expensive and difficult to create a project that will move them to very low emissions, then it might be possible to offer ‘small’ projects more generous baselines without tempting those who want to produce large numbers of credits to participate. This would be an application of the ideas on revelation mechanisms. Whether it could be

Joint implementation in climate change policy

233

applied is partly an empirical question about the correlation between high baseline emissions and ‘small’ optimal projects. 3.5

Baselines and Limitations that are Stricter than BAU

A strange baseline setting problem arises in joint implementation and nowhere else. It primarily affects countries whose emissions limitations are strictly binding. Even if the economy were saturated by projects that reduced emissions relative to business as usual, the country might still not be in compliance. Some rule has to be set to allocate the nationally required emission reductions across projects. This could be in the form of emissions limitations (for example allowances that are tradable or not) beyond which JI credit can be earned. Alternatively the government takes on the cost of making the national reduction. In the symmetric case, countries with non-binding limitations, where governments need to allocate the rewards from ‘hot air’, most people assume that governments will capture these gains. 3.6 Monitoring Under the international rules described above, AIEs approve a monitoring plan and actual monitoring reports before they verify ERUs. The draft JI decision is clear on the need to collect and archive data and have quality control processes for monitoring. The method of monitoring is not completely specified yet, however. OECD/IEA (2002) discuss monitoring and leakage in detail. As with baselines, project implementers have the best information; however, they have an incentive to understate actual emissions to increase their rewards. Hargrave et al. (2000) discuss the problems that arise with national-level monitoring and propose possible solutions. Many of these insights apply to individual JI projects as well. Basically there is a choice between deterrence and prevention. If deterrence is used, all information and methods used in the monitoring plan must be documented and auditable, but auditing could involve considerable effort. Then projects should be randomly audited and those (and their AIEs) that are found to have misled the Supervisory Committee should be punished. This allows detailed monitoring that accurately matches true emissions. It depends on being able to impose credible severe penalties. At an opposite extreme, the monitoring process could be made much more transparent so that cheating is avoided. This requires the use of simple standardized methodologies and readily observable verifiable information. Third parties could check all project reports quickly before the ERUs are verified. The advantage of this is that cheating is completely avoided. The disadvantage is that the simplicity required is likely to make the measurement of emissions

234

Environmental and resource economics 2004/2005

much more inaccurate. This imposes uncertainty on project participants who could make certain reductions but then receive a different number of credits. The participants would probably redirect their efforts to create measurable reductions, rather than real reductions so some efficiency is lost. The ideal solution is probably somewhere between these two extremes, with elements of both. One option is to have a simple default process but make detailed audits voluntary. This would reduce participants’ uncertainty and redirection of effort but would induce environmental bias because only those who expect to gain credits through the detailed audit would volunteer. Fines are limited at an international level, so simple monitoring should probably be emphasized. Ideally the process of monitoring JI projects should help countries build the monitoring infrastructure to meet their Article 5 and 7 commitments and simultaneously create domestic programs that allow them to evolve toward efficient regulation and use of IET. Thus applying the methods required by IPCC best-practice guidance for national inventories to individual projects wherever possible would be sensible. 3.6.1 Leakage Not only the effects within the project boundary (directly controllable by the participants) but also the indirect effects their actions have outside must be measured to assess the true environmental effect. The effects outside are called ‘leakage’. For example, a project that increases the efficiency of cars will directly reduce emissions but, indirectly, by lowering the cost of driving, might lead car owners to drive more and even purchase more cars. The severity of leakage will depend on the project but could be large. Assessing leakage requires monitoring emissions outside the project and also estimating a baseline for those emissions. Leakage can be assessed by using theory to identify paths of effect,41 informed judgements on their relative importance, data on the scale of relevant variables (how many cars there are and how much people drive now) and estimates of relevant elasticities. Ex post, monitoring information could be used to assess the extent of leakage (for example updating the number of cars actually affected and the real price of gasoline). The baseline could be determined explicitly or implicitly by using a model of emissions generation and assuming which key variables are affected by the project and to what extent. The appropriate baseline depends on the form of monitoring – it needs to be consistent. Project participants may have some information on the type and scale of leakage, but they have incentives to understate any leakage that would reduce the credits created and overstate any positive effects. They will not provide unbiased information. In relative terms, increasing the scale/scope of the project reduces the leakage problem because more effects are captured within the project. In absolute terms, of course, a larger project might create more leakage.42

Joint implementation in climate change policy

4.

NON-INTERNATIONAL ISSUES: DOMESTIC GOVERNMENTS AND THE PRIVATE SECTOR

4.1

Domestic Governance of JI

235

A country that engages in JI is essentially allowing the project to sell some of its AAUs. If it complies with Articles 5 and 7, it is concerned with maximizing the value of the AAUs it holds and, it is hoped, with compliance with its targets. The country needs to be sure that it is getting good value in return for the project. If the project is freeing up AAUs through genuine reductions, the government does not lose any AAUs on net. If, however, the project is overrewarded, the government loses. It either has fewer AAUs to sell through IET, or it has to buy more AAUs to comply. These countries will have a keen interest in the quality of the international approval process and may want to add additional layers of supervision if they are unsure. Their JI office will be a control point and they probably won’t want to completely delegate power to trade to legal entities. The exception to this will be where the government has a domestic emissions permit system (tradable or not) and it is confident about its level of enforcement and that there is no domestic hot air within its system. In this case it can be confident that any sales of permits outside the country will correspond to a tighter cap and hence lower emissions inside the country. In this case it could devolve trading responsibility to legal entities and simply act as a rubber stamp. If the country is not able to comply with Articles 5 and 7, then it cannot sell ‘hot air’ and may not have any concerns about compliance with targets.43 The AAUs it holds have no obvious value to it. In this case its only concern with the project might be to do with its domestic effects: technology transfer, foreign investment, environmental effects and so on. These effects could be addressed through existing regulations relating to these issues. The country may, however, be interested in promoting JI projects because they might bring in revenue, technology and transfer of skills. The country’s JI office might be focused on marketing rather than control. Thus the countries that are most likely to sell through JI are least concerned about the approval process. Even countries that are not in compliance with Articles 5 and 7 may see this as a transitional phase. They may want to use JI to reduce their future emissions by changing technology, to increase information about firm emissions and to evolve toward a situation where they do comply with Articles 5 and 7 and can have effective domestic regulation. Whether JI is the best way to achieve this is a serious question. If JI is project based while national inventories are top-down (that is, calculated from aggregate national data on production and trade flows), JI might not help build

236

Environmental and resource economics 2004/2005

monitoring systems that much. If the government is thinking of moving toward a tax on fossil fuels (at the producer/importer level) or an upstream tradable permit system, JI has few lessons to offer. It might lead inexorably toward cumbersome facility-level trading and regulatory systems. Buyers, including buyer countries, are not internationally liable for the quality of the credits they buy internationally. Therefore they should have no legal concern about approving a purchase of JI credits. A rubber stamp should be sufficient. The only exception to this would be if, for political reasons, the government wants to limit the purchase of ERUs and also select which ERUs will be purchased.44 This could arise for example if they want to use purchases of ERUs to reward (or punish) some countries for cooperation (or not) on other issues, or if they have a domestic constituency that objects on environmental or other grounds to purchases of ERUs from certain sources and countries.45 If for whatever reason a country wants to engage extensively in JI, it might want to reduce transaction costs by establishing bilateral arrangements with buyer countries to encourage investment. If buyers are concerned about the quality of ERUs for domestic reasons, these bilateral agreements might provide some efficient quality assurance. The seller government might also provide some services to potential investors to reduce the costs of producing a project design document. They might even want to facilitate sector-wide projects that could have much lower transaction costs relative to the ERUs they create. 4.1.1 European Union issues Some interesting baseline issues are raised by the advent of the European Union-wide GHG trading system and the accession of central and Eastern European countries to the European Union (EU).46 It is not clear how the EU will allocate emission allowances to the accession countries that will potentially be eligible to join the trading system as of 2005. The trading allowance allocation system being recommended for industries in Western European countries is based in part on business as usual and, after 2008, will need to be tied in some way to the overall EU commitments under Article 3. It is not clear how hot air will be treated in this case. In the case that they did not join the EU trading system, it is not clear whether Track II JI baselines for accession countries would be tied to EU standards if accession countries are required to meet those EU standards for environmental performance. This could significantly reduce the perceived benefit of investing in JI projects in accession countries. Accession countries that did not join the EU trading system but do comply with Articles 5 and 7 would certainly be able to trade their hot air outside the EU. These countries could self-certify their JI baselines so their baselines would not have to meet EU standards for environmental performance.

Joint implementation in climate change policy

4.2

237

Contract Design between Buyers and Sellers

A contract for ERUs could be as simple as a purchase contract, where the price and timing of delivery and payments are the only issues, or as complex as a joint venture. Many commentators on JI assume that the buyer is an investor involved with the project from the start. In some cases this might be efficient because the buyer might bring technology and skill to the project. In others the potential seller might be able to contract for the technology separately or purchase it locally and may have plenty of skill. Not all projects require very advanced technology. Foreign investment and particularly joint ventures are complex to organize and the extra lure of ERUs might not overcome the barriers that often limit such projects. Where the contract is a simple purchase contract, the buyer will want to pay on delivery to ensure performance while the seller might want the money in advance to ease credit constraints and facilitate the investment needed to achieve the emission reductions. Compromises between these desires might affect the price paid. If the contract is a joint venture it will need to deal with financial and inkind contributions by both parties, sharing of profits and ERUs (these are fungible if there is a market for ERUs), and risk sharing. The investor may not be interested in being the ultimate user of the ERUs; they may simply see them as another product. As JI becomes more valuable and hence more commercial, the usual contracting risks will arise and the usual problems of less capable actors being exploited by more capable ones (in which ever country these are) will arise. If buyers are systematically more informed and have access to better contracting advice, they may bargain more effectively and gain more from the contract. This is no different from any other interaction among unequal commercial players. The possibility of joint ventures is also not unique to JI. Under IET companies might still find it worthwhile to form joint ventures for specific projects in countries that can reduce emissions effectively with some technological input. Depending on the form of domestic regulation, their reward would be designated in either lower regulatory burdens or taxes or in AAUs that they can sell. The only reason there would be any difference between investor behavior under JI and IET would be if a seller government chose to give special regulatory treatment to investors that engage in JI, thus making investment easier. There is no obvious reason for seller governments to do this, however.

238

Environmental and resource economics 2004/2005

5.

COMPARISONS BETWEEN MECHANISMS

5.1

Joint Implementation and International Emissions Trading

5.1.1 When would a seller choose to use Track II JI rather than IET? This is a two-step question. First, will the country be able to use IET? Second, if they can, how great are the benefits relative to using JI? As with most twostep problems, it is easiest to think about the second question first. 5.1.1.1 How great are the benefits from using IET rather than Track II JI? The factors that affect the relative attractiveness of IET (or Track I JI) versus Track II JI include: the net benefits from allowing trade from a non-projectbased domestic regulatory system, the transaction costs of using the international Track II accreditation process relative to a domestic process, the likelihood that commitment period reserve restrictions will bind; and the effect of using Track II accreditation on public perceptions of the environmental integrity of the trade. 1. (i)

Net benefits from trading non-project-based units Gains from international trade derived from non-project-based regulation IET allows trading of AAUs that are not needed for compliance. They do not have to be associated with any specific regulation or project. This is immediately relevant for countries with hot air. Hot air can be legally traded under IET but not under JI because of the additionality requirement. It is also directly relevant for countries that can make reductions through the use of nuclear energy; these reductions cannot be traded under JI. It will also be relevant for countries with significant real potential to reduce emissions at low cost. If a country has a relatively generous target under Article 3 and relatively inefficient current use of fossil fuels (or high but controllable emissions of other GHGs), it could potentially be a large seller if it has efficient domestic regulation. Any efficient regulation will need to go beyond small projects, so the country is likely to want to be able to trade reductions from policies such as emissions trading, emissions taxes, promotion of energy efficiency, building standards, transport policies, and strategic infrastructure investments. If a country thought these policies might be used to exactly achieve Article 3 compliance but no more, it could allow any credits from projects that go beyond the effects of broad government policies to be sold. Thus broad policies would be used domestically but all international trading would be through projects. This would be hard to judge and implement, however.

Joint implementation in climate change policy

(ii)

239

Another common reason why countries might want to allow flexible entity-level trading based on broad policies, even if they are unlikely to be a large net seller (or may even be a net buyer), is if they want to use a domestic emissions trading system. They might want to avoid the problems of market power or lack of liquidity that can arise in a small country.47 They might also want to automatically link the stringency of their domestic system to the international market by allowing free trade and price equalization across systems.48 Domestic regulatory set-up costs The domestic regulatory costs of a project-based trading mechanism under either IET or Track I JI are identical. There is no reason to use different rules. Thus here we focus on the differences between IET and Track II JI. Many countries have already participated in the Activities Implemented Jointly pilot phase. Thus they are familiar with the domestic institutions required to participate in this process. Those for Track II JI will be similar. This might make it attractive. They avoid the need to be able to certify their own baselines or monitor emissions from their own projects. They already have focal points designated for giving domestic approval. They may be comfortable with the quality of the international certification process. If they do not intend to demonstrate compliance with their targets, they may not be that concerned with the true additionality of the credits and the effect of the sales on their ability to comply anyway. In any case, many have hot air that they will be unable to sell under JI so they are at little risk of non-compliance unless low environmental integrity JI projects happen on a large scale. If a country chooses to use projects under IET it must certify baselines and verify monitoring itself. It could simply copy the international JI process, prepare a PDD and hire AIEs to certify and verify. It would face slightly lower costs than under JI if it could avoid costs associated with the Article 6 Supervisory Board. Alternatively it could use this system as a base and adjust it to its own requirements. Its system could be stricter (because it really bears the environmental liability which will be measurable in the national inventory – though not associated with specific projects) if it doesn’t have confidence in the international system. It could be more relaxed if it were willing to trade off administrative simplicity, which will encourage more projects, with greater risk bearing by government. In either case it could be better adapted to the specific needs of the country.

240

Environmental and resource economics 2004/2005

If the country chooses to create a non-project-based domestic regulation and allow trading on that basis, the initial set-up costs would probably be much higher. For example, creating an effective domestic tradable permit system requires both up-front costs and ongoing monitoring and administration. These costs would need to be weighed against the likely gains both domestically and through international trading. The government could also decide to set up domestic emissions trading in some sectors but have project-based reductions in others. (iii) Uncertainty for the government under IET versus JI Two forms of uncertainty arise. The first is how many real additional reductions a specific project or action creates – how many AAUs the project frees up for the government. A domestic trading system that is well enforced offers complete security on this. It is essentially a large ‘project’ with a baseline defined as the ‘cap’. The government knows exactly how many emissions it will have from the sources covered by the trading system. If the sources choose to reduce emissions even further and sell internationally, this does not affect the government at all. These AAUs could be sold as soon as the trading system is created and the entities that sell them will be liable under domestic law if they are unable to comply with their domestic limitations after they sell AAUs. With any other sort of project (without a fixed cap) there is uncertainty on actual emissions until they occur and are monitored. Even once monitored, the real reductions from projects will always be uncertain because of uncertainty in the baseline. In a domestic trading system the baseline never needs to be formally agreed. Estimates of it will enter implicitly into negotiations on how strict the cap should be and how permits should be allocated across sources, but it creates no uncertainty for government. The target is designated in emissions levels and so is the cap. The second form of uncertainty for government is in national performance. In both JI and IET what ultimately matters for the validity/additionality of credits is national performance, not performance in any specific project or sector, so this uncertainty is identical. Any JI project (Track I or II) is only really valid if the country complies overall.49 A final issue relates to the potential timing of sales under the different systems. ERUs can be traded as soon as they are verified. Project accounting may well move faster than national accounting. If IET could not occur until national accounts were finalized, this would be a significant advantage for JI. However, AAUs could be

Joint implementation in climate change policy

2.

241

traded any time from 2008 onward. Countries with good monitoring systems will know roughly how many they are likely to have to spare quite quickly. Legal entities within domestic trading systems will also know early on. Both countries and firms can make fine adjustments at the end of the commitment/domestic regulation period. The only real limitation on early trading in IET comes from the commitment period reserve. This is likely to bind strongly only on countries that intend to be out of compliance. These are the targets of the restriction. For others this might simply alter the timing of sales. Countries that are severely affected by the restrictions on the timing of sales could choose to submit projects for verification under JI Track II to avoid the restriction. Transaction costs of each trade under JI (Track II) versus IET The time and cost burden of verifying ERUs through Ttrack II will determine the desirability of using it for JI projects on a voluntary basis. Trading under IET would likely involve lower transaction costs than JI. The transaction costs for both include finding a trading partner; getting government approval for the trade; meeting any international requirements; and monitoring emissions. For any IET trade, the identity of the trading partner is irrelevant, so this is much easier. It could be done through an anonymous market. In contrast, in JI the partner needs to get approval from its host government and because the ERUs come from an identifiable project some partners might have direct concerns about the perceived environmental integrity of the project. This makes matchmaking harder. For IET, government approval could be as simple as delegating power in a domestic tradable permit market. Alternatively it could be a rigorous process of baseline setting and verification of emissions before trade if the government uses a project-based system and is concerned about the additionality of trades. In JI, government approval will only be complex if the country decides to put additional limitations on purchases of ERUs because it does not fully trust the international system to protect the environmental integrity of the Protocol or because it has objectives in seller countries, such as environmental sustainability, that are not addressed in the Kyoto Protocol. In contrast, however, there are no international requirements (as yet) for IET whereas JI projects have to go through the international certification and verification process. Finally, if IET occurs as an adjunct to a domestic emissions trading system, no additional emissions monitoring is needed for

242

Environmental and resource economics 2004/2005

3.

international trade. Monitoring is an intrinsic part of the domestic system. If the IET trade is project-based, monitoring costs will be similar to JI if the stringency of the verification is similar to the international requirements. Otherwise it could be more or less costly. Overall, transaction costs per trade are likely to be much higher under Track II JI than under IET that links to a domestic trading system. The transaction costs under a project-based IET system could be either higher or lower than under Track II JI but these will be tailored to the needs of the specific country. If they impose higher transaction costs it will be because they want a higher level of certainty about additionality than they believe the international rules achieve. Higher JI transaction costs will reduce the overall gains that can be realized under JI. Public perceptions of environmental integrity The most reassuring thing for those who are skeptical of the trading system should be compliance with Articles 5 and 7 and then compliance with Article 3. If a seller complies with all three, the environmental integrity of a seller’s contribution to the agreement is (as far as possible) assured. If a country or company wants to avoid buying ‘hot air’, it should buy only from countries that either cancel their hot air or sell only through JI Track II. If a buyer chooses to buy from countries that might not be in compliance with at least one of Articles 5, 7 and 3, it might be concerned about perceptions of the environmental integrity of the units it purchases. If the seller country is not in compliance with Articles 5 and 7, the buyer can buy only through JI Track II so has few choices: only to buy or not to buy. If, however, the seller is considered to be at risk of non-compliance only with 3, the buyer has a choice between purchasing through IET (maybe even insisting on project-based credits) or through JI Track II. It is not legally liable for compliance of the countries it purchases from, but might be held morally liable if people think the company/country knows it is purchasing shaky credits. Track II JI might offer some assurance that the ERUs meet certain standards.

5.1.1.2 Will the country be able to use IET? The previous section discussed the pros and cons of IET relative to JI. A country will have this choice only if it complies with Articles 5 and 7. Each country needs to make a decision about how much effort to put into complying with Articles 5 and 7 by 2008. Many countries will comply so that they can prove that they have met their Kyoto targets. Others, particularly economies in transition, may find the

Joint implementation in climate change policy

243

cost of proving compliance high and may not care enough about the international rewards from proving Kyoto compliance. Compliance might be considered to be a function of capacity, but capacity is often strongly correlated with willingness to comply. A country that can make huge gains from trading under IET rather than JI will probably be able to find the resources (or get its potential buyers to provide them) to build adequate monitoring systems. If JI is seen as an unattractive high-transactioncost way to trade, those who want to trade, even if they have little or no hot air, will want to invest in monitoring systems as a precursor to gaining the benefits of IET. Some, however, will, find it either impossible or not worthwhile (probably indistinguishable) to comply by 2008 and will be forced to use JI if they want to trade at all. 5.1.2

When should buyers use one mechanism rather than the other? All tradable units are economically equivalent from the buyer’s point of view. Buyers should ultimately be willing to pay the market price minus the transaction costs they bear. Higher transaction costs make otherwise equal projects different. If buyers face different transaction costs, they will be willing to pay less for the units that involve the higher transaction costs. JI credits will tend to receive less than AAUs sold directly through IET. If the seller decides to bear the transaction costs, the seller will receive a higher price, but this should just compensate them for the higher costs they bear (unless they are more efficient at managing the trade). Some domestic governments might put restrictions on the number of both JI and CDM credits that can be used domestically in part because of concerns about the environmental integrity of these credits. These restrictions are not yet decided but will affect buyer options. Many people believe that in JI, the buyer might have a greater tendency than in IET to be an investor in the project as well as a simple purchaser of ERUs. If a buyer is an investor as well, it will prefer projects that have lower risk as well as lower prices. Projects in countries with stronger institutions and greater human capital will tend to be less risky. The role of buyer as investor may, however, be more a function of the countries involved than of the mechanisms themselves. To reduce emissions efficiently, companies in economies in transition may need capital and expertise from more advanced countries regardless of the form of trading they might use for the reductions they generate. Thus economies in transition may tend to engage in JI, may be more risky locales for projects, and may tend to have active involvement of buyers in projects but these may not be directly related.

244

Environmental and resource economics 2004/2005

5.2

Joint Implementation and the Clean Development Mechanism

5.2.1 What is the difference between the suppliers? Joint implementation will mostly involve economies in transition while the clean development mechanism involves developing countries. Neither of these is a homogeneous group. There are four key differences between the groups. First, the developing countries that will engage in CDM are non-Annex B parties. They do not have emission reduction or limitation commitments and are not subject to the same national inventory, registry and reporting requirements as Annex B parties. Even though the main participants in JI will not be in compliance with these requirements (those who can use Track II only), they might be expected to be working toward compliance. Second, the economies in transition are expected to want to behave in the near future like other Annex B parties that have strong monitoring systems and truly binding limits. Many are candidates for accession to the EU. Most aspire to be considered equal to developed OECD countries. They might be concerned about creating a reputation for compliance. Third, the potential scale of CDM provision from developing countries is much higher, so any problems in implementation will be magnified. Fourth, the ability of the government to design and effectively implement regulations without risk of corruption or incompetence varies both between and within groups. Developing countries may be generally less capable, however; some ‘developing countries’ have advanced strong governments while some economies in transition have relatively weak governments. These differences have motivated different international rules and will also affect the level of supply (and possibly demand) for CERs relative to ERUs. 5.2.2 How are the international rules different? Although both JI and the CDM are project-based mechanisms and are subject to similar additionality requirements under the Kyoto Protocol, these mechanisms and their governing rules are different. The Protocol requires that CDM projects achieve the dual purpose of assisting non-Annex B parties (that is, developing countries) in achieving sustainable development and contributing to the ultimate objective of the Convention, and of helping Annex B parties to meet a portion of their Article 3 commitment. Under the CDM, eligible project activities exclude avoided deforestation and the ‘land use, land use change and forestry’ (LULUCF) activities defined in Article 3.4. The only eligibility requirement for non-Annex B parties to host CDM projects is ratification of the Protocol. The governance of the CDM is similar to the Track II governance of ERU verification by the Article 6 Supervisory Committee, but involves different bodies. A ten-member CDM Executive Board governs the registration of CDM projects and the certification of their emission reductions. In-depth project review for validation and

Joint implementation in climate change policy

245

verification of emission reductions is undertaken by Designated Operational Entities (DOEs). JI projects can begin as early as 2000 but ERUs can be issued only for activities undertaken after the beginning of 2008. The Article 6 Supervisory Committee will not be appointed until the first meeting of the COP/MOP following entry into force of the Protocol. Once the Supervisory Committee has been appointed, it will take time to finalize the rulemaking for verifying ERUs under Track II. In contrast, the CDM was designed to start generating CERs as early as the year 2000. The first CDM Executive Board was appointed immediately after the approval of the Marrakesh Accords, and launched into designing the accreditation process for DOEs and developing the project design document and baseline guidelines. At the time of this writing, the CDM Executive Board was preparing to approve the accreditation of DOEs and begin registering the first CDM projects. The different governance structures, and underlying that, different levels of trust of sellers and different concerns about the potential scale of implementation problems, will have a major effect on the relative transaction costs borne by CDM projects relative to JI projects. The CDM process involves more subjective issues and may be more political. It could involve longer time lags and a greater demand for substantiation of project claims (for example, baselines, monitoring and sustainability). In addition the direct costs payable for international certification are likely to be different. The JI payment sharing rules are not yet defined. Under the CDM, project participants will have to pay fees to the DOEs to get their projects validated and their CERs verified. In addition, a share of the proceeds from CDM projects (2 percent) must be applied to support adaptation activities in non-Annex B countries. An additional share, as yet undecided, will be taken to cover the administrative costs of the CDM. Transaction costs might be much higher in the CDM than under JI. 5.2.3 Should buyers treat these mechanisms differently in practice? As discussed above, buyers should only treat units differently if they have different transaction costs or if they are restricted in the use of one type of tradable unit. CDM credits may well face the highest transaction costs of all the mechanisms, so the sellers will tend to receive less than the market price. Buyers are restricted by the international rules in their use of CDM credits (for example for LULUCF credits50) and individual countries may put on other limitations. Where buyers are involved as investors in the project and bear project risks they will take into account the relative riskiness of different project partners. They might find some developing countries highly risky while others are relatively good risks. Similarly they will have different attitudes to risk in different economies in transition. It is hard to say how the ‘average’ risk will vary across the two groups.

246

5.3

Environmental and resource economics 2004/2005

The Joint Market for AAUs, ERUs and CERs

The market price for tradable units will be set by demand and supply. If demand rises, the price will rise; if supply rises, the price will fall. All countries will have emissions abatement opportunities ranging from near zero cost to extremely high cost. Their marginal reduction should cost the same as the market price net of transaction costs. Thus no instrument intrinsically offers ‘cheaper reductions’. However, the marginal cost curve for reductions will be more elastic in some countries and the total level of emissions and hence potential scale of reductions varies enormously. Many people argue that because economies in transition are currently very inefficient and because they have high levels of capital turnover (because their stock is so old) they have a flatter cost curve and hence large numbers of cheap emission reduction opportunities. Developing countries may have similar opportunities – they are also extremely large in terms of population and at least some for example China) have high growth and so high levels of new investment. In addition, Russia in particular has a lot of ‘hot air’, which is essentially zero cost tradable units. Thus the supply of ERUs (and possibly AAUs from economies in transition) and CERs is potentially huge and could lower the market price enormously. How much this potential can be realized depends partly on the monitoring requirements and transactions costs involved in the mechanisms. Given the impossibility of estimating accurate baselines, few projects would actually be feasible if parties were unwilling to risk some loss of environmental integrity. If some economies in transition can comply with Articles 5 and 7, they can sell their hot air and also avoid the transaction costs involved in JI; this would greatly increase their supply. Realizing this potentially large supply also depends partly on the inherent difficulties in investing in technology change in countries with relatively undeveloped institutions and economic systems. It is not chance that these countries have less efficient technology and high emissions per unit of output. Changing that is not easy and may not be as cheap as engineering models suggest. Foreign investment in both economies in transition and also many developing countries is challenging for reasons that have nothing to do with climate change regulation. All these uncertainties, as well as the inherent difficulty of modeling the future, means that the actual level of market price is extremely hard to predict. Model estimates vary widely but real uncertainty may be greater still as models find it difficult to account for transaction costs or weak legal institutions that deter investment. All this discussion assumes that the market operates competitively. Unless countries devolve a lot of the trading to legal entity level (as the EU seems to be moving to do), trading will have a strongly political aspect. This might

Joint implementation in climate change policy

247

make ERUs more valuable because countries prefer to pay money to countries they are comfortable giving aid to. Conversely it might make them less valuable if ERUs are seen to be less environmentally sound because of problems with baselines and monitoring and concern about future commitments of economies in transition. Political influence might limit buying of CERs for similar reasons. Pressures to show serious efforts to reduce domestic emissions (partly motivated by the desire to reduce the need to buy in future) might reduce demand even though domestic reductions are much more expensive than even genuine reductions elsewhere. These political influences are hard to predict.

6.

CONCLUSION – FUTURE RESEARCH NEEDS

We have laid out the current international rules and key outstanding issues. Many questions still remain. Some require detailed knowledge of specific projects and industries and of the monitoring data involved, so are not ideal for economic research. Many, however, could be informed by good economic research. Here we suggest some research directions on baselines, leakage and the general benefits from trading. JI advocates and international regulators tend to focus on accounting methods to generate baselines. As economists we know that they have serious flaws. Our experience with domestic environmental regulation tells us that regulators are unlikely to ever have good information on firm-level behaviour. We believe it would be worth seriously exploring the possibility of using incentive mechanisms to encourage self-revelation of unbiased baselines by firms in the way that tradable permit markets encourage revelation of abatement costs. Some preliminary work has been done but much remains. This research needs to have strong theoretical foundations, good connections to the nature of the actors and the problems, and probably needs empirical support. Any solutions would have to be simple and practical. Useful empirical work could be done on estimating baselines, and comparing different formulations (varying by scale, updating and basic methodology) with out-of-sample experience.51 Evaluating the bias in actual baselines submitted to the Article 6 Supervisory Committee and working out how much could be ascribed to strategic behavior would also be useful. The research could be done using historical data in areas with no JI projects. The problem of measuring leakage would be ameliorated by good empirical work. Identifying the key paths of leakage in different types of project and summarizing our experience about their likely scale and on what that scale depends would be useful. Some research could draw together existing estimates of key elasticities from the literature with careful documentation of

248

Environmental and resource economics 2004/2005

where and how they apply. In some cases we do not have estimates of key elasticities, or we have them only in developed countries such as the USA. Elasticities may not translate well from developed to developing countries or to economies in transition. These could be estimated. The aim could be to develop standard methodologies for specific types of projects or to create rules of thumb about when leakage is and is not material. A final area of useful research that people do not commonly associate with this area would involve the economics of foreign investment. What can we learn from the literature or from existing experience about the barriers to investment, and especially foreign investment, in economies in transition? How much impact are they likely to have on the efficiency of the mechanisms? In what ways might JI or IET reduce these barriers? Could their design be changed to enhance this? Although there is currently very little JI action, there are plenty of investments, or non-investments, in the sectors that will be involved that could provide qualitative and quantitative data.

NOTES * 1.

2. 3.

4. 5. 6. 7.

8. 9. 10.

We would like to thank Ned Helme, Cathleen Kelly, Henk Folmer, Tom Tietenberg, Peter Bohm and ZhongXiang Zhang for suggestions and support, and Gina Straker for research assistance. We remain responsible for all opinions, and remaining errors and omissions. For example in the New Zealand fisheries’ individual transferrable quota market, recreational fishers are not included directly but do share the total allowable catch. In the US nitrous oxide program, large stationary sources are generally included in the state level trading programs but other sources can opt in. Parties assuming quantified emission reduction or limitation commitments under the Kyoto Protocol are identified in Annex B of the Protocol. These are the same as Annex I countries so we use the terms interchangeably. The different tradable units are assigned amount units (AAUs) that are allocated to all Annex B states, emission reduction units (ERUs) created in joint implementation, certified emission reductions (CERs) created through the clean development mechanism and removal units (RMUs) that are created through land use change in Annex B countries. For convenience in this chapter we refer to the group as ‘units’. We also use the term ‘credit’ to refer to units created through JI or CDM projects or through domestic projects. For detailed discussion of international emissions trading see Kerr (2000a). For discussion of the monitoring and enforcement requirements involved in Articles 5 and 7 see Hargrave et al. (2000). Decision 16/CP.7 of the Marrakesh Accords clearly establishes that JI projects can start as early as 2000, but that ERUs can be issued only for a crediting period starting after the beginning of 2008. The high transaction costs may be only in Track II JI – see below. Throughout this chapter, and in contrast to some international discussion, transaction costs are simply the cost per transaction or trade. They do not include the fixed costs of creating regulatory infrastructure either domestically or internationally. They also do not include the real cost of reducing emissions. We discuss why parties would end up using Track II JI in section 5.1.1. For discussion of the experience with this phase in and with the Acid Rain program more generally see Ellerman et al. (2000). For example, see Weyant (1999).

Joint implementation in climate change policy

249

11. Other sources are identified at: ‘http://unfccc.int/program/coop/aij/wwwbib.html, though this is a bit dated now. 12. unfccc.int/program/coop/aij/ 13. For details see www.senter.nl/asp/page.asp?id=i000000&alias=erupt 14. For details see prototypecarbonfund.org. 15. This is included because if future negotiations are conducted in terms of reductions from current Annex B commitments, which is not unlikely, and those commitments could be altered through trade, countries that sell credits would end up with lower future allocations of credits. This could deter trade. 16. This does not really have any bite. The essential thing is that the seller country can control approval of sales. Beyond that, if they do not care about sustainable development there is no mechanism to make sure that they require it. It is also not clear why the buyer country should need to approve the international trade. The buyer government has complete control over how ERUs are used by companies for compliance with any domestic regulation they impose. If companies want to buy ERUs that cannot be used for domestic compliance, why stop them? 17. The commitment period reserve is the requirement that each Annex B party must maintain in its national registry either 90 percent of its assigned amount or 100 percent of five times its most recently reviewed inventory, whichever is lower. The commitment period reserve is intended to guard against overselling by Annex B parties. 18. The Supervisory Committee has ten members: three from Annex I parties with transitional economies, three from other Annex I parties, three from non-Annex I parties, and one representing the Small Island Developing States. Decisions are made by consensus when possible, and otherwise by a three-fourths majority of present and voting members. 19. Unless someone (party in the project or three members of the Supervisory Committee (SC)) raises an objection, then the AIE judgment is automatically accepted by the SC. If not, then the SC conducts a review. 20. The UNFCCC Secretariat conducts administrative duties. It records the designated focal points for Annex I parties, maintains the lists of Annex I parties eligible to participate in JI and AIEs, manages public comments on PDDs, and reports the approval of PDDs and the verification of ERUs by AIEs. 21. In particular, JI projects involving forest management, cropland management, grazing land management, and revegetation are subject to the restrictions applied under Article 3.4, which include a country-specific cap on credited removals by sinks from forest management, and the crediting of the other activities on the basis of changes relative to the party’s base year. 22. Parties cannot acquire ERUs if they are not in compliance with the national inventory, registry, and reporting requirements under Articles 5 and 7 of the Protocol. The third requirement is unlikely to be binding as most buyer countries will be in compliance with Articles 5 and 7 and, if they were not, would be unlikely to be interested in buying ERUs as they would be unlikely to be planning to comply with their emission limits (they could not prove they were in compliance in any case). This is a strange limitation. Having anyone buy ERUs (or AAUs or CERs), regardless of their status, can only strengthen the agreement because it makes the effective cap tighter. Buying should really be encouraged not limited unless parties are concerned with costly overcompliance. 23. This approach was recommended in Kelly and Leining (2000). 24. JI Track I has the probably non-binding disadvantage of a restriction on nuclear activities. 25. The nuclear power restrictions will bind on this group. 26. Project participants must document the analysis of the environmental impacts of the project in accordance with host country procedures, and must complete an environmental impact assessment if the project participants or host country deem it necessary. These are at the discretion of the parties. 27. Once the PDD has been submitted to an AIE, it is made available for public comment for a 30-day period. If the AIE approves the project, it makes its decision and the rationale publicly available. Project approval becomes final 45 days later, unless a party involved in the project or three members of the Supervisory Committee request a review by the Supervisory Committee. The decision of the Supervisory Committee is final. Project

250

28.

29.

30.

31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42.

43.

44. 45. 46. 47. 48.

Environmental and resource economics 2004/2005 participants can then submit to the AIE a report documenting reduced emissions by sources or enhanced removals by sinks in accordance with the approved monitoring plan. The verification decision of the AIE is made public, and becomes final 15 days later unless a review is requested by a party involved in the project or three members of the Supervisory Committee. The issuance procedure formally consists of the conversion of assigned amount units (AAUs) or removal units (RMUs) to ERUs. An RMU is one tonne of carbon dioxide equivalent produced by the enhancement removals of sinks under Articles 3.3 and 3.4 of the Protocol. ‘Retirement’ refers to the units that are applied by the owner party to meet its national commitment, and therefore cannot be carried over or traded to another party. ‘Cancellation’ is the term applied to the units that need to get destroyed in order to adjust assigned amount for net emissions (if any) under Articles 3.3 and 3.4. ‘Cancellation’ also refers to the destruction of credits that are found to be invalid for whatever reason (fraud, etc.). Two restrictions are placed on carry-over: parties can carry over ERUs only to a maximum of 2.5 percent of its assigned amount; ERUs converted from RMUs cannot be carried over. These restrictions are not likely to bind, because ERUs, RMUs and AAUs are fungible for compliance purposes. Parties could retire ERUs or RMUs and bank AAUs instead. The parties may decide to subsidize the mechanism to faciliate participation by economies in transition. Not using Track II JI at all is one extreme solution that avoids all environmental risk. Rentz (1998) argues for the other extreme. This problem is discussed in detail by Wirl et al. (1998). For more discussion of these issues in the context of deforestation baselines see Kerr et al. (2002). For example the Anthropogenic Emissions and Policy Analysis (EPPA) Model, web.mit.edu/globalchange/www/eppa.html In contrast, in the CDM national baselines could allow for national ‘projects’ or be an input into negotiations leading to countries taking on emission limitations. This mechanism was developed for the UK emissions trading tender for allowances but was not used. This approach has been proposed in CDM through the use of ‘growth baselines’ that adjust to GDP growth. The effect of baseline updating on uncertainty is often misconceived because people forget that the shocks on baselines and actual emissions are correlated. Several papers state that updating baselines increases uncertainty for project organizers. There would be technology leakage from the regulated sectors, price leakage and bias because the companies and countries excluded from regulation would not be a random sample. Geres and Michaelowa (2002) discuss qualitative effects of leakage. However projects are designed, leakage of effects to non-Annex I countries through relocation of economic activity and changes in the terms of trade is always a problem. This leakage arises with or without trading but is probably reduced by trading because the costs of emission reduction are lower. ‘Hot air’ is AAUs that were allocated in Annex B countries but that are not needed for domestic compliance even when the country makes no effort to control emissions. The hot air arises primarily because of the unanticipated severity of the economic collapse in Eastern Europe and the Former Soviet Union. If they simply wanted to limit purchases, they could run a first-come-first-served system. Discrimination of this type could raise WTO issues, but it is not clear. Some of these issues are discussed in van der Gaast (2003). For discussion of market power in international trading markets see Kerr (2000b). Market power has been exacerbated by the withdrawal of the USA, a major buyer, from the Protocol. For discussion of the integration of domestic regulatory systems with international emissions trading see Kerr (2000c).

Joint implementation in climate change policy 49.

50. 51.

251

If a country operates under Track II during the commitment period but by the end of the period has a monitoring system that demonstrates that it is out of compliance with its Article 3 commitments, the ERUs it created are still valuable but the government is penalized for non-compliance. There is a cap on each Annex I country’s use of CDM LULUCF credits equal to 1 percent of base year emissions times five. See for example, Kerr et al. (2003) for discussion of deforestation baselines in the CDM.

REFERENCES Ausubel, Lawrence, Jeremy Bulow, Peter Cramton, Paul Klemperer and Eric Maskin (2003), ‘Auctioning greenhouse gas emission reductions: the UK experience’, draft paper, presented at AEA, Washington. Bailey, P. and T. Jackson (1999), ‘Joint implementation for controlling sulphur in Europe and possible lessons for carbon dioxide’, in S. Sorrell and J. Skea (eds), Pollution for Sale: Emissions Trading and Joint Implementation, Cheltenham, UK and Northampton, MA: Edward Elgar: pp. 255–71. Bohm, P. (1994), ‘Making carbon emission quota agreements more efficient: joint implementation versus quota tradability’, in G. Klaassen and F.R. Førsund (eds), Economic Instruments for Air Pollution Control, Boston: Kluwer Academic Publishers, pp. 187–208. Bohm, P. and B. Carlén (1999), ‘Emission quota trade among the few: laboratory evidence of joint implementation among committed countries’, Resource and Energy Economics, 21 (1), 43–66. Ellerman, A. Denny, Paul L. Joskow, Richard Schmalensee, Juan-Pablo Montero and Elizabeth M. Bailey (2000), Markets for Clean Air: The U.S. Acid Rain Program, New York: Cambridge University Press. Fischer, Carolyn (2001), ‘Rebating Environmental Policy Revenues: Output-Based Allocations and Tradable Performance Standards’, Resources for the Future discussion paper 01-22. Fischer, Carolyn (2002), ‘Determining project-based emissions baselines with incomplete information’, Resources for the Future Discussion Paper 02-23. Geres, R. and A. Michaelowa (2002), ‘A qualitative method to consider leakage effects from CDM and JI projects’, Energy Policy, 30 (6), 461–3. Hagem, C. (1996), ‘Joint implementation under asymmetric information and strategic behavior’, Environmental and Resource Economics 8, 431–47. Hanafi, A.G. (1998), ‘Joint implementation: legal and institutional issues for an effective international program to combat climate change’, The Harvard Environmental Law Review, 22 (2), 441–508. Hargrave, Tim, Suzi Kerr, Ned Helme and Tim Denne (2000), ‘Treaty compliance as background for an effective trading program’, in Suzi Kerr (ed.), Global Emissions Trading: Key Issues for Industrialized Countries, Cheltenham, UK and Northampton, MA: Edward Elgar. Janssen, J. (1999) ‘(Self-) enforcement of joint implementation and clean Development mechanism contracts’, Fondazione Eni Enrico Mattei Note di Lavoro, 14/99. Jepma, Catrinus J. (ed.) (1995), The Feasibility of Joint Implementation, Dordrecht: Kluwer. Kelly, Cathleen and Catherine Leining (2000), ‘Developing the rules and guidelines

252

Environmental and resource economics 2004/2005

for joint implementation’, in Suzi Kerr (ed.), Global Emissions Trading: Key Issues for Industrialized Countries, Cheltenham, UK and Northampton, MA: Edward Elgar. Kerr, Suzi (1995), ‘Adverse selection and participation in international environmental agreements,’ in Suzi Kerr, ‘Contracts and tradeable permit markets in international and domestic environmental Protection’, PhD thesis, Harvard University, available at www.motu.org.nz/climate.htm Kerr, Suzi C. (ed.) (2000a), Global Emissions Trading: Key Issues for Industrialized Countries, Cheltenham, UK and Northampton, MA: Edward Elgar. Kerr, Suzi (2000b), ‘Market power and Annex I trading’, in Suzi Kerr (ed.), Global Emissions Trading: Key Issues for Industrialized Countries, Cheltenham, UK and Northampton, MA: Edward Elgar. Kerr, Suzi (2000c), ‘Domestic greenhouse regulation and international emissions trading’, in Suzi Kerr (ed.), Global Emissions Trading: Key Issues for Industrialized Countries, Cheltenham, UK and Northampton, MA: Edward Elgar. Kerr, Suzi, Joanna Hendy and Alexander S.P. Pfaff (2003), ‘Uncertainty and the contribution of tropical land-use to carbon mitigation’, Motu manuscript. Kerr, Suzi, Shuguang Liu, Alex Pfaff and R. Flint Hughes (2002), ‘Carbon dynamics and land-use choices: building a regional-scale multidisciplinary model’, Motu manuscript. Lawson, Karen, Stan Kolar and Cathleen Kelly (2000), ‘ A regional approach to developing multi-project baselines for the power sector’, Center for Clean Air Policy manuscript, available at www.ccap.org/pdf/Karpow.pdf Lawson, K. and N. Helme (2000), Implementing the Additionality Requirement & Ensuring the Stringency of Project Baselines Under the CDM, Washington, DC: Center for Clean Air Policy. Mullins, Fiona and Richard Baron (1997), ‘International greenhouse gas emission trading’, Annex I Expert Group on the UNFCCC working paper No. 9, Organisation for Economic Co-operation and Development, Paris OCDE/GD (97) 76. Nentjes, A. (1994), ‘Control of reciprocal transboundary pollution and joint implementation’, in G. Klaassen and F. R. Førsund (eds), Economic Instruments for Air Pollution Control, Boston: Kluwer Academic Publishers, pp. 209–30. OECD/IEA (2000), Emission Baselines: Estimating the Unknown, Paris: Organisation for Economic Co-operation and Development. OECD/IEA (2002), Developing Guidance on Monitoring and Project Boundaries for Greenhouse Gas Projects, Information Paper, Organisation for Economic Co-operation and Development. OECD/IEA/IETA (2002), National Systems for Flexible Mechanisms: Implementation Issues in Countries with Economies in Transition, workshop report, Szentendre. Parkinson, S., K. Begg, Peter Bailey and Tim Jackson (1999), ‘JI/CDM crediting under the Kyoto Protocol: does “interim period banking” help or hinder GHG emissions reduction?’, Energy Policy, 27 (3), 129–36. Rentz, H. (1998), ‘Joint implementation and the question of “additionality” – a proposal for a pragmatic approach to identify possible joint implementation projects’, Energy Policy, 26 (4), 275–79. Ridley, M.A. (1998). Lowering the Cost of Emission Reduction: Joint Implementation in the Framework Convention on Climate Change, Boston: Kluwer Academic Publishers. Tietenberg, Tom, Michael Grubb, Byron Swift, Axel Michaelowa and ZhongXiang Zhang (1999), International Rules for Greenhouse Gas Emissions Trading:

Joint implementation in climate change policy

253

Defining the Principles, Modalities, Rules and Guidelines for Verification, Reporting and Accountability, UNCTAD/GDS/GFSB/Misc.6, United Nations Conference on Trade and Development, Geneva. United Nations Framework Convention on Climate Change draft decisions 15/CP.7 and 16/CP.7 available at: unfccc.int/resource/docs/cop7/13a02.pdf Van der Gaast, Wytze (2003), ‘The scope for joint implementation in the EU candidate countries’, International Environmental Agreements, 2 (3), 275–90. Weyant, J.P. (1999), The Costs of the Kyoto Protocol: A multi-model evaluation, The Energy Journal, Kyoto Special Edition. Wirl, F., C. Huber and I.O. Walker (1998), ‘Joint implementation: strategic reactions and possible remedies’, Environmental and Resource Economics, 12 (2), 203–24. Woerdman, E. (2000), ‘Implementing the Kyoto Protocol: why JI and CDM show more promise than international emissions trading’, Energy Policy, 28 (1), 29–38. Zhang, Z. and A. Nentjes (1999), ‘International tradable carbon permits as a strong form of joint implementation’, in S. Sorell and J. Skea (eds), Pollution for Sale: Emissions Trading and Joint Implementation, Cheltenham, UK and Northampton, MA: Edward Elgar, pp. 322–42.

7. Environmentally harmful subsidies Jean-Philippe Barde and Outi Honkatukia1 INTRODUCTION Every year, OECD countries give about US$400 billion in subsidies to different economic sectors. The objectives of governments’ subsidies to various economic activities are often presented as to promote economic growth, employment and incomes. Subsidies distort prices, affect resource allocation decisions and change the amount of goods or services produced and consumed in an economy. Policies providing subsidies are generally introduced for various social or economic reasons, but they can have unintended negative effects on the environment that are generally ignored. Subsidies can thus result in a policy failure (see below) and be harmful to the environment. In agriculture, for example, they can lead to the overuse of pesticides and fertilizers and in fisheries to the overexploitation of the fish stock. Fuel tax rebates, subsidies for road transport, and low energy prices generally stimulate the consumption of fossil fuels and greenhouse gas emissions and increase congestion and air pollution. Over the past 20 years, the OECD has made significant progress in the measurement and analysis of subsidies for sectors such as agriculture, coal production and fisheries. Factors contributing to the relatively modest progress in measuring support for other sectors range from complex methodological and data issues to a lack of political will to provide reliable and internationally comparable subsidy figures in some areas like manufacturing. Tradeoffs are made both at national and international levels as data collection is often resource intensive and aggregate subsidy estimates are only as good as the underlying data. Although methodological and data constraints severely limit comparisons across sectors, work carried out by the OECD highlights agriculture as the sector with the largest subsidies (see below). While the other sectors seem pale in comparison, it is likely that subsidies are underestimated in these sectors due to the methodologies applied (Steenblik, 1995, 2003). It should be noted that not all support measures implemented in environmentally sensitive sectors are potentially harmful. A number of support measures are specifically designed for environmental protection, such as 254

Environmentally harmful subsidies

255

support for pollution control investments (grants, soft loans, accelerated depreciation) and the development of ‘clean technologies’ or energy efficiency. Support to production inputs can be targeted on non-polluting inputs or recycled materials; pricing of transport infrastructures may be designed to facilitate public transports and so on. Whilst such support measures may be environmentally beneficial in the short/medium term, they may introduce rigidities (for example vested interest) and distort markets in the longer term. Furthermore, they can contradict the polluter-pays principle, as defined by the OECD (1972). These environmentally motivated subsidies are not analysed in this chapter. This chapter aims to identify environmentally harmful subsidies. After a brief reminder of the consequences of market and government intervention failures, it reviews the definition and measurement of different types of support measures. It then gives a brief overview of the magnitude of subsidies in OECD countries and analyses the environmental effects of these subsidies, taking into account the interlinkages between the economy and the environment. It concludes with some considerations of the political, social and technical challenges associated with the phasing out of these subsidies. It does not claim to fully cover, nor to give a final answer to, this very complex issue, but aims at providing a brief survey of the state of knowledge and main issues to be resolved. The issue of environmentally harmful subsidies will remain for a long time on the environmental, economic and political agendas.

1.

WHY DO SUBSIDIES MATTER FOR THE ENVIRONMENT?

It is now widely recognized that environmental policies should rely, to the greatest extent possible, on properly functioning markets (that is, with internalized externalities and no distorting subsidies and tax provisions). The need for an effective internalization of externalities, which economists have been promoting for several decades, is now increasingly recognized by policymakers. Although the ‘polluter pays principle’ was promulgated 30 years ago by the OECD (1972), the implementation of an economic discipline and rationale in environmental policy is fairly recent and far from fully achieved. In particular, although the use of economic instruments in environmental policy, such as charges, taxes, tradable permits and deposit refund systems, has evolved considerably over the last 15 years in OECD countries (OECD 1999, 2001a), governments continue to perpetrate innumerable ‘intervention or policy failures’2 generally in the form of inefficient regulations, subsidies and tax exemptions. Take for instance the case of taxes; most OECD countries have undertaken significant tax reforms during the 1990s, chiefly in two ways: first by

256

Environmental and resource economics 2004/2005

reducing tax rates in the higher income tax brackets (which fell on average by ten points between 1986 and 1997) and lowering corporate tax rates (down by ten points over the same period); second, by broadening the tax base, especially for indirect taxes (VAT and consumption taxes). In this context, the ‘greening’ of tax systems provides a good opportunity to correct market failures. However, energy-related taxes are flawed by a host of exemptions, with the effect that 80 per cent to 90 per cent of the burden of these taxes falls on households, thus virtually exempting the industry sector. Introducing new CO2 taxes, for instance, makes little economic sense if exemptions drastically erode the expected environmental benefit of these taxes. The OECD database on environmentally related taxes3 documents 1500 different exemptions and tax breaks. Governments have historically manipulated market prices through regulations, taxation levels, government ownership, subsidized loans, purchase commitments, direct and indirect budgetary transfers, trade barriers, set prices, and the like. Support measures are put in place to enhance the competitiveness of certain products, processes, industries, or to develop the employment and income of social groups or regions. In general, the full economic, financial, environmental and social costs of support measures are not considered, and on balance these costs may often outweigh the benefits of implementing the support measure, thus leading to significant government intervention failures. Recent experiences in OECD countries indicate that the reform or removal of many of these subsidies may not only increase economic efficiency and reduce the burden on government budgets and consumers, but can also alleviate environmental pressures (see below, section 4). Many policies providing subsidies in OECD countries are implemented in view of supporting environmentally sensitive sectors, particularly agriculture, the fishery industry, energy production, transport and heavy industries. Most of the support measures take the form of cost-reducing support (for example support to infrastructure, research and development, material and energy inputs, and so on) or the form of revenue-enhancing support (for example market price support for particular products). These support measures often lead to increased use of (possibly polluting) inputs and increased production levels as prices for the finished good fall in response to declining costs (OECD, 1998a). Encouraging production augmentation through such support measures increases the risk of environmental damage from production. In the end, market and policy failures may result in a cumulative detrimental effect on the environment. In Figure 7.1, the marginal private cost curve MPC intersects the demand curve D so that a quantity Q0 is produced at price P0. Accounting for the external cost associated with the production increases the cost so that the marginal social cost curve MSC (including

Environmentally harmful subsidies

257

Prices MSC MPC MPC – Subs

p2

B

p* p0 p1

A C

D Q2

Q* Q0

Market failure Figure 7.1

Q1

Q3

Quantities

Government failure

Market and government intervention failures

private plus external cost) intersects the demand curve at point B with a quantity Q* < Q0 produced at price P* > P0. The difference in output Q–Q* represents the market failure. If a subsidy is paid to the producer, the marginal private cost curve is shifted down to MPC – Subs. Corresponding to a quantity Q1 > Q0 and a price P1 < P0. The excess production represents the policy failure. In this particular case (other configurations can be conceived), the market and policy failure are added up. If the government sets a price at P2 to support the producer’s income, the government may need to take specific measures to guarantee the purchase of the quantity Q3–Q2 which would otherwise not be purchased at price P2. Long-term effects of subsidies are generally different from short-term ones. There is no technological change in the short term or substitution between inputs or factors of production. Thus removing subsidies will reduce profits in the short term and cause marginal firms to exit from the market. Removing a long-standing subsidy will open the way to the development and application of new technologies hitherto blocked by the subsidy. The technology lock-in effect will disappear in the long term, thus enabling substitution between factors of production and increases in efficiency.

258

Environmental and resource economics 2004/2005

2.

DEFINING AND MEASURING SUBSIDIES

2.1

Defining Subsidies

The concept of subsidy is not straightforward. While the term ‘subsidy’ is used in this chapter, it is as common to use the terms transfers, payments, support, assistance or protection associated with governmental policies in OECD work. Sometimes these terms are used interchangeably, but often they are associated with different methods of measurement and thus different economic indicators. Subsidies have been defined to ‘comprise all measures that keep prices for consumers below market level or keep prices for producers above market level or that reduce costs for consumers and producers by giving direct or indirect support’ (see, for example, de Moor and Calamai, 1997 or De Moor, 1997). This definition is consistent with the OECD approach of defining environmentally harmful subsidies and tax concessions to include ‘all kinds of financial support and regulations that are put in place to enhance the competitiveness of certain products, processes or regions, and that, together with the prevailing taxation jurisdiction, (unintentionally) discriminate against sound environmental practices’ (OECD, 1998a). It is not necessary to make a distinction between subsidies and tax expenditures as the latter can be regarded as implicit subsidies. Subsidies take different forms: budgetary payments or support involving tax expenditures (various tax provisions that reduce the tax burden of particular groups, producers or products), market price support, subsidized input prices, preferential interest rates. This is why the more generic terminology of ‘support measures’ is often used. There is, however, no international consensus: different definitions prevail for specific purposes, fields (for example agriculture or transport) or contexts (for example international trade). There has been much controversy over whether the non-internalization of external costs should be construed as a subsidy, the argument being that, as external costs are not internalized, the environment is used ‘freely’ by the users: in a sense, a public good is freely supplied to users. Those who object to such an expanded definition observe that the notion of a subsidy has traditionally connoted an explicit government intervention, not an implicit lack of intervention. In addition, for these and more practical purposes, namely the difficulty of quantifying external costs, non-internalization is not regarded as a subsidy in this chapter except for the transport sector, where this definition is currently used (Nash et al., 2002). 2.2

Measuring Subsidies

Five main methods can be used or combined to measure subsidies:

Environmentally harmful subsidies

1.

2. 3.

4.

5.

259

Programme aggregation: subsidies are measured by aggregating the value transferred to beneficiaries from government budgets (as often in the case of fisheries, agriculture and processing industries). Price gap: subsidies are measured by the differential between border and domestic market prices (as in the case of agriculture). Producer/consumer support estimate (PSE): this method combines both government expenditures to producers and market price support. The consumer support estimate (CSE) is the analogous indicator for transfers provided to (or from) consumers through government programmes and prices interventions. The OECD uses the PSE and CSE framework to measure support to agriculture and coal. The resource rent captured by users of publicly owned natural resources such as minerals, forests and water. These rents estimate the difference between the full economic rent and the price paid for exploiting a natural resource. Difference between prices and marginal social cost, in particular in the area of transport.

Although the focus of this chapter is not on the different methodologies used to estimate subsidies, it is important to highlight the extent to which methodological differences and data gaps limit the comparability of subsidy figures across sectors (or, as the case may be, within a sector). The strengths and weaknesses of the main approaches used in domestic and international subsidy assessments are summarized in Table 7.1. The approaches used to estimate subsidies differ in the amount of data required to calculate them and in the degree to which budgetary payments and market transfers are measured accurately. Programme-specific approaches capture the value of government programmes benefiting (or taxing) a particular sector, whether these benefits end up with consumers (as lower prices), producers (through higher revenues), or resource owners (through higher rents). Unless integrated into a macroeconomic model, this information says little about the ultimate incidence of the subsidy programmes and their effect on market prices. By definition, the price-gap approach highlights observed price distortions, though it misses the often substantial budgetary support that does not affect consumer prices but does affect the structure of supply. The producer and consumer support estimates provide insights into both. 1.3

Measuring Subsidies in Environmentally Relevant Sectors

These different measurements have been applied differently to the main environmentally relevant sectors. In the field of agriculture, the most commonly used definitions and

260

Environmental and resource economics 2004/2005

Table 7.1 Overview of subsidy measurement approaches Approach/description Programme aggregation Quantifies financial transfers associated with various government programmes. Aggregates programmes into overall level of support Price gap Evaluates positive or negative ‘gaps’ between the domestic and border prices. Also known as market price support

Resource rent gap Estimates the difference between the full economic rent and the price paid for exploiting a natural resource

Strengths

Limitations

Captures transfers whether or not they affect prices. Can capture the overall cost (which is higher than the direct cost) of government lending and insurance

Does not address questions of ultimate incidence of pricing distortions. Sensitive to decisions regarding inclusion of programmes. Requires programme-level data

Can be estimated with relatively few data. Useful for multi-country studies. Good indicator of pricing and trade distortions

Sensitive to assumptions regarding ‘free market’ and transport prices. Understates full value of support by ignoring transfers that do not affect end-market prices

Relevant for natural resource sectors such as forest and water

Data intensive. Sensitive to assumptions

Marginal social cost approach Estimates the difference Most comprehensive between the marginal social approach. Used for transport cost (that internalizes all externalities) and the price paid Producer/consumer support estimate Systematic method to aggregate budgetary transfers and consumer transfers (through market price support calculation) to specific industries

Integrates budgetary transfers with market price support into holistic measurement of support. Distinguishes between support to producers and consumers

Data intensive. Requires a significant amount of modelling. Sensitive to assumptions and has a wide range of uncertainty

Data intensive. Currently calculated for agriculture and coal production, but not for other sectors

Source: Based on Koplow and Dernbach (2001).

measures of subsidies are the producer support estimate (PSE), the consumer support estimate (CSE), the estimated total support estimate (TSE) calculated by the OECD and the aggregate measure of support (AMS) used in the Uruguay Round and WTO agricultural negotiations. OECD estimates cover market price support, direct payments (including those to reduce the cost of fixed capital and/or variable inputs), general services (transfers covering the

Environmentally harmful subsidies

261

costs of research, marketing an structural/infrastructure services) and consumption support (such as ‘food stamps’ in the United States). Subsidies to fisheries involve a wide range of transfers: examples include grants or loans for the construction of new fishing vessels or the refurbishment of old ones, exemptions from excise taxes on fuel, subsidized unemployment benefits for idled fishermen, non-requited payments for access to fish in another country’s exclusive economic zone; government-supported exploration for new fishing grounds; and below-cost provision of fishery-specific infrastructure. Energy subsidies are provided to producers typically through: grants or soft loans for the construction of mines, wells or generating plant; grants for the construction or operation of demonstration plants; government-brokered contracts with large consumers; reduced VAT; and government financed R and D. They are most commonly provided to consumers through differential rates of excise tax or (mainly in developing countries) administered prices for fuels or electricity. Transport subsidies can be defined in two different ways (Nash et al., 2002). One way is to compare the total social cost and total revenue for each transport mode in order to assess how far users pay the total cost. The second approach is to compare the marginal social cost and the price paid for the transport mode; the failure of prices to cover marginal social cost is regarded as a subsidy.4 The main forms of subsidies to industry are: grants and interest rate subsidies, tax exemptions, soft loans, equity investments, tax deferrals and loan guarantees. There are many forms of water subsidies. Water abstraction is subsidized when water is charged below cost recovery; water supply is also subsidized in the same way (pricing below cost recovery) or through direct financial assistance (for example for water infrastructure for agricultural or industrial water supply); irrigation subsidies are defined either as government expenditure covering all or some of the costs of installing and/or maintaining irrigation systems, or by comparing the price of water with the marginal cost of water supply. Ideally, the price should be compared with the marginal social cost of supply, including the scarcity rent.

3.

OVERVIEW OF SUBSIDIES IN OECD COUNTRIES

Many OECD countries have committed to the reform of subsidies in the particular sectors of the economy, but they have made only limited progress over the past ten years. Although the methodologies and coverage differ and consequently the subsidies data are not comparable across sectors, Table 7.2

262

Environmental and resource economics 2004/2005

and Figure 7.2 give an indication of the importance of support in different sectors. Agriculture is the sector with the highest subsidy figures, but is also the sector with the most complete data in terms of coverage comprehensiveness and methodology. Subsidies measured for the other sectors, such as transport and energy, amount to only a fraction of the figure for agriculture.

4.

WHAT MAKES A SUBSIDY ENVIRONMENTALLY HARMFUL?

4.1

A Conceptual Framework

Source:

Based on Table 7.2.

Figure 7.2

Subsidies in OECD countries (most recent years)

Agriculture

Transport

Energy

Manufacturing

Water

Fisheries

350 300 250 200 150 100 50 0

Forestry

US$ billion

Most production and consumption activities have an impact on the environment, which is accentuated or attenuated by government policies. A subsidy is deemed harmful to the environment if it ‘encourages more environmental damage to take place than what would occur without the subsidy’. In other words, the subsidy leads to higher levels of waste and emissions, including those in the previous stages of production and consumption, than what they would be without the support measure (OECD, 1998a). For instance, support to specific agricultural inputs (fertilizers, pesticides, energy) or to specific agricultural prices encourages excess production and environmental pressure; pricing of transport infrastructures below marginal social cost induces increased traffic; non-taxation of aircraft kerosene accelerates growth of air transport and related environmental damage; subsidized (or under-taxed)

Table 7.2

Subsidies in OECD countries US$ billion

Agriculture

Comments

263

1990

Most recent data (year)

351

318 (2002)

Total support estimate; includes market price support, budgetary payments and support for general services; covers all OECD countries

Equivalent to 1.2% of GDP

40 (1981)

Subsidies estimated as the difference between total revenues and total social costs; includes the European Union, Hungary and Switzerland

Nash et al. (2002) estimated that revenues cover on average 36% of rail system costs

n.a.

20–30 (1999)

Agregate estimate

Probably largely underestimated

11

5 (2000)

Includes market price support, budgetary payments and support for general services; includes France, Germany, Japan, Spain, Turkey and UK

Equivalent to US$68 per tonne of coal produced

Transport (road and rail)

Energy production of which Coal production

Coverage

Table 7.2 (continued) US$ billion 1990 Manufacturing

Coverage

Comments

Net government expenditures to industry. Figures in italics cover the EU only and include grants, interest subsidies, tax exemptions, equity participation, soft loans, tax deferrals and loan guarantees, converted into cash grant equivalents Figures in italics cover the EU only and include grants, interest subsidies, tax exemptions, equity participation, soft loans, tax deferrals and loan guarantees, converted into cash grant equivalents

Figures in italics from the EU State Aid Survey

Most recent data (year) 22 (EU)

of which Shipbuilding

.. 2.5 (1995)

0.75 (2000) 1 (2000)

Steel

2.2 (1995)

– (2000)

264

44 (1993) 49 (1992)

Includes grants, interest subsidies, tax exemptions,

Figures in italics from the EU State Aid Survey

Figures from EU State Aid Survey

equity participation, soft loans, tax deferrals and loan guarantees, converted into cash grant equivalents; EU only n.a. (9 in 1996)

6 (1999)

Government financial transfers to the marine capture fisheries; includes direct payments, cost-reducing transfers and general services. The 1999 figure excludes Australia, Belgium, Mexico, the Netherlands, Poland and Turkey

Water

–

10

Aggregate estimate

Forestry

–

6

Aggregate estimate; includes only Canada and the United States

265

Fisheries

Notes: Sources:

Equivalent to 20% of landed value

Data and calculation methods are not comparable across sectors. OECD (2001b, 2003a); European Commission (2002); IEA (2001); ECMT (2000); Nash et al. (2002); Myers and Kent (2001).

266

Environmental and resource economics 2004/2005

energy increases CO2 and other polluting emissions; underpricing of water resources results in water shortage and salinization of soil. However, these ‘volume effects’ are environmentally harmful as long as the associated externalities are not internalized; hence the influence of the ‘environmental policy filter’ (see below). Removing environmentally harmful subsidies would result in a better environment; however, this depends on the factors determining the magnitude of environmental effects of support measures (OECD, 1998a), that is: 1.

2.

3.

The level of protection from competition that support measures offer the recipient sector, in other words, the extent to which alternatives to the recipient sector are discouraged. The environmental effects of the alternative products or technologies that are discouraged by the support measure, compared with those of the supported sector. The circumstances that determine how sensitive the environment is to the particular change in emission or waste levels brought about by the support measure.

These factors highlight the division between what governments can change, such as support policies and, to some extent, the emergence and use of cleaner technologies; and what they cannot influence, such as the dose-response relationship between particular emissions and environmental quality (OECD, 1998a). The environmental impact of support measures results from complex mechanisms that are far from being fully elucidated. There is no direct linkage between the volume and nature of the subsidy and the environmental impact. In Figure 7.3, linkage 1 is the link between the support measure and the volume and composition of output; this linkage includes complex interactions: for instance, the support measure will benefit one sector or one production technique relative to others or a transfer of resources from taxpayers to producers or between consumers and producers. Furthermore, other ‘autonomous changes’ such as technical change and economic development will interfere with these linkages. Market price support (for example floor price with specific measures to guarantee purchases at this price) affects the revenue of the recipient sector (typically the case of agriculture). This may result in increased production volumes because the greater the volume of production, the higher the support received by producers. A ‘leakage effect’ is also likely, as the increased volume of production may induce extra use of inputs (for example fertilizers), thus transferring (part of) the increased revenue to the input suppliers, but also having an environmental impact by changing the production process (more intensive farming).

Environmentally harmful subsidies Linkage 1

Support

Linkage 2

Marginal cost and/or revenue in the producing sector

Demand and supply conditions

Volume and intensity of activity

Exogenous factors

267 Linkage 3

Emissions and resource use

Impact of environmental policy

Environmental damage and resource depletion

‘Absorption’ by the assimilative capacity of the environment

Environmental expenditures

Rebound effects on the economy

Note: As with all analyses, results will be dependent on the chosen assumptions, methodologies and available data such that quantitative results will always be subject to some degree of uncertainty. Source: Based on OECD (1998a).

Figure 7.3

Linkages between support measures and environmental effects

Support measures targeted on the use of specific products or inputs (for example energy tax exemptions of heavy industries), or on the use of particular production processes (for example low-interest loans for the construction of intensive livestocks), reduce production costs. Depending on market structure and international competition, products could be sold at lower prices, thus increasing consumption and possibly environmental impacts. Linkage 2 relates to the level of emissions resulting from the changes in the volume and composition of output. This is a function of production and emission abatement techniques of the polluting sector and the type and effectiveness of the environmental policy in place (the environmental policy ‘filter’). Environmental expenditure will then have a rebound effect on the economy (multiplier effect). Finally, the environmental impact of increased emissions will be a function of dose-response relationships and the assimilative capacity of the environment (Linkage 3). Those environmental impacts will also produce rebound effects on the economy, for example health effects, depletion or deterioration of natural resources, higher production costs and so on.

268

Environmental and resource economics 2004/2005

This analysis shows the complexity of the linkages between support measures and environmental impacts. A thorough assessment would require a complex set of general equilibrium analysis (to evaluate the rebound effect on the economy) and environmental impact evaluation techniques. It is important to analyse subsidies separately from their objectives. Subsidies are generally introduced for a ‘good’, or at least politically rational, purpose. But the devil, that is, the difference between a ‘good’ and a ‘bad’ subsidy, is in the details on how they are implemented (Pieters, 2003). It should be assessed whether subsidies serve the intended purpose, at what cost, how the costs and benefits are distributed and whether they are harmful for the environment. Many existing subsidy policies do not even serve effectively their intended purpose. For example, most subsidies for energy or water that are ostensibly meant to protect poorer members of society end up benefiting the rich or create major inefficiencies in the economy (see section 5). Subsidies to road transport often encourage motor vehicle overuse and increase pollution and congestion. Agricultural subsidies are an expensive and inefficient way to maintain farm incomes, especially for small farms. Thus subsidies are often inefficient and expensive policies that are environmentally harmful and impose a burden on government budgets and taxpayers. All these are strong arguments for reforming the existing subsidy policies. Decoupling subsidies from input use, production and consumption would bring economic, environmental and social benefits. Table 7.3 highlights the expected environmental effects of removing different types of subsidies. Not all subsidies are bad for the environment. Road transport and pollution would increase if the other modes of transport were not subsidized (public transports). Some subsidies are used to support the generation of environmental benefits. OECD countries are increasingly linking agricultural support payments to farmers’ action to improve the environmental performance of agriculture. Some countries pay farmers who limit the use of environmentally damaging inputs, such as certain fertilizers and pesticides, or those who use organic farming techniques. Others support farmers in planting trees to reduce agricultural runoff and provide habitat for wildlife, in removing marginal land from production, or in creating or restoring wetlands, which reduces soil erosion and creates wildlife habitat. There are also substantial programmes in OECD countries that support the development and production of renewable energy sources. However, all of these subsidies are higher than would otherwise be needed, in so far as they are used to offset the environmental damage caused by other policies that stimulate environmentally harmful production, and many are not well targeted to achieve specific environmental outcomes. Studies on the environmental impacts of subsidies use different models, assumptions and data, and consequently the estimates are not directly

Table 7.3 Categories

Environmental impacts of subsidies removal Environmental effectsa reduction Long-termb reduction in emissions or rates in emissions or rates of exploitation, due of exploitation, due to: to:

Remarks

Market price support

Lower production levels

Lower production levels

Consumer prices will drop. Lower input requirements may lead to strong environmental effects in the production of materials and energy phase. Production may shift to areas of low-cost production, leading to a possible displacement of the environmental burden

Deficiency payments, sales premiums

If marginal revenue falls below marginal cost, the least efficient production units will exit the market

With a higher product price, on the other hand, increased efficiency may lead to different modes of production that may, or may not, be more environmentally damaging

Main points of impact

Short-termb

1.

Output

269

Table 7.3

(continued)

Categories

Environmental effectsa reduction Long-termb reduction in emissions or rates in emissions or rates of exploitation, due of exploitation, due to: to:

Remarks

Materials, energy, short-lived equipment

Higher marginal costs for subsidized ‘firms’. Least efficient production units exit the market, if marginal revenues drop below marginal costs

Strong effects may be expected due to reductions in the production of materials and energy or rates of exploitation that are often relatively environmentally harmful

Particular types of fixed capital

Exit of the least efficient production units if marginal revenues drop below marginal costs

Main points of impact

Short-termb

2.

Input use

270

Lock-in effect disappears, allowing substitution and savings on inputs. If accompanied by effective environmental policies, this creates a window of opportunity for environmental improvementc Disappearance of the lock-in-effect, depending on the specificity and duration of the conditionality

If substitution of capital equipment opens the way to more efficient use of materials or energy (or the substitution of less harmful ones), strong effects upstream may be expected

Access to natural resources

Low-interest loans

Increases the price of natural resources for downstream users, increasing their resource efficiency Possibly a (limited) effect on marginal costs

Research and development 271

3.

Profit and incomed

Preferential low rates of income taxes; preferential low rates of capital taxes; debt write-off

Allowing insufficient provisions for future

Possibly somewhat lower marginal costs. If so, exit of the least efficient production units, if marginal revenues drop below marginal costs Exit of the least efficient production

Higher barrier to entry or disappearance of the least efficient production units, or both Higher barrier to entry or disappearance of the least efficient production units, or both Deployment of environmentally more benign technologies, if accompanied with effective environmental targets Higher barrier to entry. Higher prices reduce demand

Higher consumer prices and more

Strong effects on entry with possibly large beneficial effects on rates of depletion

If the subsidy is large, it may be an exploitation subsidy to capital costs in disguise. In those cases the effects are unclear

Table 7.3 (continued) Categories

Main points of impact

Environmental effectsa reduction Long-termb reduction in emissions or rates in emissions or rates of exploitation, due of exploitation, due to: to:

Remarks

Short-termb

272

4.

Demand

environmental liabilities; exemptions from environmental standards Low rate of return requirements

units, if marginal revenues drop below marginal costs

Low rates of VAT, marketing and promotion by government

Exit of the least efficient production units, if marginal revenues drop below marginal costs

environmentally benign modes of production Higher consumer prices and higher internal discount rates. The latter shortens the planning horizon of the ‘firm’ and thereby the lockin effect Undetermined, since dependent on externalities

Some ‘up stream’ effects may be expected

Provision of infrastructure below costs

The same as above

The same as above More decentralized production close to the place of consumption; different technologies

The environmental effects depend also on site-specific environmental conditions

273

Notes: aAs stated before, elements of the policy filter (quota, limitations in infrastructure) may become, or remain, the limiting factors to production and thereby to the environmental effects of subsidy removal. In this table this is ignored. bIn the short run, technology remains the same. That is, there is no substitution between factors of production or inputs. cChoosing a particular input often casts the technology in stone and vice versa. dRemoval of subsidies based on historical entitlements, or direct payments to producers in exchange for production (modes) that are environmentally beneficial have been omitted from the table, because such removal is likely to damage the environment. Source: Adapted from Pieters (2003).

274

Environmental and resource economics 2004/2005

comparable (Steenblik, 2003). However, they do give a good indication of the range of findings available from different studies on removing support in different countries, with different assumptions and timescales. Most studies show that removing support will have a positive effect on the environment, although sometimes the predicted effects are quite small. The environmental impacts of subsidies can be estimated with a partial or general equilibrium model, and the results are typically sensitive both to the model chosen and to the magnitude of the subsidies data used as model inputs. Subsidies may have different initial points of impact, such as output, input, profits and income. Initial points of impact matter for two reasons. Subsidies to inputs affect other markets than subsidies to outputs or profits and income. Generally speaking, subsidies that directly affect material flows have more direct effects on forward linkages than subsidies to output or profits and income. Such subsidies also leave fewer options for more benign modes of production than subsidies to output or income (OECD, 1998a). Also, if input subsidies are conditional on the use of particular energy carriers or materials (including water), or particular types of capital equipment that require only certain types of energy carriers or materials, they will discourage materials and energy saving, on which the success of environmental policy is highly dependent. Environmental management regimes and other elements of the ‘policy filter’ affect the environmental impacts of subsidies. If, for example, subsidies to fisheries are removed while catches are limited by other measures, or when certain types of subsidies to road or energy are removed, while infrastructure is a limiting factor, the environmental effects of removal may not be significant (Hannesson, 2002; ECMT, 2000a). 4.2

Trends in Environmentally Harmful Subsidies

Environmentally harmful subsidies are prominent in fossil fuels, road transportation, agriculture, water forestry and fisheries. Agriculture The impacts of agricultural support measures on the environment depend on their effects on farm-level decision-making concerning the intensive (input use) and extensive (land use) degree of agricultural production. These impacts result from the relationships linking land quality, production practices, input use and environmental quality defined in terms of, for example, erosion, chemical runoff, leaching, landscape and biodiversity or wildlife habitats. In general, the more a policy measure provides an incentive to increase production of specific agricultural commodities, the greater is the incentive for monoculture, intensification (greater yields), or using marginal (environmen-

Environmentally harmful subsidies

275

tally sensitive) land – and the higher is the pressure on the environment. However, some restrictions or constraints on providing support (for example environmental cross-compliance5 and regulations) may attenuate the environmental impacts of support measures. Moreover, the more a policy measure can be targeted to a specific environmental goal and situation, the greater is its potential effectiveness in achieving such a goal. The analytical work on the producer support estimate, the policy evaluation matrix (PEM) and on the impact of support on environment allow us to rank support measures according to their relative impacts on the environment (OECD, 2002b and Portugal, 2002). Details on the ranking are shown in Box 7.1. In 2002, estimated total support to agriculture amounted to US$318 billion (OECD, 2003), which represents 1.2 per cent of GDP in OECD countries. During the 1990s many OECD countries began to take steps to reduce and restructure their support policies in an effort to reduce overproduction and trade distortions, and to encourage more environmentally sound use of land, soil and water. However, the pace of these developments has been modest and subsidies remain high in many OECD countries and for some commodities, causing adverse effects on the environment. In 2002, support to farmers represented 31 per cent of the value of farm receipts, compared with 38 per cent in the 1986–88 period (OECD, 2003a). The share of market price support, output payments and input subsidies (such as water, fertilizers and energy subsidies), which are potentially the most environmentally harmful types of support, has decreased marginally since the mid-1980s, but they still account for nearly 80 per cent of total support (the bottom three categories in Figure 7.4). This share varies across countries, and is highest in the countries with the highest levels of support (Figure 7.5). While the share remains persistently high in Korea and Japan, it has decreased in Iceland, Norway and Switzerland due to a shift to less distorting support (OECD, 2003a). Fisheries According to Hanneson (2002), it is difficult to find more perverse policies, in terms of efficient resource utilization, than fisheries subsidies. The basic problem in fisheries management is that too much capital and labour are used in the industry, and subsidization only aggravates the problem. A distinctive feature of fisheries sector subsidies is the effect that overcapacity and overfishing by subsidized producers can have in limiting other producers’ access to the shared resource. The environmental impact of fisheries subsidies depends on the management regime in place or the ‘policy filter’. Most fisheries in OECD countries limit total catches, but very few apply truly effective management, which typically includes entering restrictions for particular fisheries, such as having to

276

Environmental and resource economics 2004/2005

BOX 7.1 RELATIVE POTENTIAL IMPACTS OF AGRICULTURAL PRODUCER SUPPORT MEASURES ON THE ENVIRONMENT All other things being equal, the main categories of PSE measures can be ranked according to their relative impacts on the environment as follows: Market price support and payments based on output both increase the price received by producers for a specific commodity such that the more the commodity is produced, the higher will be the support. Thus the higher these forms of support, the greater is the incentive for monoculture, for increasing the use of inputs (such as chemicals), and/or for using environmentally sensitive land, and the higher is the pressure on the environment. Moreover, these payments have the lowest effectiveness in achieving environmental goals, as they are sector-wide payments that cannot be linked with any environmental goal or situation that are generally local. Payments based on input use reduce the cost of inputs used by producers such that the more the input is used, the higher will be the support. Thus the higher these payments, the greater the incentive to use the input, and the greater the impact on production and the environment. The more the payment is specific to a variable input (for example fertilizers, pesticides), the greater the incentive for production intensification, and the pressure on the environment. For example, the environmental impact of a credit subsidy for purchasing fertilizers or pesticides is potentially higher than a credit subsidy for acquiring farm land or extending farm buildings. Therefore these payments may have a higher, equal, or a lower effect on production and the environment than an output payment depending on the type of input on which the payment is based. Payments based on area planted/animal numbers reduce the cost of land/livestock for current plantings/animal numbers. As producers have to plant a specific crop or own specific animals, these payments may be an incentive to keep environmentally sensitive land producing commodities in a non-environmentally

Environmentally harmful subsidies

277

friendly way on such land. Although these payments may be targeted to a specific environmental goal or situation, they provide an incentive to bring additional land or animals into specific production and encourage monoculture in the same way as the payments based on output. However, as producers are not encouraged to increase yields and to produce as intensively as they are with the forms of support outlined above, the environmental impact of these payments is potentially lower. Payments based on historical entitlements (that is, past support, area, animal numbers, production, or income) and payments based on overall farming income (paid on the condition that the overall farmers’ income is below a predefined level) also have the potential for retaining environmentally sensitive areas under production. However, as to receive these payments producers are not obliged to plant, own animals, or produce any particular commodities, they allow for individual choices on environmentally friendly production techniques, and do not encourage production intensification and/or monoculture. Therefore the impact of these payments on the environment are relatively benign or lower than the previous forms of support. Payments based on input constraints are paid on the condition that farmers respect certain constraints (reduction, replacement or withdrawal) on the use of inputs, often for environmental purposes. These payments may be targeted to specific environmental situations to address specific environmental issues associated with agriculture. They may contribute to offsetting the reduction of a positive environmental impact or the increase of a negative environmental impact of farming activities often benefiting from one or more of the previous forms of support. These,mainly through input constraints that reduce production intensity, encourage production diversification, or put environmentally sensitive land aside from production relative to what otherwise would occur. The environmental impacts of these payments depend on the type of constraint, but they have the potential for reducing environmental pressure and for being the most environmentally effective PSE measures. Source:

OECD (2002b) and Portugal (2002).

Environmental and resource economics 2004/2005

100 90 80 70 % 60 50 40 30 20 10 0

2001

2000

1999

1998

1997

1996

1995

1994

1993

1992

1991

1990

1989

1988

1987

Other Payments based on input constraints Payments based on historical entitlements Payments based on area planted/animal numbers Payments based on input use Payments based on output Market price support

1986 Source:

100 90 80 70 60 % 50 40 30 20 10 0 2002

278

OECD, PSE/CSE database, 2003.

Figure 7.4

Composition of producer support estimate (PSE) (1986–2002)

buy one’s way in through buying and scrapping somebody else’s licensed boat. Over time there has been a movement in many countries from catch control to effective management. The OECD countries have supported their fishing industries with significant amounts of money and over long periods of time. Subsidies for fisheries in OECD countries amounted to US$6 billion in 1999 (OECD, 2001b). This corresponds to 20 per cent of the value of landings. Japan provides the largest fisheries subsidies in the OECD, followed by the European Union, the United States, Canada, Korea, Spain and Norway. Most of the government financial transfers in the OECD countries are for general services. Expenditures on research, management and enforcement activities are important as they can contribute to ensuring the sustainable use of fish stocks and the aquatic ecosystem. In some countries, however, the bulk of the expenditure on general services is on fisheries infrastructure and fisheries enhancement programmes that can contribute to overfishing (Cox, 2002). The introduction of cost recovery programmes for some research, management and enforcement expenditure in some countries implies that some of these activities directly benefit fishers, rather than society as a whole. Capacity-reducing transfers, including vessel buyback programmes, licence retirement schemes and payments to fishers to leave the industry, have been widely used in OECD countries in response to overfishing and overcapacity (Cox, 2002; see also Porter, 2002).

Environmentally harmful subsidies 1986–88 2000–2002

New Zealand Australia Poland a Turkey Canada Slovakiaa United States Mexico Czech Republica Hungarya OECDb European Union Japan Iceland Korea Norway Switzerland

90 80 70 60 50 % 40 30 20 10 0 –10

279

Notes: Countries are ranked according to 2000–2002 levels. a For the Czech Republic, Hungary, Poland and the Slovak Republic 1986–88 is replaced by 1991–93. b For 1986–88, the Czech Republic, Hungary, Poland and the Slovak Republic are excluded. Source:

OECD, PSE/CSE database, 2003.

Figure 7.5 Producer support estimate by country (percentage of value of gross farm receipts) Energy Energy subsidies are widespread in all countries. For a long time, governments have manipulated energy prices through regulations, taxes and direct and indirect support, for the purpose of energy security, the diversification of domestic energy sources and social concerns (for example keeping low energy prices, subsidizing coal mines). Estimates of support for coal production are more systematic and complete than for other forms of energy. Total support to coal industry in the OECD countries decreased through most of the 1990s from US$11.4 billion in 1990 to US$5.4 billion in 2000 (IEA, 2001). Germany and the United Kingdom are the countries with the biggest decreases in support (Figure 7.6). This reduction in support was accompanied by an even larger reduction in coal production over the same time period and, consequently, support per tonne of coal equivalent increased in some countries (Figure 7.7). The negative environmental consequences of coal subsidies are obvious in

280

Environmental and resource economics 2004/2005

12 000

8 000 6 000

Source:

2000

1999

1998

1997

1996

1995

1994

0

1991

2 000

1993

United Kingdom Turkey Spain Japan Germany France

4 000

1992

US$ million

10 000

IEA

Figure 7.6

Support to coal in selected OECD countries (US$ million)

terms of air pollution, but also soil degradation, toxic waste and water pollution. Generally speaking, subsidies that increase fossil fuel consumption through lower prices result in higher emissions of greenhouse gases and other pollutants (SOx, NOx and so on). It should be noted that energy subsidies in non-OECD countries are twice the absolute level of OECD countries. IEA (1999) has simulated the effects of removing energy subsidies in eight non-OECD countries (China, Russia, India, Indonesia, Iran, South Africa, Venezuela and Kazakhstan), where the average subsidization is 21 per cent. The results indicate, on average, a 0.76 per cent increase in GDP, a 13 per cent decrease in energy consumption and a 16 per cent reduction of CO2 emissions. Manufacturing There seems to be a trend away from subsidies for particular industry sectors towards more horizontal objectives, including regional development, research and development (R&D) and small and medium-sized enterprises (SMEs) (OECD, 1998b). Indirect means of support, such as public procurement, R&D contracts and R&D intermediary institutions, channel far more financial resources to manufacturing industry than does direct support. Even if the support element in indirect measures only represents a very small percentage, it would still be significant. As there is no agreed methodology for measuring

Environmentally harmful subsidies

281

250 Turkey

US$/tce

200 150

Japan Germany France Spain OECD

100 50

United Kingdom 2000

1999

1998

1997

1966

1995

1994

1993

1992

1991

0

Source: IEA

Figure 7.7

Support to coal in selected OECD countries (US$/tce)

the support element in indirect support, uncertainties remain as to its role as a policy instrument and, more specifically, as a tool of support to manufacturing industry. Programmes intended to support one or selected manufacturing sectors are of special interest from the environmental point of view. Most sectoral programmes target the shipbuilding industry. Other industries where sectoral programmes are common include fish processing, textiles and the steel industry. The support for the aircraft and space industries is also channelled through R&D programmes, equity capital injections and intermediary space agencies. There have been no systematic efforts to assess the environmental impacts of manufacturing subsidies. Support to manufacturing, measured in constant prices, declined in 1986–89, reaching US$37 billion in 1989 (OECD, 1998b). The support peaked at US$45.7 billion in 1991 before declining to US$43.7 billion in 1993. There was a 24 per cent growth in support in nominal terms from 1989 to 1993, corresponding to a 4 per cent decrease in constant terms during the period (OECD, 1998b). There are, however, no recent figures available. Transport Subsidies for road and rail transport in the European Union, Hungary and Switzerland amounted to about US$40 billion in 1998 (Nash et al., 2002). The estimate is based on a broad definition of subsidies that compares total revenues with total social costs for each mode of transport. In nearly all countries,

282

Environmental and resource economics 2004/2005

revenues from road transport cover the cost of providing and maintaining the infrastructure. In many countries these revenues cover total social cost (Nash et al., 2002). However, as mentioned earlier, another approach is to compare the marginal social cost with the price paid; this shows that road transport is often charged much below marginal social cost. Rana Roy (in ECMT, 2003) calculated the optimal transport pricing in urban areas for five EU countries; for cars, this indicates, for example, that optimal prices would imply an increase in peak-period prices of about 70 per cent in the Paris suburban area (Ile de France), 95 per cent in Munich and over 150 per cent in London. According to Roy, this price increase would provide a significant fall in car passenger kilometres and significant welfare gains. On the other hand, as other modes of transport (for example rail) are heavily subsidized, phasing out these subsidies would divert traffic from rail to road. Although there might be some reduction in the total amount of transport, the increase in road transport would have negative effects on the environment. According to Nash et al. (2003), passenger and freight revenues cover, on average, 36 per cent of rail system costs. Water Water-related subsidies can take several forms and cause a series of environmentally harmful consequences (see Table 7.4). According to Mona Sur et al. (2002), farmers across the world seldom pay more than 20 per cent of the full cost of water, thus encouraging wastage, groundwater depletion, pollution, soil salinization and reappearance of virulent forms of malaria. They also claim that ‘full (cost) recovery, to the best of our knowledge, including the recovery of the full investment cost, has not been practiced anywhere’.

5.

CONCLUSION: THE POLITICAL ECONOMY OF SUBSIDY REFORM

Subsidies are at the core of the sustainable development paradigm in a complex and paradoxical manner. On the one hand, while it is largely recognized that sustainable development implies, inter alia, well-functioning and non-distorted markets, subsidies constitute a prominent form of market distortion; on the other hand, support measures may be needed to start a virtuous cycle of sustainable development, for instance to help certain economic sectors or regions; such measures should be transitory with a firm sunset clause. Subsidies also have a long, complex and somewhat chaotic history; they have been introduced over decades, often under political pressure, often without a long-term strategic vision, for a variety of economic and social purposes: protecting economic sectors from international competition,

Environmentally harmful subsidies

283

Table 7. 4 Some examples of possible environmentally harmful consequences of water-related subsidies Description of the subsidy

The mechanism through which it may harm the environment

How it may harm the environment

Agricultural price support policies

Incentives for farmers to grow water-inefficient crops in unfavourable environments

Salinization, waterlogging and/or decline in groundwater tables

Surface-water price Overuse of water and cultivation Pollution and depletion of of water-inefficient crops. Use of water bodies. Salinization, inefficient technologies elevated levels of water tables and drainage problems Electricity price

Substitution of surface water, (SW) with groundwater (GW), especially in places where SW supply is inadequate or irregular. Overuse of GW due to excessive pumping

GW levels are lowered, aquifers are depleted and contaminated via intrusion of low-quality water from adjacent aquifers or seawater intrusion

Pesticide prices

Overuse of pesticides and inefficient application management practices leading to high rates of pesticide leaching

Pesticides contaminate GW aquifers and may create irreversible health damages

Fertilizer prices

Overuse of fertilizers and inefficient application management practices leading to high rates of fertilizer leaching

Fertilizers can increase soil salinity and contaminate GW aquifers. They may also adversely affect the development of infants

Source:

Mona Sur et al. (2002).

supporting employment or income of segments of the population (coal miners, farmers and so on). Whilst it is now widely recognized that subsidies are costly, often inefficient, distorting and, in a number of instances, environmentally harmful, reforming and phasing out these subsidies faces formidable challenges. 5.1

A Political Challenge

Inherited habits, political, social and institutional barriers to subsidy reform

284

Environmental and resource economics 2004/2005

result in what van Beers and de Moor (2001) call ‘addiction to subsidies’. Removing subsidies faces strong economic barriers such as: lock-in technologies (removing the subsidy would force technical change when the subsidy is targeted on a specific technology, thus stifling innovation); rent-seeking and vested interests. The fear of loss in competitiveness is also prominent, for example concerning tax exemptions, in particular in the energy sector. Van Beers and de Moor also underline the institutional barriers such as: purchase of votes by politicians through subsidies, creation of institutions and bureaucracy in charge of the management of subsidy schemes (removing the subsidy would imply laying off the employees in charge of the scheme), and fear of political instability. For instance the Chicago Convention on Civil Aviation (1944) introduced the de-taxation of kerosene in 1953 (a typically environmentally harmful subsidy). This exemption should be phased out, but it faces strong resistance from vested interests (for example airlines). Furthermore, this would involve an international conference to seek a new consensus among all parties or the renegotiation of a large number of bilateral treaties (van Beers and de Moor, ibid., p. 73). The current difficulties and slow progress in the reform of the EU common agricultural policy is an example of political barriers; even keeping income support to farmers, while changing its nature and purpose, faces vigorous opposition. Overcoming these obstacles requires careful implementation strategies. These could include, in particular (OECD, 1998a): • Addressing the effects on equity and employment, for instance through compensatory payments to the stakeholders affected by the subsidy removal; these payments should be decoupled from output levels and be temporary, for instance to ease the transition of workers towards new employment opportunities. • Implementing a transparent and cooperative reform process, in particular through reliable and transparent data and indicators on subsidies and their detrimental effects and a clear exposition of the objectives and benefits of the reform. • Implementing a progressive reform involving all stakeholders over a carefully planned period. • Resolving the international competitiveness issue, when removing a subsidy in a given country would affect the competitiveness position of the concerned sector. A case in point is the current prevalence of energy tax exemptions of industry in OECD countries (1500 cases of exemptions recorded in the OECD database on environmentally related taxes). Many exemptions have been introduced to protect certain industry (particularly energy-intensive) sectors. Progress can only be expected if it is possible to achieve an internationally concerted action (OECD, 2001a).

Environmentally harmful subsidies

285

Also, all new subsidies should include a sunset clause, making sure that after a given period they would be automatically phased out. 5.2

A Social Challenge

Subsidies are often designed for social reasons (for example to protect poorer segments of the population), one of the three ‘pillars’ of sustainable development. This social concern may often work against sound environmental management, for instance when subsidized water prices result in wastage and salinization of soils. The social justification of water subsidies is particularly questionable in developing countries: in urban areas, subsidized water supply benefits primarily relatively wealthy families who have access to the urban water supply system, while poorer families in rural areas still lack access to potable water (de Moor, 1997). Subsidized irrigation water, often proportional to the surface of land owned, tend to benefit richer farmers. In fact, increasing water prices in developing countries are likely to generate social benefits. The willingness to pay of poor people for adequate water supply is high (World Bank, 1992). In practice, the prices of privately sold water on which poor people must rely are much higher than water supplied through proper supply infrastructures (up to 12 times higher – FAO, 1994). A 1994 World Bank study shows that 80 to 90 per cent of the richest quintile of the population in some developing countries have access to public water supply, while only 30 to 50 per cent of the lowest quintile have access (World Bank, 1994, cited by de Moor, 1997). The appropriateness of energy-related subsidies can also be questioned. For instance, coal subsidies may not be the best way to maintain social cohesion: subsidized coal mines often produce coal several times more costly than imported coal (IEA, 1999). The social implications of removing energy subsidies are not straightforward; higher energy prices can be a priori construed as socially regressive, but, in the case of motor vehicles, in developing countries car owners tend to be concentrated in the higher income segment of the population. 5.3

A Challenge for the Developing World

Subsidies in OECD countries affect the development of non-OECD countries. For instance, agricultural subsidies create a barrier to the import of agricultural products of developing countries. Subsidies in the developing world can cause the depletion of natural resources. Subsidies to fisheries result in overfishing, not only in OECD countries, but also in developing countries; for instance, the European Union has agreements with a number of developing countries to fish in their coastal waters, with subsidized, highly efficient, sophisticated fishing vessels.

286

Environmental and resource economics 2004/2005

Clearly, environmentally harmful subsidies are a global challenge for sustainable development as their economic, environmental and social effects are pervasive throughout the world. This has been recently recognized by the World Summit on Sustainable Development (Johannesburg, September 2002), which calls upon countries to ‘promote energy systems compatible with sustainable development through the use of improved market signals and by removing market distortions, including restructuring taxation and phasing out harmful subsidies’. 5.4

A Technical Challenge

As explained in this chapter, assessing the environmental impact of environmentally harmful subsidies is technically complex. Yet dealing with this complexity this is an essential step in paving the way for the reduction or removal of these subsidies. Considerable work needs to be done to get a clearer picture and develop effective analytical tools. An OECD workshop on environmentally harmful subsidies (OECD, 2003b) concludes that further work should include: • supplementing and updating existing databases on subsidies and exploring the fuller inclusion of subsidies in National Accounts; • improving the conceptual framework for analysing the environmental impact of subsidies and testing a ‘checklist’ designed to assess the environmental impacts in various sectors (such as energy, water, transport, agriculture); • strengthening cooperation between the various institutions working in this area; and • examining the role of subsidies in the broader context of sustainable development, in order to understand the possible synergies and tradeoffs in subsidy reform.

NOTES 1. Although this chapter relies heavily on past and ongoing OECD work on environmentally harmful subsidies, the opinions expressed are the authors’ own and do not necessarily reflect the views of the OECD. The authors have benefited from many useful comments and suggestions, in particular from Henk Folmer, Luis Portugal, Rana Roy, Stephen Perkins, Ronald Steenblik, Tom Tietenberg and an anonymous referee. All remaining errors are our responsibility. 2. Government intervention failures, also called ‘policy failures’, occur when governments’ interventions distort the price system and result in environmental degradation, or worsen situations when such degradation already exists. 3. www.oecd.org/env/tax-database 4. Rana Roy defines a transport subsidy as the revenue foregone relative to the revenue that would be provided by an optimal pricing. 5. Support conditional upon farmers undertaking some type of environmental compliance.

Environmentally harmful subsidies

287

REFERENCES Cox, A. (2002), ‘OECD work on defining & measuring subsidies in agriculture’, paper prepared for the OECD Workshop on Environmentally Harmful Subsidies, 7–8 November, Paris, available at http://interprod.oecd.org/agr/ehsw/index.htm De Moor, A.P.G. (1997), ‘Perverse incentives’, paper prepared for the Institute for Research on Public Expenditure (IRPE), Earth Council. De Moor, A. de and Calamai P. (1997), ‘Subsidizing unsustainable development: undermining the earth with public funds’, paper prepared for the Institute for Research on Public Expenditure (IRPE), commissioned by the Earth Council, San José, Costa Rica. European Conference of Ministers of Transport (ECMT) (2000), Efficient Transport Taxes & Charges, Paris: OECD/ECMT. European Conference of Ministers of Transport (2003) Efficient Transport Taxes and Charges 2003, Paris: OECD/ECMT. Food and Agriculture Organization of the United Nations (FAO) (1994), ‘Water Policies and Agriculture’, in The State of Food and Agriculture, Rome: FAO. Hannesson, Rögnvaldur (2002), ‘The Development of Economic Institutions in World Fisheries’, background paper for the World Development Report 2003, World Bank, Washington, DC. International Energy Agency (IEA) (1999), World Energy Outlook, 1999 – Looking at Energy Subsidies: Getting the Prices Right, Paris: OECD/IEA. International Energy Agency (2001), Energy Policies of IEA Countries, 2001 Review, Paris: IEA. International Energy Agency (2002a and previous years), Coal Information, Paris: OECD/IEA. International Energy Agency (2002b), Energy Policies of IEA Countries, 2002 Review, Paris: IEA. Koplow, D., D. Andrew, J.M. Burniaux and E. Adorrian (2001), Environmental Effects of Liberalizing Fossil Fuels Trade: Results from the OECD Green Model, Paris: OECD. Koplow, D. and Dernbach, J. (2001), ‘Federal fossil fuel subsidies and greenhouse gas emissions: a case study of increasing transparency for fiscal policy’, Annual Review of Energy and Environment, vol. 26; pp. 361–89. Mona, Sur, Dina Umali-Deininger and Ariel Dinar (2002), ‘Water-related subsidies in agriculture: environmental and equity consequences’, prepared for the OECD Workshop on Environmentally Harmful Subsidies, 7–8 November, Paris, available at http://interprod.oecd.org/agr/ehsw/index.htm Myers, N. and J. Kent (2001), Perverse Subsidies: How Tax Dollars Can Undercut the Environment and the Economy, in cooperation with the International Institute for Sustainable Development, Washington, DC: Island Press. Nash, Chris, Peter Bickel, Friedrich Rainer, Heike Link and Louise Steward (2002), ‘The Environmental Impact of Transport Subsidies’, paper prepared for the OECD Workshop on Environmentally Harmful Subsidies, 7–8 November, Paris, http://interprod.oecd.org/agr/ehsw/index.htm NIEIR (National Institute of Economic and Industry Research) (1996), ‘Subsidies to the use of natural resources’, available at http://www.ea.gov.au/pcd/economics/ subsidies/subs11.html. OECD (1972), Recommendation of the Council on Guiding Principles Concerning the International Economic Aspects of Environmental Policies, Paris: OECD.

288

Environmental and resource economics 2004/2005

OECD (1998a), Improving the Environment through Reducing Subsidies, 3 vols, Paris: OECD. OECD (1998b), Spotlight on Public Support to Industry, Paris: OECD. OECD (1999), ‘Economic instruments for pollution control and natural resources management in OECD countries: a survey’, unclassified document, Paris: OECD. OECD (2000), Transition to Responsible Fisheries: Economic and Policy Implications, Paris: OECD. OECD (2001a), Environmentally Related Taxes in OECD Countries: Issues and Strategies, Paris: OECD. OECD (2001b), Review of Fisheries in OECD Countries: Policies and Summary Statistics, Paris: OECD. OECD (2002a and previous years), Agricultural Policies in OECD Countries, Monitoring and Evaluation, Paris: OECD. OECD (2002b), ‘Methodology for the measurement of support and use in policy evaluation’ accessed at http://www.oecd.org/agr/policy. OECD (2003a – forthcoming), Agricultural Policies in OECD Countries, Monitoring and Evaluation, Paris: OECD. OECD (2003b), Environmentally Harmful Subsidies: Policy Issues and Challenges, Paris: OECD. Pieters, J. (2003), ‘What makes a subsidy environmentally harmful: developing a checklist based on the conditionality of subsidies’, in OECD, Environmentally Harmful Subsidies: Policy Issues and Challenges, Paris: OECD. Porter, G. (2002), Fisheries Subsidies and Overfishing: Toward a Structured Discussion, Geneva: United Nations Environment Programme. Portugal, Luis (2002), ‘OECD work on defining and measuring subsidies in agriculture’, paper presented at the OECD Workshop on Environmentally Harmful Subsidies, Paris, 7–8 November, available at http://interprod.oecd.org/agr/ehsw/ index.htm. Roy, Rana (2003), ‘Optimal transport pricing’, in European Conference of Ministers of Transport, Efficient Transport Taxes and Charges 2003, Paris: OECD/ECMT. Steenblik, R.P. (1995), ‘A note on the concept of subsidy’, Energy Policy, 23 (6,) 537–53. Steenblik, R.P. (2003), ‘Subsidy measurement and classification: developing a common framework’, in OECD, Environmentally Harmful Subsidies: Policy Issues and Challenges, Paris: OECD. van Beers, C. and S. de Moor (2001), Public Subsidies and Policy Failures: How Subsidies Distort the Natural Environment, Equity and Trade and How to Reform Them, Cheltenham, UK and Northampton, MA: Edward Elgar. The World Bank (1992, 1994), World Development Report, Washington, DC: World Bank.

Index 3M Company, USA 126 abatement costs 62, 65, 69–74, 97, 172 optimal timing 89, 94–5 technology 66, 67, 75, 90–93 acidification 173 Acid Rain program 248 Adamowicz, V. 27, 49, 51, 200, 212 Adams, R.M. 195, 212 Adriaans, Albert 164, 165, 181 Africa 130, 144 Aghion, P. 61, 79, 97 agriculture 101, 106, 110, 118, 160, 254, 262–3 environmental impact of payments 274–7 land 114, 116, 132 rents 107, 109 subsidies 103, 132, 172, 268, 285 air pollution 140, 151, 170, 192–3, 202, 220, 254, 280 aircraft 262 industry 280 noise 205 Alaskan oil spill 5, 13, 14, 17, 47, 48, 212 see also Exxon Valdez Alberini, A. 52, 193, 202, 203, 212 Alterman, R. 110, 112, 134 amenity values 83, 85, 88, 89, 128 American Farm Trust 133 Anderson, J.E. 106, 109, 134 Angrist, J.D. 49, 52 animal feed 159 aquatic ecosystem 278 Arellano, M. 49, 52 Aronsson, Thomas 141, 145, 181 Arrow, Kenneth J. 44, 47, 52, 64, 98, 146, 181, 198, 206, 213 Asheim, Geir B. 142, 145–6, 152, 180,181

Asian Development Bank 192, 203, 213 Atkinson, Giles 147, 150, 153, 182 Atkinson, S.E. 210, 213 atmospheric dispersion models 193 Austria 152, 157, 165 Ayres, Robert U. 160, 163, 177, 182 Bailey, P. 221, 251 Balistreri, E. 47, 52 Barde, Jean-Philippe 254–88 Barichello, R.R. 109, 134 Barro, R.J. 61, 79, 81, 82, 98 Barton, D.N 203, 205, 208, 210–13 Batabyal, A.A. 133, 134 bauxite 143 beaches 8, 34, 35, 38, 50, 308 bear, endangered species 127 Ben-Akiva, M.E. 23, 48, 52 benefit cost 69–71 transfer 194, 206–08 Bergland, O. 207, 213 Beron, K. 45, 52 Bertram, Regina 61–97 Biglaiser, G. 78, 98 Bills, N. L. 109, 134 biodiversity 127, 150, 169 conservation 79, 126, 159 damage to 192, 274 biological resources 126, 128 bioproductive land 159, 160, 163 birds deaths in oil spills 42, 43 and inter-tidal life 35 loss and injury 13, 48, 51 meadow 207 Bishop, R.C. 27, 40, 50, 52 Bockstael, N. E. 32, 50, 133, 52 Bogdanski, B.E.C. 115, 124, 134 Bohm, P. 220, 221, 251 Boisvert, R.N. 109, 134 Bovenberg, A.L. 62, 79, 81, 86, 98 289

290

Environmental and resource economics 2004/2005

Boward, W.L. 75, 98 Bowes, M.D. 121, 134 Brennan, T. J. 191, 213 British Columbia, Canada 126, 128, 129 British Petroleum 155, 182 Brookshire, D.S. 7, 46, 53 Brouwer, Roy 170, 171, 182, 205, 207, 211, 213 Brown, G. M. 7, 11, 53 Brown, T.C. 53, 125, 134 Brueckner, J.K. 108, 110, 113, 134 Brundtland Commission 126 building materials, air pollution effect 195 Buonanno, P. 90, 98 Burchell, R.W. 113, 134 Bureau of Census 40 Bureau of Labor Statistics 40, 45, 53 California 18, 20, 21, 22, 35–8 California Oil Spill (COS) 35, 42–4 Cameron, S. 113, 134 Cameron, T.A. 33, 34, 46, 48, 53 Canada 127, 128, 130, 278 cap-and-trade schemes 125 capital depreciation, natural 169 equipment 63 human 142, 143, 154 natural 155, 168, 180 Capozza, D.R. 104, 133, 134 carbon 101, 114, 119, 120, 157 atmospheric concentration 160 flux 119–20, 122 in forest land 11 fuels 162 storage 132 tax 61, 70, 91, 92 carbon-based energy industry 92 carbon dioxide 50, 116, 150, 170 abatement 65, 67 concentration 93, 97 emissions 90, 143, 162, 163, 177 pumped into deep oceans 161 cars accidents 151 efficiency of 234 zero emissions 176 Carson, R.T. 16–17, 19, 21, 41–50, 53, 190, 199, 212–3

Castañeda, B. E. 152, 182 cedar trees, poaching of 127 cement 227 Census of the USA 2000 3, 4 Center for Clear Air Policy 222 Chambers, Nicky 160, 182 Chapman, D.J. 50, 54 Chay, K. 5, 54 chemicals, harmful 106, 274 Cheshire, P.133, 134 Chestnut, L.G. 208, 214 Chiang, A. 66, 98 Chicago Convention on Civil Aviation 284 China, population 246 Cicchetti, Charles 8 cigarette smoking, health effects 46 Ciriacy-Wantrup, S.V. 44, 54 Clark, J.S. 105, 109, 134 Clarke, L.E. 61, 62, 95, 98 Clarke, Matthew 152, 182 Clean Air Act 191 Clean Development Mechanism 218, 219, 244 clear-felling 131 climate change 61, 97, 128, 156, 169, 173 policy 76, 218–53 coal mining subsidies 279, 285 production 143, 219, 254, 263 Cobb, C.W. 151, 153–7, 182 commercial buildings 110 land use 101, 103 commodities and market exchanges 32 commuters 113, 151 compensation payments 112, 190, 284 congestion 151, 254 conservation 103, 126–8, 131 Conservation Fund 110 Conservation Reserve Program, USA 102 Constanza, R. 210, 214 consumer 27, 46, 88, 155, 156, 158 expenditure 151, 180 goods, quality 139, 164, 174, 177 preference 12, 25, 34, 35 Consumer Price Index (CPI) 38, 45

Index Contingent Valuation (CV) 1–60, 194, 195, 197 survey 44, 45 Cook, T. 124, 135 copper 143 Corbett, R. 109, 135 Cotterill, R. W. 54 Coughlin, Robert E. 112, 135 countryside preserving 110, 112, 169 Cox A. 278, 287 Crooker, J. 51, 54 Cropper, M.L. 193, 214 crops and cropland 103, 159, 195, 249 Cummings, R.G. 45, 48, 54 Current Population Survey 3 Czech Republic 171 Dahme, Kai 165, 182 Daly, Herman E. 140, 151, 154–6, 163, 182 damage assessment 15, 17, 48 cost estimates 140 function 192 Dasgupta, Partha 143, 180, 182 data collection 5, 34 Davis, Graham A. 6, 143, 182 Davis, R.K. 55 DDT releases 17, 20 Dear, Michael 108, 135 defensive expenditures 154–5 deforestation 133, 244 De Haan, Mark 182 dehydration of land 173 De Moor, A.P.G. 258, 284, 287 desertification 133 Desvousges, W.H. 13, 47, 49, 55, 190, 212, 214 developed countries 127, 149, 158–60, 176 developing countries 203, 218, 246, 285–6 emission reduction 244 development rights, transfer of 111, 112 Diamond, P.A. 13–16, 23, 24, 47, 48, 55 Diefenbacher, H. 152, 155, 183 Dincer, Ibrahim 141, 183 disease 128 dispersion of water 173 disposal of waste 164

291

domestic emissions permit system 235, 241 forage 124 regulation 238, 239, 240 Dow Corporation, USA 126 Downing, M. 207, 212, 214 Downing, P.B. 76, 98 drainage systems 106 Duffield, J. W. 55 Dunford, R.W. 50, 55 Dutch ERUPT Program 223 eagles, bald, reproductive problems 20 Earth Summit, Rio de Janeiro 130 Ebert, U. 33, 34, 55 ecological deficit 159 footprints 140–41, 158–63, 176 impact of human activity 160 economic growth 61–97 integration, global 162 models 41 policy 130 restructuring 174 values 29, 30, 33 ecosystem 80, 111, 126, 127 Edmonds, J. 96, 98 education 151, 154 Eisner, Robert 180, 183 Ekins, Paul 169, 170, 183 electricity generation 190, 228 elephant, endangered species 127 Ellerman, A. Denny 248, 251 El Serafy method 143, 146–9, 155, 176, 178–9, 183 emissions 92, 163, 164, 177, 226–8, 262 abatement 246 allowances in EU 236 reduction 77, 94, 169, 230–34, 237, 241, 246–8 tax 62, 63, 65, 76–8, 238–40 endangered species 127–9 habitat protection 132 legislation 129 Endangered Species Act (ESA) 128 Endangered Species Committee 129 energy 92, 164, 262, 263 atomic 157 efficiency 90, 255

292

Environmental and resource economics 2004/2005

and environmental policy 61, 192 footprint 162, 181 resource depletion 154–9 subsidies 261, 279–80, 285 tax exemptions 267, 284 Englin, J. 24, 55 environment 24, 102, 169, 173, 179, 226, 242 amenities 120–25, 211 damage 143, 150–58, 166, 177 degradation 32, 145, 163, 286 policy and protection 61–97, 254 regulation 71–6 resources 23, 29, 31, 34, 48 subsidies harmful to 254–88 valuation estimates 101, 164, 167, 190–92 Environmental Liability, White Paper 192 environmentally-friendly technologies 78 Eom, Y.S. 25, 55 epidemiology 50 equity, intergenerational 139, 140 Ervin, D. E. 109, 135 Europe agricultural subsidies 110 Central and Eastern (CEE) 223 air pollution in 202 forest landowners in 130 European Commission 176, 182, 190–92, 214 European Common Agriculture Policy 284 European Conference of Ministers of Transport 287 European Union 165, 236, 278, 285 cutting emissions 75 environmental standards 220 eutrophication 173 Ewing, R. 133, 135 Exxon Valdez oil spill 5, 13, 14, 17, 47, 48, 212 Fankhauser, Samuel 143, 183 farming 108, 109, 110, 114, 151, 180 extensive and intensive land use 266, 274 and land development 105 subsidies 103

Faustmann, M. 103, 114, 116, 119, 131, 135 felling, clear 115 Fernandez, Linda 111, 135 Ferrari, Sylvie 141, 183 fertilizers 254, 262, 268, 283 filters on cigarettes 46 financial incentives 5, 102 fire 128 Fischel, William A. 112, 135 Fischer, Carolyn 76, 98, 228, 231, 251 Fischer-Kowalski, Marina 181, 183 fish 21, 129, 150 processing industry 281 stocks, sustainable use 254, 278 fisheries 254, 261, 265, 275–9 and sea fishing 285 fishing 33, 49, 159, 199 freshwater 205, 206 flowers, wild 207 focus groups 9, 40, 199 food 20, 155 Food and Agriculture Organization 287 forage 103 for wild ungulates 121 foreign investment 235, 237 Forest Practice Code, British Columbia 102, 129–30 forestry 118, 130, 150, 244, 265 forests 101, 114, 123, 143, 161, 180 air pollution effect on 195 amenities 115 certification 129–30 depletion 180 ecosystems 125 harvesting 121 landscape 124, 160 management 126, 132, 249 public ownership 127 recreational benefit 189, 201, 207 sustainable 102, 130 Forest Service 189 Forest Stewardship Council (FSC) 130, 132 fossil fuel 91, 118, 157, 159–61, 254 tax on 236 France 112, 171, 208 transport in 191 Freeman, A.M. 55 freshwater recreation sites 46

Index fuel 90,156, 159, 254 Fuguitt, Glenn V. 112, 135 Fujita, Masahisa 104, 106, 135 Galster, G. 135 Gardner, Bruce 106, 135 GARP see Green Accounting Research Project II gas 156 Gawel, Erik 166, 167, 183 GDP 143, 155, 157, 165, 170, 176, 203 Geres, R. 250, 251 Gerlagh, Reyer 172, 173, 184 Germany 152, 164–5, 176, 208 global atmosphere 163 human impact 163 warming 91, 101 GNP 143–4, 155, 170, 176 gold 143 goods and services consumption 177 Goulder, L.H. 62, 65–8, 70–71, 74, 89, 92–8 government intervention 257 ownership 256 policy 85, 88, 89 on intensive land use 104, 105 Gradus, R. 81, 98 grassland, temperate 127 grazing land management 249 Green Accounting Research Project II (GARP) 140, 191, 214 greenhouse gas 62, 75, 219, 280 emissions 90–93, 91, 157, 173, 218, 254 Green, M.J. B. 126, 135 GREENSTAMP 141, 168, 170–71, 174–6, 178 ‘green’ technologies 94, 95 gross domestic product see GDP Grossman, G.M. 64, 98 gross national product see GNP groundwater 205, 282, 283 growth management 110–11 Grubb, M. 61, 98 Guenno, G. 152, 155, 184 habitat, fragile 8, 111 Hagem, C. 231, 251

293

Hamilton, Bruce W. 133, 135 Hamilton, Clive 152, 155, 184 Hamilton, Kirk 142, 149, 184 Hammack, J. 7, 11, 55 Hanafi, A. G. 220, 251 Hanemann, W.M. 2, 4, 8, 11, 31, 44–8, 50, 55 Harbberger, A.C. 31, 56 Hardie, Ian 101–38 Hargrave, Tim 233, 251 Harrison, G.W. 16, 17, 28, 44, 56 Hartman, R. 115, 136 Hartwick, John M. 141, 142–7, 180, 184 harvesting of timber 114–17, 119–20, 123, 143, 180 Hausman, J.A. 1, 44, 50, 56 Hawaii, farmland conversion restriction 112 Hazell, P.B.R. 123, 136 health 151, 202, 211 expenditure 154 and property, damage to 192 of workers 79 heavy metal dispersion into water 173 Heckman, J.J. 45, 56 Heidug, Wolfgang K. 61–97 Heimlich, Ralph E. 106, 109, 136 Henderson, Vernon 133, 136 Herendeen, Robert A. 141, 184 Hicksian demand for recreation 51 welfare measures 26, 46, 51, 197 Hines, J.R. 50, 56 Hinterberger, Friedrich 163–6, 178, 181, 184 Hoehn, J.P. 210, 214 Hof, John 123, 136 Hofkes, Marjan 173, 184 Hollis, Linda E. 110, 136 Honkatukia, Outi 254–88 Hotelling, Harold 143, 147, 185 households 13, 43, 82, 103, 104, 196 Houthakker, H.S. 40, 41, 51, 56 Howitt, P. 61, 79, 97 Huang, J.C. 12, 56 Hueting, Roefie 141, 168–71, 174, 175, 185 hulls, double in oil tankers 43 Human Development Index (HDI) 165

294

Environmental and resource economics 2004/2005

human impact 127, 158, 159 hunting 7, 127, 196 hydro-electric power 159, 189 illness, serious 202 Imbens G.W. 50, 51, 56 incentives 11, 28, 50 for landowners 118 income levels 38, 154, 177 tax surcharges 36 Index of Sustainable Economic Welfare 139, 150–58, 175–7, 180 Indonesia 168 industrial land use and production 101, 103, 159 industry carbon dioxide emissions 91 subsidies to 261 infrastructure 113, 114, 131, 274 interest rates, preferential 258 International Emissions Trading (IET) 218, 238 International Energy Agency 222, 287 investment 88, 89, 91, 150 credits 111 foreign 151, 246, 248 in knowledge and research 79, 95 private 151 profit-seeking 79 projects 190 iron 143, 227 irrigation water 129, 285 Irwin, Elena 133, 136 Israel, increased population 112 Jackson, T. 152, 155, 157, 185 Jaffe, A.B. 61, 76, 99 Janssen, J. 231, 251 Japan 16, 44, 164–6, 275, 278 Jepma, Catrinus 220, 251 Johnson, F.R. 49, 56, 205, 214 Johnston, Robert A. 111, 136 Joint implementation (JI) 222–3, 226, 238, 240, 244, 247 domestic governance 235–7 Kahn, M. 5, 57 Kahneman, D. 48, 57

Kain, J.F. 45, 57 Kanninen, B. J. 46, 57 Kelly, Cathleen 220, 249, 251 Kemp, R. 76, 99 Keohane, N.O. 78, 99 kerosene, 262, 284 Kerr, Suzi 218–53 Khanna, M. 125, 136 Kiker, C.F.126, 136 Kirchhoff, S. 215 Kline, S. 95, 99 knowledge accumulation 66, 85, 86, 95 creation 63, 64 depreciation 91 spillover 92 Kobe Conference, Japan 44 Kolar, Jan 171, 185 Kolk, A. 125, 136 Kooten, G. Cornelis van 101–38 Korea 275, 278 Koskela, E. 115, 136 Krcmar, E. 111, 115, 136 Kristoffersson 208, 215 Krosnick, J.A. 17, 57 Krupnick, A.1, 57, 203, 212, 215 Krutilla, J.V. 189, 215 Kyoto Protocol 221, 241 emission reductions 218, 223–4, 248 targets 242–3 treaty on climate change 75 labor productivity 79 laboratory experiments 28 LaFrance, J.T. 50, 57 lake systems 127 land 104, 244 development projects 101, 103 prices, rural and urban 108 speculators 106, 109 use 102 change 114 decisions 101–38, 131 planning 111 spillover 125–30 values 113 landowners private 101, 128, 129 rural 109 subsidy to 118

Index landscape, natural 101, 205 Lant, C.R. 133, 136 Lapping, M. B. 108, 136 Larson, D.M. 24, 50, 57 Latin America 130, 144 Lawn, P.A. 152, 153, 185 Lawson, Karen 230, 252 Layton, D.F. 26, 27, 57 leaching onto land, harmful 274 lead 143 leakage 234, 250 learning-by-doing 64, 65, 67, 6, 94, 95 Leeworthy, V.R. 50, 57 legislation for endangered species 128 right-to-farm 108, 131 Leining, Catherine 218–53 Leon, C. J. 210, 215 Leonard, D. 123, 137 life expectancy 177 lifestyle consumers 106 light-rail transit 113 Lisansky, J. 109, 137 litigation 17 on emission standards 4 exemptions for farming108 litter 151 livestock, intensive farming 267 living, higher-density 112 logging 127–9 Loomis, J.B. 204, 206, 215 Lopez, R.A.105, 137 Louviere, J.J. 48, 57 Maastricht Treaty 191 MacNair, D. 49, 58 Magat, W.A. 76, 99 Magnussen, K. 205, 215 malaria, virulent forms 282 Mäler, K.G. 50, 58 Mansfield, C.1, 29, 45, 58 manufacturing 264, 280–81 marginal land 274–5 Markandya, Anil 140, 185 market failures 108–14 forces 63, 113 incentives 125 and non-market goods 2, 30, 31, 32

295

prices 45, 256, 258, 275 simulated 28 marketing 199 research 25, 40 Marrakesh Accords 223, 225, 226, 248 material flows (MF) 163, 167, 177, 178, 181 Matthews, Emily 164, 165, 167, 177, 185 McConnell, K.E. 21, 32, 46, 50, 58, 115, 137 McFadden, D. 23, 58 McLeod, D. 212, 215 McMillan, D.P.112, 137 Measure of Economic Welfare (MEW) 151 Meyer, A.L. 133, 137 microeconomic data 4 milk 159 Milliman, S.R. 76, 77, 99 mineral resources 159, 164 Mishan, Ezra J.154, 185 Mitchell, R.C. 2, 6, 7, 46, 58 Moffatt, I. 152, 155, 185 monetary incentives 28, 29 indicators 140, 141–58 valuation 162, 168 for health impact 203 of sustainability 174 monoculture 274 Monongahela River 7, 11, 46 Montrose chemical company 17–23, 39 Morikawa, T. 23, 48, 58 mortality and morbidity risks 197, 203 motor vehicles 285 mountain systems 127 Mullins, Fiona 220, 252 Myers, N. 287 Nash, Chris 187, 261, 282 national income 168, 171, 175 National Institute of Economic and Industry Research 287 National Oceanic and Atmospheric Administration 14–17, 35, 44, 46–7, 50, 58 National Opinion Research Center (NORC) Study 17

296

Environmental and resource economics 2004/2005

national parks 205 National Survey of Fishing and Hunting 4 natural capital 139, 140, 162 depreciation 146–7, 176 gas 143, 148, 162 habitat 110 land, scarcity of 130 resources 23, 165, 259 damage 12, 14, 47, 177, 190, 267, 285 exploitation 145 Natural Resource Damage Assessment (NRDA) 1, 190, 198, 200 Nature Conservancy 110 nature protection policies 126–7 Navrud, Ståle 189–217 Nelson, A.C. 137 Nelson, R.R. 96, 99 Nentjes, A. 220, 252 Netherlands 112, 152, 170–73, 178 empirical studies 164–6 nature reserves 211 valuation 207 Neumayer, Eric 139–88 New Zealand fisheries 248 New Jersey law 109 nickel 143 Nickerson, C.J. 107, 111, 137 NIMBY response to land use 108 nitrogen oxide 150 noise 106, 108, 151, 196 insulation 196 non-commercial use of forestry 127 non-market evaluation 33, 34, 41 Nordhaus, William D. 91, 93, 99, 150, 176, 186 Norway 206, 275, 278 nuclear energy 160, 181, 238, 249 ocean fishing 159 OíConnor, Martin 171, 178, 186 Odum, H. T. 141, 159, 186 Office of Management and Budget (OMB) 44, 58 oil 13, 43, 143, 155, 156, 159 and gas wells 161 prices 155 recovery and clean-up 43

reserves 148 spill 34, 35, 38, 42, 47 tankers 36, 42, 50 Oil Pollution Act, 1990 14, 44, 50 open space 103, 108, 109, 114, 131 Oregon farmland conversion restriction 112 riverfront improvement 29 organic compound emissions, volatile 173 farming techniques 268 Organisation for Economic Co-operation and Development (OECD) 140, 144, 186, 192, 215, 222, 227, 252, 255–6, 287 Orr, L. 71, 99 outdoor recreation 7 owls, spotted 121 owner-occupied housing 4 ozone layer 169, 173 Parkinson, S. 220, 252 Parks, Peter J. 101–38, 111, 115, 133, 137 Parry, I. W. H. 62, 71–5, 93, 99 patents 64 Pearce, David W. 142, 186 peat meadows in Netherlands 207 Peiser, R.B. 114, 137 peregrine falcons 20 Permanent Cover Program, Saskatchewan, Canada 102 permits 8, 77 tradable 76, 78, 235, 241, 255 pesticides 262, 268, 283 overuse 254 Pezzey, John C.V. 141–2, 152, 186 phospate 143 photovoltaic generators 161 Pieters, J. 268, 288 Pigouvian tax 62, 72, 73, 74 plant death in oil spills 42, 43 poaching 127 Poe, G.L. 12, 47, 58 policy instruments 76–8, 94–5 pollutants 17, 150–51, 163–4 polluter-pays principle 255 pollution 63, 268, 282 control 71, 72, 73, 75, 93–4, 181

Index incidents 190 levels 164 polychlorinated biphenyl (PCB) 17, 20 poplar plantations 120, 128 Popp, D. 62, 90, 91, 99 population 6, 80, 103, 105, 159, 246 Porter, G. 278, 288 Porter, M.E. 97, 99 Portney, P.R. 44, 58 Portugal, Luis 288 preference analysis 9, 11, 23, 31, 33, 49, 51 research 33, 41 Pressey, R. L. 127, 137 prices and incomes 31, 41 Prince William Sound 18 Proops, John L.R. 149, 186 property taxes 45, 63, 109, 131 Protected Areas Strategy, British Columbia 126 Prototype Carbon Fund 223 psychology surveys 40, 41 Public Area Recreation Visitorsí Survey (PARVS) 35 public health and the environment 64 Putnam, Robert D. 180, 186 Quebec, farmland conversion restriction 112 R & D 71, 74, 90, 93, 280, 281 in energy sector 91 environmental 94 industrial 64, 65, 69 investment in 79, 96 productivity 79 service sector 92 rail transport subsidies 281 rain forests 127 Randall, A. 2, 6, 7, 46,58 Random utility model (RUM) 27 rating issue 26 Ready R. 202, 207, 212, 216 Reaves, D.W.47, 59 recreation 36, 110, 205 activities 196, 201 sites 8, 46 in US states 207 recreational value, loss of 189 recycling 164

297

Rees, William 162, 186 reforestation option 163 regulations, command-and-control 125 rent-seeking 4, 284 Rentz, H. 252 reproduction in fish and birds 20, 21, 23 Requate, T. 78, 99 research 63, 247–8 residential land use 101, 103, 110 resources 102, 177 consumption 149, 150, 155 data 179 depletion 83, 149, 151, 155–7 exploitation 149, 155 extraction 143, 149, 156 non-renewable 164 respiratory symptoms 202 revealed preference data 3, 7 rhinoceros, endangered 127 Ridley M.A 220, 252 rights to traverse farmland 110 right-to-farm laws 109 risk sharing 237 riverfront improvement, Oregon 29 roads 103, 105 congestion 268 development 113, 133 transport subsidies 254, 268, 281, 282 Roe, B. 25, 59 Rollestone, B.S. 112, 137 Romer, D. 63, 100 Romer, P.M. 79, 100 Rosenberg, D. 95, 152, 187 Rowe, R.D. 46, 59, 190, 216 Roy, Rana 24, 282, 286, 288 Rozan, A. 208, 216 rubber 159 Ruijgrok, E. C. M. 211, 216 rural land 101, 130 preservation 105, 108–10 rural-nature interface 102 rural-urban area 104, 105, 106, 113 Russia, ‘hot air’ emissions 228, 246 Santos, J.M.L. 216 salinization of soil 266, 282, 283 salmon fishing 49, 206 Samuelson, P.A. 40, 41, 59, 117, 137

298

Environmental and resource economics 2004/2005

Santopietro, George D. 143, 187 Saskatchewan, farmland conversion restriction 112 Saudi Arabia 147–9 savings and tax 77, 141–50 Scarpa, R. 200, 216 Schipper, Y. 205, 216 Schmidt-Bleek, Friedrich 163, 187 schools 103, 105 Schulze, W.D. 29, 46, 59 scope effects 14–17, 20–21, 25, 39, 48 Scotland 152 sea bass (fish) 20 sea fences, floating 43 Seebregts, A. 96, 100 Sefton, J.A. 145, 187 Segerson, K. 125, 137 self-reliance, intraregional 162 Serôa da Motta, Ronaldo 147, 187 set-asides 128 sewage treatment plants 103, 105, 133, 196 Shi, Y. J. 105, 137 shipbuilding industry 263, 281 Shrestha, R. K. 205, 216 Sieg, H. 45, 59 silviculture 127 see also forests Sinclair, A.R.E. 127, 137 Sloan, F.A. 51, 59 slough draining 110 smell 106, 108 Smith, V. Kerry 1–60, 113, 133, 137, 205, 206, 210, 216 smoking 46 Smulders, S. 62, 80, 81, 83, 86, 88, 89, 93,100 social scientist teams 40 social welfare 80, 112, 142 soil 150, 164 contamination 173 degradation 280 erosion 164, 195, 268, salinization 282 solar energy 156, 161, 181 solid wastes 151 Solow, Robert 87, 141, 187 Southern California Bight 17 space industry 281 Spain 7, 207, 278 Spangenberg, Joachim 163, 164, 187

Spanish Peaks Wilderness Area 7 species extinction 126 spillover, high 93 sprawl 103 SS (strong sustainability) 139, 140, 175 stands of timber 124 Stavins, R.N. 125, 126, 137 steady-state solution 84 steel industry 46, 227–8, 264, 281 Steenblik, R.P. 254, 274, 288 Steenge, Albert E. 140, 187 Stockhammer, E. 152, 155, 157, 187 Sturtevant, L.A. 205, 216 subsidies 76, 77, 254, 258–61 for the environment 255–7 environmentally harmful 286 measurement 260 in OECD countries 261–2 reform 282–6 sulphur dioxide 150 Superfund legislation, USA 14, 44 Sur, Mona 282 sustainability 139–88 gap 169, 170, 178 measured by land area 158–60 by weight 163–7 sustainable development 139 forest management 130 Sustainable European Research Institute, (SERI) Vienna 165 Sustainable Human Development Index (SHDI) 165 Swait, J. 48, 60 Swallow, S.K. 115, 121, 123, 124, 126, 138 Sweden 110, 125, 152 Forestry Act 129 Switzerland 275 Takeuchi, K. 16, 60 Tamborra, M.191, 216 tax 70, 76, 78, 157, 255, 256, 261, 286 assessments, preferential 109, 110 base, local 112 concessions 258 exemptions 284 incentives 109 subsidy schemes 125 surcharge 37

Index

299

and tax breaks 118 taxpayers 42, 43 technical progress 175 technological change 61–97 progress 91, 93, 94, 145 spillovers 90, 91 technology environmental 71 transfer 219, 235 telephone surveys 47 textile industry 281 Thailand 152 thermodynamics 141 Tiebout, C.M. 138 Tietenberg, T.H. 191, 216, 220, 252 tiger, endangered species 27 timber 117, 130, 131, 159, 189 commercial 116, 119, 120–21, 123 harvest 114, 115, 116, 124 and non-timber benefits 132 poaching 127 tin 143 Titman, C.M. 138 Tjahjadi, B. 165, 187 tourist revenue 127 toxic waste 280 tradable permits 78 units 245–7 trading system in EU 236 traffic noise levels 197 trains 113 transaction costs 232–3, 243 transport 23, 113, 191, 199, 262–3, 281–2 subsidies 261 travel cost 26, 38, 51, 196 demand 9, 10 trees 12, 103, 268 clearing 129 endangered species 110 plantations with native species 120 TRIAD zoning 127–8, 132 tropical forestlands 126, 133 trout 121 Trust for Public Land, USA 110

unemployment benefits 261 ungulates, wild 121 United Nations Development Programme 165 United Nations Environment Program 192 United Nations Framework Convention on Climate Change 222–3, 253 United States 130, 278 Acid Rain program 220 Congress 176 Department of Justice 35 emissions 228 empirical studies 164–6 Environmental Protection Agency 44 Farm Bill 111 National Oceanic and Atmospheric Administration 198 Oil Spill Act 201 upland burning 110 urban containment 112 economics 102 growth 105, 112, 131 land 101, 106, 108, 130 life, disamenities 151 sprawl 109 urban-rural interface and land use 103–08 USDA Forest Service 189

UK Environment Act 191 United Kingdom 152, 170

Wackernagel, Mathis 158–60, 181, 187 Wallace, N.E. 112, 138

value transfer and environmental policy 189–217 Van Beers, C. 284, 288 Van den Bergh, Jeroen C.J.M. 160, 162, 177, 187 Van der Gaast, Wytze 250, 253 Van Kooten, G.C. 115, 119, 126, 129, 133, 138 Varian, H.R. 41, 60 vegetation 249 Verdier, T. 81, 100 Vincent, Jeffrey, R. 123, 138, 147, 187 Vitousek, Peter M. 140, 159, 187 Von Haefen, R.H. 50, 60 Von Weizsäcker, Ernst 187 Vossler, C.A. 12, 28, 29, 60

300

Environmental and resource economics 2004/2005

Walsh, R.G. 201, 205, 217 Washington State, USA 42 wastage 282, 285 waste 163, 164, 177, 262 disposal sites 196 generation 177 products 159, 163 treatment 190, 205 water 103, 150, 265 abstraction 261 flows 164 pollution 151, 193, 194, 280, 285 purifiers 196 quality 7, 32, 46, 132, 205, 206 subsidies 282–3 supply, potable 285 Waugh, F. V. 60 Weitzman, Martin L. 97, 100, 145, 188 welfare gains 71–8, 93 loss, oil spill 13 measures 151 wetlands 103, 114, 120, 205, 268 loss of 151 Wetlands mitigation banking (WTB) 111 Wetland Reserve Program, USA 111 white croaker (fish) 20 Whitehead, J.C. 60 whitewater rafting 33 Whittington, D. 217 Wicksell model 102, 103, 106, 133 Wiebe, K. 109, 138 Wigley, M.L. 61, 100 wilderness 7, 13, 47, 48, 101 wildlife 124 corridor 110 deaths in oil spills 43 habitat 121, 126, 129, 268

harm to 274 loss (birds) 13 Willey, David 162, 188 Willig, R.D. 50, 60 willingness to pay (WTP) 20–21, 25–9, 31, 33, 37, 49, 98, 108 and willingess to accept 46 Wilson, B. 125, 127, 138, 152, 155, wind energy 161, 181 Wirl, F. 230, 253 Woerdman E. 220, 253 wood products 128 wool 159 World Bank 143–50, 155, 176, 188, 192, 285, 288 Environment Department 142 Protoype Carbon Fund 223 World Business Council on Sustainable Development 181 World Commission on Environment and Development 138 World Summit on Sustainable Development 286 World Wide Fund for Nature (WWF) 130, 188 Wright, B.D. 64, 100 Wuppertal Institute for Climate, Environment and Energy 165 Zerbe, R.O. 76, 100 Zhang, X. 25, 60 Zhang, Z. 220, 253 zinc 143 Zolotas, Xenophon 151, 188 zoning 102, 112–14, 123, 133 of protected ares 126, 128 regulations 111, 129 TRIAD 127–8