Changing Stocks, Flows and Behaviors in Industrial Ecosystems

Changing Stocks, Flows and Behaviors in Industrial Ecosystems

Changing Stocks, Flows and Behaviors in Industrial Ecosystems Edited by

Matthias Ruth University of Maryland, College Park, Maryland, USA

Brynhildur Davidsdottir University of Iceland, Reykjavik, Iceland

Edward Elgar Cheltenham, UK • Northampton, MA, USA

© Matthias Ruth and Brynhildur Davidsdottir 2008 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 The Lypiatts 15 Lansdown Road Cheltenham Glos GL50 2JA UK Edward Elgar Publishing, Inc. William Pratt House 9 Dewey Court Northampton Massachusetts 01060 USA

A catalogue record for this book is available from the British Library Library of Congress Control Number: 2008927701

ISBN 978 1 84720 740 1 Printed and bound in Great Britain by MPG Books Ltd, Bodmin, Cornwall

Contents List of figures List of tables List of contributors Foreword by John R. Ehrenfeld Acknowledgments List of abbreviations

vii viii ix xv xviii xix

PART I BACKGROUND AND CONCEPTS 1 Background and concepts: an introduction Matthias Ruth and Brynhildur Davidsdottir 2 Beyond a sack of resources: nature as a model – the core feature of industrial ecology Ralf Isenmann, Christoph Bey and Martina Keitsch

3

7

PART II STOCKS AND FLOWS DYNAMICS 3 Stocks and flows dynamics: an introduction Thomas E. Graedel 4 Dynamic modeling of material stocks: a case study of in-use cement stocks in the United States Amit Kapur and Gregory A. Keoleian 5 The economic dynamics of stocks and flows Brynhildur Davidsdottir and Matthias Ruth

33

40 54

PART III AGENT-BASED ANALYSIS OF DYNAMIC INDUSTRIAL ECOSYSTEMS 6 Agent-based analysis of dynamic industrial ecosystems: an introduction Marco A. Janssen 7 Changing a firm’s environmental performance from within Clinton J. Andrews v

77 82

vi

Contents

8 Managing energy futures and greenhouse gas emissions with the help of agent-based simulation David F. Batten and George V. Grozev

101

9 Organizational dynamics in industrial ecosystems: insights from organizational theory Jennifer Howard-Grenville and Raymond Paquin

122

10 Changing stocks, flows and behaviors in industrial ecosystems: retrospect and prospect Brynhildur Davidsdottir and Matthias Ruth

140

Index

147

Figures 2.1 Scientific profile of industrial ecology in terms of an epistemological architecture 3.1 Energy stocks and flows and related spheres of action 3.2 Material stocks and flows and related spheres of action 4.1 Generic cement life cycle 4.2 Consumption of cement in the United States 4.3 Cement end-use market in the United States in the year 2003 4.4 Cumulative net addition of cement stock in the United States over the time period 1900–2005 5.1 Process nodes and flows of energy and materials in the US pulp and paper industry 5.2 Schematic representation of the capital stock 7.1 Structure of PolyModel 7.2 Formal reporting hierarchy versus social network 7.3 Cost and price trends in the polymer processing industry 7.4 Historical simulation of revenues, expenses, and profits: sample screen shot from computer model 8.1 Entropy and the economic process 8.2 The NEM as a complex adaptive system 8.3 An overview of NEMSIM 8.4 NEMSIM regional summary graph for GHG emissions 8.5 DG perturbed demand and price example 8.6 NEMSIM vertical bars of clusters of DG next to load centers

vii

12 34 35 44 45 46 49 57 59 87 89 92 93 102 107 112 116 117 118

Tables 2.1 Classification framework for understanding nature 3.1 Examples of stimuli and actions related to energy and material flow analyses 3.2 Published analyses of studies in materials actions and stimuli 4.1 Range of service lifetimes for each infrastructure use 4.2 Estimation of parameters of the lifetime distribution 7.1 Summary of results (per cent change from base case after reaching steady state) 9.1 Parallels between industrial ecology concepts and tools and those within organization theory 9.2 Network attribute definitions 9.3 Network concepts, explanations, and visualizations

viii

17 36 38 42 48 94 124 131 132

Contributors Clinton J. Andrews directs the planning program at Rutgers University’s Edward J. Bloustein School of Planning and Public Policy where he is an associate professor. He was educated at Brown and MIT as an engineer and planner. Previous experience includes working in the private sector as a design engineer and energy technology assessor, helping to launch an energy policy project at MIT and helping to found a science policy program at Princeton. At Rutgers, he has launched initiatives in energy policy, green building and innovation studies. His books include Humble Analysis (Praeger, 2002), Regulating Regional Power Systems (Quorum, 1995) and Industrial Ecology and Global Change (Cambridge, 1994, co-edited with R. Socolow, F. Berkhout and V. Thomas). David F. Batten joined CSIRO in 1970, his work in that period culminating in a PhD in regional economics and leadership of several regional development projects. In 1986, he moved to a Chair in infrastructure economics at the University of Umeå in Sweden. From 1991 to 1995, he also held the position of Professorial Fellow at the Institute for Futures Studies, a scientific think-tank in Stockholm. During his years in Sweden, Dr Batten managed teams exploring the eco-efficient design of infrastructure systems involving transport, energy, water, waste and community planning. Since returning to Melbourne, he has managed the Australian office of the TEMAPLAN Group, an international consortium of small, scientific consultancy firms specializing in industrial ecology for industry and government. He has undertaken several eco-industrial projects and others in related fields like agent-based simulation (e.g. NEMSIM), and has authored or co-edited ten books (e.g. Discovering Artificial Economics: How Agents Learn and Economies Evolve). In 2002, he was invited back to CSIRO to assist with several “One-CSIRO” activities, and is currently involved in projects funded by CSIRO’s Energy Transformed Flagship and its Centre for Complex Systems Science, as well as an industrial symbiosis project for the Victorian Government. In 2005, he was invited to join CSIRO’s Alternative Transport Fuels project as the team’s economist and international analyst. Christoph Bey has first degrees in biology and in international management and took a PhD in ecological economics at the University of Edinburgh, ix

x

Contributors

Scotland. He is a member of the research lab CERMAD for sustainable management at the ESCEM School of Management Tours/Poitiers, France, where he teaches strategic management and sustainable development. Christoph is also a research fellow in strategic management at the University of Poitiers. Other than industrial ecology, his research interests are in CSR and corporate governance. Brynhildur Davidsdottir is an Associate Professor of Environment and Natural Resources at the University of Iceland and the Director of the Master’s Program on Environment and Natural Resources at the University of Iceland. Before relocating to Iceland Dr Davidsdottir was an adjunct professor at Boston University and an associate at Abt Associates, an environmental consulting firm in Cambridge, Massachusetts. Her research focuses on dynamic modeling of social and environmental systems, with a particular focus on the dynamics and social and environmental implications of technological change in the context of transitioning towards alternative energy sources, sustainable energy development and the dynamics of industrial change in energy-intensive industries and the impact of a carbon constrained world on their behavior. John R. Ehrenfeld is Executive Director of the International Society for Industrial Ecology. He currently serves on the Council of the Society for Organizational Learning. His current research focus is on sustainability and culture change. A book on this subject is forthcoming in 2008 from the Yale Press. He retired in 2000 as the Director of the MIT Program on Technology, Business and Environment. In October 1999, the World Resources Institute honored him with a lifetime achievement award for his academic accomplishments. He holds a BS and ScD in Chemical Engineering from MIT, and is author or co-author of over 200 papers, books, reports and other publications. Thomas E. Graedel is Clifton R. Musser Professor of Industrial Ecology, Professor of Chemical Engineering, Professor of Geology and Geophysics, and Director of the Center for Industrial Ecology at Yale University, New Haven, CT, USA. His academic degrees are BS, Washington State University; MA, Kent State University; MS, PhD, University of Michigan. Professor Graedel was elected to the US National Academy of Engineering for “outstanding contributions to the theory and practice of industrial ecology, 2002”. His research is centered on developing and enhancing industrial ecology, the organizing framework for the study of the interactions of the modern technological society with the environment. His textbook, Industrial Ecology, co-written with B.R. Allenby of AT&T, was the first book in the field, and is now in its second edition. It, and his 2005

Contributors

xi

textbook, Greening the Industrial Facility, are used for courses of the same names. His current interests include studies of the stocks and flows of materials within the industrial ecosystem and the development of analytical tools to assess the environmental characteristics of products, processes and urban infrastructures. George V. Grozev is a principal research scientist with the Commonwealth Scientific and Industrial Research Organization (CSIRO) in Australia. His background is in operations research, particularly network theory and algorithms and mathematical programming, complex systems modeling, simulation and software engineering. G. Grozev’s recent research has focused on sustainable development issues of electricity markets, including approaches for modeling and simulation of future developments in Australia’s National Electricity Market (NEM). D. Batten and G. Grozev established together the NEMSIM project that aims to develop an agentbased simulation model of the NEM as a complex system of interactions between market participants, technical infrastructures and the natural environment. NEMSIM is a project within CSIRO’s Energy Transformed Flagship Research Program that involves researchers from several CSIRO divisions. In his initial years with CSIRO, after joining the organization in 1999, Dr Grozev participated actively in the development of several simulation, optimization and mapping tools for cellular mobile networks, developed in collaboration with Telstra Research Laboratories. Previously Dr Grozev worked for Siemens Research in Australia, Collaborative Information Technology Research Institute in Melbourne and the Bulgarian Academy of Sciences in Sofia. Jennifer Howard-Grenville is an Assistant Professor of Management at the University of Oregon’s Lundquist College of Business. She studies how cultural and institutional processes constrain or advance organizational change, with a focus on changes in corporate environmental practice. Jennifer’s work has been published in a number of journals including Organization Science, Organization & Environment, Law & Social Inquiry and California Management Review. She is the author of Corporate Culture and Environmental Practice (Edward Elgar, 2007), co-author (with Thomas Graedel) of Greening the Industrial Facility (Springer, 2005) and co-editor (with Frank Boons) of a forthcoming book on social science perspectives on industrial ecology. Jennifer received her PhD at MIT, her MA at Oxford University and her BSc at Queen’s University, Canada. Ralf Isenmann studied economics, business administration and industrial engineering at the University of Kaiserslautern, Germany. He received his doctorate for a thesis on the interface between Philosophy of Science and

xii

Contributors

Industrial Ecology. Currently, he is a senior lecturer at the Faculty of Business Studies and Economics at the University of Bremen, Germany, teaching sustainable management, business information systems, technology management and entrepreneurship. He is a member of the Institute for Project Management and Innovation (IPMI), and of the Research Center for Sustainability Studies (artec), University of Bremen, Germany. Marco A. Janssen has a formal training in operation research and applied mathematics (PhD, 1996, University of Maastricht, the Netherlands). He initially worked on integrated models for climate change policy in the Dutch National Institute for Public Health and the Environment, before moving to research positions at the Vrije Universiteit in Amsterdam and Indiana University. He moved in 2005 to Arizona State University and is faculty within the School of Human Evolution and Social Change and the School of Computing and Informatics. His research focuses on the coevolution of cognitive, institutional and ecological processes. He is interested in how people, their institutional rules and the environment they live in fit together in the past, the present and the future and from local to global scales. He is mainly using computational models, like agent-based models, in combination with laboratory experiments, surveys, case study analysis and stakeholder workshops. Amit Kapur is a Consultant with Five Winds International. Amit specializes in the emerging field of industrial ecology. His work focuses on life cycle assessment (LCA) and material flow analysis (MFA). Amit’s project experience in LCA includes work with clients such as International Zinc Association, Shell, and Shaw Industries. Prior to joining Five Winds International, Amit was a Research Fellow at the Center for Sustainable Systems, University of Michigan. His post-doctoral research was focused on sustainable concrete infrastructure materials and systems. Amit’s doctoral dissertation focused on building resource flow scenarios to estimate future developments at the regional and global level with respect to resource use, environmental burdens and technological changes. Amit is a member of International Society of Industrial Ecology (ISIE) and has authored numerous peer-reviewed publications. Dr Kapur holds a PhD from the Center for Industrial Ecology, Yale University and an MS in Environmental Engineering from Purdue University and BE in Civil Engineering from Delhi College of Engineering. Martina Keitsch is a senior research advisor at the Oslo School of Architecture and Design, Norway. She has a doctorate in philosophy in the field of environmental ethics and aesthetics. She has experience for over ten years in sustainable development and industrial ecology, mainly

Contributors

xiii

in eco-industrial design, CSR, design theory, environmental ethics, systems methodology, theory of science and aesthetics. From 1997 to 2007 she was a researcher and lecturer at the Norwegian University of Science and Technology (NTNU) in Trondheim (Program for Industrial Ecology and Department for Product Design). From 1998 to 2000 she was the leader of an interdisciplinary research project on sustainable wastewater systems design in the Trondheim region. From 2002 to 2004 she was Post-doc at NTNU working on sustainable development and industrial ecology. Gregory A. Keoleian holds a PhD in Chemical Engineering from the University of Michigan. He serves as co-director of the Center for Sustainable Systems and is an Associate Professor in the School of Natural Resources and Environment (SNRE) at the University of Michigan. His research focuses on the development and application of life cycle models and sustainability metrics to guide the design and implementation of products and technology. He has pioneered new methods for life cycle design, life cycle optimization of product replacement, life cycle cost analysis and life-cycle-based sustainability assessments. Systems studied include alternative vehicle technology, infrastructure and buildings, renewable energy systems such as building integrated photovoltaics and willow biomass electricity, information technology, food and agricultural systems, household appliances and packaging alternatives. Professor Keoleian teaches interdisciplinary graduate courses on Sustainable Energy Systems and Industrial Ecology. He is co-coordinator of the Engineering Sustainable Systems dual degree program between Engineering and SNRE and codirects the Graduate Certificate Program in Industry Ecology. Raymond Paquin is an Instructor at the University of Oregon’s Lundquist College of Business and a doctoral candidate in organizational behavior at the Boston University School of Management. His research interests include institutional change processes in the emergence of new organizational practices in environmental and sustainability areas. Raymond holds an MA in Education from Virginia Tech and a Bachelor of Music from North Carolina School of the Arts. Matthias Ruth is Roy F. Weston Chair in Natural Economics, founding Director of the Center for Integrative Environmental Research at the Division of Research, Director of the Environmental Policy Program at the School of Public Policy and Co-Director of the Engineering and Public Policy Program at the University of Maryland. His research focuses on dynamic modeling of natural resource use, industrial and infrastructure systems analysis, and environmental economics and policy. His theoretical

xiv

Contributors

work draws heavily on concepts from engineering, economics and ecology, while his applied research utilizes methods of non-linear dynamic modeling as well as adaptive and anticipatory management. Professor Ruth has published nine books and over one hundred papers and book chapters in the scientific literature. He collaborates extensively with scientists and policymakers in the USA, Canada, Europe, Oceania, Asia and Africa.

Foreword John R. Ehrenfeld With this two-volume collection, the editors, Matthias Ruth and Brynhildur Davidsdottir, and the other contributors have made an important and substantial contribution to the still evolving field of industrial ecology. This first volume focuses on the basic notions of industrial ecology, on the dynamics of material flows, and on the role of agents in dynamic industrial ecosystems. In the more than ten years that have transpired since the emergence of the idea that economic/industrial systems generally exhibit features analogous to natural ecosystem, the field has taken root. Industrial ecology now has associated with it activities in many universities, consultants with programmes based on industrial ecological principles, and applications of these principles showing up in corporate strategy, product design and public policy. The key principles spring from the above-mentioned ecological analogy and include such notions as industrial metabolism (flows of energy and materials), loop closing and symbiosis, all of which mimic forms and processes found in healthy ecologies. The stressed term above, healthy, lends a normative dimension to the field, beyond the merely descriptive character of analogies. Environmental management and its successor concept, sustainability, have become firmly embedded in high-level societal activities in virtually every economic sector and industrialized nation. The relevance of these terms is tied to a stillgrowing consciousness of the fragility of the Earth’s ecosystem and its criticality as the primary life support system of our species and indeed all life. International consensus about global warming and its impact on climate has now heightened interest in acting to preserve the environment for the present and for future generations. Among many potential pathways toward sustainability, one stands out as the choice of most industrial and governmental strategies: eco-efficiency. Eco-efficiency, the idea of providing more value for less impact, is contained in many other prescriptive statements, such as dematerialization, decarbonization, detoxification, factor X reduction, cradle-to-cradle, and so on. Healthy ecosystems are naturally “eco-efficient”. They recycle the nutrients found in their local environment by closing material loops. xv

xvi

Foreword

Detritivores turn the wastes produced in the food web into nutrients for species in other places in the web. The source of energy is renewable solar energy. It is only a very small jump to get from this observation to a normative possibility for industrial ecology: produce a more sustainable world by designing economic/industrial systems to look and behave more like ecosystems. This possibility has taken hold in several important areas, for example, in the design of technological artifacts (design for environment) and in the design of industrial organization (eco-industrial development). In both of these cases, analytic and design tools, based on material and energy flows, have been developed and applied. Other analytic models and tools have been developed for larger systems, such as national or regional material economies (flows), but these have not achieved the level of design applications as the above two cases. It would seem, based on a patently unscientific assessment by this author, that the “simpler” the system, as in product systems, the more the ideas of industrial ecology have found their way into practice. Simple in this sense has several aspects, temporal and organizational. Products generally have shorter lifetimes than industrial systems especially looking at common consumer products such as automobiles, mobile phones or computers. The present generation of industrial ecological models and tools largely springs from relatively static analyses. The assumptions that are made in applying the tools generally assume that the context of the analysis approximates the conditions during the actual lifetime of the system under the analyst’s lens. These tools also generally do not take into account sociological and organizational processes that are involved in putting the prescriptions into play. Again, for product systems, this limitation is not critical as to technical considerations, although it is part of the reasons that the outcome the designer or strategist had in mind may turn out differently. Furthermore, these first generation models are almost exclusively based on assumptions of linearity with respect to the technical components and on normal rationality with respect to the human elements, in those cases where consideration of actor behavior enters the analytic framework. And finally, much of the work reflects the reductionist nature of the technical disciplines on which industrial ecology rests. This statement should not be read as a criticism of this sociological fact, but merely as an argument for expanding the intellectual basis for what has been the mainstream of research and analysis within industrial ecology. Readers who are familiar with my recent writings know that I believe that the limits of the present linear models, including those representing ecosystem processes, correspondingly limit the ability of the workers in the field to muster convincing arguments that industrial ecology can be a powerful

Foreword

xvii

new frame for thinking about and acting towards sustainability. Ecoefficiency thinking is extremely important in revealing ways to stop and even reverse the apparently inexorable trajectory towards breakdown and destruction of the natural world with consequent immense potential social implications. But eco-efficiency, like efficiency in any setting, ignores possible absolute limits to growth and also the existence of feedback loops internal to the socio-economic system. William Jevons, writing in 1865, noted that coal consumption in England eventually rose in volume even after the large increase in efficiency produced by Watt’s steam engine. Jevons’ notion lives today in the current notion of the rebound effect, which implies that eco-efficiency (creating wealth in the process) will produce more investment and more consumption, eventually outstripping any gains from technological improvements. If one stops for a moment and thinks about the more complex situations mentioned above, the next generation of analytic and design tools will have to incorporate models of processes that more realistically reflect the messy way that the world does, unfortunately for analysts, really work. As the editors of this volume point out, this requires that new ideas must be injected into industrial ecological thinking and research. For example, ways to account for changes in material stocks over long periods are now being incorporated into frameworks for analyzing material flows in large and long-lasting systems as several chapters indicate. As the title indicates, the editors have deliberately taken on the challenge of extending these primarily linear models to account for both temporality and for human behavior.

Acknowledgments This book is one of two which are dealing with the dynamics of industrial ecosystems. Much like those systems, the books are the product of organic growth and development, with endless efforts by individuals and organizations loosely connected, but often working towards common goals. Special recognition in that endeavor needs to be given to Bari Levine for her relentless work helping to organize the manuscript and bringing it all together into a final publishable form and to Tara Gorvine at Edward Elgar’s Massachusetts office, for her attention to detail as well as the big picture. Last but not least, our personal appreciation and thanks go to the book’s contributors for their creativity and hard work, and to our families and friends for their love and patience throughout the years.

xviii

Abbreviations ABS ASCE BEA C&D CAS CDF CO2 CSIRO DG E/MFA EMCAS GHG IE ISIE LCA MAS MFA NEM NEMMCO NEMSIM NETA NTNU OTC PD PVC REC SFA SFE SNA USGS WUR XML

Agent-based simulation American Society of Civil Engineers Bureau of Economic Analysis Construction and demolition Complex adaptive system Cumulative probability distribution function Carbon dioxide Commonwealth Scientific and Industrial Research Organization Distributed generation Energy and material flow analyses Electricity Market Complex Adaptive System Greenhouse gases Industrial ecology International Society of Industrial Ecology Life cycle analysis/assessment Multi-agent simulation Material flow analysis (Australia’s) National Electricity Market National Electricity Market Management Company National Electricity Market Simulator New Electricity Trading Arrangements Norwegian University of Science and Technology Over the counter Prisoner dilemma Polyvinyl chloride Renewable energy certificates Substance-flow analysis model Supply function equilibrium Social network analysis United States Geological Survey Waste fiber utilization rate Extensible markup language

xix

PART I

Background and Concepts

1. Background and concepts: an introduction Matthias Ruth and Brynhildur Davidsdottir The use of materials and energy for production and consumption leads to changes in the sizes and compositions of stocks of substances in nature, including ores in the ground, synthetic materials in the water, and greenhouse gases in the atmosphere. In addition, effects can be found in stocks in the human domain, machines in factories, consumer goods purchased by households, and the knowledge embedded in materials, libraries, and institutions. Flows cause changes in stocks connecting the different subsystems of the environment, economy, and society with each other. It is those changes in stocks, the associated changes in flows, and the interdependent changes in society, economy, and environment to which the chapters of this book are dedicated. Their viewpoint is decidedly one that considers stocks, flows, and behaviors from the perspective of industrial ecology, a newly emerging area of research attempting to provide a consistent material and energetic description of human production and consumption processes in the larger context of environmental and socioeconomic change. In this volume, specific attention is given to changes in production processes and changes in their organization at the firm, industry, and larger industrial ecosystem level, where the industrial ecosystem is conceived as the interplay of producers, consumers, and regulatory agencies that exchange materials, energy, and information with each other and their environment. A companion volume (Ruth and Davidsdottir 2009) focuses on the dynamics of regions and networks in which the material, energy, and information flows occur. Many of the processes that characterize such interplay are variable and changing over space and time: new technologies emerge and old ones are replaced, new materials and energy sources are developed, consumer needs and preferences evolve, and new resources and environmental repercussions are discovered. Regulatory interventions into material and energy use by consumers and producers alter, and often are guided by changes both at the process and larger system level. For example, new understanding of the human health impacts of a material or recognition of 3

4

Background and concepts

global environmental harm from greenhouse gas emissions have prompted restrictions on the use of select substances, promotion of particular technologies, or implementation of incentives to reduce emissions. Typically neither the activities of producers, consumers, and regulatory agencies occur immediately in response to each other or to changes in the environment, nor are they typically in direct proportion to each other. Detection of environmental insult, for example, can take years or decades. Changing technologies, behaviors or institutions, likewise, is never instantaneous, and if and when they occur, they do so with respect to a combination of past situations and expectations of the future. The resulting time delay and often non-linear cause–effect relationships between actions and reactions in the industrial ecosystem make the understanding of their dynamics and the management of their behaviors daunting tasks. A myriad of connections between system components – producers, consumers, and the environment in interaction with each other – does not simply result in the complexity of industrial ecosystems; instead, such complexity is fundamentally and inherently related to our ability to comprehend and explain them through multiple disciplinary perspectives required to encompass the relevant system features. For example, an engineering perspective will provide valuable information on material and energy conversions, gaps that may exist between existing practices and ideal conversion processes, and alternatives to close those gaps. Economic, legal, and institutional analyses will be able to provide insights into opportunities and constraints for closing such spaces. Biological information will help quantify impacts of material and energy use on the living environment. Public health insights can be used to assess implications for public and community health. Computer modeling may be required to relate the various pieces of information on the dynamics of industrial ecosystems to each other and make that information relevant to decision makers in the public, private, and non-profit sectors. How that information is perceived and acted on, in turn, depends to a large degree on psychology, organizational structures of industry and government agencies, and the roles and responsibilities of civil society. The complexity of dynamic industrial ecosystems results from the several possible interactions at the physical and technological levels, the many pathways through which their ramifications permeate environmental, economic and social systems, and the numerous, diverse perceptions and actions of the individuals making up those systems. A simple change anywhere in an industrial ecology may be buffered and never exert larger-scale system impact, or it may ripple through the many interconnections among system components to ultimately determine new behaviors, new material and energy flows, and new feedbacks among those components. The breakthrough associated with

An introduction

5

the steam engine stimulated the industrial revolution, a combination of institutional and technical changes triggered the agricultural revolution, and the merger of information and telecommunication technology gave rise to an ever proliferating internet. Simple physical descriptions of the associated material and energy flows provide only one relevant piece of information to understand those changes. An extension of physical descriptions of systems change to encompass a broader socioeconomic and environmental context allows analysts to better comprehend bifurcations developmental trajectories. Choosing multiple disciplines means embracing the complexity of industrial ecosystems. Most importantly, though, a diverse, multidimensional approach allows for the identification of added degrees of freedom for system intervention to promote self organization and resilience outside the simply physical world. In recent years, heightened interest in the dynamics of industrial ecosystems has been developing. Scholars and practitioners, from varied backgrounds and with different purposes in mind, attempt to gain a deeper understanding of the important features and dynamics of industrial ecosystems, some through analogies, others through application of first principles, yet others through case studies and modeling-based inquiries. Given the novelty of their perspective and the diversity of their approaches, the products of their work are widely dispersed in individual book chapters and journal articles. Yet, as interest in these perspectives is growing among researchers and decision makers, the need to bring or hold the newly developing strands of interdisciplinary scholarship together, and to identify their relationships and differences, is rising. Without such an effort, continued fragmentation may result in lost opportunities to develop synergies among research programs and in reduced impact on thinking in industrial ecology, environmental research in general, and investment and policymaking. This volume covers basic and advanced analytical concepts and tools to explore the dynamics of industrial ecosystems at the firm, industry, and larger ecosystem levels. Rather than remaining theoretical and conceptual, the bulk of the material presented here makes its case through the application to very specific issues, keeping in mind both the needs for methodological advancement as well as issues of proof-of-concept and applicability. The volume is organized in three parts. This first part presents analogies and analytical concepts pertinent to understanding the dynamics of industrial ecosystems and offers a reflection on the use of those analogies and concepts, their limitations and potential extensions. In doing so, it provides both an historical and conceptual frame for the remainder of this book. The subsequent two parts focus on different analytical approaches and their application to important components and processes in industrial

6

Background and concepts

ecosystems. Each of these sections contains its own introduction, provided by one of the leaders in the respective field. Part II focuses on stocks and flows dynamics at the firm and industry level – one more of an engineering nature, the other more concerned with the economic issues of stocks and flows dynamics; however, both provide significant overlap in their concern and approach. Part III turns to the use of agent-based modeling and organization behavior theory to better understand and represent the dynamics within firms and the larger institutional environment within which they choose to use materials, energy, and technology to produce goods and services for sale to other firms and final consumers. Here, as elsewhere in the book, connections are made between those dynamics and the associated changes in environmental quality. The volume concludes by drawing on the knowledge gained in previous chapters and the various areas of research to which they are related, and it offers directions for further scientific inquiry and applications.

REFERENCE Ruth, M. and B. Davidsdottir (eds) (2009), The Dynamics of Regions and Networks in Industrial Ecosystems, Cheltenham, UK and Northampton, MA, USA: Edward Elgar.

2. Beyond a sack of resources: nature as a model – the core feature of industrial ecology Ralf Isenmann, Christoph Bey and Martina Keitsch INTRODUCTION Since its launch nearly two decades ago (Ayres and Simonis 1994; Erkman 1997), industrial ecology (IE) has grown from being just a smart “idea” (Frosch 1992: 800) to a “somewhat fuzzy concept” (Ehrenfeld 2000: 229), giving rise to a professional international society. It now constitutes a “powerful prism” (ISIE 2006) with numerous tools, studies, publications, resources and other characteristics that make it a discipline (Ehrenfeld 2000, 2001). Industrial ecology’s main goal is to study industrial systems and their fundamental linkage with natural ecosystems, thereby contributing to a more sustainable future. As an intellectual area, industrial ecology’s scientific community, with its professional academic culture, has a growing impact on governmental agendas, business applications in industry and higher education programs. As it develops into more of an institution, now is the vital time to assess industrial ecology’s disciplinary contours from a philosophical point of view, thus uncovering the constitutive characteristics in the field. Industrial ecology lays claim to an established intellectual area and a permanent form of institutional legitimacy. With a need “to improve the craft and to demonstrate the value of industrial ecology” (Lifset 2002: 1), clarifying industrial ecology’s identity and uniqueness, especially through defining its basics and highlighting tacit assumptions, will support intellectual and institutional development. As a result, such an effort will secure a place for industrial ecology within the scientific community. Ironically, Cohen and Howard (2006) argue, in a recent contribution to the Journal of Industrial Ecology, that efforts to foster industrial ecology institutionally could cause conflicting rebound effects with the objective to see the field become the science of sustainability, in the sense of Ehrenfeld. It is our aim 7

8

Background and concepts

here, however, to claim that industrial ecology’s characteristic understanding of nature as a model constitutes its identity and uniqueness. The contribution in this chapter is organized in three major parts. First, an argument is made with regard to the increasing importance, from both inside the community and from outside perspectives, of the discussion of what is essential for industrial ecology. This ongoing process of reflection is traced through the current literature. A number of implications regarding the evolution of industrial ecology are discussed, which are relevant for academics as well as for practitioners in the field. In the subsequent section, an analysis concerning industrial ecology’s epistemological base is presented. A philosophical point of view is useful to clarify industrial ecology’s emerging disciplinary contours, with its characteristic understanding of nature as a model as its foundation. A comparison with other fields of research reveals that the basic assumption of understanding nature makes industrial ecology unique. In the concluding section, suggestions are offered about how to communicate this tenet of “nature as a model”. In order to prevent misunderstandings due to superficial usage and to avoid unfair criticism from inside and outside the community, a solid strategy is recommended. Making philosophical thinking more accessible to anyone interested in industrial ecology is another goal of this contribution. Regardless of their background, these interested parties may include economists, engineers, ecologists, or any other researchers or practitioners aiming to do more than business as usual. Moreover, the issues considered here are also relevant for other fields and branches close to industrial ecology, such as environmental and resource economics, ecological economics, ecological engineering, cleaner production and environmental pollution prevention, as well as ecology in its different facets.

WHAT IS UNIQUE IN INDUSTRIAL ECOLOGY: INDICATORS AND IMPLICATIONS FOR AN ONGOING DIALOGUE Treating the question of distinctiveness of a field usually requires considering a number of crucial topics including philosophical basis, common language, definitions and goals, concepts, tools for applications, and no less important, aspects of institutionalization (Lifset and Graedel 2002; Weston and Ruth 1997). A brief outline follows to emphasize aspects that are important for the epistemological exploration. Setting boundaries is crucial to any field of research, school of thought, scientific community, or discipline. This idea is true for well-established

Beyond a sack of resources: nature as a model

9

entities like physics, ecology and economics, but in particular for emerging fields that are evolving from diverse intellectual roots into more formal interdisciplinary entities. There is no clear picture yet whether industrial ecology is considered (1) a field of research, (2) an academic branch, (3) a school of thought, (4) a scientific community, or (5) an umbrella approach. Usually one of the subsequent notions follow these different labels: multidisciplinary, interdisciplinary, transdisciplinary, or even metadisciplinary. These different labels are viewed as a sign that industrial ecology has several scientific roots that utilize various intellectual resources. As a rather young discipline “with a name as provocative and oxymoronic as industrial ecology” (Lifset and Graedel 2002: 3), the discourse on a unifying selfimage in industrial ecology must certainly be a continuous process. Lifset (1998) and Allenby (1998) stimulated this process nearly a decade ago. Lifset (1999: 1) is aware that “[q]uestions of definition and identity will not completely take care of themselves”. As an early perception of how industrial ecology would evolve, he maintains that its scientific profile could develop like a monolithic body of theory in the sense of a discipline or a modular structure seen as a complement to other fields (Lifset 1998: 1). Allenby (1998) argues for an opening up of the early engineering/technological and resource/industry-focused bipolar view of industrial ecology (for example Ayres and Ayres 1996) to contextual forces like culture and social aspects, as these are interrelated with the industrial system. Likewise, Ruth (1998) maintains that the role of technology needs to be seen in concert with social change and socioeconomic arrangements. His approach has some similarities to the “trefoil of development” which Müller-Merbach (2000: 172–3) proposed, arguing that any socioeconomic development is rooted in continuous interdependence between technological progress, economic prosperity and social change. The internal discourse culminated in a debate on the normative status of the field, among Allenby (1999) and Boons and Roome (2001). While Allenby (1999) makes a plea for industrial ecology to maintain objectivity and avoid normative positions, Boons and Roome (2001) argue that recommendations on what should or should not be included in a scientific field always imply a normative position. Related to the debate on the right standing of normative positions, Allen (2001) encourages community members to articulate the field, clarify its principles and tools, and uncover still hidden values and value judgments. Bey (2002) recommends paying more attention to industrial ecology’s ecological roots. He suggests taking into account current developments within ecology, not just the early milestones on ecosystem behavior that the work of E.P. Odum (1969) provides, but rather the work of Schneider and Kay (1994) in the field of theoretical ecology in particular. For Bey (2002),

10

Background and concepts

it is necessary to go beyond the basic idea of waste as food (Tibbs 1992) that mirrors the concept of the complex food webs, thus linking animals and plants representing producers, consumers, and recyclers in natural ecosystems, but that seems only a superficial similarity between the two systems. Ehrenfeld’s work (2004a, 2004b) takes note of two applicable issues for theory building in the field: (1) proper use of ecosystem metaphor and biological analogy, and (2) the field’s aim to become the science of sustainability. Faber et al. (1998) clearly state that any traditional scientific– technical–economic concept to approach sustainability by itself could not be a successful enterprise in principle due to its inherent intellectual limits, narrow perspective, and restricted point of view. In contrast, such an effort may look like a fireman with an extinguisher in one hand while pouring gasoline on the burning fire with the other. What is actually needed, they argue, is willingness towards sustainability, that is – according to Kant’s famous question, “what shall we do?”1 – the (re)incorporation of ethics into science and research, and consequently a discussion on how we want to live on a global scale, today and in the long term. In the first “Handbook on Industrial Ecology”, Lifset and Graedel (2002: 14) provide a synopsis on the consensus in the field, in terms of its goals and key definitions. While reviewing research agenda and early insights, they found that “there is no authoritative epistemology in industrial ecology”. It is clear that industrial ecology’s boundaries can hardly be defined in a purely abstract manner. This perception, however, could lead to misunderstandings or invite criticism. Some may understand it as methodological pluralism pleading for broad-mindedness. Critics of the field could argue that there is no scientific quality standard at all, concluding that epistemology is not an important matter to be taken seriously. The fact that epistemological aspects have received little recognition is not unique to industrial ecology, but the case for many emerging fields. For example, Tacconi (1998) considers that theory in ecological economics is at the beginning of a voyage of epistemological exploration. Bourg (2003) takes a strong position in the discourse regarding the development of a coherent self-image within industrial ecology. He claims that “industrial ecology has developed into a discipline in its own right” (ibid.: 13), due to its links with the larger movements of ecostructuring and industrial transformation and its close relations with ecological economics, and economic and political theory. As an input to the understanding of industrial ecology as a scientific field, Bey and Isenmann (2005) argue that the method of comparing individual companies with participants in natural ecosystems, constituting the employment of a metaphor, may become problematic, primarily resulting from a level of ecosystem analysis inappropriate for explaining the

Beyond a sack of resources: nature as a model

11

common root of these two kinds of complex systems. For a more solid foundation, they propose looking at thermodynamic system characteristics and complex systems theory. Such an approach seems to be a strategy to prevent the misuse of nature as a model, especially when it is overemphasized in the sense of nature as blueprint. Recently, Allenby (2005) started a discussion in the President’s corner of the International Society for Industrial Ecology (ISIE) News under the headline: “What is Industrial Ecology?” The key terms in Allenby’s article – industrial ecology as a metaphor or as an analogy – lead directly to the understanding of nature as a model, for these are the common methods used in the community for communicating the industrial ecology perspective on nature. Nature as a model is typically stated with an appealing natural ecosystem metaphor and based on an analogy between industrial systems and natural ecosystems. Bringing tacit assumptions to the surface with regard to nature plays an important and catalyzing role within the ongoing processes of articulating, clarifying, and defining the field. Some efforts have already been made to illustrate industrial ecology’s philosophy of nature, a philosophy that can be regarded as a feature of identity that makes the field unique (Isenmann 2003a, 2003c). To a certain extent, Allen et al. (2001) and Allenby’s (2005) considerations mentioned above reflect one of the early stated goals for industrial ecology’s research agenda: “To add coherence to the thinking about industrial ecology, by clarifying what is more and less important and what is well and poorly understood” (Socolow 1994: xix). At first glance, reasoning on the matters above may sound like a tedious task, perhaps of value for academics only. Shaping the boundaries of a field, however, is far from merely being l’art pour l’art; these efforts have farreaching implications both for theory building and practical applications. The various implications are illustrated in terms of four factors that influence identity and matter uniqueness (Figure 2.1).

INDUSTRIAL ECOLOGY UNDERSTOOD FROM AN EPISTEMOLOGICAL PERSPECTIVE An Epistemological Architecture for Industrial Ecology We consider an analysis of industrial ecology from an epistemological perspective a proper method for organizing its emerging body of theory and describing the characteristics of the field. Following Immanuel Kant, the fundamental question typical for epistemology is: “What can we know?” In their seminal book on ecological economics, Faber et al. (1998: 205) treat

12

Background and concepts

I

…books, journals, reports, Conpresentations text of statements (oral, written)

Context of phenomena (objects, e.g. products, …industrial II processes, waste, fields of metabolism, closing study like sustainable loops manufacturing, logistics, consumption, eco-industrial parks etc.

III

Context of methods …the IE research (calculations, instruments, tools, e.g. material flow toolbox analysis, life cycle assessment, design for environment, operations research tools, metaphors etc.) Context of basics (theories, axioms, etc.)

IV

…heart and

--------------------------------------------------------------------------------- core of a field “Fundamentals”, i.e. basic assumptions (human, nature, human nature, role of economy and technology, e.g. nature as model)

Figure 2.1 Scientific profile of industrial ecology in terms of an epistemological architecture this Kantian question in a modern fashion related to current problems in environmental policy and management: “What can we control? What possibilities of action do we have? What can we do?” These interpretations, they maintain, are an outcome of the interest of humankind particular in Western society to control nature and treat natural resources as an object merely to be manipulated and dominated.2 An epistemological analysis seems useful for providing a common and methodologically sound structure for industrial ecology. Obtaining order in the obvious disorder of assumptions, principles, concepts, tools and research objects is not as simple a process as it may appear at first glance. As a young and ambitious field that makes use of various intellectual roots originating from different theoretical backgrounds, this is particularly true for industrial ecology. Usually all research fields share a common perspective, whether they are also labeled an academic branch, science, school of thought, or discipline. Stated differently, any form of intellectual arrangement could be characterized through a specific view. In terms of epistemology, such a perspective/view is understood as a specific way to treat and deal with problems. These problems are phenomena that are regarded as relevant. To elucidate industrial ecology’s theoretical foundation with the help of tools of philosophy, a basic architecture is described here, represented in the form

Beyond a sack of resources: nature as a model

13

of a pyramid. This architecture serves as a rough framework for examining epistemological issues when portraying the contours of a scientific profile along four basic layers, or contexts, respectively (Figure 2.1): (1) context of statements, (2) context of phenomena, (3) context of instruments and (4) context of basics: 1.

2.

3.

4.

Context of statements: a statement or a system of statements that covers insights, information, expertise, and knowledge that is documented in articles, textbooks, journal publications, etc. Context of phenomena: different phenomena that could be considered or observed; an object or a class of objects. For example, the design of products, processes, wastes and services, as well as certain fields of study. Context of instruments: methods, instruments and calculations (for example material flow analysis, life cycle analysis) that are part of the industrial ecology toolbox. Context of basics: basic ideas, theories, principles and axioms that form the basic assumptions of understanding of human and their relationship with nature, and the role economy and technology should play in this relationship.

The epistemological architecture described here has its methodological basis in the philosophy of science (Zwierlein 1994) and sociology of science (Krüger 1987). It is also similar to a conceptual framework Vellinga et al. (1998) proposed, who made a distinction among (1) intuitions and beliefs, (2) operational principles and concepts and (3) practical applications. Moreover, according to Weston and Ruth (1997: 2), developing a new science requires the following: (1) a consistent philosophical base, (2) a common language, (3) a set of concepts and (4) methods for empirical application. Such a scheme is especially useful to illustrate the interdisciplinary common ground of various disciplines when analyzing different fields of research or comparing diverse schools of thought. The distinction between the layers/contexts (in the architecture above) could be traced back even to earlier works in philosophy (Rickert 1986), and in economics (Amonn 1927). Amonn (1927: 21–2) introduced the terms “object of experience” (German Erfahrungsobjekt) and “object of cognition” (German Erkenntnisobjekt) to distinguish between, on the one hand, a phenomenon that is considered, observed, or treated as an object of research; and on the other, the distinct method of consideration, observation, or treatment as the manner of providing insights. According to Amonn, a field is defined through a combination of certain objects of research and specific objects

14

Background and concepts

of cognition of which a set of methods, instruments and tools could be employed. This distinctive approach serves as the basic method used in modern economics and management sciences to define what to research and how to determine the preparation or manipulation needed for certain research objects. The architecture above (Figure 2.1) indicates a rough framework only. For example, layer three includes several contexts such as the context of discovery, description, explanation, justification and application (see Bey and Isenmann 2005). Further, a set of basic assumptions is situated typically at the bottom of layer four, providing an understanding of humanity, nature and the relationship between humans and nature. The viewpoint from which humans interpret nature depends on underlying presuppositions which indicate the human observer’s subjective perception employed for explaining and making sense of all that is happening on Earth. As the assumptions necessarily contain experiences, final interests, and implicit valuations and value judgments, they constitute a normative compass.3 Consequently, the anthropological basis of understanding the human (what or who is man?) is the ultimate key defining the importance of nature and influencing both the purpose and the value that nature should have. In the sense of Kant (1787/2000, ed. 3, 443 B 699), these assumptions serve as regulative ideas. They play a dominant role in research, arranging thoughts, organizing imagination and governing decision making. Kant made it clear that one is not able to extract knowledge from nature as a passive observer through his senses. But one is always able to extract knowledge in a more active manner through the construction of knowledge gained from human experience. The intellect, as Kant said, does not draw its laws from nature but imposes its laws upon nature. Comparable with Kantian interpretation, a string of famous economists described the significant influence of such underlying meta-ideas to modern economic theory and different schools of thought. For example, Schumpeter (1954) and more recently Daly (1991a, 1991b) have termed Kant’s regulative idea a pre-analytic vision, whereas Boulding (1971) called it an image. Despite conceptual differences, and though the terms used are different, the similarity is obvious. Any process of research is guided by meta-ideas serving as an underlying intellectual heuristic, much like a lens focusing light rays. In the case of industrial ecology, on the basis of such guiding meta-ideas, industrial ecologists decide to look to nature for a model, and only then can nature be analyzed as a model (that is, it is not a model by itself). Next, solutions, strategies and principles for solving environmental problems can be chosen according to the goal previously defined. In the sense of the example above,

Beyond a sack of resources: nature as a model

15

the epistemological architecture comprises (1) a statement about (2) something that is (3) represented, interpreted or manipulated by something (4) in the light of something. The epistemological architecture for industrial ecology above (Figure 2.1) has been conceptualized by certain issues that are widely discussed in the community. These issues can be taken as prototypical for the current state of the field. For example, material flow analysis, life cycle analysis, and design for environment are some of the methods that comprise the industrial ecology toolbox. They are listed here resulting from a document analysis, covering all oral and poster presentations that were held during international industrial ecology conferences, including the inaugural conference “The Science and Culture of Industrial Ecology” in Leiden 2001, the second conference “Industrial Ecology for a Sustainable Future” in Ann Arbor 2003, and the recent conference “Industrial Ecology for a Sustainable Future” in Stockholm 2005. The issues again have been grouped in terms of the four levels above (for example, statements, phenomena, methods and basics). The Basic Tenet of Nature as a Model in Industrial Ecology An in-depth analysis of industrial ecology’s understanding of nature is useful for deciding whether the notion of nature as a model could actually be regarded as the core feature that makes the field unique. The ensuing analysis is built around three steps of validation: 1. 2.

3.

Are there any implications that could be drawn from the name of the field “Industrial Ecology” as it is usually understood and interpreted? Is there empirical evidence that nature as a model constitutes a basic tenet of the field, and is it used as an identifying characteristic in the current literature? Is the notion of nature as a model in fact particular to industrial ecology, compared to the understanding of nature within other fields of research and applications for sustainability?

In his historical overview, Erkman (1997) argues that the intellectual roots go back much farther than the creation of the name for the field and its early incarnations as a scientific community. From the start, the core idea seemed a sound one: it is the assumption that nature, mainly understood as natural ecosystems, could be regarded as being a model for industrial ecological theory building. In other words, “understanding industrial systems in terms of natural systems” (Bey and Isenmann 2005) forms the central pillar of industrial ecology.

16

Background and concepts

This idea has been stated in many publications (see Bey and Isenmann 2005). For example, for Cleveland (1999: 148) a characteristic of industrial ecology is “to look to the natural world for models of highly efficient use of resources, energy and byproducts” (see also Lifset and Graedel 2002: 3). The online introduction “What is Industrial Ecology” posted on the Journal of Industrial Ecology website (JIE 2005) provides a similar illustration. The raison d’être of industrial ecology rests on a double analogy between industrial systems interpreted as living systems on the one hand, and natural ecosystems seen as a mature cyclical economy on the other. The empirical evidence of the notion “nature as a model” is quite overwhelming (see Bey and Isenmann 2005). Industrial ecologists widely use nature as a model and thus often draw heavily from natural ecosystem metaphors and biological analogies. According to Lifset (2002: 1), the utility of the biological analogy belongs to the set of substantive premises that give industrial ecology its unique identity. Likewise, Lifset and Graedel (2002: 4) call the important role metaphors and analogies are playing in industrial ecology “core elements or foci in the field”. The literature cited above makes it clear that in industrial ecology nature is employed as a model explicitly (or at least implicitly) and frequently based on a biological analogy between industrial systems and natural ecosystems. Beyond the employment of different terms, mostly linguistic expressions, nature is appreciated as an ideal model from which to learn how to balance the development of industrial systems with the constraints of natural ecosystems (Isenmann 2003a, 2003b, 2003c). As natural ecosystems are evolving, and nature in itself mirrors a certain human cognitive creation representing the result of their epistemological abilities, it is also clear that nature as a model is changing over time. Hence, nature may be best understood in terms of its dynamics, perhaps as a dynamic norm, not merely as the stationary state of the environment. A comparison of understanding nature with other schools of thought may reveal that management inspired by nature is in fact particular to industrial ecology. Moreover, in its scientific sense, industrial ecology’s characteristic understanding of nature as model indicates a groundbreaking change in the interpretation of nature. This change describes a specific development from interpreting nature as an object, or rather a sack of resources, to understanding and appreciating nature as a model or ideal system from which to draw lessons. Similarity exists between treating nature in economics as described here and the natural sciences and engineering sciences (von Gleich 1989). For example, in comparison with other branches within environmental economics schools, this change can be illustrated in a classification framework. The classification framework is represented here (Table 2.1) as a matrix with five

Beyond a sack of resources: nature as a model

17

Table 2.1 Classification framework for understanding nature Comprehension of nature (theory) Treatment of nature (practice)

Nature as object

Utilizing nature

Taking care of nature

Nature as limit

Nature as model

Nature as partner

Avoiding use of nature

Learning from nature

Co-operation with nature

Epistemological Intervene Preserve Respect for interest in in nature nature nature nature (metatheory) Interrelation human–nature

Domination

Stewardship

Environmental ethics

Anthropocentrism

Different schools of thought

Neoclassical Spaceship environmental and economics resource economics

Orientation Co-production by nature with nature

Liasion

Elucidated anthropocentrism Industrial ecology

Partnership Physiocentrism Bioeconomics

Ecological economics

columns and six rows (Isenmann 2003b). Despite being an extract of a more comprehensive typology,4 the classification provides a proper scheme for surveying the different relevance of nature and its underlying conceptual basics in general. Industrial ecology’s distinct interpretation of nature as model can be shown in a more general sense. The columns point to four different disciplinary perspectives of understanding nature, whereas the rows include five crucial indicators used to build the classification. The sixth row includes applications and examples in environmental economics and management. Perspective (1) stands for neoclassical environmental and resource economics representing mainstream economics. Perspective (2) indicates the field of spaceship economics (for example, Boulding 1971, 1992; Georgescu-Roegen 1987; Meadows et al. 1972). Perspective (3) defines the domain of industrial ecology. Perspective (4) denotes a field called bioeconomics (for example, Immler 1993; Lovelock 1982; Starik 1995). This classification, from concrete to abstract, is based on five basic indicators presented in rows: (1) comprehension of nature, (2) way of treating nature, (3) epistemological interest in nature, (4) interrelation human–nature and (5) environmental ethics.5 The comprehension of nature relates to theoretical aspects describing how nature is understood, perceived, or defined in light of a certain

18

Background and concepts

scientific methodology. The comprehension is a certain (technicaleconomic) theory of nature influenced by a set of assumptions, knowledge and attitudes, in particular concerning humans, nature and the human– nature relationship. Ultimately, these assumptions govern whether nature is understood primarily as an object, a limit, a model, or a partner. The theoretical aspects of how nature is understood are incorporated into practical aspects that indicate the role of humankind and human behavior concerning nature, and influence the manner in which humans are dealing with nature in practice. Thus, a certain comprehension of nature correlates with a corresponding way of treating nature. A certain way of treating nature can be described through the kind of operating, using, or manipulating nature, by the predominant way of handling natural resources and services. There are five ways of treating nature: use of nature, care of nature, avoiding use of nature, learning from nature and cooperation with nature. A specific epistemological interest in nature influences individual comprehension and treatment. Such an epistemological interest represents the underlying human motivation shaping the perspective from which the relationship between human and nature is seen. In the sense of Habermas (1977), an epistemological interest indicates a guiding purpose or main reason of understanding something. Thus, a certain epistemological interest in nature points to meta-theoretical aspects of understanding nature. Five types of epistemological interests in nature are distinguished: intervention into nature, conservation of nature, respect for nature, orientation by nature and co-production with nature. Understanding nature has its basis in philosophical anthropology and hence leads to understanding the human: all four perspectives of understanding nature are characterized by a certain position as to how the human–nature relationship is viewed, either as domination, stewardship, liaison, or partnership. Further, based on understanding the human, each perspective also corresponds with an environmental ethics approach. Such an approach clarifies how humans should deal with nature and its resources and services, be it anthropocentrism (human-centered) as the most rigorous and analytical position in one case, physiocentrism (life-centered) as the most systemic and holistic position in another, or something in between like an elucidated anthropocentrism. In addition to these indicators, the classification is based on the fact that a certain understanding of nature corresponds to a specific way of treating nature. As Popper said, theory governs practice. Theory and practice concerning nature, again, can be explained by the link between understanding nature and treating nature on the one hand, and a characteristic epistemological interest in nature on the other. This interest represents an

Beyond a sack of resources: nature as a model

19

underlying guiding desire for, respectively, using nature, taking care of it, avoiding using nature, learning from nature, or co-operating with nature. In short, both comprehension of nature representing conceptual aspects and a corresponding way of treating nature representing practical aspects are influenced by an equivalent epistemological interest in nature. Conversely, there is empirical evidence to support the link between a certain epistemological interest in nature (for example, whether to intervene into nature, to keep nature as it is, or to look at nature for a model in order to gain orientation) and the resulting conclusion drawn (for example utilizing nature, preserving it, or learning from nature) (Ruijgrok et al. 1999). In total, the indicators above form a characteristic perspective of understanding nature inclusive of meta-theoretical aspects (epistemological interest in nature), theoretical aspects (comprehension of nature), practical aspects (treatment of nature), anthropological aspects (interrelation human–nature) and ethical aspects (environmental ethics). Perspective (1) represents the position of neoclassical environmental and resource economics. This perspective indicates considering nature as an object either focusing on the utilization of natural resources and services or on the care for nature for which humans may feel responsible. The position of caring for nature emerges when nature’s welfare-generating capacity decreases, as a result of scarce resources (input), finite sinks (output) and fragile self-organized cycles (throughput). The underlying epistemological interest concerning the use of nature is explained through humanity’s will to take control of and seize power on nature via the intervention into nature in a dominating sense (Leiss 1994; Merchant 1994). Otherwise, the desire to conserve nature as a whole predominantly motivates the responsible care for natural resources and services. In short, nature as an object exemplifies understanding nature analogous to a mechanistic figure of a complex machine, still influencing the mainstream of economics. As such, nature is more or less degraded to mere material and energy that can be used unlimited, as free gifts of nature. Perspective (2) represents spaceship economics in the sense of Boulding (1971, 1992), Georgescu-Roegen (1987) and Meadows et al. (1972). This perspective understands nature as a limit requiring respect to its inherent scarce resources, its restrictions regarding carrying capacity and its biophysical limits concerning fragile natural ecosystems’ services. Here, Ayres (1993: 189) makes the point: “The most important scarcities . . . are largely outside the market domain: soil fertility, clean fresh water, clean fresh air, unspoiled landscapes, climatic stability, biological diversity, biological nutrient recycling and environmental waste assimilative capacity. There are no plausible technological substitutes for these.” For example, there are limits to biophysical throughput of resources from natural ecosystems, through

20

Background and concepts

the industrial system and back to nature as waste. This perspective reminds us to respect the constraints of natural ecosystems and to avoid using nature as a mere resource for three reasons. First, natural resource stocks are declining worldwide and may be exhausted. Second, environmental pollution appears to approach or even exceed the absorption capacity of natural sinks. Third, as a long-term danger the fragile self-organizing cycles of natural ecosystems may be destroyed. Consequently, according to spaceship economics, utilization of natural resources is limited to the natural rate of reproduction. Further, it implies inflicting no more damage than natural resilience allows. In total, spaceship economics’ perspective on nature strongly recommends not surpassing nature’s inherent biophysical limits and ecosystems’ implicit carrying capacity. Perspective (3) represents understanding nature as a model, which is characteristic of industrial ecology. Whereas perspective (1) illustrates the most rigorous and analytic interpretation of nature, perspective (2) includes some holistic or systemic elements. Industrial ecology, however, goes beyond the other two. Its perspective is refreshingly different, though not a substitute for the other, more traditional ones. Perspective (3) seems reasonable due to its interdisciplinary basis, including insights from philosophy, economics, ecology, biophysics and engineering sciences. Further, this perspective is useful as a powerful heuristic for studying the links between industrial systems and natural ecosystems and as a source of environmental innovations. The question of where or how exactly nature can be a model for industrial ecology remains open. Industrial ecologists are often guilty of looking to relationships between species (plants and animals) for an understanding of natural systems and from there deriving clues for industrial organization. As argued elsewhere (Bey and Isenmann 2005), the relationships between species are perceived to be the superficial expression of a much more important process, the dynamic development of natural ecosystems towards maximizing their capacity to degrade incoming solar energy. The make-up and development of one natural ecosystem’s flora and fauna is not the cause of its dynamics, but the result of those dynamics. It would thus be a grave mistake to only consider predator–prey relationships or symbioses to derive insights for redesigning industrial processes. Resulting from perspective (3), industrial ecology is a notable departure from the traditional economic viewpoints. The field has the merit of recognizing that the dynamic systemic connections between ecosystem participants form an important plank of inquiry towards sustainable development. Organic and, to a certain extent, mineral components of the biogeosphere are engaged in dynamic behavior, contributing to global flows of material and energy. This insight of nature as a model in the sense of a

Beyond a sack of resources: nature as a model

21

dynamic norm transcends the traditional perspective of the environment as a sum of stocks of resources. It is, in fact, the in-depth understanding of this conception of nature and dynamic ecosystem behavior that has implications for current work in industrial ecology. There is also a fourth perspective (4) that could be identified in the literature, primarily in the field of bioeconomics: the perspective of understanding nature as a partner. Despite its clever sound, this perspective seems to be more problematic as it often implies ambitious claims (see Schneider et al. 2004; Henrich 2002). For example, what is the difference (if any) between humankind and nature when interpreting nature as partner (see, for example, Sheldrake 1993; Lovelock 1982). For example, Starik (1995) suggests viewing nature in the sense of a stakeholder, whereas Commoner regards nature as a consultant advising what to do stated in his slogan “nature knows best” (Commoner 1973: 45). Due to such a challenging perspective of understanding nature, this perspective is not discussed further (for details see Isenmann 2003b). In total, industrial ecology’s characteristic perspective indicates a fundamental shift of understanding nature, moving away from an interpretation of nature as a mere limited source of materials and energy towards a hypothetical model appropriate for deriving ecological innovations. In a broader sense, such groundbreaking change could be called a paradigm shift (Ehrenfeld 1997, 2000; Gladwin et al. 1995). In addition, nature could be appreciated as serving a proper heuristic to adapt, apply or learn from its phenomena, functionalities, strategies, principles and in particular its dynamics. It is these dynamic characteristics that are of crucial importance for balancing industrial systems and natural ecosystems (Ring 1997). In the words of Simonis (1993), the essence of industrial ecology’s perspective of understanding nature is to learn from the wisdom of nature. Reasoning on the issues mentioned above is not restricted to the emerging field of industrial ecology; it is also of particular relevance for economics (Becker et al. 2005; Becker and Manstetten 2004), engineering sciences (Levine 2000, 2003; Tilley 2003; Vincent 2000; von Gleich 1998), ecology (Cuddington 2001) and biology (Haila 1995, 2000).

CONCLUSIONS Communicating Nature as a Model in Industrial Ecology Industrial ecology’s characteristic understanding of nature as a model can be regarded as the field’s unique identity. The classification framework (Table 2.1) presents the different perspectives on nature. The contrast in the

22

Background and concepts

perspectives outlined in the table helps make clearer various perspectives on nature, even the difference between Boulding’s source-and-sink view and industrial ecology’s perspective on nature as a model. All in all, the four perspectives on nature (object, limit, model, partner) indicate a development that could be read from left to right. Perspective 1, “nature as an object”, represents the most rigorous and analytic interpretation of nature, followed by perspective 2, in which nature is understood as a limit. In this second perspective, some holistic or systemic elements are included. Perspective 3, “nature as a model”, goes beyond the other two. This perspective can be seen as an additional perspective, as a supplement, and not as a substitute for the latter traditional ones. Hence, industrial ecology’s perspective represents an enrichment of the orthodox perspectives of nature traditionally used in economics. In conclusion, some suggestions are made on how to communicate this identity. From one perspective, if this identity is being used too rigidly, interested members in science, government, and business may be discouraged from what would otherwise be a refreshingly different perspective. Conversely, superficial communication may create misunderstandings and invite criticism. In order to prevent either, from inside or outside the community, a solid three-part strategy is proposed. As a first recommendation, pitfalls, fallacies and shortcomings in argumentation should be avoided, especially regarding the philosophy of nature, ethical issues, and in terms of epistemological standards. These potential dangers may occur when a discussion of understanding nature as a model is based exclusively on rhetoric, fashionable linguistic expressions, trendy metaphors or pictographic analogies, but without any foundation (see examples in Isenmann 2003a). A second suggestion would be to clarify the proper role of the natural ecosystem metaphor and biological analogy in industrial ecology (see Bey and Isenmann 2005; Isenmann 2003a). Allen et al. (2001) illustrate with a number of examples that metaphor and analogy are excellent vehicles to recognize and communicate scientific work on the environment. Metaphors and iconic analogies are also useful in making industrial ecology’s ambitious goal accessible that could otherwise seem unobtainable. Moreover, these vehicles are powerful tools to incite change and deliver new perceptions on sustainability. Use of metaphor and analogy are seen as proper and often highly useful instruments for theory building, but only as long as what is involved is primarily the elucidation of a given proposition. In epistemological terms, metaphors and iconic analogies are assigned to the context of discovery as an “inducer of new ideas” (Johansson 2002: 70) in one sense, and for illustrative or pedagogical purposes for communicating new insights in the

Beyond a sack of resources: nature as a model

23

context of application in another. If, however, one tries to use them for proving a proposition or even to establish a hypothesis in the context of justification, potentially severe errors might ensue. This matter of epistemological thinking becomes particularly important when nature as a model is used in the realm of the sciences because here the crucial criterion is to make transparent how well grounded is industrial ecology’s perspective on nature. As the third suggestion, a few basic philosophical prerequisites are presented for consideration when industrial ecologists employ nature as a model in a reasonable way. As nature does not automatically or clearly speak to humans, and nature as a whole is ultimately impossible to perceive, humans must translate nature with their language and into their language through several mediating steps. Keeping this in mind, industrial ecology could become a proper scientific way for carrying out this translation. As a new field, industrial ecology consists of assumptions, concepts, tools and applications. While some aspects of the field may have attained a well-defined status, other issues still seem to be in a status nascendi. This interim status is not viewed as a weakness in principle, but rather as an opportunity for fruitful development. For that purpose, an epistemological architecture is proposed, highlighting the features of industrial ecology’s identity. A closer look at the history and current literature in the field reveals that nature as a model is widely used in the industrial ecology community. A more detailed philosophical analysis shows that this basic tenet provides the foundation of industrial ecology’s scientific profile. When communicating industrial ecology’s specific understanding of nature as a unifying feature of self-image, we suggest a specific strategy, making clear the refreshingly unorthodox perspective, but also taking care to prevent misunderstandings from inside and outside the community. All in all, the emerging interest in matters of identity, differentiation, and uniqueness is a sign for the dynamics of the industrial ecology community picking up these issues as a critical theme and considering the implications of industrial ecology’s central ideas. It is seen that matters of clear argument, proper vocabulary and fair communication relevant for industrial ecology both in terms of the intellectual heritage of its proponents and the development of the community as a whole. Although some may deem these issues as useless in comparison to more practical or empirical research, such discussions are crucial for recognizing that these issues have consequences. They shape concepts, instruments and vocabularies that are used to approach research problems. For this reason, if no other, industrial ecology’s intellectual heritage needs to be examined. Through a philosophical exploration, finding important threads can shed light on present problems. At the very least, from this exploration, it is possible to better understand both nature and ourselves.

24

Background and concepts

ACKNOWLEDGMENTS The authors are grateful to the reviewers for detailed comments and to the editors, Matthias Ruth and Brynhildur Davidsdottir, for their motivating assistance throughout the editorial process when fine-tuning this manuscript. The contribution is based on two presentations, initially held at the International Lecture Series on “Industrial Ecology – Sustainable Development of Industrial Systems” (http://www.industrialecology.de), at the University of Bremen (Germany), and later on at the 3rd Conference of the International Society for Industrial Ecology (ISIE) on: “Industrial Ecology for a Sustainable Future 2005” (http://www.isie-2005.org), 12–15 June 2005, Stockholm (Sweden).

NOTES 1. According to a famous introduction given by Immanuel Kant in his lecture on logic (1800/1996: A 25), philosophy could be described along four fundamental questions, that is (1) “What can we know?” (realm of epistemology), (2) “What shall we do?” (realm of ethics), (3) “What may we hope?” (realm of religion) and (4) “What or who is man?” (realm of philosophical anthropology). Given this basic description, epistemology, or the philosophy of science, is a certain branch of academic philosophy. It studies the nature of knowledge, its presuppositions and foundations, and its extent and validity. The main subjects are theory building and theory development. 2. It is precisely on the predominant instrumental orientation where consensus is reached in environmental philosophy. Such a narrow view on nature is thought to be destructive (Dryzek 1996; Zwierlein and Isenmann 1996). A mere instrumental orientation is rather typical for neoclassical environmental and resource economics. The presupposition underlying the instrumental orientation again is that nature supposedly lacks of purpose. Natural teleology is the idea that everything in nature exists to serve some purpose (for example the sky rains in order for corn to grow). Teleology, however, has been banished from the neo-Darwinian’s mechanistic system and its successors applied in the natural sciences, particularly in physics and biology: “Nothing in biology makes sense except in the light of evolution” (Dobzhansky 1973: 125). This definition has far-reaching consequences for mainstream economics. For example, following Ghiselin (1999), the mechanical causation seems to be everything in nature whereas the existence of final causation, purposes and ends are unthinkable. For a number of leading thinkers in the realm of ecological economics, however (for example Herman Daly, Malte Faber, Reiner Manstetten and John Proops) a critical discussion on right purpose is seen as the central focus to solve what is called the environmental crisis of modern society (Isenmann 2003b). 3. The roots of how to understand nature make up the foundation of the basic layer four. Put more simply, understanding nature, dealing with and treating nature is based on being human and thus finally meets philosophical anthropology. From being human every interpretation starts. Zwierlein (1994: 294) called this fundamental process of ultimate understanding “logical anthropomorphism”. Further, here it becomes clear that the claim for the supposedly strong difference between the realm of facts and norms seems to be obsolete. Borrowing from Myrdal (1978: 778): “There can be no view without a viewpoint. In the questions raised, valuations are implied.” Given this basic architecture, it could be a proper starting point also to resolve discussions on the right standing of normative positions.

Beyond a sack of resources: nature as a model

25

4. Typology is a research methodology used both in the social sciences and humanities and in the natural sciences. It is the methodology of arranging entities, objects, and phenomena. In contrast to a simple classification, which is usually based on one criterion alone, a typology uses at least two criteria that form certain perspectives existing in practice or that could be observed in reality. The overall aim of a typology is to obtain order in a field with various specifications. 5. Environmental ethics is the branch of philosophy that examines the right place of humankind in nature, and studies how humans should treat nature and deal with natural resources and services. For an introduction see Gruen and Jamieson (1994).

REFERENCES Allen, D. (2001), “A set of core principles and tools?”, Journal of Industrial Ecology, 4(4): 1–2. Allen P., C. Bonazzi and D. Gee (eds) (2001), Metaphors for Change: Partnerships, Tools and Civil Action for Sustainability, Sheffield, UK: Greenleaf Publishing. Allenby, B.R. (1998), “Context is everything”, Journal of Industrial Ecology, 2(2): 6–8. Allenby, B.R. (1999), “Culture and industrial ecology”, Journal of Industrial Ecology, 3(1): 2–4. Allenby, B.R. (2005), “The great game: what is industrial ecology?”, ISIE News, 5(2): 3 and 6. Amonn, A. (1927), Objekt und Grundbegriffe der theoretischen Nationalökonomie, 2nd edition, Leipzig, Germany and Vienna, Austria: Franz Deuticke. Ayres, R.U. (1993), “Cowboys, cornucopians and long-run sustainability”, Ecological Economics, 8(3): 189–207. Ayres, R.U. and L.W. Ayres (1996), Industrial Ecology. Towards Closing the Material Cycle, Cheltenham, UK and Brookfield, USA: Edward Elgar. Ayres, R.U. and U.E. Simonis (eds) (1994), Industrial Metabolism. Restructuring for Sustainable Development, Tokyo, Japan: United Nations University Press. Becker, C. and R. Manstetten (2004), “Nature as a you: Novalis’ philosophical thought and the modern ecological crisis”, Environmental Values, 13: 101–18. Becker C., M. Faber, K. Hertel and R. Manstetten (2005), “Maltus vs. Wordsworth: perspectives on humankind, nature and economy. A contribution to the history and the foundations of ecological economics”, Ecological Economics, 53: 299–310. Bey, C. (2002), “Quo vadis industrial ecology? Realigning the discipline with its roots”, Greener Management International, 34: 35–42. Bey, C. and R. Isenmann (2005), “Human systems in terms of natural systems? Employing non-equilibrium thermodynamics for evaluating industrial ecology’s ecosystem metaphor”, International Journal of Sustainable Development, 8(3): 189–206. Boons, F. and N. Roome (2001), “Industrial ecology as a cultural phenomenon”, Journal of Industrial Ecology, 4(2): 49–54. Boulding, K.E. (1971), The Image, Ann Arbor, MI: University of Michigan Press. Boulding, K.E. (1992), Towards a New Economics: Critical Essays on Ecology, Distribution and Other Themes, Aldershot, UK and Brookfield, US: Edward Elgar. Bourg, D. (2003), “Introduction: perspectives on industrial ecology”, in D. Bourg and S. Ekman (eds), Perspectives on Industrial Ecology, Sheffield, UK: Greenleaf, pp. 13–18.

26

Background and concepts

Cleveland, C. (1999), “Biophysical economics”, in K. Mayumi and J. Gowdy (eds), Physiocracy to Ecological Economics and Industrial Ecology. Bioeconomics and Sustainability: Essays in Honor of Nicholas Georgescu-Roegen, Cheltenham, UK and Northampton, MA, USA: Edward Elgar, pp. 125–54. Cohen, M. and J. Howard (2006), “Success and its price: the institutionalization and political relevance of industrial ecology”, Journal of Industrial Ecology, 10(1/2): 79–88. Commoner, B. (1973), Wachstumswahn und Umweltkris (The Closing Circle: Nature, Man and Technology), München, Germany: Bertelsmann. Cuddington, K. (2001), “The ‘balance of nature’ metaphor and equilibrium in population ecology”, Biology and Philosophy, 16: 463–79. Daly, H.E. (1991a), “Elements of environmental macroeconomics”, in R. Constanza (ed.), Ecological Economics: The Science and Management of Sustainability, New York, NY: Columbia University Press, pp. 32–46. Daly, H.E. (1991b), Steady-State Economics, 2nd edition, Washington, DC, Covelo, CA: Island Press. Dobzhansky, T. (1973), “Nothing in biology makes sense except in the light of evolution”, American Biology Teacher, 35: 125–9. Dryzek, J.S. (1996), “Foundations for environmental political economy: the search for homo ecologicus?”, New Political Economy, 1(1): 27–40. Ehrenfeld, J.R. (1997), “Industrial ecology: a framework for product and process design”, Journal for Cleaner Production, 5(1/2), 87–95. Ehrenfeld, J.R. (2000), “Industrial ecology. Paradigm shift or normal science?”, American Behaviour Scientist, 44(2): 229–44. Ehrenfeld, J.R. (2001), “Industrial ecology begets a society”, Journal of Industrial Ecology, 4(3): 1–2. Ehrenfeld, J.R. (2004a), “Industrial ecology: a new field or only a metaphor?”, Journal of Cleaner Production, 12: 825–31. Ehrenfeld, J.R. (2004b), “Can industrial ecology be the ‘science of sustainability?’ ”, Journal of Industrial Ecology, 8(1/2): 1–3. Erkman, S. (1997), “Industrial ecology: a historical view”, Journal of Cleaner Production, 5(1/2): 1–10. Faber, M., R. Manstetten and J. Proops (1998), Ecological Economics: Concepts and Methods, Cheltenham, UK and Northampton, MA, USA: Reprinted edition, Edward Elgar. Frosch, R. (1992), “Industrial ecology: a philosophical introduction”, Proceedings of the National Academy of Science USA, 89(2): 800–803. Georgescu-Roegen, N. (1987), “The entropy law and the economic process in retrospect”, Schriftenreihe 5/87, Berlin, Germany: Institute for Ecological Economy Research. Ghiselin, M. (1999), “Darwinian monism: the economy of nature sociobiology and bioeconomics”, in P. Koslowski (ed.), The Theory of Evolution in Biological and Economic Theory, Berlin, Germany: Springer, pp. 7–24. Gladwin, T.N., J.J. Kennelly and T. Krause (1995), “Shifting paradigms for sustainable development: implications for management theory and research”, Academy of Management Review, 20(4): 874–907. Gruen, L. and D. Jamieson (1994), Reflecting on Nature: Readings in Environmental Philosophy, Oxford, UK and New York, USA: Oxford University Press. Habermas, J. (1977), Erkenntnis und Interesse, 4th edition, Frankfurt am Main, Germany: Suhrkamp.

Beyond a sack of resources: nature as a model

27

Haila, Y. (1995), “Natural dynamics as a model for management: is the analogue practicable?”, in A. Sippola, P. Alarandanjoki, B. Forbes and V. Hallikainen (eds), Northern Wilderness Areas: Ecology, Sustainability, Values, International Conference on Northern Wilderness Areas, 7–9 December 1994, Artic Centre, Rovaniemi, Finland: University of Lapland, pp. 9–26. Haila, Y. (2000), “Beyond the nature–culture dualism”, Biology and Philosophy, 15: 155–75. Henrich, K. (2002), “Gaia infiltrata: the anthroposphere as a complex autoparasitic system”, Environmental Values, 11(4): 489–507. Immler, H. (1993), Welche Wirtschaft braucht die Natur? Mit Okonomie die Okokrise losen (What Kind of Economy Does Nature Need? Solving the Ecological Crisis with Help from the Economy), Frankurt am Main, Germany: Fischer. International Society of Industrial Ecology (ISIE) (2006), A History of Industrial Ecology. A Field for the Future, http://www.is 4ie.org/history.html. Last accessed 1 December 2006. Isenmann, R. (2003a), “Further efforts to clarify industrial ecology’s hidden philosophy of nature”, Journal of Industrial Ecology, 6(3/4): 27–48. Isenmann, R. (2003b), Natur als Vorbild. Plädoyer für ein differenziertes und erweitertes Verständnis der Natur in der Ökonomie (Nature as Model Pleading for a more Comprehensive and Sophisticated Understanding of Nature in Economics), Marburg, Germany: Metropolis. Isenmann, R. (2003c), “Industrial ecology: shedding more light on its perspective of understanding nature as model”, Sustainable Development, 11(3): 143–58. Johansson, A. (2002), “Industrial ecology and industrial metabolism: use and misuse of metaphors”, in R.U. Ayres and L.W. Ayres (eds), Handbook of Industrial Ecology, Cheltenham, UK and Northampton, MA, USA: Edward Elgar, pp. 70–75. Journal of Industrial Ecology (JIE) (2005), What is Industrial Ecology?, http:// mitpress.mit.edu/journals/JIEC/jie-call.html#whatis. Last accessed 27 May 2005. Kant, I. (1787/2000), Kritik der reinen Vernunft (1 und 2) (Critique of Pure Reason), Werkausgabe in 12 Bänden. Band III und Band IV. W. Weischedel (ed.), 14th edition, Frankfurt am Main, Germany: Suhrkamp. Kant, I. (1800/1996), Schriften zur Metaphysik und Logik (2) (Lectures on Metaphysics and Logic), Werkausgabe in 12 Bänden. Band VI. W. Weischedel (ed.), 9th edition, Frankfurt am Main, Germany: Suhrkamp. Krüger, L. (1987), Einheit der Welt – Vielheit der Wissenschaften (United World – Various Sciences). Interdisziplinarität: Praxis – Herausforderung – Ideologie, J. Kocka (ed.), Frankfurt am Main, Germany: Suhrkamp, pp. 106–25. Leiss, W. (1994), The Domination of Nature, Montreal, Canada: McGill-Queen’s University Press. Levine, S.H. (2000), “Products and ecological models. A population ecology perspective”, Journal of Industrial Ecology, 3(2/3): 47–62. Levine, S.H. (2003), “Comparing products and production in ecological and industrial systems”, Journal of Industrial Ecology, 7(2): 33–42. Lifset, R. (1998), “Setting the boundaries?”, Journal of Industrial Ecology, 2(2): 1–2. Lifset, R. (1999), “On becoming an industrial ecologist”, Journal of Industrial Ecology, 2(3): 1–3. Lifset, R. (2002), “Closing the loop and honing our tools”, Journal of Industrial Ecology, 5(4): 1–2.

28

Background and concepts

Lifset, R. and T.E. Graedel (2002), “Industrial ecology: goals and definitions”, in R.U. Ayres and L.W. Ayres (eds), A Handbook of Industrial Ecology, Cheltenham, UK and Northampton, MA, USA: Edward Elgar, pp. 3–15. Lovelock, J.E. (1982), GAIA: Eine optimistische Ökologie, München, Germany: Piper. Meadows, D.L., D.H. Meadows and E. Zahn (1972), Die Grenzen des Wachstums. Bericht des Club of Rome zur Lage der Menschheit, Stuttgart, Germany: Deutsche Verlags-Anstalt. Merchant, C. (1994), Der Tod der Natur: Ökologie, Frauen und neuzeitliches Naturwissen, 2nd edition, München, Germany: Beck. Müller-Merbach, H. (2000), “Five concepts of holistic understanding: generalists as critical success factor of nations”, in J. Thépot, M. Godet, F. Roubelat and A.E. Saab (eds), Décision Prospective Auto-organisation. Mélanges en l’honneur de Jacques Lesourne, Paris, France: Dunod, pp. 170–84. Myrdal, G. (1978), “Institutional economics”, Journal of Economic Issues, 12(4): 771–83. Odum, E.P. (1969), “The strategy of ecosystem development”, Nature, 18: 262–70. Rickert, H. (1986), Kulturwissenschaft und Naturwissenschaft, Stuttgart, Germany: Reclam. Ring, I. (1997), “Evolutionary strategies in environmental policy”, Ecological Economics, 23(3): 237–49. Ruijgrok, E., P. Vellinga and H. Goosen (1999), “Dealing with nature”, Ecological Economics, 28(3): 347–62. Ruth, M. (1998), “Mensch and mesh. Perspectives on industrial ecology”, Journal of Industrial Ecology, 2(2): 13–22. Schneider, E.D. and J.J. Kay (1994), “Complexity and thermodynamics – towards a new ecology”, Futures, 26: 626–47. Schneider, S.H. et al. (2004), Scientists debate Gaia. The Next Century, Cambridge, MA, USA: MIT Press. Schumpeter, J.A. (1954), History of Economic Analysis, New York, NY: Oxford University Press. Sheldrake, R. (1993), Die Wiedergeburt der Natur. Wissenschaftliche Grundlagen eines neuen Verständnisses der Lebendigkeit und Heiligkeit der Natur (The Rebirth of Nature. The Greening of Science and God), Bern, Switzerland: Scherz. Simonis, U.E. (1993), “Von der Natur lernen? Über drei Bedingungen zukunftsfähiger Wirtschaftsentwicklung” (“Learning from nature? On three prerequisites of sustainable development”), in E. Zwierlein (ed.), Natur als Vorbild. Was können wir von der Natur zur Lösung unserer Probleme lernen, Schulz-Kirchner, Germany: Idstein, pp. 115–33. Socolow, R. (1994), “Preface”, in R. Socolow, C. Andrews and F. Berkhout (eds), Industrial Ecology and Global Change, Cambridge, UK: Cambridge University Press, pp. xv–xx. Starik, M. (1995), “Should trees have managerial standing? Toward stakeholder status for non-human nature”, Journal of Business Ethics, 14: 207–17. Tacconi, L. (1998), “Scientific methodology for ecological economics”, Ecological Economics, 27(1): 91–105. Tibbs, H.B.C. (1992), Industrial Ecology. An Environmental Agenda for Industry, Center of Excellence for Sustainable Development, http://www.sustainable.doe. gov/articles/indecol.shtml. Last accessed 1 September 2000.

Beyond a sack of resources: nature as a model

29

Tilley, D. (2003), “Industrial ecology and ecological engineering – opportunities for symbiosis”, Journal of Industrial Ecology, 7(2): 13–32. Vellinga, P., F. Berkhout and J. Gupta (eds) (1998), Managing a Material World: Perspectives in Industrial Ecology, Dordrecht, Netherlands: Kluwer Academic Publishers. Vincent, J.F.V. (2000), “Deployable structures in nature: potential for biomimicking. Proceedings of the I MECH E Part C”, Journal of Mechanical Engineering Science, 214(1): 1–10. von Gleich, A. (1989) Der wissenschaftliche Umgang mit der Natur. Über die Vielfalt harter und sanfter Naturwissenschaften (Dealing with and Treating Nature in the Sciences), Frankfurt am Main, Germany: Campus. von Gleich, A. (1998): “Was können und was sollen wir von der Natur lernen?” (“what can and what shall we learn from nature”), in A. von Gleich (ed.), Bionik: Ökologische Technik nach dem Vorbild der Natur?, Stuttgart, Germany: Teubner, pp. 7–34. Weston, R.F. and M. Ruth (1997), “A dynamic, hierarchical approach to understanding and managing natural economic systems”, Ecological Economics, 21: 1–17. Zwierlein, E. (1994), “The paradigm of self-organization and its philosophical foundations”, in R.K. Mishra, D. Maass and E. Zwierlein (eds), On SelfOrganization. An Interdisciplinary Search for a Unifying Principle, Berlin, Germany: Springer, pp. 288–98. Zwierlein E. and R. Isenmann (1996), “Umweltphilosophie” (“Environmental philosophy”), in P. Eichhorn (ed.), Umweltorientierte Marktwirtschaft, Zusammenhänge – Probleme – Konzepte, Wiesbaden, Germany: Gabler, pp. 133– 51.

PART II

Stocks and Flows Dynamics

3. Stocks and flows dynamics: an introduction Thomas E. Graedel INTRODUCTION Prior to the 20th century, works of art were largely static: classic marble statues from Greece, pastoral French countrysides, and so forth. In the decade of 1900–1910, several young Italians developed what is now termed “Futurism”, art designed to communicate the concept of “universal dynamism”, recognition that motion and change, not stasis, are the constants of the world. The results of their efforts are now the treasures of museums throughout the world, perhaps best exemplified by Giacomo Balla’s Dynamism of a Dog on a Leash (Albright-Knox Art Gallery, Buffalo, NY) and Umberto Boccioni’s The City Rises (Museum of Modern Art, New York, NY). This volume and the chapters in this part begin to do for industrial ecology what the Futurists did for art: move the field from the valuable, but essentially static, analysis of product design or life-cycle assessment to studies of the universal dynamism that pervades the relationship between technology and the environment. We are moving from regarding products in use as objects of no immediate scholarly interest to understanding them as “transient embodiments of materials and energy”, as Frosch (1995) so aptly states. Dynamism has come to industrial ecology. As has been the case with the Futurists, the approach is likely to be celebrated years hence as a new window to understanding.

THE MATERIALS–POLICY SEQUENCE If industrial ecology is dynamic, then its related driving forces and policy constraints are dynamic as well. In this part of the book, Davidsdottir and Ruth, in Chapter 5, indicate that traditional energy and material flow analyses (E/MFA) “often ha[ve] had little relevance to the policy community”. In the case of energy, this approach is often true at the local level, where 33

34

Stocks and flows dynamics Environmental and sustainability impacts

ES2

Energy stocks and flows

ES3

EA3

ES1

Economic and behavioral actions

EA2

ES4

Policy development

EA1

EA4

Note: EA signifies actions, ES stimuli.

Figure 3.1 Energy stocks and flows and related spheres of action

price is usually the principal constraint on the rate of energy use. It is less true at the global level, where the threat of fossil-fuel-induced climate change has individuals, individual firms, and some national governments beginning to induce energy-limiting activities. To explore why this seemingly obvious coupling is less effective than might be imagined, let us take a closer look at the links between EFA, its societal drivers, and the policy development that results (or could result). Consider the sequence in Figure 3.1. Various economic and behavioral actions are indicated by EA1 (the decision to build a factory, the decision to purchase an automobile), and those actions stimulate (ES1) the consumption of energy (EA2). Energy consumption generates new stimuli (ES2) that have various implications for active responses related to the environment and sustainability (EA3). Seeing these implications (ES3), policymakers may decide to act (EA4) and thus influence (ES4) the economic and behavioral actions that began the sequence. Table 3.1 gives typical examples of actions and stimuli. The situation with the Figure 3.1 sequence is interesting in that it is relatively straightforward, even if some of the steps in the sequence develop only over relatively long periods of time. The EA1 → ES1 → EA2 → ES2 links have been clear all along. The science related to ES2 → EA3 is complicated, with many remaining details remain to be worked out, but the basic situation has been obvious for some time. It has led to the Kyoto Protocol and other such actions for accomplishing the EA3 → ES3 → EA4 links, and governments are now in the process of developing EA4 to generate ES4. The process has significant analytical and practical advantages from both an EFA and a policy standpoint: the stressor is relatively easy to

An introduction

35

Environmental and sustainability impacts

MS2

MS3

MA3

Material stocks and flows

ES1

Economic and behavioral actions

MA2

ES4 Policy development

MA1

MA4

Note: MA signifies actions, MS stimuli.

Figure 3.2 Material stocks and flows and related spheres of action

understand, and the effects are global in scope, encouraging people and governments around the world to work together, thus able to understand the challenges and apply pressure for action. The sequence for materials rather than energy, as shown in Figure 3.2, is the same. As exemplified in Table 3.1, however, many things about the situation are much different in practice. Consider action MA2, the manufacture of a product. Rather than consuming one of a very small set of fossil fuels, and doing so irremediably, the manufacturer employs the whole cupboard to materials and the whole sweep of modern technology. This action creates a host of different consequences, and the ecosystem responses are similarly diverse. Especially for issues related to chronic effects and the slow approach of sustainability constraints, policy actions as a result of the links between economic activity and environmental and sustainability implications are quite rare. Thus, the ability of the MFA community to link with the policy community is a greater challenge than is the case for EFA researchers because of the much greater breadth of the issues and the added challenge of clearly communicating the concerns related to this enhanced scope.

THE AVAILABILITY OF INFORMATION If the level of complexity related to materials does not itself present an insurmountable challenge for the researcher, a further stumbling block is the paucity of available data. This limitation does not apply so much to energy use, which is widely measured and reported. It is certainly a

36

Stocks and flows dynamics

Table 3.1 Examples of stimuli and actions related to energy and material flow analyses Action or Stimulus*

Example

EA1 ES1

An individual desires home appliance use. The electric power industry sees a need to generate electricity. Fossil fuels are burned to generate electricity. Carbon dioxide is released to the atmosphere. As a result of the enhanced CO2 climate change occurs. Policymakers notice climate change implications. Climate change limitation policies are developed. Policies are applied to influence economic and behavioral actions. An individual desires private transportation. The individual orders an automobile. An automobile is manufactured, utilizing a large fraction of the elements in the periodic table, abundant or rare, benign or toxic. The consequences of vehicle manufacture include land disruption, emissions to air, water, and land, loss of inuse material to the environment (brake linings, lead balance weights, etc.). Ecosystem responses to the multitude of stimuli cover the entire range of chemical and physical processes and of spatial and temporal scales. Policymakers see dramatic short-term events (large mercury emissions, for example), but tend not to notice chronic longer-term ecosystem effects or implications for sustainability. Policies are highly specific and directed to the short term. Activity modifications are inadequate to the challenges imposed by economic and behavioral actions.

EA2 ES2 EA3 ES3 EA4 ES4 MA1 MS1 MA2

MS2

MA3

MS3

MA4 MS4

Note: * Symbols are shown in Figures 3.1 and 3.2.

factor in studies of material stocks and flows: the data are often regarded as proprietary, some potentially interesting data may never have been acquired at all, or the data may be available only at inappropriate spatial levels (only national, for example, while environmental effects often occur on the local scale). Data paucity is particularly common for the flows from manufacture to use or export, and flows of discards to reuse, recy-

An introduction

37

cling, or landfilling. Similarly, few quantitative data for industrial or national stockpiles are available, and estimates of in-use stocks are relatively rare. Data and analyses related to the sustainability of resources, arguably an important component of the E/MFA relationship with policy, is almost completely absent. In this regard, dynamic MFA studies like those of Kapur and Keoleian, Chapter 4 in this volume, are beginning to provide some of the needed data to the field. Similar efforts will form the foundation upon which further work can be built.

THE UTILITY OF E/MFA Although the discussion above emphasizes limitations to analysis, useful examples of dynamism in material analysis nonetheless exist, and their usefulness is readily apparent in showing us features of industrial systems that may not be readily apparent. For example, a dynamic study of lead flows in the Netherlands (Elshkaki et al. 2004) demonstrates that secondary resources can completely meet future lead demands. In a second example, Spatari et al. (2005) show that the amount of copper in landfill, tailings, and slag reservoirs in North America, exceeds the amount in current use, which could influence how copper is supplied over the next several decades. Dynamic E/MFA studies are challenging because they are limited by data, scale issues, and many other factors; however, such studies are becoming increasingly useful. Similarly, a number of examples exist of contributions that begin to speak to the use of dynamism in industrial ecology as it relates or could relate to policy, and an eclectic collection of examples is listed in Table 3.2. (Readers are invited to locate other examples or, even better, to generate their own.) The utility of these efforts for policy can readily be shown by the demonstration that a large peak in PVC waste can be anticipated in the Netherlands long after PVC use is phased out (Kleijn et al. 2000), or that per capita use of iron may show a saturation phenomenon in highly developed countries (Müller et al. 2006). As such studies become more numerous and conducted with increasing care, the industrial ecologist will progressively see studies of energy and material stocks and flows influence industrial and public policy in beneficial ways. It is still early days in these efforts, but industrial ecologists are making good progress, as seen in the following chapters and elsewhere.

38

Stocks and flows dynamics

Table 3.2 Published analyses of studies in materials actions and stimuli MS1

MA2

MS2

MA3

MA4

Binder et al. (2006) – relates metal flows to economic/societal drivers. Davidsdottir and Ruth (this volume) – overview of stocks and flows economic dynamics. Murakami et al. (2004) – product-focused study of flows of nineteen metals in Japan. Nishiyama (2005) – dynamic study of copper production and consumption in different countries over time. Kapur and Keoleian (this volume) – global scale, dynamic analysis of flows of a specific material. Brunner (2001) – local scale, static analysis flows of a specific substance. Moore and Luoma (1990) – case study of environmental implications of mining. Guinée et al. (1999) – country-level emissions of specific substances from manufacturing and use. Gordon et al. (2006) – US assessment of total copper stocks in landfills. van Leeuwen et al. (2005) – comprehensive review of metals in aquatic systems. Jeffree et al. (2001) – relates fish viability to aquatic metal loads. Tilton (2003) – an economist’s view of resource sustainability. Lahl and Hawxwell (2006) – discussion of the REACH legislation.

REFERENCES Binder, C.R., T.E. Graedel and B. Reck (2006), “Explanatory variables for per capita stocks and flows of copper and zinc: a comparative statistical analysis”, Journal of Industrial Ecology, 10(1–2): 111–32. Elshkaki, A., E. van der Voet, M. van Holderbeke and V. Timmermans (2004), “The environmental and economic consequences of the developments of lead stocks in the Dutch economic system”, Resources, Conservation and Recycling, 42: 133–54. Frosch, R.A. (1995), “Industrial ecology: adapting technology for a sustainable world”, Environment, 37(10): 16–24, 34–7. Gordon, R.B., M. Bertram and T.E. Graedel (2006), “Metal stocks and sustainability”, Proceedings of the National Academy of Sciences of the US, 103: 1209–14. Guinée, J.B., J.C.J.M. van den Bergh, J. Boelens, P.J. Fraanje, G. Huppes, P.P.A.A.H. Kandelaars, T.M. Lexmond, S.W. Moolenaar, A.A. Olsthoorn, H.A. Udo de Haes, E. Verkuijlen and E. van der Voet (1999), “Evaluation of risks of metal flows and accumulation in economy and environment”, Ecological Economics, 30: 47–65. Jeffree, R.A., J.R. Twining and J. Thomson (2001), “Recovery of fish communities in the Finniss River, Northern Australia, following remediation of the Rum Jungle uranium/copper mine site”, Environmental Science & Technology, 35: 2932–41.

An introduction

39

Kleijn, R., R. Huele and E. van der Voet (2000), “Dynamic substance flow analysis: the delaying mechanism of stocks, with the case of PVC in Sweden”, Ecological Economics, 32: 241–54. Lahl, U. and K.A. Hawxwell (2006), “REACH – the new European chemicals law”, Environmental Science & Technology, 40: 7115–21. Moore, J.N. and S.N. Luoma (1990), “Hazardous wastes, from large-scale metal extraction”, Environmental Science & Technology, 24: 1278–85. Müller, D.B., T. Wang, B. Duval and T.E. Graedel (2006), “Exploring the engine of anthropogenic iron cycles”, Proceedings of the National Academy of Sciences, 103: 16111–16. Murakami, S., M. Yamanoi, T. Adachi, G. Mogi and J. Yamatomi (2004), “Material flow accounting for metals in Japan”, Materials Transactions, 45: 3184–93. Nishiyama, T. (2005), “The roles of Asia and Chile in the world copper market”, Resources Policy, 30: 131–9. Obernosterer, R. and P.H. Brunner (2001), “Urban metal management: the example of lead, water, air, and soil pollution”, Focus, 1: 241–53. Spatari, S., M. Bertram, R.B. Gordon, K. Henderson and T.E. Graedel (2005), “Twentieth century copper stocks and flows in North America: a dynamic analysis”, Ecological Economics, 54: 37–51. Tilton, J. (2003), On Borrowed Time? Assessing the Threat of Mineral Depletion, Washington, DC: Resources for the Future. van Leeuwen, H.P., R.M. Town, J. Buffle, R.F.M.J. Cleven, W. Davidson, J. Puy, W.H. van Riemsdijk and L. Sigg (2005), “Dynamic speciation analysis and bioavailability of metals in aquatic systems”, Environmental Science & Technology, 39: 8545–56.

4. Dynamic modeling of material stocks: a case study of in-use cement stocks in the United States Amit Kapur and Gregory A. Keoleian INTRODUCTION Rapid industrialization and the rise of consumerism have transformed the global landscape. Since the middle of the last century in North America, Western Europe, Japan, and a few other countries, the culture of “overconsumption” has grown exponentially, bringing with it an unprecedented appetite for physical goods and the materials from which they are made (Young and Sachs 1994). In highly populated countries such as China, the levels of consumption of basic commodities such as grain, meat, coal, and steel have surpassed the previous highest consumer, the United States (Brown 2005). Material-based economies today face the challenge of how to make our needs and use of materials sustainable. The assessment of sustainability with regard to resource use and management can be broken down into three components: (1) relationship between rate of resource use and overall stock of resources, (2) effectiveness of resources in providing essential services, and (3) the proportion of resources that are lost from the economy and their impacts on the environment. The first two topics reflect the sustainability of supply, and the third affects the sustainability and health of receiving ecosystems where exiting, no longer in use, resources from the economic system accumulate and reside. The approach to establish and quantify the relationship between societal resource management and sustainability is an emerging subject of research in the field of industrial ecology (Graedel and Klee 2002). The concept of “industrial metabolism” (first introduced by Robert Ayres in Ayres 1989a; Ayres et al. 1989b; Ayres and Simonis 1994), establishes the analogy between the economy and ecosystems on a material level. Societies mobilize different types of materials from the earth’s crust to create “technomass”, similar to nature’s way of creating “biomass”. The organisms in the biosphere close the loop by cycling resources and wastes interchangeably, 40

Dynamic modeling of material stocks

41

but humans in the “anthroposphere” utilize resources as an “open loop” generating significant amounts of wastes. Industrial ecologists have begun to assess the physical economy through the lens of ecological principles (Ayres and Ayres 1996; Graedel and Allenby 2002; Socolow et al. 1994). Determining the anthropogenic contribution to natural material flows, the causal factors, and the spatial and temporal distribution of environmental concerns are the objectives of such analytical studies in IE. As a result of geochemical processes in nature, the number of available resources are not representative of the complete set of resources that are or will be utilized to meet future demand. Geochemical resources are divided into concentrated ores, many of which are currently mined, and other resources that are distributed in lower concentrations in rocks. Ultimate recovery will depend on geology, technology, and economics. It is also important to consider some other stocks of resources – employed and expended stocks (Kapur and Graedel 2006). Employed stock consists of the amount of resources extracted and mobilized from the earth and put into use, but not yet discarded. Expended stock consists of resources which were put in use but have been discarded or dissipated during use. The employed stock is categorized into in-use stock and hibernating stock components. In-use stock is the amount of a resource that is still in active use. Examples of in-use stocks include use of copper wiring, steel girders, and concrete walls in buildings. The portion of in-use stock of materials that has been removed from service but not discarded completely is referred to as hibernating stock. Examples of hibernating stocks include obsolete computers stored in closets and lead sheathing on telephone cables that are still in place but no longer connected to the network.

QUANTIFYING EMPLOYED STOCK Static and dynamic modeling are two basic approaches to quantifying employed stock. In the static method, after choosing the spatial system boundary for analysis, all principal reservoirs (for example buildings, roads, bridges) that contain the material under consideration are identified, and thereafter, the amount of material in each reservoir is determined (for example amount of cement in a building or bridge, expressed either in absolute weight terms or as a fraction of the total mass of the reservoir). After determining the amount of material in each reservoir, all the reservoir values are aggregated to determine the total material within the system boundary. The static approach is illustrated with an example to determine the in-use cement stock in building infrastructure in the United States (Table 4.1). The first step is to make an inventory of residential buildings, commercial

42

Stocks and flows dynamics

Table 4.1 Range of service lifetimes for each infrastructure use Infrastructure Residential buildings Commercial buildings Public buildings Utilities Streets and highways Water and waste management1 Farm construction2 Others3

Range of service life time (years) 50–80 60–80 80–100 30–60 40–60 60–80 60–100 40–80

Notes: 1. Includes construction, maintenance and repair of dams and reservoirs, river and harbor development and control, water supply systems, sanitary/storm sewers, water/sewer tunnels, and waste treatment facilities. 2. Includes construction, maintenance, and repair of farm service facilities. 3. Includes parks/stadiums/athletic fields, airport runways/taxiways/lighting, defense/space facilities, railroads/tunnels/signal systems, oil and gas wells, mining and quarrying.

buildings, and public buildings in terms of their number (N) and floor area (A). The next step is to determine the amount of cement in each type of building either expressed as the average amount of cement per unit floor area of the building (Wa) or as overall average weight of cement in a building (Wn). In a simplified form, the stock of cement in buildings could be expressed as Ni*Win or Ai*Wia where i  type of building. Usually, there is a lack of empirical data on parameters that quantify the material composition (such as Wa or Wn) of reservoirs. In addition, there are a number of factors such as affluence, population density, and so on, that can bring a high degree of variability to the amount of material contained in a reservoir (for example affluent people may have bigger homes). In the absence of data, reasonable approximations serve as proxy values for the material composition parameters representative of the whole reservoir, strengthened. Such informed estimates are based upon opinion of experts in the field. In the dynamic approach, material inflows to the system boundary under consideration are categorized into different end-uses (for example the amount of cement used for residential, commercial, and public buildings). Each of the end-uses are assigned a service lifetime. The lifetime determines the delay between the material inflows and material outflows in form of discards. The difference between the material inputs and discards is the net addition to in-use stock. The static approach provides a single snapshot of stocks and flows, whereas the dynamic model can be used to characterize the net addition or depletion of stocks over time. Characterization of

Dynamic modeling of material stocks

43

stocks over time can also be used to estimate future discards or emissions (Kleijn et al. 2000). Such information is useful to formulate end-of-life strategies and management systems. Resource flows for the built environment account for 70 per cent of the nonfuel, nonagricultural, and nonfood material flows in the United States (Wagner 2002; Wernick et al. 1997). Besides the built environment in the form of buildings, the critical infrastructure systems of streets and highways, water and waste management, and utilities provide the essential network of linkages to develop and sustain the viability of industrial systems. In this study, we present the methodology of dynamic modeling of stocks, and we model results for the “in-use” cement stock in the built environment (building, roads, highways, and bridges) in the United States over the period 1900–2005. There are growing concerns about the condition and performance of concrete infrastructure in the United States, with estimates showing that $1.6 trillion of investment will be required over the next five years to replace and/or rehabilitate existing structures (ASCE 2005). Twelve categories of infrastructure in the United States scored a grade of D overall in an assessment by the American Society of Civil Engineers (ASCE 2005). The state of infrastructure and the challenges policymakers and administrators in the United States face are quite indicative for the rest of the developed countries as well. A large section of people living in developing countries still lack access to basic infrastructure and usually inadequate maintenance deteriorates their condition prematurely (World Bank 1994). While previous material flow studies have examined the flows of construction materials including cement (Kelly 1998), no study has yet characterized the accumulation of cement in-use stock in the form of buildings and civil infrastructure. Information on in-use cement stocks is pertinent for stakeholders of US infrastructure such as the Federal Highway Administration, the Department of Transportation, public and private utilities, and the construction and cement industries. Since the beginning of the last century, the United States Geological Survey (USGS) has reported annual time series data on production and consumption of different types of commodities in the United States. A dynamic substance-flow analysis model (SFA) is developed using USGS data on cement consumption, and lifetime distributions for each cement end-use infrastructure application.

CEMENT USE Cement, one of the primary construction material resources, is a fine gray powder produced by heating limestone with sand or clay in a rotary kiln

44

Stocks and flows dynamics

Import/Export

Clinker/ cement

Use

New construction

Addition to stock

In-Use stock

Production Repair/ Renovation

Apparent consumption

Road Infrastructure, Building Infrastructure, Other Infrastructure Demolition

Raw materials

Production residues

Retirement of in-use stock

Construction & demolition debris

End-ofLife

Construction & demolition debris

Environment

Figure 4.1 Generic cement life cycle and grinding it with gypsum. Portland cement is the most common type of cement used. The material life cycle of cement as shown in Figure 4.1 consists of three stages: production (including raw material extraction), use, and end-of-life (waste management). At each life cycle stage, material can be exchanged between reservoirs, including environment and material imports and exports. Cement is used in new construction activities and in repair and renovation of different types of infrastructure. Each infrastructure type has a certain useful lifetime. During the lifetime of each structure, there are several cycles of repair and renovation. At end-of-life, different components of infrastructure systems are either partially or fully demolished. New construction, repair and renovation, and end-of-life activities generate construction and demolition (C&D) debris. The C&D debris is often used as filler material in road construction or as landfill for low lying areas. Cement, when combined with water, forms a paste that binds sand and gravel into a solid compound material known as concrete. Concrete usually consists of 11 to 14 per cent of cement on a weight basis. Most of the Portland cement produced in the United States (approximately 86–91 Tg/year) is utilized to make concrete (van Oss 2003). The consumption of

Consumption of cement (Tg/year)

Dynamic modeling of material stocks

45

140

140

120

120

100

100

80

80

60

60

40

40

20

20

0 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 Year

0

Source: van Oss and Kelly (2004).

Figure 4.2 Consumption of cement in the United States cement in the United States has increased six-fold over the last 50 years (Figure 4.2). Over the time period 1900–2005, total cement consumption in the United States was approximately 5.1 Pg. Concrete is the most widely used manufactured material for the construction of buildings, bridges, streets, and highways.

METHODOLOGY This study characterizes the stocks and flows of cement (as a proxy for concrete) utilized during the 20th century (1900–2005) in the United States. The law of conservation of mass is the basic principle upon which a stocks and flows model is constructed. The underlying equation can be expressed as df dt = ∑ Fi − ∑ Fo = Δ stock

(4.1)

where Fi and Fo are material inputs and outputs of the reservoir under consideration within the system boundary of the material cycle. The balance between inputs and outputs, or lack thereof, can lead to the following three conditions for change in stock (stock):

∑ Fi > ∑ Fo

(Build up of material stock)

(4.2)

∑ Fi < ∑ Fo

(Decline of material stock)

(4.3)

46

Stocks and flows dynamics

∑ Fi = ∑ Fo

(No change in material stock/steady state)

(4.4)

In the present technological society, there is generally net accumulation of material stock in the “Use” reservoir, as material inflows exceed material outflows. The material inflows and outflows for cement are described in the following two sections. Inflows The input-flow distribution for cement comprises two sub-flows, consisting of consumption of cement and its end-use distribution. Based on historical and contemporary data available from the Portland Cement Association, the cement consumption data was partitioned into various end-use markets such as roads, bridges, highways, buildings (residential, commercial, and public), and water and wastewater utilities. The contemporary cement enduse market in the United States for the year 2003 is shown in Figure 4.3. The use of cement for road infrastructure accounts for about one-third of the cement tonnage consumed. The cement consumption data divided in different end-uses is not available for all the years. These data gaps were bridged by assuming that cement end-use fractions do not change radically over several decades. Yearly changes in a few percentage points for each end-use fraction did not affect the results significantly, and the lifetime distribution was the most sensitive variable. Others 6% Farm Construction 5% Public Buildings 6% Streets and Highways 33%

Utilities 1%

Residential Buildings 31% Commercial Buildings 10% Water and Waste Management 8%

Source: Portland Cement Association (2004).

Figure 4.3 Cement end-use market in the United States in the year 2003

Dynamic modeling of material stocks

47

Outflows The outflow of cement discard streams depends upon the time delay between the input and output flows of cement. The residence time of the product usually determines the time delay. The residence-time distribution (life-span distribution) of the product is usually assumed to be a Weibull, Gamma, or Lognormal distribution (Elshkaki et al. 2005; Melo 1999; OECD 2001). Using a particular lifetime distribution, the rate of cement discards exiting the “Use” reservoir can be determined using the following equation: 8

(4.5)

j −1900

FiW , j = ∑ i =1 ∑ j =1900 ∑ k = 0 2000

FiU, j − k di, k

where  cement discard flow from sector i in the year j; Fi,Wj Fi,Ujk  cement flow into sector i in the year j  k; i  sector category index, varying from 1 to 8 for sectors including streets and highways, residential buildings, commercial buildings, public buildings, water and waste management, utilities, farm construction, and other uses; j  year of cement discard, 1900  j  2005; k  index for invoking the lifetime distribution function to estimate the retirements of in-use cement from sector i that entered use in years represented by k, 0  k  j 1900; di, k  lifetime distribution density value. There is a lack of literature and/or data that quantify actual lifetimes of various infrastructure applications. A probability of failure approach was adopted to estimate the parameters required to mathematically define a lifetime distribution. The probability of failure is the probability at which a particular infrastructure will fail to deliver the desired performance and is subsequently completely taken out of service (Nowak and Collins 2000). The three lifetime distributions – Weibull, Gamma, and Lognormal – can be characterized by at least two parameters and in order to determine them mathematically, values were assumed for the 50th and 90th percentile for the cumulative probability distribution function (CDF) (Table 4.2). An expert opinion survey1 sought a consensus on the assumed range of CDF values for each infrastructure use. The 50th percentile value was centered on the mean service life value. The Bureau of Economic Analysis in

48

Notes: 1. Derived from BEA 2003. 2. Informed estimate.

Residential Commercial Public Utilities Streets and highways Water and waste management Farm Other

End-use category

80 70 90 75 45 60 70 60

Mean service life (years) 70–90 60–80 80–100 70–80 40–50 50–70 65–75 55–65

Range for 50th percentile of cumulative probability of failure

Table 4.2 Estimation of parameters of the lifetime distribution

90–100 80–100 100–120 80–90 50–60 70–90 75–85 65–75

Percentile range of cumulative 85.8, 5.5 75.1, 4.8 95.6, 6.1 78.0, 9.9 63.2, 1.5 65.6, 4.2 72.9, 9.2 62.9, 8.0

Weilbull parameters

32.9, 2.7 26.2, 3.0 40.7, 2.4 106.6, 0.8 41.4, 1.2 19.3, 3.5 94.4, 0.8 69.9, 0.9

Gamma distribution parameters

4.38, 0.16 4.24, 0.2 4.53, 0.11 4.32, 0.10 3.8, 0.13 4.14, 0.16 4.3, 0.08 4.1, 0.11

Lognormal distribution parameters

Dynamic modeling of material stocks

49

the United States, in their assessment of the stocks and depreciation of fixed assets and consumer durables, reports mean service life values of different types of infrastructure (BEA 2003). A range of values were selected to account for plausible uncertainties in the lifetime data. Further, to characterize a single discrete lifetime distribution for each type of distribution and for each end-use, Monte Carlo simulations were run over the range of 50th and 90th percentile CDF values. The “in-use” cement stock was estimated as the difference between cement entering the “Use” reservoir and cement discards exiting the same reservoir.

RESULTS AND DISCUSSION The model-derived estimate of the in-use cement stocks in the United States is in the range of 4.2–4.4 Pg for the year 2005 as shown in Figure 4.4. This indicates that 82–87 per cent of cement utilized during the last century is still in use. On a per capita basis, this is equivalent to 14.3 to 15.0 Mg of in-use cement stock per person. To determine the accumulation of cement discards by each end-use category, the dynamic model can be further disaggregated into sub-models. Streets and highways account for the largest proportion of cement discards,

Cumulative net addition of cement in-use stock (Tg)

6000

5000

6000 Weibull Gamma Lognormal Consumption

5000

4000

4000

3000

3000

2000

2000

1000

1000

0 1900

0 1910

1920

1930

1940

1950 1960 Year

1970

1980

1990

2000

Figure 4.4 Cumulative net addition of cement stock in the United States over the time period 1900–2005

50

Stocks and flows dynamics

followed by residential and commercial buildings. The dynamic model also determines the age distribution of in-use cement stock and cement discards. For streets and highways in the year 2005, the largest amount of cement discards are from streets and highways built during 1945–75, whereas for residential buildings it is over the 1910–45 time period. This difference implies that there are older homes in existence today that have undergone several cycles of reconstruction in comparison to older streets and highways. The amount of in-use cement stock in streets and highways older than 30 years is approximately 25 per cent of the total stock. The in-use cement stock in streets and highways older than 30 years is expected to require reconstruction over the next 10–15 years. The estimates of stock assume that cement discards exit the economy completely at end-of-life. At times, however, old structures are not completely demolished and a part of the structure still remains “in-use”, such as foundations for a concrete bridge (Personal communication, Mr Hendrik G. van Oss). In the United States the number of abandoned housing units is on the rise, although reliable statistics on their number are scarce (Cohen 2001). According to a survey of over 100 cities in the United States conducted by Miami University and the University of South Carolina, more than 18 per cent of urban structures are unused (Interfire 2005). The Insurance Service Office has estimated that there are more than 21 000 “idle” properties over 15 000 sq ft in the United States (Interfire 2005). The complete picture on hibernating stocks of cement in the United States is not known, and there is further need for empirical research in this regard.

FINAL REMARKS A substance flow model is presented to illustrate dynamic modeling of stocks. Using the model, the accumulation of in-use cement stocks and cement discards over the time period 1900–2005 in the United States have been estimated. This study makes an assessment of the complete infrastructure of cement stock whereas earlier SFA studies (Brattebo et al. 2005; Müller 2006) have focused on building and housing stocks. The stock accumulation of cement is modeled as a function of cement consumption inflows into the economy, the distribution of cement end-uses, and three lifetime distributions for each of the end-uses. In the dynamic substance flow analysis model, the lifetime distribution is the most sensitive variable. The service life of an infrastructure is a measure of its durability, which is a key parameter that influences the sustainability of infrastructure systems. Often, however, there is lack of data on actual lifetime distribu-

Dynamic modeling of material stocks

51

tions. Therefore, for this study to address the uncertainty associated with lifetime distribution, the adoption of the failure approach characterized the parameters for each lifetime distribution. The results of the study indicate that the three lifetime distributions yield results in a very narrow range due to the longer time horizon of the study, smoothing out the differences between the three distributions. The dynamic model can be used to characterize the in-use stocks at different spatial levels; yet currently it cannot be used to estimate the amount of hibernating stocks. Hibernating stocks usually depend on consumer behavioral patterns and can be best estimated using comprehensive field surveys. Assuming a growth rate for cement consumption and the end-use distribution, the model can make projections of future discards. The estimates of the age distribution of the in-use stock and vintage of present and future discards provide useful information for different stakeholders. These stakeholders include the Federal Highway Administration, Department of Transportation, cement industry, and construction and demolition industry for lifecycle management of infrastructure in the United States. The model can also be utilized to project demand for materials based on the need to repair or replace ageing stock. Currently, these projections are based upon trends in the economy. The framework of the model developed is robust enough to be applied to characterize the stocks and flow of other infrastructure materials in industrial systems.

ACKNOWLEDGMENTS This research was funded through a National Science Foundation MUSES (Material Use: Science, Engineering, and Society) Biocomplexity Program Grant (Nos. CMS-0223971 and CMS-0329416). The authors would also like to thank Hendrik G. van Oss of the United States Geological Survey for his help on technical information and data for this study, and Melinda Tomaino Flores of Associated General Contractors of America for sharing data on recycling of construction and demolition debris and all those who participated in the expert opinion survey as part of this research.

NOTE 1. The experts were selected from academia, industry, and government with their expertise in the fields of civil and structural engineering, industrial ecology, material science, risk

52

Stocks and flows dynamics analysis, geology, cement industry, and economics.

REFERENCES ASCE (2005), Report Card for America’s Infrastructure, http://www.asce.org/ reportcard/2005/index.cfm. Last accessed 17 April 2005. Ayres, R.U. (1989), “Industrial metabolism”, in Technology and Environment, J.H. Ausubel and H.E. Sladovich (eds), Washington, DC: National Academy Press, pp. 23–49. Ayres, R.U. and L.W. Ayres (1996), Industrial Ecology: Towards Closing the Materials Cycle, Cheltenham, UK and Brookfield, USA: Edward Elgar. Ayres, R.U., V. Norberg-Bohm, J. Prince, W.M. Stigliani and J. Yanowitz (1989), “Industrial metabolism, the environment and application of material balance principles for selected chemicals”, IIASA Report RR-89-11, Laxenburg, Austria, p. 118. Ayres, R.U. and U.K. Simonis (1994), Industrial Metabolism, Restructuring for Sustainable Development, Tokyo, Japan: United Nations Press. BEA (2003), Fixed Assets and Consumer Durables in the United States, 1925–97, Bureau of Economic Analysis, US Department of Commerce, Washington, DC. Brattebo, H., H. Bergsdad, R. Bohne and D. Müller (2005), “MFA Dynamics of the Norwegian Dwelling Stocks”, http://www.indecol.ntnu.no/indecolwebnew/ publications/newsletter/webnews/documents/brattebo_etal_stockholm.pdf. Last accessed 21 March 2006. Brown, L. (2005), “China replacing the United States as world’s leading consumer”, http://www.earth-policy.org/Updates/Update45.htm. Last accessed 11 June 2005. Cohen, J.R. (2001), “Abandoned housing: exploring lessons from Baltimore”, Housing Policy Debate, 12(3): 415–48. Elshkaki, A., E. van der Voet, V. Timmermans and M. van Holderbeke (2005), “Dynamic stock modeling: a method for the identification and estimation of future waste streams and emissions based on past production and product stock characteristics”, Energy, 30(8): 1353–63. Graedel, T.E. and B.R. Allenby (2002), Industrial Ecology, New Jersey: Prentice Hall. Graedel, T.E. and R. Klee (2002), “Getting serious about sustainability”, Environmental Science and Technology, 36: 526–9. Interfire (2005), IAAI/USFA Vacant and Abandoned Buildings Community Group Presentation Outline, http://www.interfire.org/features/community_talk.asp. Last accessed 27 April 2005. Kapur, A. and T.E. Graedel (2006), “Copper mines above and below the ground”, Environmental Science and Technology, 40(10): 3135–41. Kelly, T. (1998), “Crushed cement concrete substitution for construction aggregates – a material flow analysis”, United States Geological Survey Circular 1177. Kleijn, R., R. Huele and E. van der Voet (2000), “Dynamic substance flow analysis: the delaying mechanism of stocks with the case of PVC in Sweden”, Ecological Economics, 32(2): 241–54. Melo, M.T. (1999), “Statistical analysis of metal scrap generation: the case of aluminum in Germany”, Resources, Conservation and Recycling, 26(2): 91–113.

Dynamic modeling of material stocks

53

Müller, D.B. (2006), “Stock dynamics for forecasting material flows – case study for housing in The Netherlands”, Ecological Economics, 59(1): 142–56. Nowak, A.S. and K.R. Collins (2000), Reliability of Structures, McGraw Hill Science/Engineering/Math, 1st edition, p. 360. OECD (2001), “Measuring capital”, OECD Manual, Measurement of Capital Stocks, Consumption of Fixed Capital and Capital Services, Paris: OECD, p. 13. Portland Cement Association (2004), http://www.cement.org/market/. Last accessed on 12 October 2004. Socolow, R., C. Andrews, F. Berkhout and V. Thomas (1994), Industrial Ecology and Global Change, Cambridge, UK: Cambridge University Press. van Oss, H.G. (2003), Cement – 2002, Minerals Yearbook, Washington, DC: United States Geological Survey. van Oss, H.G. and T. Kelly (2004), “Cement statistics: table on U.S. Geological Survey cement commodity”, http://minerals.usgs.gov/minerals/pubs/of01-006/ cement.xls. Last accessed 18 April 2005. Wagner, L.A. (2002), Materials in the Economy – Material Flows, Scarcity and the Environment, US Geological Survey Circular 1221. Wernick I., R. Herman, S. Govind and J. Ausubel (1997), “Materialization and dematerialization: measures and trends”, in Technological Trajectories and the Human Environment, J. Ausubel and H.D. Langford (eds), Washington, DC: National Academy Press, pp. 135–56. World Bank (1994), World Development Report 1994: Infrastructure for Development – Executive Summary, Volume I, Washington, DC: The World Bank.

5. The economic dynamics of stocks and flows Brynhildur Davidsdottir and Matthias Ruth INTRODUCTION Industrial ecosystems absorb large quantities of energy and materials, which are transformed by production processes into useful outflows (output) and waste. This transformation has been called industrial metabolism (Ayres 1989; Ayres and Simonis 1994). The industrial ecology community has been productive in quantifying industrial metabolism of energy and materials at various scales ranging from particular production processes (Brunner and Rechberger 2004), to specific industries or a system of industries (Brunner and Rechberger 2004; Schenk et al. 2004), to regions and cities (Kennedy et al. 2007) up to assessments of entire biogeochemical cycles (Graedel et al. 2004; Graedel and Ellis 2005). The basic principle of any energy and material flow analyses (E/MFA) is based on the law of conservation of matter. This law indicates that the sum of inflows must equal the sum of outflows, yet allowing for potential timedelays in various reservoirs (Brunner and Rechberger 2004). In the past E/MFA were mostly static or comparative static, providing snapshots of energy and material flows, and the accumulation in associated reservoirs at particular points in time. Recently the field has begun moving towards dynamic E/MFA (see for example Chapter 4 in this volume). In dynamic E/MFA inflows, outflows and accumulation in various reservoirs are linked in time as a function of quantitative assessments of service life of capital. The information contained in static and dynamic energy and material flow analyses certainly is and has been useful in various settings that range from process managers looking for wasteful nodes (processes) in an industrial system to policymakers that need formal assessment of the accumulation of materials in reservoirs at levels potentially harmful to human health (for example mercury) and/or ecosystem health (for example greenhouse gases). In these contexts, E/MFA analyses have been very useful. However, it can be argued that it may be time to add another layer of information to E/MFA, such that E/MFA analyses not only assess the quantity 54

The economic dynamics of stocks and flows

55

and physical dynamics of stocks and flows, but also enable assessment of what within and outside the industrial system influences change in stock and flow dynamics and thereby changes in stocks and flows. This new layer thus enables a move from only asking questions regarding physical size and physical drivers, towards asking how to influence flows and thereby identifying behavioral leverage points for change in the system. Industrial metabolism, and its associated physical material and energy flows, never exists and certainly does not change in isolation from economic and behavioral parameters (van den Bergh and Janssen 2004; Davidsdottir and Ruth 2005). Economic decision making and agent behavior in addition to the economic and physical realities of the system, influence change in physical flows, whether it is to improve energy or material efficiency, change the material or energy mix, or change output structure, and thereby affect energy and material flows and accumulation in reservoirs. As a result, the traditional E/MFA paradigm does not easily lend itself to assessing the impact of industrial change on changes in energy and material flows such as in a policy analyses context. Yet, similar to the lack of reference to behavior and economics in traditional E/MFA analyses, economic descriptions of industrial change largely are removed from physical flow and stock parameters, misrepresenting, in some contexts, the physical realities of production along with the associated physical properties of energy and material flows (Ayres 1978). Since physical realities influence economic decision making, economic analyses that are completely removed from physical realities may not be able to provide realistic descriptions of industrial behavior and industrial futures and the associated environmental impacts (Ruth 1993). Given the rich material already accumulated in various E/MFA analyses coupled with the long tradition in economic modeling of industrial change, great potentials lie in linking the physical flow and stock information embedded in various E/MFA studies to economic and behavioral variables. Such fusion could provide powerful descriptions of the economic and physical dynamics of industrial ecosystems, the dynamics of industrial metabolism and its in–outflows and associated environmental loads. Such rich industrial descriptions are likely to become particularly helpful in describing alternative industrial futures such as in the context of enhanced sustainability. They also are likely to become very useful to the community of energy, material and environmental policy analysts, who in many cases are relegated the task to design and assess the expected effectiveness of policies that utilize leverage points expected to reduce the environmental load of a particular system. Since traditional E/MFA analyses largely are removed from economic and behavioral parameters, they have not reached a widespread audience in this particular community of policy analysts.

56

Stocks and flows dynamics

This chapter illustrates how it is possible to link physical and economic information in the analyses of industrial systems, first by addressing various important concepts and parameters that need to be captured when linking economic parameters and physical flow analysis, and second, by presenting an empirical framework based on capital vintage accounting that has proven to be useful when linking economic and physical information. The framework is then implemented through a case study of the US pulp and paper industry. Illustrations from other industries can be found, for example, in Ruth (1998), Ruth and Amato (2002) or Ruth et al. (2004).

STOCKS AND FLOWS Industries process large quantities of energy and materials that flow into the system usually from an outside source and then flow out of the system in the form of useful output or waste. Over the long run, very little accumulates. Within the system, materials flow through various nodes that in most cases represent different stages in the production process. Within each production node materials temporarily accumulate and then flow onwards towards their next destination. The pulp and paper industry is a good example of an industrial system that has several well-defined nodes, where materials temporarily accumulate and then continue onwards to the next phase. Economic and engineering decisions at each point interact with the physical realities in the system to determine the overall performance of the industry (Figure 5.1). For example, if a paper company is using virgin fibers, the fibers come from forestry operations flowing to a node where fibers are initially prepared through debarking and chipping processes, creating pulpwood. At the next node the pulpwood is pulped, a process where fibers are separated and treated to produce wet woodpulp. The primary purpose of pulping is to free the fibers from the lignin that binds the fibers together and/or to separate the fibers in water. Different pulping processes are used depending on the fiber material and the desired end product. Also, this step in many cases contains bleaching and chemical recovery, depending on the pulping process. The wet pulp is then converted into paper at the final node in the system (Smith 1997; Ruth and Harrington 1998; Davidsdottir 2004). At each production node, materials are discarded or leak from the production cycle, creating a waste flow. The size of the initial in-flow drives the flow volume at each node, which in turn is determined by economic demand for the final product and the various input efficiencies at each node. The size of the inflow furthermore is limited by the productive capacity of the

The economic dynamics of stocks and flows Wastepaper Collection

Wastepaper

57

Paper and Paperboard

Use and Discard

Paper and Paperboard Production

Collected Wastepaper

Heat, Steam Electricity

Energy Recovery

Incineration Pulp

Landfill

Pulping Liquor

Bleaching

Pulping

Chemical Recovery

De-inked Wastepaper

De-inking Wood waste

Purchased Energy Wood Preparation

Pulpwood

Virgin Wood Legend: Forestry

Process Flow Material Flow Direction

Energy Flow Direction System Boundaries

Figure 5.1 Process nodes and flows of energy and materials in the US pulp and paper industry capital stock. Similarly, outflows, whether representing useful output or waste are affected by those same parameters. In the same vein, energy flows through each node in the system. Unlike material flows, separate energy inflows exist for each production node as fuels are frequently used as process fuels rather than as feedstock. Exceptions include production in the chemicals industry such as ethylene production that uses fossil fuels as feedstock (Ruth et al. 2004). Various parameters govern the scale and structure of energy flows, such as production levels, technology choice such as choices that dictate the use of particular materials, price of energy, and structure of the capital stock. If a paper company, for example, decides to use virgin fibers, it is likely to invest in an energy-intensive pulping process called chemical pulping. Recovery of pulping chemicals and lignin from pulping enables the industry to rely mostly on self-generated, bio-based energy. If waste fibers are the predominant fiber type, less energy is used; however, the fuel of choice in many cases is purchased electricity based on non-renewable fossil fuels (Ruth and Harrington 1998; Davidsdottir 2004; Davidsdottir and Ruth 2005).

58

Stocks and flows dynamics

As explained before, conventional material and energy flow analysis captures the flow of materials and/or energy through each production node using mass and energy balance equations and engineering parameters such as input intensity. Change in input and output flows in most cases is captured by an exogenous change in engineering parameters, assuming that each node and its associated capital equipment is perfectly malleable. Thus fixed capital (for example the machines, buildings and associated infrastructure) is assumed to be readily replaceable by new and possibly different capital, and, as a result, changes in the system can easily and instantaneously occur. When assessing the energy and material flow implications of different technologies, regardless of the dynamics of the system, the assumed malleability is appropriate. In reality, however an instantaneous switch between technologies cannot occur, but rather the economics of change and structure of the already installed capital stock influence and limit such actions. The following section reviews several key issues relevant to the economics of stock and flow adjustments.

THE ECONOMICS OF CHANGE The scale of physical flows can be thought of as a function of input intensity, defined as total flow size (volume or weight), divided by total output, and output volume. The economics of change then facilitate change in each parameter over time. Change in input intensity commonly is decomposed into three separate components: change driven by technological change, structural change, and input substitution (Farla et al. 1997), all of which are affected by capital vintage dynamics in addition to other economic and engineering factors. Other factors, such as capital inertia and path dependency or technology lock-in, limit opportunities for change. Output demand for various products also influences change in the scale and structure of physical flows, but again, capital stock dynamics define the opportunity for change. For example, the US pulp and paper industry operates at over 90 per cent capacity and production equipment is rather specialized to specific paper grades. As a result, the industry has little room to increase output volume or change production lines (change output structure) without changing or expanding the existing capital stock. As a result, any change or increase in most cases requires substantial capital investment (Davidsdottir and Ruth 2004; Smith 1997). In addition, most products within the pulp and paper industry are linked to a particular production process and thus to a particular type of capital, indicating that a shift in output structure, as well, requires substantial capital investment.

The economic dynamics of stocks and flows

Leakage

59

Leakage Old Capital

New Capital

Retired

Added Age 1

Age 2

Age 3

Figure 5.2 Schematic representation of the capital stock Consequently, whether the focus is on changes in intensity or the impact of changes in demand, investment behavior and capital vintage dynamics are major components in any analysis of the economic dynamics of energy and material flows. Capital Vintage Dynamics Each industry or industrial system consists of different age cohorts of capital, called capital vintage, just as the human population consists of different age cohorts of individuals. Each capital vintage is characterized by vintage specific attributes such as input intensity, output volume, output structure and other factors, such as capital utilization and depreciation rates (Davidsdottir and Ruth 2004, 2005). An industrial system evolves as new capital is added to the existing stock of capital that contains different attributes than pre-existing vintages and when old capital is retired and thus removed from the capital stock (Figure 5.2). Retirement in most cases is gradual and thus leakage occurs from each vintage, until the vintage is completely retired. Unlike nature, industrial evolution is neither spontaneous nor random but driven by goal specific and path dependent investment behavior. In addition, the pre-existing capital stock often shapes the character of new capital being installed. Consequently, the evolution of any industrial system is closely linked to the character of existing capital vintage, and without explicit descriptions of its structure, models are likely to overstate the potential for change. Capital Investment Capital investments are often distinguished by their primary productive purpose, either as expansion investment or replacement investment. Both add to the front end of the capital stock and are represented on Figure 5.2 as new capital added. Expansion investment enlarges the productive capacity of the capital stock, whereas replacement investment replaces old, retired capital. New capital investment changes the average character of the

60

Stocks and flows dynamics

capital stock, in most cases as new capital carries slightly different attributes from older capital. Investment theory describes different parameters that influence the extent of expansion investment. Four competing theories have been constructed regarding the estimation of expansion investment: capacity utilization theory, neoclassical theory, liquidation theory and the profits theory. All four demonstrate expansion investment as a function of the difference between ideal and actual size of the capital stock. According to capacity utilization theory, ideal size is proportional to output levels. The neoclassical theory describes ideal size as a function of the value of output from the firm, the cost of capital, and lagged capital investment. The liquidity theory describes ideal capacity to be proportional to the flow of internal funds that are available for investment, and the expected profits theory relates ideal capacity as a function of the expected value of the firm, with expansion investment increasing as the value of the firm increases (see an excellent and still valid overview in Jorgenson (1996)). Jorgenson (1996) compared the predictive powers of each of the four theories for various industrial sectors in the United States and confirmed that the capacity utilization and the neoclassical theories outperform the others. Regardless of what drives the increases in expansion investment, by definition, it expands the size of the capital stock and increases production capabilities in the system. As a result, expansion investment, ceteris paribus, expands the size or volume of energy and material flows that flow through the system. Research shows that industries tend to invest in slightly more efficient capital than before, and thus expansion investment is expected to somewhat improve overall average material and energy efficiency in the system (de Beer 1998; Lempert et al. 2002; Sakellaris and Wilson 2004). Replacement investment does not expand the size of the physical capital stock as it directly replaces retired capital and thus physically equals the productive capacity of retired capital or capital that has been removed from the capital stock via leakage in each given time-period. Thus, replacement investment consists of both the removal of old, less efficient capital, and the addition of new, more efficient capital. Since industries tend to invest in more efficient capital that what is being removed from the stock, replacement investment is expected to more extensively increase energy and material efficiency than expansion investment, as well as, ceteris paribus, also reduce the total flow of material and energy inputs. Given the close link between replacement and retirement, vintage specific retirement rates govern the extent of replacement investment. In sum, industrial systems, driven by replacement and expansion investment, gradually change or “evolve”, modifying the attributes of each vintage, such as input efficiency. Vintage attributes, combined with output

The economic dynamics of stocks and flows

61

levels and output structure, determine the total flow size of energy and materials moving through the system. Structural Change Structural change is defined as a change in the mix of outputs coming from an industry. For example, structural change in the pulp and paper industry could encompass a shift away from producing newsprint towards producing more printing and writing papers. Since the fiber and energy requirements differ substantially for those two products as they are produced using different pulping processes (Ruth and Harrington 1998; Davidsdottir 2004), such a shift would influence both energy and material intensity as well as the character of those flows. A modification in the structure of output, in most cases, is a response to changes in both demand and in the external economic environment, such as changes in the price of factor inputs. The rapidly rising price of energy, for example, is likely to lead the US automobile industry to shift its output lines towards increased production of hybrid vehicles, a direct response to changes in demand. Yet if a shift from one output product to another one requires a change in the structure of the capital stock, an industry cannot instantaneously shift from one output product to another. As a result, the vintage structure of the existing capital stock greatly limits the output flexibility of industrial systems. Input Substitution Substitution between factor inputs also facilitates change in material and energy flows and intensity. Input substitution is defined as an increase in the use of one input and a decline in another one, holding output structure and volume constant. Input substitution can also occur within specific factor inputs, for example in the US paper industry from residual fuel oil to natural gas. Factor inputs consist of energy, material, capital and labor, and the ease of substitution differs greatly. For example, substitution between materials and energy, or materials and labor, in most cases is difficult. Capital can, however, frequently accommodate some substitution between energy and labor. The laws of thermodynamics dictate the ultimate extent of possible substitutions among factor inputs, as a minimum amount of both energy and materials is required for industrial systems to function. As in the case of structural change, the relative price of factor inputs drives input substitution, and the structure of existing and new capital vintages limits substitution possibilities. Three main variants of input substitution exist: putty–putty, putty–clay, and clay–clay, where the difference

62

Stocks and flows dynamics

between those three is determined by the ease of substitution within different age classes of capital (Atkeson and Kehoe 1999). First, putty– putty substitution is where it is equally feasible to substitute inputs within both new and old capital vintages, and thus the capital stock is completely malleable. In other words, the old and new capital stock is equally adjustable. This kind of substitution is most applicable when the systems analysis is performed at a very aggregate, possibly economy-wide level and when the substitution in question is not between energy (or labor and capital) and materials, but rather between energy, labor, and capital. The second variant is putty–clay. In this case, substitution is only feasible for new vintages (putty), but factor proportions are fixed for older capital. This type of substitution is best applicable for industry-wide analysis, where different technologies within particular production processes are not disaggregated, and the substitution in question is not between materials and any of the other factors of production. The third kind of substitution is called clay–clay, where substitution for new or old vintages is not possible. As a result, the proportions of factor inputs are fixed. The relationship between energy and materials in disaggregated industrial systems analysis, where the unit of analysis is individual technologies, is commonly modeled as clay–clay, since those inputs are not easily substitutable for one another. Intra-input substitution, however, is in most cases easier than substitution between different input types. Consequently, substitution within energy and material bundles is often best classified as putty–semi-clay (Ruth et al. 2004). Technological Change Technological change is defined as a reduction in the use of an input, holding the volume and mix of other inputs and output constant. Technological change can either be embodied or disembodied (Solow 1957; Meijer 1994; Berndt et al. 1993; Davidsdottir and Ruth 2004), where embodied change only influences new capital vintages and disembodied change only influences the efficiency of already installed “existing” capital. Both have significant impact on the efficiency of energy and material use, and thus both need to be included in a dynamic analysis of material and energy flows. Given the different character of each type, a disaggregation of the two is preferred when assessing the dynamics of technological change. Embodied technological change implicitly includes the three stages of technological change: invention, innovation, and diffusion (Schumpeter 1939). Invention is defined as the creation of a new technology. Innovation is the commercialization of a newly invented technology and through

The economic dynamics of stocks and flows

63

diffusion, which occurs as investment, the innovated invention becomes embodied into the capital stock. Each stage requires substantial capital investment and thus embodied technological change occurs through significant direct and indirect capital investments. Since embodied change occurs through new capital investment, it only influences the input efficiency of the youngest capital vintages. Both expansion and replacement investment facilitate embodied change as both add to the front-end of the productive capital stock. As stated before, replacement investment reduces the size of existing vintage but adds new capital to the capital stock facilitating embodiment. Thus the rate of capital replacement and the corresponding volume of replacement investment has a significant impact on the rate of embodied technological change. Sakellaris and Wilson (2004) present evidence of substantial embodiment. They find that each vintage is about 12 per cent more productive on average than older equipment on average and that embodied technological change accounts for two-thirds of all output growth in the United States. Similarly, Greenwood et al. (1997) demonstrate that embodied technological change accounted for 58 per cent output growth in the US between 1954 and 1990. Significant investmentdriven technological change indicates the importance of embodiment when modeling economic dynamics of stocks and flows. In addition, it reinforces the importance of investment driven environmental policy when regulating changes in the environmental load of industrial processes, and at the same time illustrates the danger of adopting policy measures that will reduce the rate of embodied change (Davidsdottir and Ruth 2005). Apart from being influenced by investment behavior, the extent of embodied technological change also depends on technology choice. Technology choice is affected by various factors, but the most important ones are input prices and profitability, risk and path dependency (Arthur 1994; Unruh 2000; Kuper and Soest 2003). Disembodied technological change is defined as a low or no cost change in the input efficiency of the “existing” capital stock (Ross 1991). Therefore disembodied changes only can influence the efficiency of “older” vintages. Disembodied changes occur not as a result of retrofits or replacement investment but as a function of regular minor maintenance (Lempert et al. 2002) and of low- or no-cost operational changes (Ross 1991). Such changes can be in the form of improved housekeeping practices and typically do not require a substantial investment in fixed capital structures. Examples include improved insulation and the use of more energy-efficient light bulbs. Since disembodied change is relatively cheap and is easily implemented, short-term fluctuations in input prices coupled by learning-bydoing are key influences on this change. For example, when energy prices increase, managers try to conserve energy by using more energy efficient

64

Stocks and flows dynamics

light bulbs. Because of disembodied changes, recently installed capital equipment can be more efficient in practice than when it had just been installed (Meijer 1994; Pakes and Griliches 1984; Lempert et al. 2002). According to Sakellaris and Wilson (2004), disembodied technological change accounted for about one-third of all productivity increase in the United States between 1972 and 1996. Learning, Path Dependency, and Capital Inertia Important additional parameters that influence the speed and direction of technological change are path dependency, technology lock-in, and capital inertia in addition to learning. All parameters need to be accounted for when modeling the economic dynamics of stocks and flows. Path dependency characterizes technology choice in mature industrial systems and manifests itself in repetitive investment patterns, where industries invest in similar technologies as invested in before (Arthur 1994; Unruh 2000). Due to path dependency, investment choices made early on in the development of an industrial system gradually rigidify, become “locked-in”, and thus past investments and the existing capital stock defines the structure of future investments. Accordingly, investment behavior and each capital unit installed influence energy and material flows far into the future (Norberg-Bohm and Rossi 1998; Davidsdottir and Ruth 2005). This feature can be seen in most energy-intensive industries in the United States, such as the pulp and paper industry, cement, iron and steel industries, and chemical production industries (Ruth et al. 2004, 2005). The combination of path dependency, technology lock-in, and a longlived capital stock creates capital inertia. A long-lived capital stock means that the vintage structure of the capital stock is long, and this length functions like a buffer to any change that occurs as a result of new investment. Combined with path dependency, capital vintage often acts as a deterrent to change, reducing the ability of any industrial system to react swiftly to changes in the external economic and environmental environment, and limiting the ability to switch from one technology to another. Capital inertia therefore reduces the rate of technological change, making it more gradual than radical and ensuring that the physical input efficiency of the capital stock changes at a slow gradual phase (Norberg-Bohm and Rossi 1998; Davidsdottir and Ruth 2005). Learning influences both embodied and disembodied technological change (Davidsdottir and Ruth 2005) and is characterized ether as learningby-doing or investment learning. Learning-by-doing has a central role in disembodied technological change (Meijer 1994; Ruth et al. 2004; Davidsdottir and Ruth 2004) and describes that the efficiency of production

The economic dynamics of stocks and flows

65

processes increase when individuals become more adept at doing certain tasks as they gain experience. Investment learning mostly influences embodied technological change. Such learning occurs when investment in a certain process increases, and experience is gained. This increase in experience results in declining operational and investment costs and in addition reduces the risks involved for future investors. As a result diffusion rates increase.

CAPITAL VINTAGE MODELS Background In the previous section, a review was conducted on some relevant components when integrating the economics of stocks and flows into physical flow analysis. The section revealed that the structure of the capital stock (past investment behavior) and capital vintage dynamics define the dynamics of energy and material flows. Consequently, to enable a dynamic description of material and energy flows, the following must be taken into account: the structure and “evolution” of the capital stock, capturing capital vintage specific features such as input efficiency, and the size of each vintage and capital utilization rates. The challenge is to find a paradigm that enables simultaneous modeling of the flows of energy and materials and capital vintage dynamics. Fortunately this can be done using capital vintage models. Capital vintage models were first developed in the 1950s and 1960s (for example Johansen 1959; Kaldor and Mirrlees 1962) and have recently been revived to analyze physical energy and material flows in industrial systems, effectively linking economic and physical information (for example Ruth et al. 2004; Ruth and Amato 2002; Davidsdottir and Ruth 2004, 2005, 2006; Davidsdottir 2004). The stated purpose of the new generation of vintage models was “to inform policy and investment decisions at the industry level” (Ruth et al. 2004). When applying capital vintage models to assess the dynamics of material and energy flows, it is initially important to identify the nodes pertinent to the material and/or energy flow analysis. The next step is to explicitly break the capital stock into specific age cohorts (vintages) by each previously identified process node and to establish the structure of the existing capital stock by assigning attributes. Each preexisting capital stock unit is assigned vintage-specific attributes which are either econometrically estimated or based on engineering data, such as production capacity, rate of capital replacement, input efficiency, and input substitution flexibility.

66

Stocks and flows dynamics

After identifying the initial structure and attributes of the existing capital stock by system node, the system “evolves” over time as the capital stock changes expansion investment and through the gradual replacement of old, obsolete, or worn out structures (replacement investment). New attributes are assigned to new capital equipment, which due to path dependency partly depend on the structure of the already embodied capital stock. The attributes of the existing capital stock also gradually change as a function of new economic realities and also as a function of age. For example, replacement rates are age-specific and as each capital unit becomes older, more and more of the initial investment is gradually replaced, until the entire investment has been retired. Input efficiency of each already installed capital unit also changes as a function of disembodied changes in efficiency. As the capital stock changes over time, the energy and material input efficiency of each node in the system changes as well. Also, in combination with output levels, capacity utilization and output structure determine the dynamic changes in the size of material and energy flows in addition to waste flows flowing through the system. Illustration As stated above, capital vintage models have recently been adapted to link physical flow modeling to the economics of change (Ruth et al. 2004, 2005; Ruth and Amato 2002; Davidsdottir and Ruth 2004, 2005, 2006; Davidsdottir 2004). This section briefly summarizes one of those models, a model of the US pulp and paper industry and some of the implications that illustrate the importance of the approach. This section is based in part on previously published analyses (Davidsdottir and Ruth 2004, 2005, 2006). General System Description – the US Pulp and Paper Industry The US pulp and paper industry is regionally heterogenous, mature, and capital-intensive (Smith 1997; Davidsdottir 2004). For this particular model the overall industrial system is broken into the eight census regions in the United States, and production within each region is disaggregated into four paper categories (newsprint, tissue, printing and writing, packaging and industrial paper), and four paperboard categories (kraft, bleached kraft, semichemical and recycled). The industry is one of the most capital-intensive industries in the United States with capacity utilization rates averaging over 90 per cent. In this study, the regional size and physical vintage structure of the capital stock is accounted for in annual increments dating back to 1950. Region-specific

The economic dynamics of stocks and flows

67

rates of expansion and replacement investment, coupled with vintage and process-specific input efficiencies, drive the “evolution” of the system. Material inputs primarily consist of waste or virgin fibers with a waste fiber utilization rate (WUR) averaging 35 per cent, while varying regionally from 20 to 56 per cent. The model captures the physical flow of materials by type. Virgin materials flow from forestry operations to the pulp and paper-making process and flow out of the system as wastepaper. The model captures the flow of wastepaper and traces its fate, either into the industrial system again or into the solid waste stream. After wastepaper ends up in the solid waste stream, it is either incinerated or put into landfills. Total regional physical energy use by type is also captured, broken to selfgenerated energy (most commonly as a byproduct to chemical pulping processes) and five different purchased fuels. Fuel intensity and fuel mix is regionally heterogeneous, where the percentage share of self-generated energy in total fuel use ranges from 70 to 30 per cent. In sum, the model traces simultaneously, by region, the physical flow of energy and materials by type and captures the regional structure of the capital stock and output levels/mix. Waste flows consist in this case of carbon dioxide from the burning of fossil fuels and methane and carbon dioxide emissions from paper decay in landfills. Energy used in the transport of paper products to the market, in the collection of wastepaper and in the forestry sector, is considered outside of system boundaries (also see Figure 5.1). Model Structure The model contains the following five interacting modules: 1.

2.

Production module – simulates regional production levels/growth by product type using a reduced form production function, where output is a function of lagged input prices, regional income discounted by distance to a demand region, lagged production levels, and subject to total installed regional productive capacity. Physical vintage module – describes the regional size of each capital vintage, initially broken into regional annual investments back to 1950 that still are potentially productive. Changes in the size of existing vintages are assessed using physical perpetual inventory where the size of each new vintage (gross investment) is a function of replacement and expansion investment. Vintage specific replacement rates are econometrically estimated using a Gumpertz curve and are a function of the age of capital (thus vintage-specific) and input prices. Thus the capital stock changes as a function of replacement and expansion investment.

68

3.

4.

5.

Stocks and flows dynamics

Input intensity module – each vintage as it enters the capital stock is assigned a set of vintage and region-specific input intensity parameters. The input intensity module thus relates, by vintage, the physical input intensity of each input type to the size of each vintage, giving the total intensity of the capital stock, and, in combination with the production module, giving total use of each input type. The input intensity of each new vintage (embodied change) depends on process-specific relative energy intensities of new to old capital, the relative importance of each process in total production, and the weighted average embodied intensity of existing capital in addition to the size of new investment. Incorporating all those factors captures both the impact of path dependency and learning on the future direction and speed of embodied change. Input intensity also changes as a function of disembodied change, which is econometrically estimated as a function of learningby-doing and input prices. Total change in intensity is thus a function of both embodied and disembodied technological change. Input mix module – breaks total energy and fiber use into vectors of different energy and fiber types, and then simulating changes in the input mix, for example the switch between process fuels and from virgin to waste fibers. The switch between energy types is econometrically estimated and simulated as the simultaneous change in the fractional shares of each energy type, driven by relative prices and output mix. The switch between waste and virgin fibers is estimated as an incremental movement of the fractional share of wastepulp towards (or away from) region-specific maximum waste paper utilization rates (Ruth and Harrington 1998). The path towards that maximum is estimated econometrically using the Fisher and Pry technology substitution model and driven by input prices, cumulative wastepaper use and recycling legislation. To achieve mass balance in the system, waste fibers are assumed to only originate in US-produced sources and thus produced and used paper products go back into the potential pool of fibers with a one year time lag, where wastepaper use cannot exceed last year’s production of paper. The wastepaper that is not recycled is either incinerated or landfilled. Greenhouse gas emissions module – describes emissions of methane and carbon dioxide from landfilled paper using the EMCON Methane Generation Model (EMCON 1982) and emissions from wastepaper incineration and emissions from the use of process fuels by type using fixed carbon emissions coefficients.

Each module exists for each census region and contains region-specific equations, which are solved simultaneously for each future year to yield

The economic dynamics of stocks and flows

69

trajectories for system inputs and outputs. Parameter estimates are based on time-series analysis of 27 years of historical data (1972–98), vintagebased capital analysis and engineering/physical information. Variants of this model have been used to assess, for example, the impact of climate change policies on the industry (Ruth et al. 2000), capital inertia and climate change policies (Davidsdottir and Ruth 2004), the feedback between energy and material flows and capital vintage and implications for policy design (Davidsdottir 2002) and the impact of industrial inertia on environmental performance (Davidsdottir and Ruth 2006). The next section highlights some of the insights gained from using this approach. Model Insights Various insights were gained during parameter estimation and model simulations. In particular, it became clear that for a simultaneous modeling of energy and material flows and economic variables, capital vintage is a central component. Furthermore, the regional differences in the physical character of production make a difference when assessing the economic behavior of the industry, especially in the light of policy interventions. Parameter estimation revealed that energy prices are instrumental in shaping production and investment decisions, in addition to being the main driver behind input substitution. Energy prices are seen to negatively affect expansion investment but positively affect replacement investment. Thus depending on the structure of the capital stock an increase in energy prices may or may not increase embodied technological change and thereby reduce energy and material flows. Since embodied technological change accounts for, on average, 78–83 per cent of total technological change in the US pulp and paper industry, with the remainder accounted for by disembodied technological change, this clearly illustrates the potential negative effects of an increase in energy prices for any reason. Energy prices are, however, seen to positively affect disembodied technological change. An increase in energy prices is also expected to increase production levels in areas that use more virgin fibers, but regions that rely on waste fibers and thereby self-generate less of their own energy needs will experience a reduction in production levels with the associated reduction in waste fiber utilization rates (WURs). Energy prices and energy use also significantly influence the fiber mix, whereas an increase in energy prices slows down the expansion to increased wastepaper use. Overall, an increase in regional energy prices is seen to reduce WURs, stimulating an increase in the share of virgin fibers and an increase in the share of self-generated energy. Thus, feedback relationships

70

Stocks and flows dynamics

clearly exist between energy use, energy prices, and material use that only would be seen in an analysis that simultaneously captures energy and material flows and links those to economic variables. This observation has important implications for policy design and the analysis of regional impact of policy as the physical character of the industry differs greatly between different regions in the United States. A change in energy intensity is impacted by both embodied and disembodied technological change, but on average, energy intensity is expected to change as a result of technological change, at a region-specific rate of 0.3– 0.7 per cent annually. Furthermore, since embodied technological change accounts for, on average, 78–83 per cent of total technological change in the US pulp and paper industry, investment (either replacement or expansion) in new capital is the main driver of technological change and thus embodied technological change is an important driver in overall industrial change and thus changes in energy and material flows. The long lifetime of capital, however, in this industrial system greatly limits the speed by which embodied technological change occurs. The analyses clearly indicate that an increase in the rate of capital turnover is the most significant factor in permanently changing energy and material use profiles in the pulp and paper industry (see similar conclusions in for example Nystrom and Cornland 2003 and Worrell and Biermans 2005) due to the long lifetime of capital in the sector, low rates of capital turnover and high capital intensity. The immense capital intensity and capital inertia ensure that the capital stock will change slowly and thereby only gradually improve energy, carbon, and fiber efficiency. This indicates substantial capital vintage effects and highlights the importance of taking into account the physical realities of industrial systems when assessing the potential for change and the potential response to various energy, material or environmental policies. In conclusion, this illustration of a capital vintage model of the US pulp and paper model shows that policies aimed at enhancing the longterm sustainability of this particular industrial system need to facilitate faster turnover of old capital, which could simultaneously increase the efficiency of energy and material use. Given a continued growth in the industry without incentives to replace existing capital, technological change will remain incremental and material and energy use and waste emissions such as carbon emissions from the pulp and paper industry will continue to increase into the foreseeable future. Yet such policies are likely to go unnoticed if the modeling tools used for policy analysis do not capture both energy and material flows, the physical realities of the industrial system, and the economic parameters that affect change in the system.

The economic dynamics of stocks and flows

71

CONCLUSION This chapter has reviewed selected economic parameters that should be considered when combining the economics of stocks and flows into physical flow analysis of stocks and flows. An examination of the economic dynamics of stocks and flows was made, with a particular emphasis on capital vintage and the economics of change. A capital vintage framework for incorporating economic and physical variables was introduced which clearly enabled easy integration of physical and economic variables. An implementation of a capital vintage model was introduced through a case study of the US pulp and paper industry. Finally some of the implications of the case study were discussed and important insights gained from the approach were highlighted. The advantages for policy analysis of integrating economic and physical modeling and enabling the simultaneous dynamic modeling of energy and material flows, are clear. To be able to envision the potential impact and response to policy and thereby to envision changes in industrial futures due to a change in the policy environment necessitates that, at a minimum, we: (1) clearly understand the limitations of an industry to change as a result of for example capital vintage effects; (2) understand feedback relationships between energy and material flows and the links to economic variables such as energy prices; (3) understand the relative importance of the various economic factors that influence changes in physical energy and material efficiencies. As a result, an integration of E/MFA and the economics of stocks and flows may make industrial analysis more relevant, or simply necessary in the identification and selection of industrial policy options and to the discussion of how to reduce the environmental load of various industries through the means of policy.

REFERENCES Ayres, R.U. (1978), “Application of physical principles to economics”, in R. Costanza, C. Perrings and C. Cleveland (eds), Development of Ecological Economics, 1997, Aldershot, UK and Brookfield, US: Edward Elgar. Ayres, R.U. (1989), “Industrial metabolism”, in J.H. Ausubel and H.E. Sladovich (eds), Technology and Environment, Washington, DC: National Academy Press, pp. 23–49. Ayres, R.U. and U. Simonis (eds) (1994), Industrial Metabolism: Restructuring for Sustainable Development, Tokyo: United Nations University Press. Arthur, B. (1994), Increasing Returns and Path Dependence in the Economy, Ann Arbor, MI: The University of Michigan Press. Atkeson, A. and P.J. Kehoe (1999), “Models of energy use: putty–putty versus putty–clay”, American Economic Review, 89(4): 1028–43.

72

Stocks and flows dynamics

Berndt, E., C. Kolstad and J.L. Lee (1993), “Measuring energy efficiency and productivity impacts of embodied technological change”, Energy Journal, 14(1): 33–55. Brunner, P. and H. Rechberger (2004), Practical Handbook of Material Flow Analysis, Boca Raton, FL: Lewis Publishers. Davidsdottir, B. (2002), “A vintage analysis of regional energy and fiber use, technology change and greenhouse gas emissions”, Boston, MA, Boston University, PhD dissertation. Davidsdottir, B. (2004), “Forest products and energy”, in C. Cleveland (ed.), Encyclopedia of Energy, Volume 2, Amsterdam and Boston, MA: Elsevier Academic Press. Davidsdottir, B. and M. Ruth (2004), “Capital vintage and climate change policies: the case of US pulp and paper”, Environmental Science and Policy, 7(3): 221–33. Davidsdottir, B. and M. Ruth (2005), “Pulp nonfiction: regionalized dynamic model of the U.S. pulp and paper industry”, Journal of Industrial Ecology, 9(3): 1–21. Davidsdottir, B. and M. Ruth (2006), “Industrial inertia and environmental performance: opportunities and constraints for environmental policy”, Special Issue of International Journal of Environmental Technology and Policy, submitted. de Beer, J.G. (1998), Potential for Energy-Efficiency Improvement in the Long-Term, Utrecht, Utrecht University, The Netherlands, PhD Thesis. EMCON (1982), Methane Generation and Recovery from Landfills, Ann Arbor, MI: EMCON Associates, Ann Arbor Science Publishers, Inc. Farla, J., K. Blok and L. Schipper (1997), “Energy efficiency developments in the pulp and paper industry”, Energy Policy, 25(7–9): 745–58. Graedel, T.E. and D. Ellis (2005), “Copper resource dynamics in the 20th and 21st centuries”, Copper in the 21st Century, The Minerals, Metals, and Materials Society, Warrendale, PA. Graedel, T.E., D. van Beers, M. Bertram, K. Fuse, R.B. Gordon, A. Gritsinin, A. Kapur, R. Klee, R. Lifset, L. Memon, H. Rechberger, S. Spatari and D. Vexler (2004), “The multilevel cycle of anthropogenic copper”, Environmental Science & Technology, 38: 1253–61. Greenwood, J., Z. Hercowitz and P. Krusell (1997), “Long-run implications of investment specific technological change”, American Economic Review, 87(3): 342–62. Johansen, L. (1959), “Substitution versus fixed production function coefficients in the theory of economic growth: a synthesis”, Econometrica, 27(2): 157–76. Jorgenson, D.W. (1996), Investment, Cambridge, MA: MIT Press. Kaldor, N. and J.A. Mirrlees (1962), “A new model of economic growth”, Review of Economic Studies, 29: 174–92. Kennedy, C., J. Cuddihy and J. Engel-Yan (2007), “The changing metabolism of cities”, Journal of Industrial Ecology, 11(2): 43–59. Kuper, G.H. and D.P. Soest (2003), “Path dependency and input substitution, implications for energy policy modeling”, Energy Economics, 25: 397–407. Lempert, R.J., S.W. Popper and S.A. Resetar (2002), Capital Cycles and the Timing of Climate Change Policy, Pew Center on Global Climate Change, Washington, DC. Meijer, H. (1994), On the Diffusion of Technologies in a Vintage Framework, Maastricht, University of Maastricht, unpublished PhD dissertation. Norberg-Bohm, V. and M. Rossi (1998), “The power of incrementalism: environmental regulation and technological change”, Technology Analysis and Strategic Management, 10(2): 225–45.

The economic dynamics of stocks and flows

73

Nystrom I. and D. Cornland (2003), “Strategic choices: Swedish climate intervention policies and the forest industry’s role in reducing CO2 emissions”, Energy Policy, 31: 937–50. Pakes A. and Z. Griliches (1984), “Estimating distributed lags in short panels with an application to the specification of depreciation patterns and capital stock constructs”, Review of Economic Studies, 51(2): 243–62. Ross, M. (1991), “Efficient energy use in manufacturing”, Proceedings of the National Academy of Science, 89: 827–31. Ruth, M. (1993), Integrating Economics, Ecology and Thermodynamics, Dortrecht, The Netherlands: Kluwer Academic Publishers. Ruth, M. (1998), “Energy use and CO2 emissions in a dematerializing economy: examples from five US metals sectors”, Resources Policy, 24: 1–18. Ruth, M. and A. Amato (2002), “Vintage structure dynamics and climate change policies: the case of US iron and steel”, Energy Policy, 30: 41–52. Ruth, M. and T. Harrington (1998), “Dynamics of material and energy use in US pulp and paper manufacturing”, Journal of Industrial Ecology, 1(3): 147–68. Ruth, M., A. Amato and B. Davidsdottir (2002b), “Carbon emissions from US ethylene production under climate change policies”, Environmental Science and Technology, 36(2): 119–24. Ruth, M., B. Davidsdottir and S. Laitner (2000), “Impacts of energy and carbon taxes on the US pulp and paper industry”, Energy Policy, 28: 259–70. Ruth, M., B. Davidsdottir and A. Amato (2004), “Climate change policies and capital vintage effects: the cases of US pulp and paper, iron and steel and ethylene”, Journal of Environmental Management, 70: 235–52. Sakellaris, P. and D.J. Wilson (2004), “Quantifying embodied technological change”, Review of Economic Dynamics, 7: 1–26. Schenk, N.J., H.C. Moll and J. Potting (2004), “The nonlinear relationship between paper recycling and primary pulp requirements: modeling paper production and recycling in Europe”, The Journal of Industrial Ecology, 8(3): 141–62. Schumpeter, J.A. (1939), Business Cycles: A Theoretical Historical and Statistical Analysis of the Capitalist Process, New York, NY: McGraw Hill. Smith, M. (1997), The US Paper Industry and Sustainable Production: An Argument for Restructuring, Cambridge, MA: MIT Press. Solow, R.W. (1957), “Technical change and the aggregate production function”, Review of Economics and Statistics, 39: 312–20. Unruh, G.C. (2000), “Understanding carbon lock-in”, Energy Policy, 28: 817–30. van den Bergh, J. and M.A. Janssen (2004), Economics of Industrial Ecology, Cambridge, MA: MIT Press. Worrell E. and G. Biermans (2005), “Move over! Stock turnover, retrofit and industrial energy efficiency”, Energy Policy, 33(7): 949–62.

PART III

Agent-based Analysis of Dynamic Industrial Ecosystems

6. Agent-based analysis of dynamic industrial ecosystems: an introduction Marco A. Janssen INTRODUCTION The agent-based approach, which is used in the chapters of this part, explicitly studies the emergent macro-level phenomena from interactions at the microlevel between autonomous agents. Individual attributes and strategies of the agents can influence the emergent patterns, the information derived by the agents, and the structure of the network of agents. Agent-based analysis can be performed by conceptual (Chapter 9) as well as computational approaches (Chapters 7 and 8). The computational approach is referred to as agent-based modeling. Agent-based modeling within social science has its roots in the early 1970s. Initially developed during the 1940s, the technical methodology of computational models of multiple interacting agents arose when John von Neumann started to work on cellular automata (von Neumann 1966). A cellular automata is a set of cells where each cell can be in one of multiple predefined states, such as forest land or farm land. Changes in the state of a cell occur based on the prior states of the cell’s own history and the history of neighboring cells. Cellular automata became more popular in light of a creative application by John Conway, named the Game of Life (Gardner 1970). The Game of Life illustrated how the following simple rules of local interaction could lead to the emergence of complex global patterns. In contrast to cellular automata, agent-based models enable a researcher to examine the heterogeneity of agents beyond their specific location and history. A pioneering contribution is the work of economist Thomas Schelling (1971, 1978) who developed an early agent-based model by moving pennies and dimes on a chessboard according to certain simple rules. His model had a surprising result: although each agent tolerated neighbors who are different (being a penny or a dime) and thus only moved when they had many dissimilar neighbors, the population ended up in 77

78

Agent-based analysis of dynamic industrial ecosystems

segregated groups. Political scientist Robert Axelrod made a major contribution with his repeated prisoner dilemmas (PD) tournaments (Axelrod 1984). Axelrod invited scholars from all over the world to submit strategies that would be programmed to play repeated prisoner dilemma games against other submitted strategies. The winner in two successive experiments (submitted by Anatol Rapoport) was the simple rule, tit-for-tat. Players in a two-person repeated, prisoner dilemma who followed this strategy would start with cooperation. In subsequent rounds, one player would then copy the action of the other during the previous round. Thus if both players continued to cooperate in any one round, both would continue to cooperate in the next round, until one defected ( not cooperated) leading to a defection by the other. After the tournament, using an agent-based simulation, Axelrod showed why tit-for-tat strategies can evolve starting from various distributions of initial strategy populations. The founding contributions of agent-based models were thus theoretical and abstract. They showed how simple rules of interaction could explain certain macro-level phenomena such as spatial patterns and levels of cooperation. During the last 20 years, the number of publications of simulations of populations of interacting agents who play games and exchange information has grown substantially. In the last ten years, we see an increasing use of agent-based models in various application areas like economics (Tesfatsion and Judd 2006), geography (Parker et al. 2003), sociology (Macy and Willer 2002), political science (Kollman et al. 2003), anthropology (Lansing 2003) and cognitive science (Goldstone and Janssen 2005). During the last five years, agent-based modeling has been applied within industrial ecology (for example Axtell et al. 2001; Janssen and Jager 2002; Schwoon 2006; Kraines and Wallace 2006). Agent-based approaches are used within industrial ecology to address a number of fundamental themes that will be discussed in further detail: (1) collective action, (2) innovation processes, and (3) system analysis. Collective Action Firms who like to exchange waste flows experience a social dilemma: are firms willing to make the investments now to become more dependent on the performance of other firms and have monetary benefits in the longer term? What if a firm invests their production structure to be able to use the waste flows of a neighboring firm, who may move, go bankrupt, or not produce the same waste flows in the future? Within the study of industrial ecosystems, the industrial symbiosis in the Danish village, Kalundborg, has been an illuminating example of a successful industrial park (Ehrenfeld and Gertler 1997). Since the 1960s six

An introduction

79

industrial plants invested in useful exchange of waste streams (like water, heat, stream, fly ash, gypsum, etc.), leading to significant savings of energy and natural resources in addition to monetary benefits. The success of Kalundborg has been attempted to be copied in other places, without much success. A potential reason is attributed to the unique situation of Kalundborg, a small village where the managers of most of the firms came from the local community who regularly met each other. From the study of collective action, it is known that repeated interaction contributes to the self governance of common resources and the provision of public goods (Dietz et al. 2003). Imposing the exchange of waste flows between firms, while seeming logical on paper, may not function well in practice due to a lack of social capital to solve the social dilemmas. Agent-based analysis (Howard-Grenville and Paquin, Chapter 9, this volume) and agent-based modeling might help to understand the conditions under which industrial ecosystems might emerge. Innovation Processes Agent-based modeling is used to study innovation processes to address the conditions under which new innovations evolve and how new products and practices diffuse in a population of producers and consumers (Andrews, Chapter 7, this volume; Janssen and Jager 2002; Schwoon 2006; Ma and Nakamori 2005). This use of agent-based modeling refers back to the field of evolutionary economics (for example Nelson and Winter 1982), where simulation models are used to simulate innovation, diffusion and learning of firms and organizations. Evolutionary economics traditionally focuses on innovations affecting the monetary performance of the firms involved who have various types of capital and labor inputs. Firms balance the need for innovation and imitating good innovations from competitors. Although investments in research and development might be costly, there are benefits of being ahead of the curve and shaping the technological landscape. Evolutionary processes are used to investigate which strategy types employed by firms are more effective in different types of markets. Applying this type of analysis to industrial ecosystems requires the explicit inclusion of material and energy flows (inputs and outputs). System Analysis In a similar vein as the study of innovation processes, agent-based models can be used as a computational laboratory to test the implications of various policies and future scenarios, as done by Batten and Grozev

80

Agent-based analysis of dynamic industrial ecosystems

(Chapter 8, this volume). Agent-based models for policy analysis have been used for such exploration in economics (Tesfatsion and Judd 2006) and land use change (Parker et al. 2003). The goal of such an analysis is to explore, not predict, possible future given scenarios of explicit assumptions of agent behavioral strategies, rules and regulations and external perturbations (such as price shocks). Based on the state-of-the-art knowledge, simulation models of industrial ecosystems can be constructed, like the electricity market in Batten and Grosev, to understand the complex interactions and to investigate the consequences of possible policies. Such models can be used in participatory processes to enhance learning and explore future scenarios of complex social–ecological systems (Downing et al. 2000).

CONCLUSION Agent-based modeling and analysis is a fruitful way to include more social science in the study of industrial ecosystems (Janssen et al. 2001). Industrial ecology has traditionally been dominated by an engineering focus, but more emphasis is given in recent years on the social science perspective. Since agent-based modeling is an increasingly used tool for the analysis of complex social systems, it is no surprise that applications are appearing on industrial ecosystems. Agent-based models can especially provide new insights in collective action, innovation analysis and policy analysis for industrial ecosystems through the combination of the increasingly grounded social simulation models developed in other disciplines with detailed data collected on social processes of industrial ecosystems. The chapters in this section provide a stimulating contribution to reaching this ambition.

REFERENCES Axelrod, R. (1984), Evolution of Cooperation, New York, NY: Basic Books. Axtell, R.L., C.J. Andrews and M.J. Small (2001), “Agent-based modeling and industrial ecology”, Journal of Industrial Ecology, 5(4): 10–13. Dietz, T., E. Ostrom and P. Stern (2003), “The struggle to govern the commons”, Science, 302: 1907– 12. Downing, T.E., S. Moss and C. Pahl Wostl (2000), “Understanding climate policy using participatory agent-based social simulation”, in S. Moss and P. Davidsson (eds), Multi Agent Based Social Simulation, Berlin, Germany: Springer, 1979: 198–213. Ehrenfeld, J. and N. Gertler (1997), “Industrial ecology in practice”, Journal of Industrial Ecology, 1(1): 67–79.

An introduction

81

Gardner, M. (1970), “The fantastic combinations of John Conway’s New Solitaire game life”, Scientific American, 223: 120–23. Goldstone, R.L. and M.A. Janssen (2005), “Computational models of collective behaviour”, Trends in Cognitive Science, 9(9): 424–30. Janssen, M.A., J.C.J.M. van den Bergh, P.J.H. van Beukering and R. Hoekstra (2001), “Changing industrial metabolism: methods for analysis”, Population and Environment, 23: 139–56. Janssen, M.A. and W. Jager (2002), “Stimulating diffusion of green products, coevolution between firms and consumers”, Journal of Evolutionary Economics, 12: 283–306. Kollman, K., J.H. Miller and S.E. Page (eds) (2003), Computational Models in Political Economy, Cambridge, MA: MIT Press. Kraines, S. and D. Wallace (2006), “Applying agent-based simulation in industrial ecology”, Journal of Industrial Ecology, 10(1–2): 15–18. Lansing, J.S. (2003), “Complex adaptive systems”, Annual Review of Anthropology, 32: 183–204. Ma, T.J. and Y. Nakamori (2005), “Agent-based modelling on technological innovation as an evolutionary process”, European Journal of Operational Research, 166(3): 741–55. Macy, M.W. and R. Willer (2002), “From factor to actors: computational sociology and agent-based modeling”, Annual Review of Sociology, 28: 143–66. Nelson, R.R. and S.G. Winter (1982), An Evolutionary Theory of Economic Change, Cambridge, MA: Belknap Press. Parker, D.C., S.M. Manson, M.A. Janssen, M.J. Hoffman and P. Deadman (2003), “Multi-agent systems for the simulation of land-use and land-cover change: a review”, Annals of the Association of American Geographers, 93(2): 316–40. Schelling, T.C. (1971), “Dynamic models of segregation”, Journal of Mathematical Sociology, 1: 143–86. Schelling, T.C. (1978), Micromotives and Macrobehavior, New York, NY: W.W. Norton & Company. Schwoon, M. (2006), “Simulating the adoption of fuel cell vehicles”, Journal of Evolutionary Economics, 16(4): 435–72. Tesfatsion, L. and K.L. Judd (2006), Handbook of Computational Economics II: Agent-Based Computational Economics, Amsterdam and Oxford: Elsevier. von Neumann J. (A. Burks ed.) (1966), Theory of Self-Reproduction Automata, Champaign, IL: University of Illinois Press.

7. Changing a firm’s environmental performance from within Clinton J. Andrews INTRODUCTION Industrial ecologists are becoming interested in better understanding the influences on industrial structure and performance. It is clear that agency exists at several levels, with the fundamental sources of agency in industrial ecosystems being individuals acting as citizens, employees, investors, and consumers. Nations are the key actors on the global stage, but national policies emerge in part from interactions among citizens and organizations. Firms are key actors within sectors, supply chains, and symbioses. Corporate behavior, to some extent, emerges from interactions among employees, and market outcomes emerge from the myriad choices of individual consumers. Individuals are not truly independent actors, however (Scott 2001). Institutions and organizations place formal, regulative constraints on individual choices: the chain of command, the order of work, and legal requirements. They also impose informal, normative constraints, such as a code of professional conduct, a work ethic, or an expectation of environmental stewardship. Some constraints are entirely unwritten and operate as cultural framing assumptions or cognitive biases: for example, humans should be fruitful and multiply, and humans should satisfy their immediate survival needs before addressing abstract concerns that are distant in time, space, or genetic similarity. The conundrum of whether agency determines structure, or vice versa, disappears if one recognizes that they interact (Giddens 1984). Individuals have some leeway to change the organizations where they work and also their governing institutions through quitting their jobs, voting politicians out of office, and buying different products. Yet how much influence do agents have on structures at the time scales of interest to industrial ecologists? This chapter uses a simulation modeling framework to explore agency relationships in a particular corporate environmental management context. 82

Changing a firm’s environmental performance from within

83

COMPUTATIONAL STUDIES OF ORGANIZATIONS Organizational behavior is a multidisciplinary field that describes, explains, and prescribes how agency and structure interrelate in firms. The origins of the field lie in the sociology of Weber’s bureaucracies, the economics of Smith’s specialized pin makers, and the engineering of Taylor’s scientific managers. Modern studies of organizations draw on systems thinking and social network theory for inspiration. Following World War II, theorists characterized organizations as controllable systems (Ashby 1956) with map-able structures and information feedbacks (Forrester 1961). Development of “open systems” theory represented a major breakthrough that emerged in stages. Economists noticed that organizational structures could be viewed as responses to varying external environments (Lawrence and Lorsch 1967). Social psychologists claimed that organizations were open systems that interfaced in multiple ways with the external environment (Katz and Kahn 1978). Systems dynamicists, in a nod to game theory, observed the explanatory power of “double interacts”, involving linked decision makers (Weick 1979). Philosophers identified organizations as interacting assemblages of individuals, objects, and exogenous forces (Churchman 1979). Those attentive to the second law of thermodynamics noticed how “organizations are dissipative structures that can only be maintained when members are induced to contribute energy to them” (Anderson 1999; Barnard 1938; Prigogine and Stengers 1984). Organizations are structured by their networks of internal and external relationships. Formal relationships between principals and agents, that is, owners and employees, only elicit desired behaviors if appropriate incentives are in place (Kerr 1975; Panayotou and Zinnes 1994; Gibbons 1998). More broadly, the economic characteristics of contracting internal and external relationships help researchers understand why there are firms and what their boundaries should be (Coase 1937; Williamson 1979; Holmström and Roberts 1998). Friendship networks and other informal group processes are important as well (Locke and Schweiger 1979). There are several ways to study organizational behavior and none is fully satisfactory. Aggregate statistical analysis of firms’ performance often yields trivial results and may fail to address the multi-level nature of the phenomenon. Case studies of individual firms can capture rich detail but in a static, retrospective snapshot. Experimental studies of social psychology in the workplace must abstract drastically from complex reality in their pursuit of controlled conditions. Longitudinal studies that track employees and firms over time are revealing but time consuming and expensive. Studies focusing on individual behavior in the context of external

84

Agent-based analysis of dynamic industrial ecosystems

pressures are helpful for studying leadership but do not really advance understanding of the links between agency and structure. Multi-agent simulation (MAS) modeling is a complementary way to conduct formal theorizing about organizational behavior, to explore interactions among the moving parts in a system, and to inform managerial decision making in a prospective, progressive manner. Although it too suffers from weaknesses, in a relatively few years it has become a useful and widely used method for organizational research. It complements causal modeling (Snijders 1998), case studies (Lin 2000), experiments (Burton 2003), and training activities (Fridsma and Thomsen 1998). Unlike its antecedents in game theory and systems dynamics, MAS supports speculation about links between agency and structure in organizations (Levitt 2004). Most MAS researchers with training in economics do bottom-up modeling, meaning that they specify agent rules for interacting with one another and the external environment, then start the simulation and observe what happens. They hope to see social structure emerge from the interactions of individuals. They build in autonomy, decision making, and the possibility of adaptation and agent evolution. Researchers with training in sociology are more willing to specify structure in their models, acknowledging that in the real world, firms already exist, and employees have no choice but to work within existing structures. Some researchers take a Lamarckian view and allow firms to evolve in response to their employees and their external environment. See Conte et al. (2001) for an extended discussion. In the 1990s, MAS was limited to modeling highly stylized versions of phenomena such as cognition (Cooper et al. 1996), interactions among members of generic teams (Carley and Prietula 1998), and social network dynamics (Zeggelink et al. 1996). Recently, the field has been able to tackle more applied topics such as absenteeism (Sanders and Hoekstra 1998), employee turnover (Harrison and Carroll 2002), employee promotion (Phelan and Lin 2001), affinity and animosity (Costa and de Matos 2002), and the imposition of quality standards (Torenvlied and Velner 1998). This chapter presents an applied MAS model of organizational behavior affecting environmental management, tests its sensitivity to key assumptions, and shows the effects of changing policy variables.

A STYLIZED FIRM This chapter looks at selected internal dynamics within a stylized firm. It formalizes certain abstractions of organization theory in a multi-agent simulation model of a firm that includes a factory, its employees, and the external environment. The factory provides structure by specifying the

Changing a firm’s environmental performance from within

85

physical characteristics of the technology and the regulative constraints of the production process. The employees have structured, hierarchal relationships among the plant manager, shift supervisors, and machine operators. But employees also have informal social network ties that influence their behavior. The external environment provides inputs and exogenous, driving factors to which the employees (and thus, the firm) must respond, and it also accepts outputs, including products and pollution, from the firm. The time tick of the simulation model is hourly in order to capture the short time scale at which human behavior is the dominant dynamic. The model can run, however, for tens of thousands of time ticks, entering the multi-year time scale over which technological change takes place. The model thus allows exploration of the relationship between human resource management and technology management within firms. To provide an empirical anchor for the research, this modeling effort focuses on a specific industry and context, chosen for its relative simplicity. It models a small polymer processing firm that manufactures injectionmolded products such as cafeteria trays for fast food restaurants, molded housings for consumer electronic products, and plastic coffee mugs. Prior to developing the model, the research team visited and conducted detailed case studies of several actual firms in New Jersey, US, and Suzhou, China, summarized in Andrews (2006). The production technology for such a firm is fairly simple, involving a set of injection molding machines, a costly stainless steel mold for each product type, and a generic factory building with a warehouse for raw materials, a factory floor for the many parallel production lines, and a shipping area. The production process is also simple: plastic pellets are heated; the melted plastic is injected into a mold; once cooled, the product is released from the mold; the product is cleaned and finished, and it is shipped. The human resource relationships in such a firm are also not too complex: a plant manager hires, fires, and makes capital investments; a marketing manager brings in new business; shift supervisors assist machine operators and also report on their performance to the plant manager; machine operators operate the injection molding equipment; a materials mixer delivers plastic pellets to the production lines; a shipping clerk sends completed products out the factory door; and a janitor regularly cleans the factory floor.

DESCRIPTION OF MODEL At this idealized firm, maximum capacity is fixed at ten production lines. This avoids the complexity of expansion planning and allows a focus on managing existing assets. Inputs to the production process include labor,

86

Agent-based analysis of dynamic industrial ecosystems

electricity, and plastic pellets. Outputs include products (here called “widgets”), scrap plastic, and air pollution. Scrap plastic can be recycled, and air pollution is an increasing function of process temperature. Widget quality is a non-monotonic function of process temperature, that is, there is a quality-maximizing setpoint, so there is a tradeoff between pollution reduction and product quality objectives. Machine operators set the process temperature at what they view as the optimal point. It is possible to automate parts of the production process, thereby eliminating much labor, reducing human error, and improving product quality. The plant manager will invest in automation when it appears to be economically prudent given the relative costs of inputs, especially labor, and the prices obtained for the widgets that are the factory’s outputs. Automation is the sole technological choice available to the plant manager. The plant manager makes several types of human resource decisions. Based on shift supervisors’ reports of machine operators’ attendances and absences, and the frequency with which they need assistance in operating their molding machines, the plant manager can fire underperforming machine operators. The plant manager can also hire replacement machine operators from a pre-established pool using criteria including aptitude, attitude, experience, and cultural background. Finally, the plant manager can promote a well-performing machine operator to become a shift supervisor when an opening arises. Individual employees also have the autonomy to make several decisions. They choose whether to come to work each morning and whether to alter their attitude toward work when their supervisors admonish them for misconduct. Machine operators each choose what they consider to be the optimal temperature set point for their injection molding machine, weighing the firm’s desire for high quality widgets against their own possible interest in reducing air pollution. Individuals are quite heterogeneous in terms of innate aptitude and preferences, socially-influenced attitudes, and acquired experience. These factors influence their decisions and the plant manager’s responses, leading in some cases to changes in firm-wide performance.

IMPLEMENTING THE MODEL This model was implemented in Java using the Ascape multi-agent simulation framework. Object-oriented programming languages such as Java and C++ provide a straightforward pathway to agent-based modeling, because agents can be implemented as software objects that interact with one another. Ascape was developed at Brookings Institution, serving as one of

Changing a firm’s environmental performance from within PolyModel

Population

87

Persons

Factory Jobs External Environment

Employees Owner (passive) Plant Manager Marketer Maintenance Technician Janitor Shift Supervisor

Production Lines

Measure of the firm’s performance

Machine Operators

Warehouse

Materials Mixer

Shipping Department

Shipping Clerk

Note: Solid lines represent the modeling hierarchy, specifying which objects inherit traits of other objects, dashed lines represent key modeled interactions among objects.

Figure 7.1 Structure of PolyModel the first libraries of Java routines to facilitate research on social science topics (Parker 2001). Well known competing frameworks include Swarm and RePast (Swarm 2006; SourceForge 2006). The model includes 22 classes representing the categories of objects that interact (see Figure 7.1). These include PolyModel (the overall framework for the model), the external environment, the factory (containing production lines), the people (some of whom acquire jobs), and miscellaneous supporting objects. Both the raw code and executable versions of the model are available online (Andrews 2006). Key elements of the model’s logic are the bounded rationality and biases of the employees, the way employees relate in social networks, the factors that drive employees’ decisions, the hiring and firing processes used by the plant manager, the production process, and the plant manager’s automation decision process.

88

Agent-based analysis of dynamic industrial ecosystems

Employee Decisions It is common when modeling principal–agent problems to assert that agents know their utility functions and act to optimize them. A familiar formulation posits a labor–leisure tradeoff (Block and Heineke 1975). An alternative view is that employees have bounded rationality and do not fully understand their own utility functions or the firm’s cost function (Simon 1947). Adopting this latter view brings behavioral realism that approximates utility maximization under some circumstances but diverges in interesting ways under others, as employees employ heuristics and decide without complete knowledge (Cattaneo and Robinson 2000). The following are key heuristics in PolyModel: ●

●

●

An employee decides to go to work if he feels healthy enough (random draw), acts responsibly enough (random draw), and a large enough fraction of his friends are also going to work (based on a poll of his social network). An employee decides to quit his job if the net present value of quitting exceeds that of staying, given the difference between his wage and the prevailing wage, the expected time to re-employment based on the current unemployment rate relative to the historical rate, and his discount rate. A machine operator decides what temperature to set his injection molding machine at based on his environmental views. If he is an environmentalist, he will attempt to set the temperature lower to minimize air pollution. If not, he will attempt to set the temperature higher to minimize product defects. The degree to which he hits these target temperatures is a function of worker error.

Employee Characteristics The standard rational actor model, homo economicus, adopts a Hobbesian view of individuals as atomistic, pre-social agents. Yet a commitment to methodological individualism does not imply adherence to such a naïve model, and there is every reason to include a richer social dimension (Heath 2001). As Elster (1989: 13) observes, “the elementary unit of social life is the individual human action”. Likewise, whereas the simple agent models typical of game theory typically employ the unrealistic assumption that agents are homogenous, there is much benefit to relaxing that assumption and allowing heterogeneity (Lempert 2002). It is also useful to characterize agents as “computational” in the sense that they have bounded information processing abilities (Carley 2002) that give rise to mistakes. Thus

Changing a firm’s environmental performance from within

89

Owner (passive) Plant Manager

Marketer

Shift Supervisor

Materials Mixer Machine Operators Maintenance Technician Shipping Clerk Janitor Environmentalism

Work Attitude

Friendship

Figure 7.2 Formal reporting hierarchy versus social network PolyModel characterizes employees and their firm further as follows (see also Figure 7.2): ●

●

Worker error affects most stages of the production process. Many production parameters are set by drawing a number from a random distribution, and higher worker error increases the dispersion around the mean, “target” value of the production parameter. Worker error is a weighted, decreasing function of the employee’s aptitude, experience, and happiness, and an increasing function of tiredness. Employee aptitude is assigned randomly at the beginning of the simulation; their work experience grows over time from an initial random endowment; and tiredness increases with number of consecutive hours worked without a break. Happiness increases with income and friendships on a normalized, weighted basis, and is adjusted slightly for cleanliness of the workplace. Happiness weights

90

Agent-based analysis of dynamic industrial ecosystems

●

●

are assigned exogenously so that the model user can adjust the fraction of “Type A” persons (0.9 money, 0.1 friends) and “Type B” persons (0.1 money, 0.9 friends) in the population. The normalized number of friends, which is calculated every Saturday night, itself involves weighting and screening calculations, such that acquaintances with more similar cultural backgrounds (for example “environmentalist”, “non-environmentalist”) count more heavily than those with different backgrounds, people with different attitudes (for example “good”, “bad”) cannot become friends, and people become better friends with people they have previously and repeatedly met at work. Employees can change both their attitude and culture. If an employee with a “bad” attitude is threatened with firing by the plant manager, he chooses (with some probability) to adopt a “good” attitude. If a majority of other employees are “environmentalists”, then an employee chooses (with some probability) also to be one, and vice versa.

Hiring and Firing Each day, the plant manager evaluates all of the employees. He asks the shift supervisor (who transmits information with some probability of error) about the employee’s technical performance (based on amount of worker error), absenteeism, and number of times the shift supervisor has had to help the employee correct their mistakes. He fires some fraction of the employees whose performance falls below acceptable thresholds on these three dimensions. When a job opening arises because an employee quit or was fired, the plant manager hires someone. He draws a random sample of potential employees from the population, ranks the potential employees according to an exogenously set hiring bias (favoring some weighted combination of aptitude, attitude, experience, and culture), and hires the highest ranked person. Automation Every year the plant manager decides if any production line should be automated. If there are enough funds and a line is not automated, then he estimates whether the cost of automation is lower than the annual cost of running the existing line. The running cost of a line consists of the machine operator’s salary, the cost of supplying plastic pellets and electricity, the cost of disposing of waste, and the cost of recycling. Automated lines do not suffer from worker error.

Changing a firm’s environmental performance from within

91

MODEL CALIBRATION AND VALIDATION Model calibration and validation was a multi-step process. The first step was to derive initial conditions of the models from the case study evidence; second, drive the models using exogenous conditions identified in the case studies and finally, validate the models against outcomes from the case studies. The case studies are available online (Andrews 2006). Criteria for determining the acceptability of the multi-agent simulation model included ease of implementation and use, validity, and reliability. Ease of use was tested by asking several groups of non-programmer graduate students to conduct a classroom exercise using the model. By following good programming practices, reliability was ensured. Model validation proved to be the least straightforward task. Since this was a highly stylized model of an extremely complex reality, it was unreasonable to expect to replicate in detail the experiences documented in the case studies. Instead, “we were mostly concerned with validating the insights we gained . . . from the simulation” (Shannon 1975: 29). The goal was to achieve enough face validity to approximate the major features of the case studies and enough robustness to perform plausible “what-if ?” simulations that help users understand both the model’s relative sensitivity to different assumptions and how the model behaves after altering policy variables. This chapter typically reports relative results that compare a simulation to a base case that approximates the case study experience.

RESULTS This section shares illustrative modeling results that provide an insight into the effective sources of agency in firms. The first set of results shows the base case, in which the model responds to historical data. The second set of results shows four policy scenarios applied to a steady state version of the base case in which the exogenous drivers are fixed. Base Case The base case mimics performance of the industry as observed in the case studies, and includes parameter values calibrated to be “reasonable”. The key assumptions in the base case are that (1) there is no hiring or firing bias, (2) the plant manager fires only one-fifth of those employees whose performance is inadequate in any given month, (3) when a shift supervisor helps a machine operator this is the equivalent of increasing the operator’s

Agent-based analysis of dynamic industrial ecosystems

Price Index (2001 = 100)

92 200 190 180 170 160 150 140 130 120 110 100 90 2001

Labor ($/hr) Electricity (cents/kWh) Plastic ($/lb) Widget ($/item)

2002

2003

2004

2005

Year

Figure 7.3 Cost and price trends in the polymer processing industry experience by 5 per cent, (4) one-fifth of the population call themselves environmentalists, and (5) one-third of the population has utility functions weighted towards preferring friendships over money. Time trends for the model’s exogenous drivers, for example, current costs of inputs and prices of products shipped, are shown in Figure 7.3 for the years 2001–05. Prevailing wages of labor have increased steadily since the 1950s, electricity costs increased rapidly during the 1970s before recently leveling off, plastic pellets have risen in cost sporadically and have shown remarkable acceleration in recent years, and the price obtained for the product outputs (“widgets”) has increased only modestly over time, with a recent significant flattening in prices. Since 2001, the competitive position of the US polymer processing industry has deteriorated dramatically. New low-cost manufacturers have sprung up in Asia, and a significant amount of US manufacturing capacity has shut down or moved off shore. More than ever, US manufacturers are price takers, and the average price per widget has dropped in real terms. At the same time, the cost of inputs has increased. Labor costs have slowly but steadily increased over decades, whereas plastic costs have skyrocketed. This squeeze has threatened the viability of the US injection molding industry. Figure 7.4 shows how the model responds to the exogenous drivers. Profits drop dramatically over time, with the firm on the verge of losing money. Most other variables (not shown) achieve steady performance, although they exhibit substantial amounts of random noise caused by worker error, absenteeism, equipment malfunctions, and similar factors. These results are plausible and indicate that the model has face validity.

Changing a firm’s environmental performance from within

93

Figure 7.4 Historical simulation of revenues, expenses, and profits: sample screen shot from computer model Steady State Policy Scenarios Extending from the base case by (1) freezing the values of the exogenous drivers at 2005 levels, and (2) changing selected parameters of the model, this section compares four possible ways to reduce the environmental impacts of economic activity. Specifically, these four scenarios portray different approaches to reducing the amount of air pollution produced by the plastic injection-molding firm. Table 7.1 shows average per cent differences between the base case and each policy scenario. Results are based on 20 repeated simulations of 10 000 hours for each scenario, and they are evaluated for significance using a paired, two-tailed difference-ofmeans test. The values of the table entries should be treated as indicative rather than precise measures of differences among scenarios. Simulation #1: promote environmentalism as a social movement A bottom-up approach to changing the environmental performance of the firm is to promote environmentalism in the population. If the proportion of environmentalists increases, then they will eventually infiltrate the firm

94

1 1 26*** 18***

5***

2***

4***

Worker Error (average per employee)

8***

Pollution/Widget (average per widget shipped)

Notes: * Difference of means test is significant at p = 0.05. ** Difference of means test is significant at p = 0.01. *** Difference of means test is significant at p = 0.001.

Promote Environmentalism Hire Environmentalists Impose Quality Control Automate Production

Scenario Metric

63***

4

217***

240***

Environmentalists Present (among employees)

22***

9***

0

2

Employee Experience

16

2

3

3

Employee Happiness

Table 7.1 Summary of results (per cent change from base case after reaching steady state)

14***

4***

1**

2***

Profits (monthly net)

0

0

7***

8***

Air Pollution (monthly emissions total)

Changing a firm’s environmental performance from within

95

through normal hiring practices. Bansal (2003) argues that individual concerns can translate into an organizational response provided that employees enjoy adequate discretion and the organization enjoys excess resource slack. Thus the simulation explores what happens when those hired as machine operators have discretion to adjust the temperature set points of their injection molding machines to reduce air pollution. In this simulation, the probability of encountering an environmentalist in the population is increased exogenously from 20 per cent to 100 per cent. Through normal firing and hiring practices, environmentalists soon infiltrate the firm and operate the injection molding machines in a way that reduces overall air pollution and pollution per widget produced by 8 per cent. Profits decrease an average of 2 per cent. The spread of environmentalism has insignificant effects on employee happiness, experience, and worker error rates. Simulation #2: selectively hire environmentalists Champions can powerfully influence organizational environmental performance (Andersson and Bateman 2000), especially if the champions have senior management positions (Howard-Grenville 2005). In this scenario, it is the plant manager who joins the environmental movement. Instead of waiting for society as a whole to become more environmentally conscious, she preferentially hires them, accelerating the infiltration of environmentalists into the firm. This top-down initiative ensures that more machine operators will optimize their injection molding machines to reduce air pollution. The environmentalist hiring bias is implemented in the model by exogenously increasing the weight given to the cultural attribute (environmentalism) relative to other attributes such as aptitude, attitude, and experience. When hiring, the plant manager screens job applicants on that basis. Some employees are susceptible to peer pressure and will change their environmental views if they are outnumbered. Peer pressure thus amplifies the preferential hiring strategy once it has advanced far enough. This hiring strategy succeeds in reducing total air pollution by 7 per cent and pollution per unit of production by 5 per cent. The strategy fails, however, to capture the indirect benefits of cultural homogeneity (increased happiness, reduced error, increased profits). It also hinders profitability (by 1 per cent) by displacing high aptitude new hires with environmentalists who may or may not have other good qualities. Simulation #3: impose a strict quality control regime Eco-efficiency arguments from industrial ecology assert that much pollution is unintended, and that tighter quality control and reduced worker error rates can simultaneously reduce pollution and increase profits

96

Agent-based analysis of dynamic industrial ecosystems

(Manahan 1999). Similar assertions appear in the quality management literature (Evans and Lindsay 2005). This simulation explores a top-down initiative that focuses solely on reducing the worker error rate. This strategy is implemented in the model by exogenously setting (1) the employee hiring bias to favor candidates with high aptitude, and (2) increasing the amount of helpful supervision that shift supervisors provide to machine operators. The environmental results are visible only at the margin: total air pollution does not change at all relative to the base case, and pollution per unit of production drops by 2 per cent. On the other hand, profits increase by 4 per cent. Employee experience also increases because better supervision leads to fewer firings. These factors together lead to a large 26 per cent drop in the worker error rate. Simulation #4: automate the production process Rather than rely on human resource management tools, the plant manager could instead choose a technical fix calculated to promote eco-efficiency. Automating portions of the production process would displace error-prone and expensive employees, thereby reducing pollution and increasing profits, assert systems analysts such as Reimann and Sarkis (1996). The model has a switch that allows the plant manager to choose automation when it appears to be cost-effective. Once this feature is exogenously switched on, the plant manager evaluates current input costs (labor, electricity, plastic pellets), product prices, and capital costs, proceeding to automate a production line when the expected net present value of the new technology investment exceeds the status quo option. The results show that total pollution does not drop but pollution per widget produced does decrease by 4 per cent. Monthly profits shoot up by 14 per cent and worker error drops by 18 per cent relative to the base case.

DISCUSSION The four scenarios described above represent pure management strategies, whereas real managers would combine elements in their pursuit of ecoefficiency, profits, and environmental improvement. Nevertheless, the results reveal interesting subtleties and tradeoffs that hint at the complexity of the managerial challenge. Long ago, Miles and Rosenberg (1982: 26) noted that “organizations that fail to redesign internal roles and relationships to tap the underutilized capabilities of human resources not only experience built-in inefficiencies, but also strain our social fabric”. Indeed, the results shown here suggest

Changing a firm’s environmental performance from within

97

that human resource tools of hiring, training, supervision, and firing are most effective when coordinated to achieve clear objectives. The underlying labor pool bounds the firm’s potential. A rogue manager who preferentially hires environmentalists (or cronies, or people from a particular ethnic group) can affect the overall performance of the firm. The most profitable strategies emphasize quality control or automation, and they provide minor environmental benefits given the characteristics of this injection molding technology. For other technologies, quality management may pay higher environmental dividends. It is noteworthy that the behavioral/human resource fixes provided benefits that were of similar magnitude to the technical fix (automation). This result supports previous claims that organizational practices deserve attention from industrial ecologists (for example Andrews and Swain 2000).

CONCLUSIONS Locating the sources of agency in the industrial ecosystem is a high priority task, because environmental improvements depend wholly on their actions. This chapter has examined the relationships among actors within the firm, and it has found that social movements, champions, organizational rules and procedures, and technological change can each affect the firm’s overall performance. More generally, as Carley (2002: 7262) observes: In many cases, simply building the model brings value in and of itself to the policy maker or manager as it lays bare hidden assumptions and potential limitations in the current system. As we move toward more realistic agents and tasks, the value of these models can only grow. Such models are just beginning to be taxonomized and standardized (Chang and Harrington 2005) as this research community expands. The multi-agent simulation model described here has demonstrated its value as a tool for formal theorizing about complex, hard to study organizational dynamics. PolyModel can also be used as a management training tool. In the latter role, models like it may help future managers become more effective sources of agency within the industrial ecosystem.

ACKNOWLEDGMENTS The author gratefully thanks Preetham Mysore and Shawn Patton for their programming efforts, Ana Baptista for performing case study research,

98

Agent-based analysis of dynamic industrial ecosystems

Robert Axtell for his collegiality, and the Science To Achieve Results program of the US Environmental Protection Agency for financial support.

REFERENCES Anderson, P. (1999), “Complexity theory and organization science”, Organization Science, 10(3): 216–32. Andersson, L.M. and T.S. Bateman (2000), “Individual environmental initiative: championing natural environmental issues in US business organizations”, Academy of Management Journal, 43(4): 548–70. Andrews, C.J. (2006), PolyModel Home Page, http://policy.rutgers.edu/andrews/ projects/abm. Last accessed 2 September 2007. Andrews, C.J. and M. Swain (2000), “Institutional factors affecting life-cycle impacts of microcomputers”, Resources, Conservation and Recycling, 31: 171–88. Ashby, R. (1956), An Introduction to Cybernetics, London: Chapman and Hall. Bansal, P. (2003), “From issues to actions: the importance of individual concerns and organizational values in responding to natural environmental issues”, Organization Science, 14(5): 510–27. Barnard, C.I. (1938), The Functions of the Executive, Cambridge, MA: Harvard University Press. Block, M.K. and J.M. Heineke (1975), “Factor allocations under uncertainty: an extension”, Southern Economic Journal, 41(3): 526–30. Burton, R.M. (2003), “Computational laboratories for organization science: questions, validity, and docking”, Computational & Mathematical Organization Theory, 9: 91–108. Carley, K.M. (2002), “Computational organization science: a new frontier”, Proceedings of the National Academy of Sciences, 99(3): 7257–62. Carley, K. and M. Prietula (eds) (1998), Simulating Organizations: Computational Models of Institutions and Groups, Cambridge, MA: MIT Press. Cattaneo, A. and S. Robinson (2000), “Empirical models, rules, and optimization: turning positive economics on its head”, TMD Discussion Paper No. 57, Washington, DC: International Food Policy Research Institute, http:// www.ifpri.org/divs/tmd/dp/papers/tmdp 53. Last accessed 2 September 2007. Chang, M.H. and J.E. Harrington, Jr. (2005), “Agent-Based Models of Organizations”, Economics Working Paper 515, Johns Hopkins University, http://www.econ.jhu.edu/papers.html. Last accessed 2 September 2007. Churchman, C. (1979), The Systems Approach and Its Enemies, New York, NY: Basic Books. Coase, R. (1937), “The nature of the firm”, Economica, 4: 386–405. Conte, R., B. Edmonds, S. Moss and R.K. Sawyer (2001), “Sociology and social theory in agent-based social simulation: a symposium”, Computational & Mathematical Organization Theory, 7: 183–205. Cooper, R., J. Fox, J. Farringdon and T. Shallice (1996), “a systematic methodology for cognitive modeling”, Artificial Intelligence, 85: 3–44. Costa, L.A. and J.A. de Matos (2002), “Toward an organizational model of attitude change”, Computational & Mathematical Organization Theory, 8: 315–35. Elster, J. (1989), Nuts and Bolts for the Social Sciences, Cambridge, UK: Cambridge University Press.

Changing a firm’s environmental performance from within

99

Evans, J.R. and W.M. Lindsay (2005), Management and Control of Quality, Mason, OH: Thomson Southwestern, p. 28. Forrester, J. (1961), Industrial Dynamics, Cambridge, MA: MIT Press. Fridsma, D.B. and J. Thomsen (1998), “Representing medical protocols for organizational simulation: an information processing approach”, Computational & Mathematical Organization Theory, 4(1): 71–95. Gibbons, R. (1998), “Incentives in organizations”, Journal of Economic Perspectives, 12(4): 115–32. Giddens, A. (1984), The Constitution of Society: Outline of the Theory of Structuration, Cambridge, UK: Polity Press. Harrison, J.R. and G.R. Carroll (2002), “The dynamics of cultural influence networks”, Computational & Mathematical Organization Theory, 8: 5–30. Heath, J. (2001), Communicative Action and Rational Choice, Cambridge, MA: MIT Press. Holmström, B. and J. Roberts (1998), “The boundaries of the firm revisited”, Journal of Economic Perspectives, 12(4): 73–94. Howard-Grenville, J.A. (2005), “Review Essay: explaining shades of green: why do companies act differently on similar environmental issues?”, Law and Social Inquiry, 30(3): 551–81. Katz, D. and R.L. Kahn (1978), The Social Psychology of Organizations, 2nd edition, New York, NY: Wiley. Kerr, S. (1975), “On the folly of rewarding A, while hoping for B”, Academy of Management Review, 18: 769–83. Lawrence, P. and J. Lorsch (1967), Organization and Environment, Cambridge, MA: Harvard University Press. Lempert, R. (2002), “Agent-based modeling as organizational and public policy simulators”, Proceedings of the National Academy of Sciences, 99(3): 7195–6. Levitt, R.E. (2004), “Computational modeling of organizations comes of age”, Computational & Mathematical Organization Theory, 10: 127–45. Lin, Z. (2000), “Organizational performance under critical situations: exploring the role of computer modeling in crisis case analysis”, Computational & Mathematical Organization Theory, 6(3): 277–310. Locke, E. and D. Schweiger (1979), “Participation in decision making: one more look”, Research in Organizational Behavior, 1: 265–339. Manahan, S.E. (1999), Industrial Ecology: Environmental Chemistry and Hazardous Wastes, Boca Raton, FL: Lewis/CRC Press, p. 53. Miles, R.E. and H.R. Rosenberg (1982), “The human resources approach to management: second-generation issues”, Organizational Dynamics, Winter: 26–41. Panayotou, T. and C. Zinnes (1994), “Free-lunch economics for industrial ecologists”, in R.H. Socolow, C.J. Andrews, F. Berkhout and V.M. Thomas (eds), Industrial Ecology and Global Change, Cambridge, UK: Cambridge University Press, pp. 383–97. Parker, M. (2001), “What is Ascape and why should you care?”, Journal of Artificial Societies and Social Simulation, 4(1): 5–20. Phelan, S.E. and Z. Lin (2001), “Promotion systems and organizational performance: a contingency approach”, Computational & Mathematical Organization Theory, 7: 207–32. Prigogine, I. and I. Stengers (1984), Order Out of Chaos: Man’s New Dialog with Nature, New York, NY: Bantam Books.

100

Agent-based analysis of dynamic industrial ecosystems

Reimann, M.D. and J. Sarkis (1996), “An infrastructure for agile product development systems”, in L.F. McGuinness et al. (eds), Flexible Automation and Intelligent Manufacturing, New York, NY: Begell House, pp. 57–66. Sanders, K. and S.K. Hoekstra (1998), “Informal networks and absenteeism within an organization”, Computational & Mathematical Organization Theory, 4(2): 149–63. Scott, W.R. (2001), Institutions and Organizations, 2nd edition, Thousand Oaks, CA: Sage Publications. Shannon, R.E. (1975), Systems Simulation: The Art and Science, Englewood Cliffs, NJ: Prentice-Hall. Simon, H.A. (1947), Administrative Behavior, New York, NY: Macmillan. Snijders, T.A.B. (1998), “Methodological issues in studying effects of networks in organizations”, Computational & Mathematical Organization Theory, 4: 205–15. SourceForge (2006) Repast Home Page, https://sourceforge.net/projects/repast. Last accessed 2 September 2007. Swarm Development Group (SDG) (2006), Swarm Development Group Wiki, http://www.swarm.org. Last accessed 2 September 2007. Torenvlied, R. and G. Velner (1998), “Informal networks and resistance to organizational change: the introduction of quality standards in a transport company”, Computational & Mathematical Organization Theory, 4(2): 165–88. Weick, K.E. (1979), The Social Psychology of Organizing, Reading, MA: AddisonWesley. Williamson, O.E. (1979), “Transaction-cost economics: the governance of contractual relationships”, Journal of Law and Economics, 22: 233–71. Zeggelink, E.P.H., R. van Oosten and F.N. Stokman (1996), “Object oriented modeling of social networks”, Computational & Mathematical Organization Theory, 2(2): 115–38.

8. Managing energy futures and greenhouse gas emissions with the help of agent-based simulation David F. Batten and George V. Grozev INTRODUCTION A close relationship exists between industrial ecology, complex systems science, and the adaptive management of natural resources, such as energy and water. Such relationships each involve interdependencies between system elements and flow dynamics across networks of various kinds. The pivotal role of energy systems in industrial ecology and vice versa, suggests that the time is ripe for some of the tools and techniques of complex systems science to be focused on eco-industrial research, involving complex energy flows across transmission, distribution, and recycling networks. In this chapter, power markets, their associated physical networks of infrastructure, and the natural environment in which they operate are viewed as key components of a complex adaptive system (CAS). In particular, it is shown how agent-based simulation of such a CAS helps discover and explore alternative energy futures. Energy industries have been undergoing regulatory reform worldwide for more than a decade, aiming to improve economic efficiency. In many places, these changes have culminated in the appearance of wholesale and forward contracting power markets. There are contradictory conclusions about the performance of these restructured electricity markets. Market performance depends largely on how each market participant responds to market design, including bidding and operational rules, market observations, operating procedures, and information revelation. It is often the case that generator firms have gained more from these restructured markets than retailers, local communities or households, partly because they have increasingly more opportunities to adapt their behavior to changing market conditions in profit maximizing ways. Unfortunately, some of this profit-driven behavior exacerbates GHG emissions. 101

102

Agent-based analysis of dynamic industrial ecosystems High Entropy (Bound Energy) Waste Outputs Low Entropy (Free Energy) Resource Inputs

THE ECONOMIC PROCESS

Land Water Food, Energy Building materials Other resources

Solid wastes Liquid wastes Toxic wastes Sewage Greenhouse gases Other waste gases Waste heat Enjoyment of Life

Figure 8.1 Entropy and the economic process At the same time, interest has increased in use of market-based policy instruments for environmental purposes, for example, to introduce climate change regulation into the energy sector and reduce greenhouse gas emissions. Australia is an early adopter of electricity industry restructuring and market-based environmental instruments. Yet the mixed performance of these schemes to date confirms that considerable care is needed in the design of market-based approaches (MacGill et al. 2006). For example, the opportunities for market participants to exercise market power are real and observable (Hu et al. 2005). The broad goals of a greener energy policy and industrial ecology are similar, for both aim to achieve an eco-efficient future by increasing the value derived from our resources and retaining them in the economy for as long as possible. According to Georgescu-Roegen (1971), the real output of the economic process is not a material flow but a psychological flux: the enjoyment of life (see Figure 8.1). This flux, or steadily increasing “quality” of human life can be maintained for only as long as it can feed continuously on natural resources (for example low entropy from the environment). There are two important lessons to be gained from this analysis. First, much of our economic struggle hinges on dwindling supplies of nonrenewable natural resources. Ironically, there is an abundant alternative energy source: the sun. It is not possible to harness much of the sun’s energy at a cost that is competitive with other energy sources. Second, many natural resources are growing scarcer in a different sense than land. Although land and the various resources it contains are available in limited amounts, the second law of thermodynamics dictates that resources like coal cannot be used (as coal) twice. When a piece of coal is burned, its initial energy dissipates (in the form of heat, smoke, soot and ash) to such an extent that it cannot be reused as

Managing energy futures and greenhouse gas emissions

103

coal. Thus, recycling is not an option for coal resources. If the productive cycle of conversion is to continue, new uses for the waste products (heat, smoke, soot and ash) must be found in addition to increased use of renewable feedstocks or cleaner ways of processing coal. The pressing challenge is to employ renewable resources and residual forms of bound energy in new, productive, and ecologically benign ways. Nowadays, convincing electricity market models must consider the interrelated dynamics of at least three processes: (1) market participants’ behavior; (2) the physical limitations of production and transmission assets; and (3) the environmental outcomes associated with electricity generation, transmission, distribution, and consumption. This chapter examines the use of agent-based simulation models to represent electricity markets and their physical networks as evolving systems of complex interactions between human behavior in markets, technical infrastructures, and the natural environment. In the next section, Australia’s National Electricity Market (NEM) is described as an introduction to the domain of complex energy markets. In section three, an explanation is given as to why electricity systems are complex adaptive systems. Section four briefly summarizes three types of electricity modeling approaches, while section five presents a basic introduction to agent-based models of electricity markets. Finally, section six briefly describes the NEMSIM model in this study and illustrates its use for generating various eco-efficient scenarios of future energy systems, such as clusters of low-emissions distributed generation.

AUSTRALIA’S NATIONAL ELECTRICITY MARKET Launched on 13 December 1998, Australia’s National Electricity Market (NEM) now incorporates five States and one Territory: Queensland, New South Wales, Australian Capital Territory, Victoria, South Australia, and Tasmania. The NEM is not truly a national market, since Western Australia and the Northern Territory are not physically connected. It is a gross pool-type market operating under the administration of the National Electricity Market Management Company (NEMMCO). As such, the NEM displays many features in common with other pool-type markets after restructuring, for example the original England/Wales pool prior to the New Electricity Trading Arrangements (NETA). Close to 100 market participants were registered in the market at the end of 2005. They included generator companies, network service (transmission and distribution) providers, market customers and traders. There are about 340 physical generating units in the NEM and more than 180 of them

104

Agent-based analysis of dynamic industrial ecosystems

are market scheduled (many small generating units that are aggregated for market purposes, and others that are non-scheduled, using for example, wind and biomass). The total NEM installed capacity is around 37 gigawatts (GW). The NEM is operated in the following manner: 1.

2.

3.

Scheduled generators submit offers in ten price bands, stacked in an increasing order in ten incremental quantity bands over a settlement day. These quantity offers correspond to the ten price bands for each of the 48 settlement intervals, and must be received by NEMMCO before noon of the previous day (for example one day before the real dispatch). Regional network operators prepare demand forecasts, then NEMMCO runs a linear program to dispatch generation and meet demand every five minutes. Less expensive generating units are dispatched first, but the price offered by the most expensive generating unit dispatched determines the price for a dispatch interval. The settlement price for each settlement interval is the average of all prices over the six dispatch intervals in the settlement interval.

Market information plays an important role in both the participants’ decision making and ensuring reliable dispatch operations. NEMMCO provides market participants with information such as load forecasts, predispatch (and price sensitivity analysis) data, dispatch data, as well as medium- (seven days) and long-term (two years) forecasting of supply scenarios or system adequacy. Market participants are permitted to adjust their bids in response to the latest information. For example, they can re-bid the quantity for a previously bid settlement interval up to five minutes before dispatch, although the price bands are fixed for the entire settlement day. On its website NEMMCO publishes the bidding and dispatch data for all scheduled generating units shortly after the end of a settlement day. The daily dispatch files include regional price, demand, and dispatch data. Although Barmack (2003) has expressed concerns about the revelation of bidding data to the public, in some earlier work it is demonstrated that there are several positive aspects of such data revelation (Hu et al. 2005). Recent research has identified various strategies that generators use in the market (Hu et al. 2005; Batten and Grozev 2006). The data examined reveals that two-thirds of the generating units are relatively inactive in terms of changing bidding strategies. The minority of generators that do respond actively to changes in market conditions are base-load units that tend to dominate the supply side. These 30 or so active generating units are the “giants” in their regions. Being coal-based generators, they contribute disproportionately to Australia’s extremely high levels of GHG emissions.

Managing energy futures and greenhouse gas emissions

105

Their greater size affords them more opportunities to try different combinations of capacity offers in different price bands, as well as a more secure and confident position. They can more easily afford to try different strategic bids since part of the bulk of their capacity can be committed to very low price bands, safe in the knowledge that it is almost certain to be dispatched. This strategy releases the remainder of their capacity for experimental bidding in strategic price bands. Also, research shows that larger generators are more likely to use quantity offers rather than price offers to improve their market positions. This result is especially true in periods of peak demand. In terms of traditional game theory, such a result suggests that NEM generators may behave more like players in Cournot competition (by using quantity as a strategic variable). Whether deliberate or not, the strategy of generators causing capacity withholding has the effect of raising prices. Such outcomes have been observed in several deregulated electricity markets, like California (Borenstein et al. 2002), Britain (Wolfram 1999) and Australia (Short and Swan 2002). In the NEM, capacity withholding can be accomplished by offering capacity only in very high price bands or making generating units unavailable. Research shows that capacity withholding is used by several larger firms who own the larger generating units (Hu et al. 2005). The fact that these generating units have the capability to influence regional prices may serve as a partial explanation for why they engage in strategic bidding. Generators can re-bid their capacity commitment in response to load changes and other factors. Thus, they can take advantage of the information provided by NEMMCO in the pre-dispatch phases to improve revenue streams. There is concern over re-bidding, partly because it may help certain generators to exercise market power. Some peaking units use rebidding to adapt to changes in market conditions, instead of using the daily offer/bid opportunities. They offer identical quantity bands through all settlement intervals and then re-bid at short notice. Moreover, these re-bids can occur several times within a single settlement interval. This action highlights the volatility of demand and the rapidity of changes in market prices, which are not easy to predict and are further compounded by the technical advantages of those generating units that can start up rapidly.

ELECTRICITY SYSTEMS AS COMPLEX ADAPTIVE SYSTEMS It is not only the exercise of market power that can cause dramatic price fluctuations from day to day. Temperature variations and network congestion play a similar role. Since retail prices are pegged, some retailers

106

Agent-based analysis of dynamic industrial ecosystems

and market customers find their profit margins squeezed. Each generating unit (for example coal-fired, gas-fired, hydroelectric, and renewable) has different start-up costs, start-up times, and performance variables. In a climate of uncertainty, choosing a generator unit and figuring out a profitable price at which to service newly posted demand is both challenging and risky. Clearly, the market for electricity has become far more complex. In an evolving world of power generation, transmission, and distribution, individual firms make use of an increasingly specific and timely feedback of bid prices, costs, realized demand, and operating information to enhance managerial decisions. Solely possessing information holds little competitive advantage. Private firms and government organizations invest in various analytical and decision-making tools as well as other sophisticated devices in order to gather pertinent information. Faster feedback requires faster adaptive reaction. Those who receive and respond to feedback more quickly gain competitive advantage (for example revising their bids or choosing not to bid at a time of volatile price movements). The challenge is to respond to information quickly yet profitably. Economic volatility is one aspect of the problem. A need to mitigate ecological impacts, especially greenhouse gas emissions, has highlighted the need to develop methods capable of addressing economic and ecological uncertainties consistently within an integrated framework. It is now believed that greenhouse gases contribute significantly to global warming. In Australia, the stationary energy sector accounts for almost 50 per cent of these emissions. Electricity generation dominates this sector, with 66 per cent of the emissions, thus acting as the major culprit in terms of total emissions. An additional complexity is that daily bidding strategies in the NEM are affected by hedged positions in other contract markets, such as the Over the Counter (OTC) Electricity Market (mainly Swaps and Caps, with some Swaptions and other Options) and the use of Renewable Energy Certificates (RECs). The United States have introduced other jurisdictional schemes to reduce greenhouse gas emissions (MacGill et al. 2006). In order to understand planning and decision making over different time horizons, these new schemes must be considered. Interdependencies between the spot (NEM) and contract markets are important factors in determining hedging decisions of generators, retailers, and other market participants. Hedged positions are important because they influence incremental investment in the medium term and thus the closure decisions of agents. Electricity markets are an evolving system of complex interactions between nature, physical structures, market rules, and participants (Figure 8.2). Participants face risk and volatility as they pursue their

Managing energy futures and greenhouse gas emissions

107

Changes in weather and weather forecasts contribute to congestion and price volatility in the NEM Markets = NEM + OTC + RECs

Climate & Weather

Networks = NEM + DG

GHG Emissions Calculator

Socio-economic and Technical System

Stationary energy accounts for about 50% of all GHG emissions

Natural System

Figure 8.2 The NEM as a complex adaptive system goals and make decisions based on limited information and their mental models of how they believe the system operates. There is a wide diversity of agents: agents use different strategies, have different capacities, use different generation technologies, have different ownership, are physically located at different locations, and face different grid constraints. In summary, their objectives, beliefs, and decision processes vary markedly. Such a diversity of inputs may be expected to lead to a rich diversity of market outcomes. The NEM possesses the intrinsic features of a complex adaptive system: a large number of intelligent and reactive agents, interacting on the basis of limited information and reacting to changes in demand (due to weather and consumer needs). As no single agent is in control, some (for example generators) may profit more than others. The result may be a considerable degree of price volatility, inadequate reserves, demand uncertainty, and unacceptably high levels of greenhouse gas emissions. A key question then arises: what kinds of energy-economy models (if any) are capable of handling these complexities? In the next section, several approaches are reviewed in an attempt to answer this question.

APPROACHES TO ELECTRICITY MARKET MODELING From a structural viewpoint, the approaches to electricity market modeling reported in the technical literature can be grouped into three main classes: optimization, equilibrium, and simulation models (Ventosa et al. 2005):

108

1. 2. 3.

Agent-based analysis of dynamic industrial ecosystems

Optimization models consider a single firm, dealing with exogenous price and demand price functions. Market equilibrium models consider all firms, exploring Cournot equilibrium and supply function equilibrium. Simulation models include equilibrium models and agent-based models.

Optimization models focus on the profit maximization problem for a single firm competing in the market, while equilibrium models represent the overall market behavior, taking into consideration competition among all participants. Simulation models are regarded increasingly as an alternative to equilibrium models when the problem under consideration is too complex (for example there are too many nonlinear interdependencies) to be addressed within a traditional equilibrium framework. Optimization models are better for short-term applications, such as daily bidding and pricing, as they can deal with more difficult and detailed problems. Equilibrium models are better for long-term planning applications since they consider all participants. Simulation models are appropriate for both short- and long-term applications due to their flexible modeling capabilities. Previous review articles (Kahn 1998; Day et al. 2002; Ventosa et al. 2005) have focused mostly on the equilibrium models found in game theory. Kahn’s survey is limited to two types of equilibrium resulting from firms in oligopolistic competition: (1) Cournot equilibrium, where firms compete on a quantity basis; and (2) supply function equilibrium (SFE), where they compete on both quantity and price. Although both models are based on the Nash equilibrium concept, the Cournot approach is usually regarded as being more flexible and tractable. Kahn (1998) also admits that the equilibrium models are less accurate about the price formation details than production simulation models. Day et al. (2002) conducted a more detailed survey of the modeling literature, listing the strategic interactions that have been, or could be, included in power market models. These interactions include (1) pure competition, (2) generalized Bertrand strategy (game in prices), (3) Cournot strategy (game in quantities), (4) collusion, (5) Stackelberg (leader–follower games), (6) supply function equilibria, (7) general conjectural variations, and (8) conjectured supply function (CSF) equilibria. Their conclusion found that the CSF approach to modeling oligopolistic competition is more flexible than the Cournot assumption and more computationally feasible for larger systems than the standard supply function equilibrium models. Simulation models are an alternative to equilibrium models when the problem under consideration is too complex to be addressed within a

Managing energy futures and greenhouse gas emissions

109

formal equilibrium framework. Equilibrium models also impose limitations on the representation of competition between agents in an electricity market. In addition, the resulting set of equations is frequently difficult or impossible to solve. The fact that power systems are based on the operation of generation units with complex physical constraints further complicates the situation. This is certainly the case in the medium to long term, when investment decisions, hedging strategies and learning processes become important endogenous variables. It is also vital when agent–environment interactions are important, such as in models that compute levels of greenhouse gas emissions. Simulation models typically represent each agent’s strategic decision dynamics by a set of sequential rules that can range from scheduling generation units to constructing offer curves that include a reaction to previous offers that competitors submit. The great advantage of a simulation approach lies in the flexibility it provides to implement almost any kind of strategic behavior, including feedback loops between the auction market and forward contract markets. This freedom, however, requires that the assumptions embedded in the simulation be more carefully (and empirically) justified. Some simulation models are closely related to equilibrium models. For example, Day and Bunn (2001) devise a simulation model that constructs optimal supply functions to analyze the potential for market power in the England and Wales pool. This approach is similar to the SFE scheme, but it provides a more flexible framework that enables us to consider actual marginal cost data and the asymmetric behavior of firms. In this simulation model, each generation company assumes that its competitors will keep the same supply functions that they submitted the previous day. Uncertainty about the residual demand curve is due to demand variation throughout the day. In the next section, a subfield within the realm of simulation models is explored, attracting increased attention for the modeling of electricity markets: agent-based simulation.

AGENT-BASED MODELS OF ELECTRICITY MARKETS The interactions within an electricity market constitute a repeated game, whereby a process of experimentation and learning changes the behavior of the firms in the market (Roth and Erev 1995). What is needed is a computational technique that can reflect such learning processes as well as modeling the structure, market clearing mechanism, and environmental

110

Agent-based analysis of dynamic industrial ecosystems

interactions (with a high level of detail). The most promising technique at this point in time is agent-based simulation, or ABS, for short (see Batten 2000). ABS provides a dynamic framework to explore the influence that the repetitive interaction of participants exerts on the evolution of wholesale electricity markets like the NEM. Static models neglect the fact that agents base their decisions on the historic information accumulated due to the daily operation of market mechanisms. In other words, they have good memories and learn from past experiences (and mistakes) to improve their decision making and adapt to changes in several environments (economic, physical, institutional, and natural). Adaptive ABS techniques can shed light on some of the dynamic features of electricity markets that static equilibrium models are unable to discern. A Brief Review of Earlier Agent-based Models Bower and Bunn (2000) present an agent-based simulation model in which generation companies are represented as autonomous adaptive agents that participate in a repetitive daily market and search for strategies that maximize their profit based on the results obtained in the previous session. Each company expresses its strategic decisions by means of the prices at which it offers the output of its plants. Every day, companies are assumed to pursue two main objectives: a minimum rate of utilization for their generation portfolio and a higher profit than that of the previous day. The only information available to each generation company consists of its own profits and the hourly output of its generating units. As usual in these models, a linear demand curve represents the demand side. Bunn and Oliveira (2001) developed a simulation platform that represents, in more detail than the above model, the way that market clearing in NETA was designed to function. This platform (1) models the interactions between the power exchange and balancing mechanism, (2) considers that generators may own different types of technologies, (3) considers an active demand side, including suppliers, and (4) takes into account the learning dynamics underlying these markets as a process by which a player selects the policy to use in the game, through interactions with components. In later work, the authors adapt and extend this simulation platform to analyze whether the two particular generators in the competition commission inquiry gained enough market power to operate against the public interest (Bunn and Oliveira 2003). Researchers at Argonne National Laboratory in Chicago have developed the Electricity Market Complex Adaptive System (EMCAS) model (Veselka et al. 2002). Like the above-mentioned simulation models developed at the

Managing energy futures and greenhouse gas emissions

111

London Business School, the EMCAS model is an electronic laboratory that probes the possible effects of market rules through a simulation of the strategic behavior of participants. EMCAS agents learn from their previous experiences and modify their behavior based on the success or failure of their previous strategies. Genetic algorithms are used to drive the adaptive learning of some agents, and pool, bilateral contract, and ancillary services markets are included. The EMCAS model is arguably the most sophisticated agent-based electricity model to date, embodying more development hours than other simulation models of its type. At Iowa State University, Leigh Tesfatsion and her colleagues have examined market power experimentally in an agent-based computational wholesale electricity market operating under different concentration and capacity conditions (Nicolaisen et al. 2001). Double auction with discriminatory midpoint pricing determines pricing. Buyers and sellers use a modified Roth–Erev individual reinforcement learning algorithm to determine their price and quantity offers in each auction round. Taylor et al. (2003) developed an agent-based model to simulate the complexity of the large-scale Victorian gas market in south-eastern Australia. The model can be used to elicit possible emergent behavior that could not be elicited otherwise under an uncertain future of deregulation and restructuring. Like an electricity market, the complexity in the gas market derives from the uncertain effects of a multiplicity of possible participant interactions in numerous segments, such as production, storage, transmission, distribution, retailing, service differentiation, wholesale trading, power generation, and risk management. A similar platform has been adopted for our National Electricity Market Simulator (NEMSIM).

NEMSIM: THE NATIONAL ELECTRICITY MARKET SIMULATOR NEMSIM is an ABS model that represents Australia’s National Electricity Market as an evolving system of complex interactions between human behavior in markets, technical infrastructures, and the natural environment. This simulator is the first of its kind in Australia. Users of NEMSIM will be able to explore various evolutionary pathways of the NEM under different assumptions about trading and investment opportunities, institutional changes, and technological futures, including alternative learning patterns as participants grow and change. The simulated outcomes will help the user to identify futures that are eco-efficient (for example that maximize profits in a carbon-constrained future). Questions about sustainable development, market stability, infrastructure security, price volatility, and

112

Agent-based analysis of dynamic industrial ecosystems

Figure 8.3 An overview of NEMSIM greenhouse gas emissions could be explored with the help of the simulation system. An overview of NEMSIM is shown in Figure 8.3. The NEMSIM project is part of CSIRO’s Energy Transformed Flagship research program, which aims to provide innovative solutions for Australia’s pressing energy needs. Motivation for the project is the Flagship’s mission to develop low emission energy systems and technologies. Learning from Experience The development of NEMSIM is facilitated by six years of historical market data on demand, pricing and power dispatch for the NEM. Being a dynamic, repetitive and information-rich system, the NEM offers large amounts of data. NEMSIM uses this data to extract representative patterns of regional demand (on a daily, weekly, and seasonal basis), regional prices, supply and demand growth, and so on. The historical database includes an extensive time series of bidding data, an essential source of information about market participants and their trading strategies. In NEMSIM, silicon agents (for example representing generator firms, network service providers, retail companies, the market operator and

Managing energy futures and greenhouse gas emissions

113

others) buy and sell electricity in a simulated trading environment. Agents have different goals; in addition to maximizing profits, some may wish to increase market share, diversify generation sources, or work more closely with end-users. Short-term strategies (for example bidding tactics in the NEM) are affected by medium-term strategies (for example hedged positions in the OTC markets). In turn, both are affected by investment decisions and other changes over the longer term. NEMSIM treats agents as being uniquely intelligent, making operational and strategic decisions using the individual information available to them. They are also adaptive and can learn to modify their behavior in order to realize their goals. Learning algorithms can allow agents to look back (learn from their historical performance), look sideways (learn from other participants’ strategies), and look ahead (take future plans and forecasts into account). In an adaptive market like the NEM, no single agent has control over what all the other agents are doing. The market is structured in a way that allows the marginal bidder to exert more influence on market outcomes. The overall outcome is not always obvious since it depends on many factors. NEMSIM provides a platform on which simulated agents interact, constrained only by realistic rules and the physical grid system. Agents’ individual and collective behaviors co-evolve from the bottom up, producing both expected and unexpected emergent outcomes at the system level. NEMSIM Environments Simulated agent life in NEMSIM unfolds in three “environments”: (1) a trading environment in which transactions can occur in interlinked spot and forward contract markets, (2), a physical grid of sites, generation units, lines and inter-connectors across which electricity flows, and (3), a natural environment, which provides energy resources and accumulates greenhouse gas emissions. Each “environment” is separate from the agents, on which the agents operate and with which they interact. The physical environment of transmission and distribution infrastructure imposes several constraints on electricity market operations. For example, the transmission capacity between the two regions constrains the quantity of power sold from one region to another. NEMSIM represents the diversity of objects and attributes associated with generation and transmission infrastructure in a simplified form. Some key objects (with corresponding attributes in parentheses) are generating plants (location, unit composition), generating units (maximum capacity, generation technology, ramp rates, efficiency and emission factors), and interconnectors (transmission technology, adjacent regions, losses).

114

Agent-based analysis of dynamic industrial ecosystems

NEMSIM as a Greenhouse Gas Emissions Calculator The effects of anthropogenic and natural greenhouse gas (GHG) emissions on the absorption of terrestrial radiation and global warming have been extensively studied in the last two decades. Changes in the concentrations of GHGs can alter the balance of energy transfers between the atmosphere, land, oceans, and space. Higher concentrations of GHGs increase the Earth’s energy absorption, thus producing global warming. Carbon dioxide (CO2) is the main component of GHG emissions. Over the last 150 years, the concentration of atmospheric CO2 has increased from about 280 parts per million by volume (ppmv) to about 372 ppmv (US Environmental Protection Agency 2004). Without intervention, it will exceed 550 ppmv by the end of this century. Efforts to mitigate or reduce GHG emissions include international initiatives such as the UN Framework Convention on Climate Change and the Kyoto Protocol. The total anthropogenic GHG emissions, however, are expected to continuously grow irrespective of current initiatives, with one of the main causes being increasing demand for power generation. Electricity generation is a major source of greenhouse gas emissions. According to the Australian Greenhouse Office (2006), it contributed 195.2 Mt of carbon dioxide equivalent emissions (Mt CO2-e) in 2004 or 34.56 per cent of net national emissions (564.7 Mt CO2-e) in Australia. Emissions from electricity generation increased by 65.8 Mt (50.8 per cent) from 1990 to 2004. The net emissions are calculated across all sectors under the accounting provisions of Kyoto Protocol for Australia. Graham et al. (2003) provide a method of calculating the potential GHG emissions from future electricity generation in Australia. This method aims to evaluate the impact of future development options of the electricity market in terms of GHG emissions and costs. It is based on a portfolio simulation framework and CSIRO’s Electricity Market Model (Graham and Williams 2003) that is similar to the bottom-up type of dynamic optimization models such as MARKAL. MARKAL, developed by the International Energy Agency over a period of more than 20 years, is a generic model that aims to represent the evolution of a specific energy system over a period of up to 50 years at the national, state, or regional level. Graham et al. (2003) studied five GHG targets (business as usual, two moderate and two extreme emissions targets) in conjunction with six key data assumptions (more abundant gas available, CO2 capture and sequestration feasible/infeasible, high/low demand growth, etc.). As an agent-based simulation tool, NEMSIM can calculate the GHG emissions associated with electricity generation of a given simulation scenario by way of bottom-up type aggregation. The advantages of NEMSIM

Managing energy futures and greenhouse gas emissions

115

lie in its simulation framework, allowing precise modeling of emissions to the level of each generating unit, accommodating a variety of changes in the operational, technological, company, market, and regulatory values and parameters. The ABS framework has the advantage of being able to examine the nested effects of slow changes (accumulating over long periods of time) and sudden changes that arise from emergent events, based purely on the decision making and interactions of the participating agents. GHG emissions calculations from electricity generation are approximated in NEMSIM as fossil fuel consumption, using fuel-specific (generation technology) emission factors. A set of generation technologies are modeled, including conventional black coal/pulverized fuel, conventional brown coal/pulverized fuel, natural gas simple cycle, and so on. The set of generation technologies is specified in an XML input file that allows changes to generation technologies and their parameters. New generation technology such as supercritical black coal can be added easily. The main attributes of each generation technology are the emission GHG factor and the net energy efficiency that describes how much of the embodied energy of the fossil fuel is transformed into electricity. The net energy efficiency can be defined for a selected generating unit to allow flexibility over the longer term, when the efficiency may change. Emissions associated with extraction and production of fossil fuels are not considered. This approach allows easier comparison between different plants, companies, and regions. It is a first approximation, however, as it does not include indirect emissions that usually show moderate variability by region, company, and technology. Figure 8.4 presents a regional summary graph of GHG emissions. Again the numbers are only illustrative. Other options for output windows and reports pertaining to GHG emissions are available within NEMSIM. Distributed Generation For environmental and security reasons, distributed generation (DG) is expected to grow in importance in the near future. A comprehensive list of several definitions of DG in the context of competitive electricity markets can be found in Ackermann et al. (2001). DG is loosely defined as a set of small-scale power generation technologies (up to 10 000 kW) located close to where electricity is used (for example a home or business). It provides an alternative to, or an enhancement of, the traditional, centralized electric power system. Small generators have been used as back-up generators and on-site power systems for a long time. Thus, DG is not a new idea. In the early years of electricity generation, the power plant supplied only local customers and even the grid was based on direct current. On the other hand, newer DG technologies (for example microturbines, fuel cells,

116

Agent-based analysis of dynamic industrial ecosystems

1.2

60 000

1.1

55 000

1.0

50 000

0.9

45 000

0.8

40 000

0.7

35 000

0.6

30 000

0.5

25 000

0.4

20 000

0.3

15 000 10 000

0.2

5 000

0.1

0

NSW

OLD

SNOWY

SA

TAS

VIC

Specific CO2 emissions (t/MWh)

CO2 emissions volume (kt)

Regional emissions, 1/1/2004–31/12/2004 65 000

0.0

Region CO2 emissions volume (kt)

Specific CO2 emissions (t/MWh)

Figure 8.4 NEMSIM regional summary graph for GHG emissions photovoltaic, and wind systems) have been developed recently. Distributed generators are connected to the distribution part (11–33 kV) of the electricity networks, at or near the end user. Potential benefits from DG are lower cost, higher service reliability, higher power quality, and increased energy efficiency (see Pepermans et al. 2005). DG is a promising solution for the security of electricity supply, providing distributed and diverse energy source infrastructure. The use of certain types of renewable DG can also provide a significant environmental benefit in terms of reducing GHG emissions. Not all DG is beneficial for the environment, as smaller generating plants are usually less efficient in terms of thermal efficiency and fuel use. DG can be beneficial to both electricity consumers and energy utilities. In combination with demand side management, it can offer electric utilities alternatives to system capacity investment in large central generation, transmission and distribution (Hoff et al. 1996). DG can help manage peak load demands, thus having the potential to reduce price volatility and companies’ market power. Utility deregulation is an additional reason for the high level of interest in DG. Network integration of DG is a complex issue, significantly different from the conventional networking of power generating units and transmission lines (Ackermann et al. 2001). For example, the traditional distribution network infrastructure allows power to flow only in one direction. In the

Managing energy futures and greenhouse gas emissions 14000 Regional demand

10000

1 000

8000 320

6000 Regional price

4000

100

2000

Wholesale regional price [$/MWh]

10 000

12000 Regional demand [MW]

117

NSW – 1/12/2004 0 4:00

8:00

12:00

16:00 Hours

20:00

00:00

10 4:00

Figure 8.5 DG perturbed demand and price example case of DG, the power should be able to flow in both directions, to and from the customer. The additional cost of the network protection system, which has to be designed and implemented, must be taken into account when assessing economic impacts and conducting reliability analysis of DG. NEMSIM can aid in understanding these advantages and disadvantages, and the economic impacts of growth in the uptake of DG technologies. For example, Figure 8.5 shows the demand and price graphs for NSW on 1 December 2004 (a very hot day). Around midday, demand peaked, thereby stressing the supply system and resulting in the market’s regional pool price, jumping to its maximum allowable value of $10 000 per Megawatt-hour ($/MWh). If significant DG had been available on that day, it could have been engaged mid-morning (when the price exceeded a threshold value of 320 $/MWh), thereby reducing demand and preventing or reducing price spikes. The dotted lines in Figure 8.6 represent the simulated influence of DG on demand and price curves. NEMSIM capabilities to model DG are under development. NEMSIM aggregates a number of homogeneous DG units into clusters. Usually, but not necessarily, each cluster has a number of identical units. For example, a cluster may aggregate DG units that use the same generation technology and possess similar operational characteristics, but are not necessarily identical in terms of capacity or manufacturer. Aggregation is not only based on similarity of generation parameters, but also on location (for example a load center). For different scenarios of DG growth, different aggregation strategies are possible. Each cluster has an aggregated generation capacity

118

Agent-based analysis of dynamic industrial ecosystems

Figure 8.6 NEMSIM vertical bars of clusters of DG next to load centers and specific rules for switching it on and off; these rules are usually related to wholesale price thresholds. A cluster may switch on when prices exceed 100 $/MWh and switch off when the price drops below 80 $/MWh. An example of a NEMSIM graphical window for a cluster of DG units is shown in Figure 8.6. Circles in this diagram represent loads that are connected to busses (thick horizontal lines). The lines between the busses represent transmission lines. DG clusters are depicted by vertical bars next to the corresponding load. During the simulation, the DG clusters/bars that are colored gray are fully or partially dispatched, whereas those colored white are not dispatched. If a cluster is dispatched, the generation capacity of that cluster offsets the corresponding regional demand. In this way, all clusters play a “negative demand” role when they are switched on.

CONCLUSIONS The purpose of agent-based simulation models such as NEMSIM and EMCAS is to generate and explore, but not to predict, alternative futures

Managing energy futures and greenhouse gas emissions

119

that may develop under different conditions. These simulators can explore various “what-if ” scenarios under different eco-efficiency goals. They can show the possible evolutionary trajectories of a given scenario under given conditions. For example, the introduction of more distributed generation into the marketplace involves a transition from the current paradigm of the centrally dispatched electricity grid to new, more decentralized ones. This may involve new markets, new brokers, new technologies and new grid structures. Some of these distributed strategies can reduce the level of GHG emissions considerably. NEMSIM and EMCAS are generative simulation tools that can identify the transition states needed to reach specific final states, including more eco-efficient futures. The main challenge for these agent-based simulation tools remains as an adequate representation of adaptive behaviors of company agents, including employment of proper learning algorithms.

ACKNOWLEDGMENTS Financial support from CSIRO Centre for Complex Systems Science and CSIRO Land and Water (for the first author to attend the ISIE-2005 Conference in Stockholm and to prepare the first draft of this chapter) is gratefully acknowledged. Both authors are grateful to our colleagues Paul Graham, Geoff Lewis, John Mo, Shanon McQuay, Marcus Thatcher, Miles Anderson, Chi-hsiang Wang and Scott Maves for their contributions towards the development of NEMSIM. The ABM platform used to build NEMSIM was provided by Swinburne University of Technology, Melbourne, Australia.

REFERENCES Ackermann, T., G. Andersson and L. Söder (2001), “Distributed generation: a definition”, Electric Power Systems Research, 57 (3): 195–204. Australian Greenhouse Office (2006), Australian Greenhouse Accounts: National Inventory Report 2004 – Volume 1, Canberra: Commonwealth of Australia. http://www.greenhouse.gov.au/inventory/. Barmack, M.A. (2003), “What do the ISOs public bid data reveal about the California market?”, The Electricity Journal, 24: 63–73. Batten, D.F. (2000), Discovering Artificial Economics: How Agents Learn and Economies Evolve, New York, NY: Westview Press. Batten, D.F. and G. Grozev (2007), “NEMSIM: finding ways to reduce greenhouse gas emissions using multi-agent electricity modeling”, in P. Pascal and D. Batten (eds), Complexity Science for a Complex World: Exploring Human Ecosystems with Agents, Canberra: ANU ePress.

120

Agent-based analysis of dynamic industrial ecosystems

Borenstein, S., J. Bushnell and F. Wolak (2002), “Measuring market inefficiencies in Californias restructured wholesale electricity market”, American Economic Review, 92(5): 1376–405. Bower, J. and D.W. Bunn (2000), “A model-based comparison of pool and bilateral market mechanisms for electricity trading”, The Energy Journal, 21(3): 1–29. Bunn, D.W. and F.S. Oliveira (2001), “Agent-based simulation: an application to new electricity trading arrangements of England and Wales”, IEEE Transactions on Evolutionary Computation, 5(5): 493–503. Bunn, D.W. and F.S. Oliveira (2003), “Evaluating individual market power in electricity markets via agent-based simulation”, Annals of Operations Research, 121: 57–77. Day, C. and D. Bunn (2001), “Divestiture of generation assets in the electricity pool of England and Wales: a computational approach to analyzing market power”, Journal of Regulatory Economics, 19(2): 123–41. Day, C.J., B.F. Hobbs and J.S. Pang (2002), “Oligopolistic competition in power networks: a conjectured supply function approach”, IEEE Transactions on Power Systems, 17(3): 597–607. Georgescu-Roegen, N. (1971), The Entropy Law and the Economic Process, Cambridge, MA: Harvard University Press. Graham, P. and D.J. Williams (2003), “Optimal technological choices in meeting Australian energy policy goals”, Energy Economics, 25(6): 691–712. Graham, P., N. Dave, P. Coombes, D. Vincent and G. Duffy (2003), Options for Electricity Generation in Australia, Technology Assessment Report 31, Cooperative Research Centre for Coal in Sustainable Development. https:// ejournal.csiro. Hoff, T.E., H.J. Wenger and B.K. Farmer (1996), “Distributed generation: an alternative to electric utility investment in system capacity”, Energy Policy, 24(2): 137– 47. Hu, X., G. Grozev and D. Batten (2005), “Empirical observations of bidding patterns in Australia’s national electricity market”, Energy Policy, 33(16): 2075–86. Kahn, E.P. (1998), “Numerical techniques for analyzing market power in electricity”, The Electricity Journal, 11(6): 34–43. MacGill, I., H. Outhred and K. Nolles (2006), “Some design lessons from marketbased greenhouse gas regulation in the restructured Australian electricity industry”, Energy Policy, 34(1), 11–25. Nicolaisen, J., V. Petrov and L. Tesfatsion (2001), “Market power and efficiency in a computational electricity market with discriminatory double auction pricing”, ISU Economic Report No. 52, August. Pepermans, G., J. Driesen, D. Haeseldonckx, R. Belmans and W. D’haeseleer (2005), “Distributed generation: definition, benefits and issues”, Energy Policy, 33(6): 787–98. Roth, A.E. and I. Erev (1995), “Learning in extensive form games: experimental data and simple dynamic models in the intermediate term”, Games and Economic Behavior, 8:164–212. Short, C. and A. Swan (2002), “Competition in the Australian national electricity market”, ABARE Current Issues, January, 1–12. Taylor, N.F., M.P. Harding, G.S. Lewis and M.G. Nicholls (2003), “Agent-based simulation of strategic competition in the deregulating Victorian gas market”, Paper presented at the 2003 WDSI Conference.

Managing energy futures and greenhouse gas emissions

121

US Environmental Protection Agency (2004), US Emissions Inventory 2004: Inventory of U.S. Greenhouse Gas Emissions and Sinks: 1990–2002, Washington, DC: US Environmental Protection Agency. Ventosa, M., A. Baillo, A. Ramos and M. Rivier (2005), “Electricity market modeling trends”, Energy Policy, 33(7): 897–913. Veselka, T., G. Boyd, G. Conzelmann, V. Koritarov, C. Macal, M. North, B. Schoepfle and P. Thimmapuram (2002), “Simulating the behavior of electricity markets with an agent-based methodology: the electricity market complex adaptive system (EMCAS) model”, International Association for Energy Economists (IAEE) North American Conference, 6–8 October, Vancouver, B.C. Wolfram, C.D. (1999), “Measuring duopoly power in the British electricity market”, American Economic Review, 89(4): 805–26.

9. Organizational dynamics in industrial ecosystems: insights from organizational theory Jennifer Howard-Grenville and Raymond Paquin INTRODUCTION Industrial ecology’s strength, as a theoretical and practical approach, is its attention to systemic interactions between organizations and the natural environment itself (Graedel and Allenby 1995). The metaphor of sustainable natural ecosystems provides a vision for the transformation of industrial production through optimizing the flow of materials and energy at the local, regional, and global scale (Chertow 2000; Korhonen et al. 2004). This perspective has shifted attention away from improving or optimizing the immediate environmental impact of a single facility, as typically demanded by pollution prevention and other traditional environmental management approaches, towards a more holistic view of the company as a part of a larger network of exchanges. This physical network, however, must also be understood as embedded within a social system that influences individual and organizational action, which in turn critically contributes to the development and dynamics of industrial ecosystems. This chapter presents three important and interrelated perspectives in organizational theory – institutional theory, field theory, and social network theory – that together shed light on the organizational dynamics inherent in and critical to the development of industrial ecosystems. Within the social science literature, these theories provide a language and set of conceptual tools for holistically analyzing the formal and informal influences of the broader social environment on a company, its possible actions within this environment, and associated outcomes. Such an understanding also sheds light on the constraints and opportunities that individual decision makers face and enables a better understanding of agents’ behavior, whether that of individuals or organizations, in bringing about the changes that industrial ecology demands. 122

Organizational dynamics in industrial ecosystems

123

Understanding organizational action is particularly important for the approach to industrial ecology known as industrial symbiosis. In industrial symbiosis, companies physically exchange materials, energy, water, and byproducts to realize net environmental and economic gains (Chertow 2000). Critical to industrial symbiosis is collaboration between disparate companies, who may share only relative geographic proximity. Empirical accounts point to numerous non-technical reasons for the failure of industrial symbiosis efforts, including failures of communication, coordination, trust, and reciprocity (Heeres et al. 2004; Gibbs et al. 2005) as well as numerous explanations for successes, including effective brokering, education, repeated interactions, and learning over time (Baas and Boons 2004; Chertow and Lombardi 2005; Ehrenfeld and Chertow 2002; Jacobsen 2005; Malmborg 2004). All of these reasons are rooted in the social behaviors and interactions of individuals and organizations. Indeed, scholars have called for greater attention to the “social side” of industrial ecology, suggesting that understanding individual, organizational, and interorganizational behaviors are critical to understanding how industrial ecosystems emerge and develop (Andrews 2001; Ehrenfeld 2000, 2004; Hoffman 2003; Korhonen et al. 2004). Several researchers have begun to draw directly from organizational theory, including institutional theory and social network theory to analyze and interpret particular examples of industrial symbiosis and industrial ecology (Baas and Boons 2004; Jacobsen 2005; Malmborg 2004). Despite this, there is no comprehensive effort to lay out concepts and tools from the social sciences for assessing the dynamics of industrial ecosystems. At many levels of analysis, parallels exist between the concepts and tools currently prevalent within the field of industrial ecology and those introduced here. First, at the conceptual or metaphorical level, industrial ecology “requires that an industrial system be viewed not in isolation from its surrounding systems, but in concert with them. It is a systems view in which one seeks to optimize the total materials cycle, from virgin materials, to finished material, to component, to product, to obsolete product, and to ultimate disposal” (Graedel and Allenby 1995: 9). Similarly, attention to the social side of industrial ecology must also embrace an “open systems” view as articulated in the social sciences, which sees organizations and their actions as embedded in, and at least partially defined by, their external environments (Hoffman 2003). Second, research on industrial symbiosis in particular finds that specific organizational, local, and regional characteristics make a difference to how the systemic influences are felt and acted upon, resulting in a variety of symbiotic configurations and approaches (Chertow 2000; Heeres et al. 2004). Similarly, institutional theory and field theory draw attention on the social side to the various configurations of

124

Agent-based analysis of dynamic industrial ecosystems

Table 9.1 Parallels between industrial ecology concepts and tools and those within organization theory Industrial Ecology Concepts and Tools

Organizational Theory Concepts and Tools

Conceptual/ Metaphorical Image

Systemic flows of material/ “Open systems” view of energy between industrial organizations – organizational organizations action inherently shaped by interactions with other Metaphor of sustainable organizations and broader natural ecosystem society

Configurations of Industrial Ecosystems

Empirical attention to local, regional, and national factors that contribute to particular configurations Empirical attention to the development/emergence of linkages

Analytic Tools to Map/Optimize Linkages and Exchanges

Materials flow analysis, life cycle analysis (LCA)

Field theory characterizes relevant organizational fields and the institutional norms operating within them Social and cultural capital explain who can take actions within fields, and how fields might evolve Social network analysis

corporate behavior that emerge within an open system, and provide explanations for why, within such configurations, individual companies have different incentives and opportunities to act on industrial ecology issues. Finally, industrial ecology has the tools of materials flow analysis and life cycle analysis at its disposal (Graedel 1998; Korhonen et al. 2004) to map and optimize linkages between individual organizations. The analogous tool in the social sciences is social network analysis, which allows for the mapping and analysis of interorganizational social linkages. Table 9.1 summarizes these parallels. Institutional theory, the theory of fields, and network theory, when taken together, can produce a robust, holistic approach to analyzing organizational dynamics within industrial ecosystems and understanding how individual and company choices shape the development of industrial ecosystems over time. The next section briefly reviews these organizational theories. We begin with institutional theory and field theory which focus on how social systems shape opportunity and action for individual organizations. Following this, we introduce social network analysis which can be

Organizational dynamics in industrial ecosystems

125

used to probe particular configurations of organizational arrangements. Throughout, examples from the industrial symbiosis literature are used to illustrate how insights from these organizational theories can complement and extend current industrial ecosystem analyses. Finally, this chapter concludes with an outline of some productive directions for exploiting insights from these theories to develop a better understanding of the behavioral aspects of industrial ecology.

ORGANIZATIONAL THEORY AND SOCIAL SYSTEMS A longstanding debate within organizational theory concerns the question of whether an organization’s social context constrains its managers’ choices and actions, or whether managers freely chart courses of action for their organizations. Mirroring a similar debate within sociology, Astley and Van de Ven stated that one perspective “views individual action as the derivative of the social system, and the other views the social system as the derivative of individual action” (1983: 251). Many scholars see neither extreme of “structure” nor “agency” as adequate to explain the actions of organizations and regard both as important. Institutional Theory Institutional theory recognizes the interplay of social structure and individual agency, making it well suited to exploring the social and organizational dynamics of industrial ecosystems. Institutional theory attends to the social norms that shape organizational action, and to how organizations and individuals can purposefully reshape these social norms over time (Greenwood et al. 2002; Lawrence 1999; Maguire et al. 2004). The term “institution” refers to often tacit, taken-for-granted norms or “rules of the game” that shape organizational and individual behavior. Formally defined, institutions “consist of cognitive, normative, and regulative structures and activities that provide stability and meaning to social behavior” (Scott 1995: 33). Less formally, institutions are “rules, norms and beliefs that describe reality for the organization, explaining what is and what is not, what can be acted upon and what cannot” (Hoffman 1999: 351). This understanding of institutions sheds light on empirical observations that industrial ecosystem interactions only “make sense” to a limited number of companies because they go against the norms for what constitutes “business as usual” in a number of contexts (Chertow 2000; Ehrenfeld and Gertler 1997; Heeres et al. 2004). Chertow observes that

126

Agent-based analysis of dynamic industrial ecosystems

“even explaining industrial symbiosis – the educational component – is arduous because industrial symbiosis is not business as usual, and requires a significant change to dominant, rugged individualist mental models” (2000: 332). Institutional theory provides a framework for understanding such mental models not just as individual perspectives, but as products of a number of broader social forces. The influencing forces likely include the particular regulatory climate for business, the historical trajectory of industrial interactions within a country or region, the contribution of communities, local authorities or other outsiders to business decision making, and the financial and economic pressures that shape how companies’ actions are valued and over what time period. Institutional Norms and Organizational Fields Institutional norms have meaning for a collective of organizations and individuals who comprise a relevant organizational “field”. A field is “a community of organizations that partakes of a common meaning system and whose participants interact more frequently and fatefully with one another than with actors outside of the field” (Scott 1995: 56). In contrast to an industry, a field may include regulators, pressure groups, communities, and/or businesses engaged in quite different activities. Any given organization is typically subject to many institutional norms, and is a member of many fields. For example, members of an industrial ecosystem will continue to operate in a field defined by their industry; they will still interact with their suppliers, distributors, customers, and competitors. However, these firms will also need to interact within a new field, one that may include members of local and regional government agencies, local industries, businesses operating in other sectors, environmental advocacy groups, and community organizations (Heeres et al. 2004). Fields take on different forms (Maguire et al. 2004). Some fields are mature or stable with dominant actors and strongly held institutional norms (Greenwood et al. 2002). Other fields, where norms and relationships are in flux, are considered emerging or fragmented (Fligstein 1997). There is likely much greater uncertainty surrounding what constitutes acceptable action in emerging and fragmented fields precisely because the norms are not well defined. Finally, mature fields may be “in crisis”. Previously stable relationships and norms are sharply disrupted in fields in crisis (Hensmans 2003; Maguire et al. 2004) and need to be re-established to the satisfaction of many groups within and outside the field. For example, following the Bhopal toxic gas release in the mid-1980s, the chemical industry faced a major loss of public confidence that ultimately led to the reconfiguration of its interactions with communities, regulators, and suppliers.

Organizational dynamics in industrial ecosystems

127

The development of industrial ecosystems often calls for fundamentally new arrangements, which may involve companies acting outside their traditional fields. As Heeres et al. observed, the exchanges demanded by industrial ecology involve “new unexpected connections between heterogeneous classes of industries or even outside industrial production” (2004: 987, emphasis in original). Reluctance to get involved in industrial symbiosis arrangements or eco-industrial parks reflects the uncertainty surrounding environmental and economic gains in such newly emerging fields (Chertow and Lombardi 2005; Heeres et al. 2004). Norms of engagement must be actively worked out as relationships between organizations are themselves developed within industrial ecosystems. For example, firms from different industries who share no prior relationships will need to develop norms surrounding the sharing of confidential operational information such as manufacturing plans or details on material and energy consumption. On the other hand, when prior shared norms exist, they may enable interaction and collaboration. Local business associations have been found to be important to the development of industrial symbiosis projects (Chertow and Lombardi 2005; Ehrenfeld and Chertow 2002; Jacobsen 2005). Membership in such associations likely provides at least the initial contours of a field for interaction. For example, in the By-Product Synergy Project in Tampico, Mexico, 18 of the 21 participating organizations were already involved in the local industry association, thus already operating in a common organizational field (Chertow 2000). This arrangement likely facilitated the 13 early industrial symbiosis projects at the site. In cases where a nascent local or regional field exists, establishing industrial symbiotic linkages is less a question of “making” a new field and its associated norms, and more one of extending the existing field to include new types of interactions and associated norms. In each type of field – mature, emerging, fragmented, or “in crisis” – the strategies and skills individual organizations use to shape the field and influence or perpetuate norms will be quite different (Fligstein 1997, 2001; Rao et al. 2000). Those seeking change need to identify the type of institutional norms that are influential, the fields in which they operate, and the type of fields that the relevant organizations are embedded in. This knowledge then enables an understanding of the opportunities, uncertainties, and risks that are present for companies seeking to depart from “business as usual” in a given setting. Agency within Organizational Fields Attention to institutional norms and fields reminds us that any organization’s actions are constrained – there is not an infinite range of choices

128

Agent-based analysis of dynamic industrial ecosystems

available, nor are choices determined solely by internal organizational factors. Yet, while institutional norms constrain organizational choice, they do not determine it. Early institutionalists argued that conformance with institutional norms led to isomorphism (DiMaggio and Powell 1983), resulting in members of a given field looking and acting like one another (Tolbert and Zucker 1983). More recent institutional research, however, attends to the more active role organizations play in creating and changing institutional norms through contestation, negotiation, and debate (Fligstein 2001; Greenwood et al. 2002; Lawrence 1999; Maguire et al. 2004). The role of institutional change – or the reshaping of rules and norms – is no longer only attributed to government (DiMaggio 1991) or social movement organizations (Benford and Snow 2000), but increasingly to organizations themselves as they strategically and purposively pursue their own agendas (Fligstein 2001; Lawrence 1999). Institutional entrepreneurs are organizations or individuals who possess or gain the resources and social skills to influence a field in a way to realize their own interests (DiMaggio 1988; Fligstein 2001; Maguire et al. 2004). An important empirical observation is that some actors are more successful than others at bringing about institutional change (Howard-Grenville et al. 2007; Maguire et al. 2004). Especially within emerging fields, effective institutional entrepreneurs are capable of bringing together disparate interests, connecting the “old” institutional logics with new ones and embedding these logics within ongoing practices (Maguire et al. 2004). As individual or organizational actors, they work with persistence, accumulating “small wins” over time and building on these to demonstrate to others the value of the new approaches or arrangements (Creed et al. 2002; Reay et al. 2005). Which organizations and individuals are best able to act as institutional entrepreneurs? This question is important for industrial ecology because for change to occur some organizations and individuals must act as early change agents, enabling the wider adoption of key practices like industrial symbiosis. Change agents, or institutional entrepreneurs, differ from others in a field because they hold greater power and influence. Power need not be simply economic or market clout, however. Social theorists think of power as having many facets and use the label “capital” to refer to multiple potential sources of power. Capital “represents a power over the field (at a given moment)” (Bourdieu 1985: 724) and may include cultural, economic, and/or social forms. Each form of capital represents a type of asset that an organization may hold, accumulate, and use over time. Although capital can be used to influence action, not all types of capital are equally valuable for doing so. For example, economic capital may not easily convert into the necessary cultural capital, which is often acquired slowly over time as an organization builds up trust and respect from others in the field.

Organizational dynamics in industrial ecosystems

129

Particularly important to the development of a number of local or regional industrial ecosystems have been the non-economic forms of capital (for example cultural and social capital) held by key members of the field. For example, the emergence of the Kalundborg industrial ecosystem is attributed to the “short mental distance” between participants (a form of cultural capital) and the enabling social connections (a form of social capital) achieved through the Rotary Club (Ehrenfeld and Gertler 1997; Jacobsen 2005). When initiators or participants in industrial symbiosis projects do not possess capital that others value, they will be ineffective at bringing others on board and ensuring their participation. For example, local government agencies provided funding for a number of US ecoindustrial parks, but companies were reluctant to participate because they did not see the agencies as trusting partners (Heeres et al. 2004). In this case, economic capital that government agencies provided could not substitute for necessary cultural or social capital. This example also nicely illustrates the strong connections between institutional norms and capital. What gives particular capital value (or not) within a given field are the institutional norms operating within that field. In the US example (Heeres et al. 2004), companies tended to value corporate anchor tenants over regulatory agencies. This preference is likely due to the broader institutional norm of government regulators as enforcers of inflexible, “command and control” style environmental regulation. This type of environmental regulation tends to limit the incentives for individual companies to innovate or cooperate with government agencies around environmental solutions, or to share “internal” information as necessary for environmental collaborations such as industrial symbiosis (Chertow 2000; Ehrenfeld and Chertow 2002; Ehrenfeld and Gertler 1997; Porter and Linde 1999). In Europe, however, a historically more collaborative relationship between organizations and regulatory agencies has shaped institutional norms where firms regard local agencies as more helpful in bringing together corporate partners in eco-industrial parks (Heeres et al. 2004). To understand the operation of capital, particularly social capital, in producing (or not producing) successful configurations for industrial ecosystems, it is necessary to turn to network analysis as a conceptual tool for focusing on interorganizational interactions within and across fields. Social capital is the “sum of resources . . . that accrue to an [actor] by virtue of possessing a durable network of . . . relationships of mutual acquaintance and recognition” (Bourdieu and Wacquant 1992: 119). As it exists inherently through network relationships, further understanding social capital demands a more detailed understanding of social networks themselves.

130

Agent-based analysis of dynamic industrial ecosystems

Social Network Theory A network describes the web of relationships existing between a set of individual or organizational actors. More formally, “a network is a metaphor to characterize a form of economic organization in which organizations have . . . permeable boundaries, and numerous connections to other organizations” (Smith-Doerr and Powell 2005: 380). Of course, the image of a network is prevalent in research on industrial ecology, primarily as applied to material and resource exchanges within industrial ecosystems. Such exchanges, however, are simply one set of formal connections existing between organizations. Social network analysis draws attention to the fact that “beneath most formal ties . . . lies a sea of informal relations” (Powell et al. 1996: 120). It is through these informal interorganizational relations that norms of action, reciprocity, and trust – the bedrock of social capital – develop. Social network theory and the analytic techniques of social network analysis (SNA) offer powerful ways of visualizing, analyzing, and comparing network structures and relationships across industrial ecosystems, and within the same industrial ecosystem over time. Using SNA, it is possible to identify the organizations (or individuals) who are more or less well connected within a network and make predictions about their capacity to act as change agents. SNA also enables visualization and analysis of the whole network and its evolution over time. Network Structure and Dynamics Social networks are conceptualized in two primary ways. The first conceptualization focuses on individual ties between network actors, and the second on aggregate relationships comprising the overall network. Tables 9.2 and 9.3 summarize a number of basic social network concepts discussed below. First, ties refer to connections that exist between two focal entities (individuals or organizations). For the purposes of this chapter, organizations are the primary entity of interest. Formal ties refer to interorganizational connections that are explicitly defined and agreed to, such as a contractual arrangement (for example supply chain relationships). Informal ties refer to the less structured connections between organizations, which may include common membership in professional associations, or simply friendships between individuals in different organizations. It is typically through informal ties, rather than formal ones, that norms of trust and reciprocity develop over time (Uzzi 1996). Further, formal ties often grown out of preexisting informal ties (Powell et al. 1996; Uzzi 1996; Kilduff and Tsai 2003). For example, the “short mental distance” described as essential to the development of the Kalundborg industrial ecosystem

Organizational dynamics in industrial ecosystems

131

Table 9.2 Network attribute definitions Attribute

Definition

Formal tie

Explicit connection between organizations in which information, goods and/or services are exchanged. Tacit, social connection between members of organizations, which develops prior to or in tandem with formal ties. A direct connection between two organizations. A connection between two organizations through a common third party, when the two organizations are not directly connected. A count of the number of steps required to connect two organizations in a network. The proportion of actual ties in a network relative to the total possible number of ties. How closely tied (directly or indirectly) an organization is to all other organizations in a network. Variety of types of organizations within a network.

Informal tie Direct tie Indirect tie Path length Density Centrality Diversity

(Ehrenfeld and Gertler 1997; Jacobsen 2005) suggests that informal ties existed between managers which facilitated early, formal industrial symbiosis exchanges. When two organizations share a direct tie to each other, it means they interact directly with each other (for example a direct supplier–customer relationship, industrial symbiosis project collaboration). When organizations share an indirect tie, they interact only via an intermediary organization (for example through a third-party “broker”). When examining indirect ties, it is also useful to consider path length – the number of others separating two organizations within a network. An easy way to understand path length is through the popular notion of “six degrees of separation” which captures the maximum number of ties posited to exist between any members of a social community. For organizations with direct ties, path length is one. When there is one organization between the two focal organizations, then path length equals two, and so forth. Organizations with direct ties and/or shorter path lengths are more likely to work together than those with longer path lengths and/or no ties at all (Kilduff and Tsai 2003; Uzzi 1996). The second way to view a network is through the aggregate relationships of the organizations involved, shedding light on the structure of the network itself and the relative power, or capital, of individual organizations within it. Two network level measures, density and centrality, show complementary dimensions of the level of connectedness between organizations in a network. Density describes how closely connected all organizations in a

132

Agent-based analysis of dynamic industrial ecosystems

Table 9.3 Network concepts, explanations, and visualizations Visualization

B

Direct & Indirect Ties: A and C have direct ties to B; A and C are indirectly tied to each through B. Path Length: A to B = 3; A to C = 6.

A

A

C

B

C

Density: From left to right, density increases from completely unconnected to partially connected to fully connected networks. Centrality: This network has one central organization (in the middle), and four surrounding organizations with less centrality, which are connected to the remaining organizations on the periphery. Diversity: Despite identical structures, the network on the right has greater diversity than the one of the left because it connects different types of organizations.

network are to each other, by measuring the ratio of actual interorganizational ties to the maximum potential ties in the network (if every organization was connected to every other). Although higher density networks tend to have stronger norms of action and reciprocity, organizations are often more insulated from external changes than are those in lower density networks. As a result, high density networks are regarded as less adaptable. The second measure, centrality, describes how much of a network’s density is organized around one or a few focal organizations. Centrality measures the ratio of the number of ties each organization has to the total number of ties in the network (Scott 2000). Central organizations, because of their higher proportion of ties, tend to have greater information flow from and influence on the network than organizations with less centrality (Powell et al. 1996; Smith-Doerr and Powell 2005). Industrial ecosystems with higher network density would be expected to have a larger number of symbiotic exchanges, and firms with higher centrality may serve as important “anchors” (Chertow 2000; Heeres et al. 2004), enabling the flow of

Organizational dynamics in industrial ecosystems

133

information, materials, resources, or “know-how” to link less connected companies. Diversity, another network-level attribute, describes the variety of the types of organizations within a network. For example, a network might be considered more diverse if it includes companies from multiple industries, small and large size companies, public and not-for-profit organizations, and/or single site firms along with facilities run by multinational parents. Overall, because diverse networks are more likely to expose organizations to different ways of interpreting issues and operating, organizations within them tend to be better at learning, creating new knowledge, and adapting to the external changes than those in homogeneous networks (Boons and Berends 2001; Goerzen and Beamish 2005; Powell et al. 1996). This suggests that network diversity may be an important consideration for the resilience of an industrial ecosystem as it develops over time. Embeddedness and Collaboration Organizations within networks often share multiple ties and multiple types of ties. Equally important as the structure of these ties are the nature, strength, duration, and overlap of the various relationships. As the number and type of ties between organizations increase over time, the organizations are said to become increasingly “embedded”. Embedded organizations are more likely to interact with each other, rather than with other organizations, even when economic considerations do not necessarily justify such interactions (Uzzi 1996). Given that industrial symbiosis projects often require organizations with little or no prior history to collaborate on projects whose economic justifications may be ambiguous (Ehrenfeld and Gertler 1997), embeddedness is likely an important attribute of many interorganizational relationships in successful industrial symbiosis arrangements. At least three aspects of embeddedness influence the extent and nature of collaborations between organizations. First, embeddedness increases interorganizational trust, potentially decreasing the cost of working together in the future (Uzzi 1996). Second, embeddedness increases information sharing. Organizations with embedded relationships are more likely to share necessary project information, even when it is sensitive in nature. As information sharing increases, the information is also perceived as more credible and valuable to those receiving it because it is coming from a trusted partnership. Finally, embeddedness increases joint problem solving. This stems directly from the previous two points – trust increases information sharing, which in turn also broadens the perspectives of the organizations involved. Overall, joint problem solving tends to increase organization-level learning, performance, and adaptability (Boons and Berends 2001; Powell et al. 1996; Uzzi 1996).

134

Agent-based analysis of dynamic industrial ecosystems

These findings from the social networks literature support many of the empirical observations of those who study industrial symbiosis and industrial ecosystems. Prior research on industrial ecosystems shows trust, communication, and collaborative problem solving as keys to success (Baas and Boons 2004; Ehrenfeld and Gertler 1997). Boons and Berends (2001) also suggest that the value of increased information sharing is further enhanced when sharing occurs within a heterogeneous, or diverse, network. For the particularly complex technical, operational, economic, social and regulatory problems encountered in the development of industrial ecosystems, the collaborative problem solving enabled by embedded network relationships also increases the chances of finding novel solutions amenable to all parties (Baas and Boons 2004; Ehrenfeld and Gertler 1997).

IMPLICATIONS To understand the emergence and development of industrial ecosystems, a robust, holistic understanding of the social systems in which individuals and organizations act must exist. Organizational and individual agents often need to change their current perspectives and relationships, implying changes in institutional norms and associated configurations of organizational fields. Furthermore, some agents will be more successful at bringing about intended changes than others. The language of capital and social networks allows researchers and practitioners to understand what gives rise to these differences and to identify characteristics of agents and broader social systems that enable the development and resilience of industrial ecosystems. Several important implications for research and practice follow. Agent-based Modeling of Organizational Interactions Individual managers and companies as a whole are agents whose decisions and actions critically shape whether and how industrial ecosystems emerge and develop. Social network analysis can be used to systematically analyze relationships within industrial ecosystems and compare relationships across several industrial ecosystems or within a single ecosystem over time. It provides a powerful way of representing the relative position and power, or capital, of various agents within the ecosystem, and for observing and predicting consequences of change over time. Some of the specific questions social network analysis can help answer include: How does the centrality of certain agents within an industrial ecosystem network influence the adoption of material exchanges or resource sharing arrangements? What role do preexisting ties between

Organizational dynamics in industrial ecosystems

135

agents play in the development of an industrial ecosystem and/or the emergence of particular exchanges? While social network analysis can capture the position of, and opportunities available to, agents, it must be enhanced by an understanding of broader institutional norms and the nature of particular ties in order to qualitatively assess the meaning of various network relationships and understand why and how they arose in a particular social context. For example, a highly central organization may posses certain types of capital, enabling it to exercise power over others in the network. The types of capital and their value, however, must be determined empirically and not simply assumed from a centrality measure. Brokering in Industrial Ecosystems The findings from the organizational theories discussed suggest it is difficult to develop new, collaborative arrangements between companies with no preexisting relationships; and that it takes time and repeated interactions to develop new norms supporting such arrangements (Uzzi 1996; Smith-Doerr and Powell 2005). In these settings, third party brokers, who facilitate introductions between organizations who previously did not know one another, may be particularly important. Although the term broker per se is not prominent in the industrial ecology literature (see Malmborg 2004 for an exception), researchers have suggested the value of local associations, anchor tenants, and local champions in drawing organizations together in industrial ecosystems (Gibbs 2001; Baas and Boons 2004; Chertow and Lombardi 2005; Heeres et al. 2004). Yet successful brokering is not merely about making connections. To be successful, the organizations involved must feel that the connections made are relevant and valuable (Burt 2002) and these organizations must trust the broker (Granovetter 1985). Finally, the broker must understand the context in which the brokering is taking place (Bourdieu 1986). Brokering takes different forms, with some brokers acting as passive conduits of information and others much more actively involved in transactions (Andrews and Mauer 2001). A distinction is drawn between information brokering, in which a third party collects and disseminates information to others, and relationship brokering in which the third party seeks to understand more deeply the needs of two companies, working with them to identify and implement appropriate solutions. While both are likely helpful in the development of industrial ecosystems, the latter may result in more robust, meaningful, and strongly embedded ties between companies. Several specific questions arise regarding the role and effectiveness of brokers (who may be individual or organizational agents) in influencing the organizational dynamics in an industrial ecosystem. For example, does the presence of a broker change the structure and operation of an

136

Agent-based analysis of dynamic industrial ecosystems

organizational network? Under what social and organizational conditions are information and/or relationship brokering valuable? How does the role of the broker change over time as the network structure and nature of relationships change? Social network analysis in combination with analyses of the institutional and field conditions can address these questions. Development of Industrial Ecosystems over Time Finally, the organizational theories discussed here make predictions about how organizational arrangements will change over time in industrial ecosystems, in response to the decisions and actions of particular individual and organizational agents. Mapping network structure over time, and seeing how ties between organizations change and evolve, can capture the organizational dynamics of the industrial ecosystem. Interesting questions include: What factors influence the nature and speed of change in interorganizational relationships within industrial ecosystems? How are the development of interorganizational ties fostered by the absence or presence of various types of brokers, regulatory incentives, economic conditions, and/or preexisting ties in emerging industrial ecosystems? These questions carry important practical implications. The most prominent model of an industrial ecosystem is Kalundborg, Denmark, which developed organically over about four decades. Given today’s pressing environmental needs, the prevailing opinion is that industrial ecosystems need to develop more quickly than this. However, as Ehrenfeld and Gertler (1997: 77) cautioned, “designing an industrial ecosystem from the ground up is different and cannot follow the evolutionary path that contributed so strongly to Kalundborg’s positive development”. Planned industrial ecosystems are often less effective than anticipated (Baas and Boons 2004; Heeres et al. 2004; Gibbs et al. 2005), suggesting that a middle ground between unassisted emergence and full planning is needed. Indeed, empirical research suggests that successful arrangements often grow from small projects, expanding in scope and participation over time as new projects prove successful (Baas and Boons 2004; Chertow 2000; Ehrenfeld and Gertler 1997). These findings fit well with the organizational theories outlined in this chapter, suggesting the value such theories can bring to informing and analyzing the development of industrial ecosystems.

CONCLUSION There is considerable potential for insights from organizational theories – such as institutional theory, field theory, and social network theory – to

Organizational dynamics in industrial ecosystems

137

contribute to a more holistic understanding of the social and organizational dynamics of industrial ecosystems. As Cohen-Rosenthal (2000: 245) observed, “knowledge of kinds of waste streams can provide a means to determine potential linkages. But this does not link them; decisions by people do.” This chapter has argued for paying attention to social and organizational factors that shape people’s decisions on these crucial matters. By attending to institutional norms and the fields and networks in which they operate, industrial ecology researchers can unite the long-held view of industrial facilities as embedded within systems of material and resource flows with a view of them as also embedded in systems of social interactions Both perspectives need attention in order to understand and influence the development of robust industrial ecosystems.

REFERENCES Andrews, C. (2001), “Building a micro foundation for industrial ecology”, Journal of Industrial Ecology, 4(3): 35–51. Andrews, C. and J. Mauer (2001), “Materials exchanges: an exploratory US survey”, Local Environment, 6(2): 149–68. Astley, W.G. and A.H. van de Ven (1983), “Central perspectives and debates in organization theory”, Administrative Science Quarterly, 28(2): 245. Baas, L.W. and F.A. Boons (2004), “An industrial ecology project in practice: exploring the boundaries of decision-making levels in regional industrial systems”, Journal of Cleaner Production, 12: 1073–85. Benford, R.D. and D.A. Snow (2000), “Framing processes and social movements: an overview and assessment”, Annual Review of Sociology, 26: 611. Boons, F. and M. Berends (2001), “Stretching the boundary: the possibilities of flexibility as an organizational capability in industrial ecology”, Business Strategy and the Environment, 10(2): 115–24. Bourdieu, P. (1985), “Social space and the genesis of groups”, Theory and Society, 14: 723–44. Bourdieu, P. (1986), “The forms of capital”, in J.G. Richardson (ed.), Handbook of Theory and Research for the Sociology of Education, Westport, CT: Greenwood Press, pp. 241–58. Bourdieu, P. and L.J.D. Wacquant (1992), An Invitation to Reflexive Sociology, Chicago, IL: The University of Chicago Press. Burt, R.S. (2002), “The social capital of structural holes”, in M.F. Guillen, R. Collins, P. England and M. Meyer (eds), New Directions in Economic Sociology, New York, NY: Sage. Chertow, M.R. (2000), “Industrial symbiosis: literature and taxonomy”, Annual Review of Energy and the Environment, 25: 313–37. Chertow, M.R. and D.R. Lombardi (2005), “Quantifying economic and environmental benefits of co-located firms”, Environmental Science & Technology, 39(17): 6535–41. Cohen-Rosenthal, E. (2000), “A walk on the human side of industrial ecology”, The American Behavioral Scientist, 44(2): 245.

138

Agent-based analysis of dynamic industrial ecosystems

Creed, W.E.D., M.A. Scully and J.R. Austin (2002), “Clothes make the person? The tailoring of legitimating accounts and the social construction of identity”, Organization Science, 13(5): 475–96. DiMaggio, P.J. (1988), “Interest and agency in institutional theory”, in L.G. Zucker (ed.), Institutional Patterns and Organizations, Cambridge, MA: Ballinger, pp. 3–22. DiMaggio, P.J. (1991), “Constructing an organizational field as a professional project: U.S. art museums, 1920–1940”, in W.W. Powell and P.J. DiMaggio (eds), The New Institutionalism in Organizational Analysis, Chicago, IL: University of Chicago Press, pp. 267–92. DiMaggio, P.J. and W.W. Powell (1983), “The iron cage revisited: institutional isomorphism and collective rationality in organizational fields”, American Sociological Review, 48(2): 147–60. Ehrenfeld, J.R. (2000), “Industrial ecology: paradigm shift or normal science?”, The American Behavioral Scientist, 44(2): 229–44. Ehrenfeld, J.R. (2004), “Industrial ecology: a new field or only a metaphor?”, Journal of Cleaner Production, 12: 825–31. Ehrenfeld, J.R. and M.R. Chertow (2002), “Industrial symbiosis: the legacy of Kalundborg”, in R.U. Ayres and L.W. Ayres (eds), Handbook of Industrial Ecology, Cheltenham, UK and Northampton, MA, USA: Edward Elgar. Ehrenfeld, J.R. and N. Gertler (1997), “Industrial ecology in practice: the evolution of interdependence at Kalundborg”, Journal of Industrial Ecology, 1(1): 67–79. Fligstein, N. (1997), “Social skill and institutional theory”, The American Behavioral Scientist, 40(4): 397. Fligstein, N. (2001), “Social skill and the theory of fields”, Sociological Theory, 19(2): 105–25. Gibbs, D. (2001), “Trust and networking in inter-firm relations: the case of ecoindustrial development”, Local Economy, 18: 222–36. Gibbs et al. (2005), “Industrial ecology and eco-industrial development: a potential paradigm for local and regional development?”, Regional Studies, 39: 171–83. Goerzen, A. and P.W. Beamish (2005), “The effect of alliance network diversity on multinational enterprise performance”, Strategic Management Journal, 26(4): 333–54. Graedel, T.E. (1998), Streamlined Life-cycle Assessment, Upper Saddle River, NJ: Prentice Hall. Graedel, T.E. and B. Allenby (1995), Industrial Ecology, Englewood Cliffs, NJ: Prentice Hall. Granovetter, M.S. (1985), “Economic action and social structure: the problem of embeddedness”, American Journal of Sociology, 91(3): 481–510. Greenwood, R., R. Suddaby and C.R. Hinings (2002), “Theorizing change: the role of professional associations in the transformation of institutional fields”, Academy of Management Journal, 45(1): 58–80. Heeres, R.R., W.J.V. Vermeulen and F.B. de Walle (2004), “Eco-industrial part initiatives in the USA and the Netherlands: first lessons”, Journal of Cleaner Production, 12: 985–95. Hensmans, M. (2003), “Social movement organizations: a metaphor for strategic actors in institutional fields”, Organization Studies, 24(3): 355. Hoffman, A.J. (1999), “Institutional evolution and change: environmentalism and the U.S. chemical industry”, Academy of Management Journal, 42: 351–71.

Organizational dynamics in industrial ecosystems

139

Hoffman, A.J. (2003), “Linking social systems analysis to the industrial ecology framework”, Organizations & Environment, 16(1): 66–86. Howard-Grenville, J., A.J. Hoffman and C.B. Bhattacharya (2007), “Who can act on sustainability issues? Corporate capital and the configuration of organizational fields as enablers”, in S. Sharma, M. Starik and B. Husted (eds), Organizations and the Sustainability Mosaic: Crafting Long-term Ecological and Societal Solutions, Cheltenham, UK and Northampton, MA, USA: Edward Elgar. Jacobsen, N.B. (2005), “Industrial symbiosis in the making: bridging social and technical explanations – The case of Kalundborg, Denmark”, Paper presentation, International Sustainable Development Research Conference, Helsinki, Finland, June 2005. Kilduff, M. and W. Tsai (2003), Social Networks and Organizations, Thousand Oaks, CA: Sage. Korhonen, J., D. Huisingh and A.S.F. Chiu (2004), “Applications of industrial ecology – an overview of the special issue”, Journal of Cleaner Production, 12: 803–7. Lawrence, T.B. (1999), “Institutional strategy”, Journal of Management, 25: 161–87. Maguire, S., C. Hardy and T.B. Lawrence (2004), “Institutional entrepreneurship in emerging fields: HIV/AIDS treatment advocacy in Canada”, Academy of Management Journal, 47(5): 657–79. Malmborg, F.V. (2004), “Networking for knowledge transfer: towards an understanding of local authority roles in regional industrial ecosystem management”, Business Strategy and the Environment, 13(5): 334. Porter, M.E. and C.V.D. Linde (1999), “Green and competitive: ending the stalemate”, Journal of Business Administration and Policy Analysis, January, 215–29. Powell, W.W., K.W. Koput and L. Smith-Doerr (1996), “Interorganizational collaboration and the locus of innovation: networks of learning in biotechnology”, Administrative Science Quarterly, 41: 116–45. Rao, H., C. Morrill and M.N. Zald (2000), “Power plays: how social movements and collective action create new organizational forms”, Research in Organizational Behavior, 22: 237–81. Reay, T., K. Golden-Biddle and K. GermAnn (2005), “Legitimizing a new role: small wins and micro-processes of change”, Academy of Management Journal, 49(5): 977–98. Scott, J. (2000), Social Network Analysis: A Handbook, 2nd edition, Thousand Oaks, CA: Sage. Scott, W.R. (1995), Institutions and Organizations, London: Sage Publications. Smith-Doerr, L. and W.W. Powell (2005), “Networks and economic life”, in N.J. Smelser and R. Swedberg (eds), The Handbook of Economic Sociology, 2nd edition, Princeton, NJ: Sage/Princeton University Press. Tolbert, P. and L. Zucker (1983), “Institutional sources of change in the formal structure of organizations: the diffusion of civil service reform, 1880–1935”, Administrative Science Quarterly, 28: 22–39. Uzzi, B. (1996), “The sources and consequences of embeddedness for the economic performance of organizations: the network effect”, American Sociological Review, 61: 674–98.

10. Changing stocks, flows and behaviors in industrial ecosystems: retrospect and prospect Brynhildur Davidsdottir and Matthias Ruth RETROSPECT Industrial systems, built of physical and natural components, are complex adaptive systems that individual agents control. These agents, such as people, households, firms, and agencies interact with each other through intricate networks, often cooperating or competing with their self-interest in mind. A change in one part of the system, for example a technology used in an industry or a constraint imposed by regulatory agencies, influences other parts of the system through dynamic, often non-linear, lagged and feedback-driven processes. Industrial systems closely interact with their economic, social, and environmental surroundings, cohesively forming an industrial ecosystem. As fundamental components of human society and cornerstones of human economies, the development of industrial ecosystems shapes regional and national economies and societies while influencing regional and national growth. They absorb material and energy inputs and emit wastes into the environment, thereby altering and influencing the natural environment. Thus as the structure and functioning of industrial ecosystems changes, the extent and quality of the interactions within them and with their surroundings modify as well. Given the importance of industrial ecosystems to human and natural systems, an understanding of the dynamics of industrial ecosystems is pertinent to successful industrial, economic, and environmental management and for successful planning of sustainable futures. The scholarship of the dynamics of industrial ecosystems is still in its infancy and scattered throughout the literature. This volume has brought together various fragments, contributing to a comprehensive paradigm ranging from concepts and analogies to the various modeling approaches used to analyze the dynamics of industrial ecosystems. 140

Retrospect and prospect

141

The volume is composed of three sections, with each providing a specific perspective to and illustrations of the dynamics of industrial ecosystems. Here, a review of the individual contributions to each section is given with a summary of the main conclusions. The chapter concludes with some thoughts on further developments in the field. Background and Concepts The use of nature as a model for the emergence and development of industrial ecosystems is often a basic and coordinating precept in the study of industrial ecosystems. This precept initially required identifying various analogies, which are now slowly developing into an explicit body of theory. In their chapter, “Beyond a sack of resources: nature as a model – the core feature of industrial ecology”, Ralf Isenmann, Christoph Bey and Martina Keitsch described how industrial ecology views nature as a model for the development of industrial ecosystems and how this feature of industrial ecology creates a unique identity for the field. The chapter identified the attributes that are the important features of this specific identity and examined what makes it unique. To accomplish this task, the authors reviewed the evolution and epistemological base of industrial ecology. Their review illustrated that nature is employed as a model within industrial ecology both implicitly and explicitly, and nature is frequently based on biological analogies between industrial systems and natural ecosystems. The authors also compared industrial ecology with other schools of thought, such as neoclassical economics and biophysical economics in addition to how industrial ecology views and understands nature. They concluded that the industrial ecology perspective indicates a fundamental departure from understanding nature’s contributions to human economies through energy and material flows or waste absorption capacities towards nature as a model for innovation, an entity from which to learn in the move towards sustainability. With their historically-informed exposition and analysis of the state and potential future direction of industrial ecology, Ralf Isenmann and his colleagues provide a valuable context for the application-oriented chapters that follow. Several of those clearly transcend the traditional analogy from which industrial ecology has sprung, exploring what the changes observed at one part of the overall system imply for other system components and for the industrial ecosystem as a whole. Stocks and Flows Dynamics Part II, Stocks and Flows Dynamics, contains two chapters illustrating the use of stocks and flows analysis and economic dynamics when modeling

142

Changing stocks, flows and behaviors in industrial ecosystems

the dynamics of industrial ecosystems. Stocks can be comprised of materials that entered a nation’s economy as engineered commodities, which accumulate over time as “in-use” stock of materials. Each commodity has a useful lifetime during which it provides the desired service. At the end of its useful lifetime, the commodity, in theory, in many cases exits the economy as waste, which typically accumulates in the environment. Information on growth rates of new stocks and retirement rates of old stocks is useful for key stakeholders when analyzing resource use and environmental implications of industrial ecosystems. Modeling tools, such as material and substance stock and flow analysis, provide useful approaches to understanding such stock dynamics. Stocks and flows analysis can either be static or dynamic. The static approach provides a single snapshot of stocks and flows, whereas the dynamic model can be used to characterize the net addition or depletion of stocks over time. In their chapter, “Dynamic modeling of material stocks: a case study of in-use cement stocks in the United States”, Amit Kapur and Gregory Keoleian illustrated the differences between static and dynamic material flow analysis and developed a dynamic substance flow model. To illustrate practical use, the authors also implemented the model with the aim of assessing the accumulation and discards of employed cement stocks from 1900 to 2005 in the United States. Employed cement stocks include in-use stocks, for example a stock that still is in active use, and hibernating stocks, for example stocks that have not been fully discarded. In quantifying employed stocks, the authors categorized material inflows into specific end-use categories with end-use specific service life-times. These lifetimes determine the delay between the material inflow and material outflows in the form of discards, and can be used to calculate material accumulation and de-accumulation. Their model and findings illustrate the importance of knowledge about the age distribution of in-use cement stocks for anyone responsible for the life cycle management of infrastructure. Stocks change in size as a function of economic and social variables. As industrial ecologists become increasingly interested in what influences change, industrial structure, evolution and performance, the links to the external environment become essential. In their chapter, “The economic dynamics of stocks and flows”, Brynhildur Davidsdottir and Matthias Ruth focused on the need to integrate economic dynamics into conventional analysis of stocks and flows. The authors argued that this increases the relevance of the modeling of stocks and flows to the policy community because accumulation and flows never exist in isolation from economic and behavioral parameters. Davidsdottir and Ruth presented in their chapter a selected set of components that are needed when integrating economic

Retrospect and prospect

143

dynamics with stocks and flows analysis. Their approach was illustrated with an applied capital vintage model of the US pulp and paper industry. In their model, each capital vintage is characterized by specific attributes such as rates of depreciation, input efficiencies and output structure. An industrial system evolves as new capital is added to the stock that has different attributes from pre-existing vintages. The authors, through their case study of the US pulp and paper industry, implemented a model that simultaneously captures capital vintage dynamics, technological change, economic variables and physical flow variables. The results of the model illustrate the potential contributions of capital vintage dynamics, capital inertia and path dependency to the dynamics of industrial ecosystems. Agent-based Analysis Davidsdottir and Ruth’s chapter illustrated a top-down model of industrial behavior. The level of aggregation is at the regional industry level, which easily lends itself to policy analysis and enables economic decisions to be linked to physical variables. Agency exists at various levels, however, and to some extent corporate behavior and changes in industrial ecosystems emerge, for example, from interactions between employees and from pressures from consumers. Such interactions are rarely, if at all, accounted for in industry-wide analyses. In many ways, individuals acting as citizens, employees, investors and consumers are the fundamental agents of change in industrial ecosystems. Part III of this volume includes three chapters that present agent-based analysis in the modeling of industrial ecosystems. Clinton Andrews, in his chapter entitled “Changing a firm’s environmental performance from within”, investigated the behavioral and organizational questions associated with the environmental performance of industrial firms. The study presents a bottom-up view of industrial ecosystems through the examination of the intrapersonal dynamics influencing corporate environmental behavior using a multi-agent simulation modeling approach. The chapter is based on a generalized case study of an injection molding factory and its employees. Results show that a more realistic account of corporate environmental behavior, and thus the dynamics of industrial ecosystems, depend on representing employees as agents having bounded rationality, subject to social influences. Employees operate within the formal, regulative structures of the firm and government, as well as in the informal, normative, or cultural structures of social networks that all need to be taken into account. The results also illustrate that changes in the human behavioral underpinnings of industry performance may provide similar benefits to technical change.

144

Changing stocks, flows and behaviors in industrial ecosystems

So far in the volume, the emphasis has mostly been on individual material or substance flows. Frequently obvious but often subtle interdependencies exist among them, and the resulting structure and function of industrial ecosystems. Given energy’s important place in industrial ecology, and in the dynamics of industrial ecosystems, it is important to incorporate eco-industrial research that focuses on complex, interrelated energy and material flows and recycling networks. David Batten and George Grozev in their chapter entitled “Managing energy futures and greenhouse gas emissions with the help of agent-based simulation”, modeled the Australian electricity market and physical network as a complex adaptive system (CAS). Agent-based simulation was used to explore alternative energy futures with respect to energy markets, in particular increased use of distributed generation and greenhouse gas emissions. The authors demonstrated that generators are able to respond strategically to changes in market conditions, and adapting their behavior in profit-driven ways tend to exacerbate air pollution and may cause dramatic price fluctuations. Clearly the electricity market operates as a complex adaptive system where a large number of agents interact and react to various stimuli, without a single agent in control. Furthermore, behavior is characterized by repeated games and learning. As a result, the authors argued that it may be appropriate to use agent-based simulations in order to capture many of those complexities and to ask “what if ” questions regarding potential future system performance. In an effort to answer such questions with an agentbased model of the Australian Electricity Market, Batten and Grozev illustrated that increased use of distributed generation requires new market structures, new clusters of small generation units with recyclable feedstocks, new grid systems, new participants (for example aggregators) and new rules. In the third chapter in Part III of this volume, “Organizational dynamics in industrial ecosystems: insights from organizational theory”, Jennifer Howard-Grenville and Raymond Paquin demonstrated how theories of organizational behavior and institutional change inform understanding of the organizational dynamics that are necessary for the development and evolution of industrial ecosystems, in particular the development of symbiotic relationships. The authors illustrated with the use of three related organizational behavioral theories: institutional theory, field theory and social networks theory. In addition, at least two critical changes must be made by a company in order to establish and maintain such relationships. First, a company must develop new or expanded relationships with other organizations within a broader organizational field. This change in behavior may involve working collaboratively with competitors or developing new ties with non-traditional public or private partners. Second, the

Retrospect and prospect

145

company may need to alter its own and others’ broadly accepted norms of business practice within a field. This may involve, for example, developing new forms of contracts, or divulging competitive information to others while remaining a trusted and legitimate member of the field. The authors concluded that in order to understand the emergence and evolution of industrial ecosystems, we must have a holistic understanding of the social systems in which individuals and organizations act. With their argument, they bring us full circle back to the opening chapter of this book, where we have pointed to the need for a comprehensive understanding of production and consumption processes in a broader environmental and social context.

PROSPECT Although tremendous strength exists in carrying out static analysis of industrial ecosystems for comparative purposes, the very nature of sustainability issues requires an understanding of the time varying behavior of industrial, socioeconomic, and environmental systems in their interrelationship. This volume presents some recent efforts to gain such understanding, and as a collection of works at the forefront dealing with the dynamics of industrial ecosystems, suggests at least the following three avenues for further developments. First, as valuable as analogies are to enhance understanding of industrial systems as part of larger social and ecological systems, to date, the choice of analogies has largely been limited to those from mid-20th century biology and ecology. Significantly less attention has been paid to recent advancements in ecosystem theory, especially where it concerns complexity theory, non-equilibrium thermodynamics and related bodies of knowledge. Even more important for the development of industrial ecology, though, may be a move beyond analogies to actual applications of concepts and tools from those areas of research. A few first steps in that direction are presented in this volume and more are likely to come as the community of researchers expands. Proliferation of methods and tools across disciplinary boundaries is encouraged and supported as the first insights from those first steps bear fruit in industry, economy and society. Second, as the field of industrial ecology matures, methods are finetuned and their usefulness for analysis, as well as investment and policy support, are demonstrated through a growing number of case studies. One trend already apparent in this volume, and likely increasing over the next few years, is the confluence of methods and tools. For example, life cycle analyses are linked to dynamic simulation models, agent-based models become increasingly used in conjunction with organizational theory, and

146

Changing stocks, flows and behaviors in industrial ecosystems

economics and econometrics make it into dynamic models of mass and energy flow within and across industry boundaries. In addition dynamic computer simulations of large industrial systems are opened up to allow for, and facilitate, dialog with stakeholders, learning, experimental data generation and direct investment and policy support. Third, it is increasingly recognized that sustainable solutions to engineering challenges require that economic, environmental and social constraints are not just acknowledged but explicitly dealt with as part of the industrial ecologist’s repertoire. Integrating engineering insights with models and tools from the environmental and social sciences in itself is an intellectual challenge for theory building and modeling. Using those models in investment and policymaking requires attention to the combined limitations of each constituent part, error propagation when they are assembled, and effective communication of remaining uncertainties as they are communicated to decision makers. Several of the studies presented in this volume are not only cognizant of the need for integration and sensitive to its challenges, but make efforts to move the field forward by attempting to do what has been called for elsewhere. There clearly is no panacea for the social, economic and environmental challenges associated with industrial activities at the large scales required to meet globally growing human needs and wants. There are, however, smart and creative approaches to better understand and manage individual aspects of these challenges and clear avenues to pursue to reduce pressures at all levels of system organization – from local and regional to global. The diversity of studies assembled in this volume should provide hope that development of flourishing industrial ecosystems is possible over the long haul and a lot can be accomplished between now and then.

Index accumulation 43, 46, 49–50, 54–5, 142 agency/agencies 3, 4, 82–4, 91, 97, 125–7, 129, 140, 143 Environmental Protection Agency 98 International Energy Agency 114 agent-based analysis 77, 79, 143 agent-based modeling 6, 77–80, 86, 134 analogy/analogies 5, 10–11, 16, 22, 40, 140–41, 145 anthropocentrism 17–18 anthroposphere 41 application(s) 5–8, 11, 13–15, 17, 23, 43, 47, 77–8, 80, 108, 141, 145 architecture 11–15, 23, 24 automation 86–7, 90, 96–7 balance 16, 36, 45, 58, 68, 79, 114 bidding 101, 104–6, 108, 112–13 bioeconomics 17, 21 biomass 40, 104 bounded rationality 87–8, 143 brokering 123, 135–6 building(s) 41–7, 49–51, 58, 85, 97, 102 building (verb) 128 theory building 10, 11, 15, 22, 24, 146 byproduct(s) 16, 67, 123 capacity 19–20, 56, 58–60, 65–7, 85, 92, 104–5, 111, 113, 116–18, 130 capital 54, 58–63, 66–70, 79, 96, 124, 128–30, 134–5, 143 capital inertia 58, 64, 69–70, 143 capital investment 58–60, 63, 85 capital stock 57–70 capital vintage 56, 58–9, 61–7, 69–72, 143 cellular automata 77 cement 40–47, 49–51 centrality 131–2, 134–5 coal 40, 102–6, 115 collaboration 123, 127, 131, 133

collective action 78–80 community 4, 8, 11, 15, 22–3, 55, 79, 97, 126, 131, 145 industrial ecology 54 MFA 35 policy 33, 142 scientific 7–9, 15 complex 10–11, 19, 77, 80, 83, 85, 91, 97, 101, 103, 105–9, 111, 116, 119, 134, 140, 144; see also complex adaptive system; complexity complex adaptive system 101, 107, 144 complexity 4–5, 35, 85, 96, 106, 111, 145; see also complex computer modeling 4 computer(s) 4, 41, 146 concentration(s) 41, 111, 114 concrete 17, 41, 43–5, 50 connectedness 131 constraint(s) 4, 16, 20, 33–5, 82, 85, 107, 109, 113, 122, 140, 146 construction 14, 42–7, 50–51 consumer(s) 3–4, 6, 10, 40, 49, 51, 79, 82, 85, 107, 116, 143 consumerism 40 consumption 3, 12, 34, 38, 40, 43–6, 49–51, 103, 115, 127, 145 consumption process(es) 3, 145 context 3, 5, 9, 12–14, 22–3, 54–5, 82–3, 85, 115, 125, 135, 141, 145 copper 37–8, 41 co-production 17–18 Cournot equilibrium 108 demand 37, 41, 51, 56, 58–9, 61, 67, 104–10, 112, 114, 116–19, 122, 127, 129 density 42, 47, 131–2 discards 36, 42–3, 47, 49–51, 142 disorder 12 distributed generation 103, 115, 119, 144

147

148

Index

distribution 41, 43, 46–51, 78, 89, 101, 103, 106, 111, 113, 116, 142 diversity 5, 19, 107, 113, 131–3, 146 dynamics 3–6, 16, 20–21, 23, 33, 38, 54–5, 58–9, 61–5, 71, 84, 97, 101, 103, 109–10, 122–5, 130, 135–7, 140–45 dynamism 33, 37

fiber 56–7, 61, 67, 68, 69–70 flow(s) 3–6, 12–13, 15, 20, 33–8, 41–3, 45–7, 50–51, 54–71, 78–9, 101–2, 113, 116–17, 122, 124, 132, 137, 141–4, 146; see also waste flow food web(s) 10 fuel 57, 61, 67–8, 115–16 fossil fuel 34–6, 57, 67, 115

economics 9, 13–14, 16, 17, 19–22, 24, 41, 51, 55, 58, 65–6, 71, 78, 79–80, 83–4, 141, 146 ecological economics 8, 10–11, 17, 24 environmental economics 16–17 resource economics 8, 17, 19, 24 spaceship economics 17, 19–20 see also bioeconomics economy 3, 12–13, 16, 40–41, 50–51, 62, 102, 107, 142, 145 ecosystem(s) 3–6, 7, 9–11, 15–16, 19–22, 35–6, 40, 54, 55, 77–80, 82, 97, 122–7, 129–30, 132–7, 140–46 effectiveness 40, 135 effectiveness of policy/policies 55 efficiency 55, 60, 62, 64, 66, 70, 95–6, 113, 115–16, 119 economic efficiency 101 energy efficiency 60, 115–16 input efficiency 60, 63–6 embeddedness 133 employee aptitude 89 energy 3–6, 16, 19–21, 33–7, 55–67, 68, 69–71, 79, 83, 101–3, 106–7, 112–16, 122, 127, 140–41, 144, 146; see also efficiency, energy efficiency entropy 102 environment 3–6, 8, 12, 14–22, 24–5, 33–6, 38, 40–41, 43–4, 55, 61, 63–4, 69, 70–71, 82–5, 87–90, 92, 101–3, 109, 111, 113, 116, 122, 140, 142 environmental load(s) 55, 63, 71 environmental quality 6 epistemology 10–12, 24 ethics 10, 17–19, 24–5

grain 40 greenhouse gase(s) 3–4, 54, 68, 101–3, 106–7, 109, 112–14, 144 gypsum 44, 79

feedback 4, 69, 71, 83, 106, 109, 140 feedstock 57, 103, 144

happiness 89, 94–5 health 40, 54, 88 ecosystem health 54 health impact(s) 3 human health 3, 54 public health 4 hedged 106, 113 hedging 106, 109 history 23, 77, 133 industrial ecologist 14, 16, 20, 23, 37, 41, 82, 87, 142, 146 industrial ecology 3–5, 7–17, 20–24, 33, 37, 40, 51, 54, 78, 80, 95, 101–2, 122–5, 127–8, 130, 135, 137, 141, 144–5 industrial ecosystem 3–5, 54–5, 77–80, 82, 97, 122–7, 129–30, 132, 134–7, 140–46 industry 3–6, 7, 9, 36, 59, 61, 62, 65, 69–71, 85, 91, 126–7, 140, 143, 145–6 automobile industry 61 chemical industry 126 cement industry 51 electricity industry 102 injection molding industry 92 polymer processing industry 92 pulp and paper industry 56–8, 61, 64, 66, 143 information 3–5, 13, 35, 43, 51, 54–6, 65, 69, 77–8, 83, 88, 90, 101, 104–7, 110, 112–13, 127, 129, 131–6, 142, 145 infrastructure(s) 41–4, 46–7, 49–51, 58, 101, 103, 111, 113, 116, 142

Index

149

innovation 20–21, 62, 78–80, 141 institution(s) 3–6, 7–8, 82, 86, 110–11, 122–5, 127–9, 134–7, 144 institutional theory 122–6, 136, 144 inventory 41, 67

population 42, 59, 77–9, 87, 90, 92–3, 95 producer(s) 1, 3–4, 10, 79 psychology 4, 83 public policy 37; see also policy

law(s) of thermodynamics 61, 83, 102, 145 leakage 59–60 learning-by-doing 64 life cycle analysis 13, 15, 124 lignin 56–7 limestone 43

realism 88 regulatory reform 101 research 3, 5, 6, 8–15, 23, 25, 35, 40, 50–51, 60, 77, 79, 83–5, 87, 97, 101, 104–5, 110, 112, 123, 128, 130, 134–7, 144–5 resource(s) 3, 7–9, 12, 16–20, 24–5, 37–8, 40–41, 43, 79, 85–6, 95–7, 101–3, 113, 128–9, 130, 133–4, 137, 141–2 Revolution Agricultural 5 Industrial 5

market power 102, 105, 109–11, 116 material composition 42 material flow analysis 12–13, 15, 142 material(s) 3–6, 12–13, 15, 19–21, 33, 35–8, 40–46, 51, 54–63, 64–7, 69–71, 79, 85, 87, 89, 102, 122–5, 127, 130, 133–4, 137, 140–42, 144 metaphor 10–12, 16, 22, 122–4, 130 model 5–6, 7–8, 11, 14–24, 42–3, 45, 49–51, 59, 62–71, 77–80, 82, 84–93, 95–7, 103, 107–11, 114–15, 117–18, 126, 134, 136, 140–46; see also computer modeling; simulation motion 4, 33, 84 natural gas 61, 115 nature 3, 6, 7–8, 11–23, 24–5, 40, 59, 83, 106, 133, 135–6, 141, 145 network network analysis 124, 129–30, 134–6 network dynamics 84 network integration 116 network structure 130, 136 oil 42, 61 order 12, 25, 82, 104 organizational behavior 83–4, 144 paradigm shift 21 path dependency 58, 63–4, 66, 68, 143 policy 33–7, 55, 63, 65, 69–71, 80, 84, 91, 93, 97, 102, 110, 142–3, 145–6 energy policy 102 environmental policy 12, 63 see also community policy

scale 4, 10, 36–8, 54, 57–8, 82, 85, 111, 115, 122, 146 scholarship 5, 140 science(s) 7, 10, 12–15, 22–3, 34, 51, 98, 101, 119 biophysical science(s) 20 engineering science(s) 16, 20, 22, 35 National Science Foundation 51 natural science(s) 16, 24–5 social science(s) 77–8, 80, 87, 122–4, 146 service(s) 6, 13, 18–19, 25, 40–42, 47–50, 54, 103, 106, 111–12, 116, 131, 142 simulation 49, 69, 78–80, 82, 84–6, 89, 91, 93, 95–7, 101, 103, 107–12, 114–15, 118–19, 143–6; see also computer modeling; model society 3–4, 7, 11–12, 24, 46, 51, 95, 124, 140, 145 American Society of Civil Engineers 43 stasis 33 steel 40–41, 64, 85 stock(s) 3, 6, 20–21, 34–8, 40–51, 55–70, 141–4; see also feedstock structure 9, 12, 44, 50, 55, 57, 58–9, 61, 64–7, 69, 77–8, 82–4, 87, 109, 125, 130–31, 133, 135–6, 140, 142–4

150

Index

substitution 58, 61–2, 65, 68–9 supply 40, 42, 82, 104–5, 108–9, 112, 116–17, 130 supply chain 130 sustainability 7, 10, 15, 22, 34–8, 40, 50, 55, 70, 141, 145 system(s) 3–5, 9–11, 13, 15–16, 20, 38, 40–45, 50, 55–60, 62, 66–9, 78–80, 83–4, 96–7, 101, 103–4, 106–9, 110–17, 119, 122–6, 134, 137, 140–42, 144–6 industrial 7, 11, 15–16, 20–21, 24, 37, 43, 51, 54–6, 59, 60–62, 64–6, 70, 123, 140, 143, 145–6 technical change 143; see also technological change technological change 58, 62–5, 68–70, 85, 97, 143 theory 6, 9–11, 15, 17–18, 22, 24, 60, 83–5, 88, 105, 108, 122–6, 130, 136, 141, 143–6 economic theory 14

tool(s) 5, 7–9, 12, 14, 22–3, 70, 80, 96–7, 101, 106, 114, 119, 122–4, 129, 142, 145 toolbox 12, 13, 15 transmission 101, 103, 106, 111, 113, 116, 118 value(s), moral 7, 9, 11, 14, 60, 88–9, 96–7, 102, 115, 126, 128–9, 134–6 value judgment(s) 9, 14 value(s), numerical 41–2, 47, 49, 91, 93, 117 volatility 105–7, 111, 116 waste flow 56, 66–7, 78–9 waste(s) 10, 12–13, 19–20, 37, 40–44, 46–8, 54, 56–7, 66–70, 90, 102–3, 137, 140–42 wastepaper 47, 67–9 wastewater 46 water 3, 19, 36, 42–4, 46–8, 56, 79, 101–2, 119, 123 weight 36, 41–2, 44, 58, 89–90, 95 weighted 68, 89–90, 92