Micro-Pollutant Regulation in the River Rhine: Cooperation in a Common-Pool Resource Problem Setting 303036769X, 9783030367695


132 59 11MB

English Pages [354]

Report DMCA / Copyright

DOWNLOAD PDF FILE

Table of contents :
Preface
Acknowledgments
About the Book
Contents
About the Author
Abbreviations
Chapter 1: The Compelling Nature of Water Pollution as a Common-Pool Resource Problem
1.1 The CPR and the Policy Problem of Micro-pollutants in Surface Water
1.2 Defining the Key Terms: Management Process, Collaboration, and Cooperation
1.3 Theoretical Explanations for Cooperation and Methodological Approach
1.4 Societal Relevance of the Research
1.5 Structure of the Book
References
Primary Literature
Secondary Literature
Chapter 2: Theories and Theoretical Contribution
2.1 Theories of Noncooperation
2.1.1 The Prisoner’s Dilemma: A Non-cooperative, Nonzero-Sum Game
2.1.2 The Tragedy of the Commons
2.1.3 The Theory of Collective Inaction
2.2 Theoretical Concepts on Collective Action and Common-Pool Resources
2.2.1 A Theoretical Concept of Cooperation’s Core Relationships
2.2.2 Preconditions for Collective Action in CPR Settings
2.2.3 The IAD Framework: An Analytical Tool to Assess Collective Action
2.3 Factors Supporting Cooperation in CPR Settings
2.3.1 Recognizing the Environmental Problem
2.3.2 The Ecology of Games Framework (EGF): How Forums Matter
2.3.3 The Advocacy Coalition Framework (ACF): Actors’ Shared Beliefs
2.3.4 Theoretical Relevance of the Research
References
Primary Literature
Secondary Literature
Chapter 3: Methods and Cases
3.1 A Public Policy Analysis
3.1.1 The Policy-Making Process
3.1.2 The Policy Problem
3.1.3 The Solutions to a Policy Problem
3.1.4 Policy Networks
3.2 The Conceptual Framework Guiding the Research
3.2.1 The Social-Ecological System Framework (SESF) and Its Critique
3.2.2 Conceptualizing the Variables
3.2.2.1 The Dependent Variable Cooperation
3.2.2.2 The Independent Variables
3.2.2.3 The Control Variables
3.3 Case Study Design and Case Studies
3.3.1 The Case Study Design: A Mixed-Method Study
3.3.2 The Case Studies and Their Selection Criteria
3.3.2.1 The Three Case Studies
3.3.2.2 The SESF’s Characteristics Informing the Case Studies
3.3.3 The Micro-pollutant Management Process at the Rhine Basin at Basel
3.3.3.1 The Different Water Uses
3.3.3.2 The Laws and Instruments
Water Policy Regarding Micro-pollutants at the National Level
Water Policy Regarding Micro-pollutants in the Cantons Basel City and Basel Country
3.3.4 The Micro-pollutant Management Process in the Ruhr Region
3.3.4.1 The Different Water Uses
3.3.4.2 The Laws and Instruments
The European Union Water Framework Directive (WFD)
Water Policy Regarding Micro-pollutants in the Federal Republic of Germany
Water Policy Regarding Micro-pollutants in the Ruhr Region
3.3.5 The Micro-pollutant Management Process in the Moselle Basin
3.3.5.1 The Different Water Uses
…in Luxembourg
…in Germany
3.3.5.2 The Laws and Instruments
Water Policy Regarding Micro-pollutants in Rhineland-Palatinate and Saarland
Water Policy Regarding Micro-pollutants in Luxembourg
Water Policy Coordination at the Level of the IKSMS
3.3.6 Similarities and Differences Between the Case Studies
3.4 The Data Collection Process
3.4.1 Document Analysis
3.4.2 Actor Identification
3.4.2.1 Based on the SESF
3.4.2.2 Through the Reputational Approach
3.4.3 Expert Interviews
3.4.4 The Survey
3.4.4.1 Collecting Data on the Dependent Variable
3.4.4.2 Collecting Data on the Independent Variables
3.4.4.3 Collecting Data on the Control Variables
3.4.5 Response Rates and Handling Missing Data
3.4.5.1 The Respondents
3.4.5.2 The Non-respondents
Basel Case Study
Ruhr Case Study
Moselle Case Study
3.4.5.3 Handling Missing Data
3.5 The Data Analysis Methods
3.5.1 Descriptive Social Network Analysis
3.5.1.1 A Graph, an Edge, and Vertices
Walks, Trails, and Paths
Components
Factions
Density
Degree Centrality
Core-Periphery
3.5.2 Exponential Random Graph Models (ERGM)
3.5.3 The Qualitative Analysis: A Case Comparison
References
Sources
Primary Literature
Secondary Literature
Chapter 4: Empirical Analysis I: On Cooperation
4.1 The Constituting Elements of Cooperation
4.1.1 Aiming Towards the Same Goal
4.1.2 Coordinating Each Other’s Actions: Actors’ Collaboration
4.1.3 Exchanging Resources
4.1.4 Relating the Three Elements
4.2 A Network Perspective on Collaboration, the Core of Cooperation
4.2.1 The Macro-level: Reciprocity, Fragmentation, and Components
4.2.2 The Meso-level: Factions
4.2.3 The Micro-level: Core, Important, and Peripheral Actors
4.2.3.1 The Collaboration Networks’ Cores
4.2.3.2 The Collaboration Networks’ Most Important Actors
4.2.3.3 The Collaboration Networks’ Most Peripheral Actors
4.3 Actors’ Perceptions of Cooperation
4.3.1 … in the Basel Case Study
4.3.2 … in the Ruhr Case Study
4.3.3 … in the Moselle Case Study
4.4 Qualitative Comparison I: Cooperation at Different Stages
References
Sources
Primary Literature
Secondary Literature
Chapter 5: Empirical Analysis II: On the Emergence of Cooperation
5.1 The Exponential Random Graph Model
5.2 Problem Perception and Cooperation
5.2.1 Problem Perception as Initial Trigger: The Moselle Case Study
5.2.1.1 Uncertainties Regarding the Management of Micro-pollutants in the Moselle Basin
5.2.2 Knowing the CPR Problem: The Ruhr Case Study
5.2.3 Mastering the CPR Problem: The Basel Case Study
5.2.4 Actors’ Similar Viewpoint on the CPR Problem
5.3 Actors’ Participation in Forums and Cooperation
5.4 Actors’ Shared Beliefs and Cooperation
5.5 Actors’ Attributes and Cooperation
References
Sources
Primary Literature
Secondary Literature
Chapter 6: Empirical Analysis III: On the Consolidation of Cooperation
6.1 The Forums for Water Quality Management in the Rhine Basin
6.1.1 The Diversity of Forums Actors Attend
6.1.2 The Key Forums for Micro-pollutant Management in the Rhine Basin
6.2 Forums’ Bridging and Bonding Capital
6.2.1 … in the Basel Case
6.2.2 … in the Ruhr Case
6.2.3 … in the Moselle Case
6.3 How Forums Reinforce Cooperation in CPR Problem Situations
6.4 Qualitative Comparison II: The Factor Time and Triggers for Cooperation
6.4.1 Cooperation in an Evolving CPR Management Process
6.4.2 Cooperation in an Established CPR Management Process
6.4.3 Controlling Factors and Cooperation in CPR Situations
6.4.4 Contextual Factors and Cooperation in CPR Situations
References
Primary Literature
Secondary Literature
Chapter 7: Conclusion I: Theoretical Insights on Cooperation in CPR Settings
7.1 Factors Contributing to Cooperation in a CPR Problem Setting
7.2 Explanatory Strength of the Results
7.2.1 Construct Validity
7.2.2 Internal Validity
7.2.3 External Validity
7.2.4 Reliability
7.3 Discussing the Methods
7.4 Contributions to Theory
References
Chapter 8: Conclusion II: Insights for Practitioners
8.1 Actors in the Center of Attention
8.2 Cooperation Across Borders
8.3 Forums as Transmitters of Knowledge and Trust
Annex
Annex I: The Prisoner’s Dilemma (Table 1)
Annex II: The Core Relationships (Fig. 1)
Annex III: The Social-Ecological System Framework’s Second-Tier Characteristics (Table 2)
Annex IV: Expert Interviews (Table 3)
Annex V: Actor Lists (Tables 4, 5 and 6)
Annex VI: Share of Actor Types Across the Case Studies (Table 7, 8 and 9)
Annex VII: Questionnaire (Fig. 2)
Annex VIII: Data Preparation
DV: Networks of Collaboration and Information Exchange and Similar Goal Index
IV 1a: Problem Perception
IV 1b: Similar Problem Perception
IV 2a: Participation in Forums
IV 2b: Co-participation in Forums
IV 3: Same Belief
CV 1 and 2: Actor Has Regulatory Power and Actor Is an Implementer
CV 3: Reputation
CV 4: Pollution-Sensitive Water Use
CV 5: Territoriality
Annex IX: Lists of Non-respondents (Tables 10, 11 and 12)
Annex X: Network Statistics (Tables 13, 14, 15, 16, 17, 18, 19, 20 and 21; Figs. 3, 4, 5 and 6)
Annex XI: ERGM Results—Models 1–5 (Tables 22, 23, 24, 25 and 26; Figs. 7, 8 and 9)
Annex XII: The Model Robustness Checks and Its Goodness-of-Fit
Annex XII.1: The Model Robustness Checks
Annex XII.2: The ERGMs’ Goodness-of-Fit
Annex XIII: Lists of Forums (Table 27, 28 and 29)
References
Recommend Papers

Micro-Pollutant Regulation in the River Rhine: Cooperation in a Common-Pool Resource Problem Setting
 303036769X, 9783030367695

  • 0 0 0
  • Like this paper and download? You can publish your own PDF file online for free in a few minutes! Sign Up
File loading please wait...
Citation preview

Laura Mae Jacqueline Herzog

Micro-Pollutant Regulation in the River Rhine Cooperation in a Common-Pool Resource Problem Setting

Micro-Pollutant Regulation in the River Rhine

Laura Mae Jacqueline Herzog

Micro-Pollutant Regulation in the River Rhine Cooperation in a Common-Pool Resource Problem Setting

Laura Mae Jacqueline Herzog Institute of Environmental Systems Research Osnabrück University Osnabrück, Germany

Dissertation, University of Bern, 2018 Inauguraldissertation zur Erlangung der Würde eines Doctor rerum oeconomicarum (resp. rerum socialium) der Wirtschafts- und Sozialwissenschaftlichen Fakultät der Universität Bern. Die Fakultät hat diese Arbeit am 08.11.2018 auf Antrag der beiden Gutachter Prof`in Dr. Karin Ingold und Dr. Dimitris Christopoulos als Dissertation angenommen, ohne damit zu den darin angesprochenen Auffassungen Stellung nehmen zu wollen. ISBN 978-3-030-36769-5    ISBN 978-3-030-36770-1 (eBook) https://doi.org/10.1007/978-3-030-36770-1 © Springer Nature Switzerland AG 2020 This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors, and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Cover illustration: Photo by Feteme Fuentes. This Springer imprint is published by the registered company Springer Nature Switzerland AG. The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

It’s easy to throw something into the river, but hard to get it out again. Indian proverb Surface water is like an opened heart. Everyone has access to it.1 Interview N° 9

 Original wording: Das Oberflächengewässer ist wie ein offenes Herz. Alle haben Zugang dazu.

1

To my sister, Lola

Preface

In a world with a growing population that demands more food and more water, the question of how we use and protect the vital resource of water is fundamental. In a world that is further facing the threat of an increase in temperature, which will increase the pressure of water stress and water scarcity around the world, the question of how we use water sustainably is crucial. I conducted the research that led to this book because I am interested in how people use water as a resource and in how we can detect water contamination, what measures exist to prevent this contamination at its source, and what solutions we have once the pollution is in the world, i.e., in water. Moreover, I am interested in how humans interact with each other when they face a water pollution problem. It is the people who influence nature, often in a negative way. So it is the people who have to work out ways together that reduce this negative influence and turn it into a careful and sustained handling with nature and its resources. This book therefore focuses on the interactions of collective actors who are affected by water pollution and are supposed to deal with this environmental problem, in this case, the contamination of surface water with micro-pollutants. Micro-pollutants are a relevant issue as we humans are constantly and increasingly producing them through our use of chemicals in our daily lives and the manufacturing processes of the goods we consume. The chemical substances emerging from these actions are released to the environment, ending up in the water cycle. One of the most inspiring aspects of my investigation of actor cooperation in water quality management was the commitment of the people I spoke to. They had all made water the focus of their work; they all took care of its good condition. I want this book to be an inspiration as well for graduates and Ph.D. students trying to find their way in the middle of their research, for practitioners of water management looking for examples of effective water quality management, for

ix

x

Preface

scholars of common-pool resources wanting to gain insights into collective action in a large-scale common-pool resource setting, for researchers applying Social Network Analysis and looking for an exemplary application, and for anyone interested in managing the world’s most valuable resource, water. Osnabrück, Germany September 2019

Laura Mae Jacqueline Herzog

Acknowledgments

I want to thank my sister, without whose conviction I would not have started this endeavor. I want to thank my parents for their unwavering faith in my abilities. I want to thank Simon, who has always been supportive, understanding, and patient with me during the time of conducting this research. I want to thank my supervisor, Karin Ingold, for welcoming me to her research group with open arms. Thinking her way into my research idea, leaving me freedom to develop my thoughts, guiding me, and supporting me all the way through this task have been incredibly enriching and encouraging. I am deeply thankful for her constant motivation, her trust in me, her valuable comments on my manuscript, and our shared passion for having fun. I want to thank my supervisor, Christophe Sohn, for always supporting me, even from afar, and for his helpful and insightful comments on my work. I want to thank my supervisor, Dimitris Christopoulos, who has brought in yet another perspective on how to look at things and whose opinions and thoughts about my research I highly appreciate. I want to thank Andreas Moser, Christian Stamm, Hans-Peter Bader, and Ruth Scheidegger who have introduced me to the world of micro-pollutants and mass flow analysis and Ursina Roffler who has been an outstanding help for the data transcription. I want to thank my dear colleagues who over the course of the years became dear friends: Lenzi, Lorenz, Florence, Anik, Mario, Claudi, Ruth, Anna, and Marlene. They made my workplace! I thank them for their ideas, comments, and critiques on my dissertation and for all the fun moments they shared with me. A special thank you goes to Lenzi, my personal method queen. I would never have made it without her support. I want to thank my uncle, Hermann Ehninger, for his passionate engagement with my topic and his valuable insights about the world of chemicals in general and the chemical industry in Basel with all the downsides to it in particular. I want to thank my friends, Vroni, Chrissi, Marit, Sarah, Nina, Lisa, Hannes, Farisa, Rachael, Sophia, and Anaïs, who have physically not been around me much in the last years but who supported me emotionally. xi

xii

Acknowledgments

I want to thank my friends, Ale and Hannes, for being in my life. They are the lights that shine. I want to thank Andrea Thode for her outstanding work on making this book understandable and readable. I want to thank Bettina Engels and Kristina Dietz who have educated me to head toward a Ph.D. I want to thank the PEGO and the CrossWater Group for being this bundle of exceptional unique people, all dedicated to overlaps, fish perspectives, loads and concentrations, transitions, perceptions, indices and simulations, and environmental problems. I want to thank the Luxembourg National Research Fund (FNR) and the Swiss National Science Foundation (SNF) for having made this research possible. I want to thank Margaret Deignan who has made the process of publishing my work such a smooth and encouraging experience. Finally, I deeply want to thank all the experts who have found the time to listen to and answer my questions and all the individuals who took their time to respond to my questionnaire. This book would not exist without their commitments.

About the Book

This book analyzes cooperation among actors involved in the management of a specific common-pool resource problem: the pollution of the resource surface water. It enhances theories on cooperation and collective action in common-pool resource settings that until now have mainly focused on the maintenance and success of cooperation but less so on its emergence. The book asks: Why do actors cooperate in the management of a common-pool resource problem of over-appropriation? Borrowing from studies on the severeness of a resource’s threat and theories of the ecology of games framework (EGF) and the Advocacy Coalition Framework (ACF) to make assumptions about the causes for cooperation, the present study enriches common approaches of CPR theory. To structure the study of the social-­ ecological system that surface water, its appropriators, and their rules-in-use form, the research employs the social-ecological system framework (SESF) and tests its range of applicability. The study conceptualizes the social phenomenon of cooperation as consisting of three different elements: actors’ joint goal, actors’ collaborative interactions, and actors’ exchange of information. Actors collaborative interactions and their exchange of information are each operationalized as actor networks. The research assesses the factors triggering two actors to engage in collaborative interactions through inferential network analysis, applying an exponential random graph model (ERGM) technique. A descriptive social network analysis (SNA) examines the patterns of actors’ collaboration and information exchange networks and assesses the explaining factors more thoroughly. Combined with a qualitative analysis of the management processes’ contexts, these analyses enable the interpretation of the model’s results. The research applies the methodological approach to three cases of surface water pollution management in three different sub-catchment areas of the river Rhine: the Moselle basin, the Ruhr catchment area, and the Rhine at Basel. The book is of interest for scholars of collective action, common-pool resources, water management, and the ecology of games framework, for undergraduate and graduate students keen on gaining a deeper understanding of a sound research design, for students of public policy analysis, for researchers applying Social xiii

xiv

About the Book

Network Analysis, for scientists who apply the social-ecological system framework (SESF) in their research, and for practitioners in water quality management. More precisely, the book offers the following insights for the respective interest group: • Scholars working with the theory of the commons or researching on cooperation in the context of environmental problems find an overview of the research field’s evolution in Chap. 2, an applied conceptualization of cooperation in the context of water quality management in Sect. 4.1, and a discussion about new insights to this field of research in Sects. 7.1 and 7.4. • Undergraduate and graduate students find useful insights on the superstructure and the content of a neat research design, on the features of a mixed-method case study design, and on the process of case study selection and the procedure of gathering their own data in Chap. 3 and an overview of a research’s evaluation criteria in Sect. 7.2. • Scholars and students working with Social Network Analysis find an introduction to this analytical method in Sect. 3.5. and its exemplary application in Sects. 4.2 and 5.1. • Section 3.1 offers an overview of the key features of public policy analysis. • Scholars applying the social-ecological system framework (SESF) get an insight on the framework’s contribution to case study selection and actor identification in case study research in Sects. 3.2, 3.3, and 3.4. • Scholars researching on the threat of environmental problems and actors’ perception thereof find an exemplary study of actors’ perception of an environmental problem and its implication for collective action in Sect. 5.2. • Scholars of the ecology of games framework (EGF) are invited to follow the identification of forums’ bridging and bonding capital in the context of water quality management in Chap. 6. • Scholars working on water management and water laws find their share in the description of the European, the Swiss, the German, and the Luxembourgian water legislation in the case study descriptions in Sect. 3.3 and in the overview of stakeholders’ perception of water quality management in the Rhine basin in Sect. 4.3. • Practitioners of water management are invited to read Chaps. 5 and 6 that inform on the specific contexts of water quality management in three regions of the Rhine basin and the factors enhancing actor cooperation therein and the insights for practitioners in Chap. 8. The book finds that depending on the current stage of the actors’ management process and actors’ cooperation therein, cooperation is triggered by different factors. At the beginning of actors’ cooperation within a CPR management process, the actors’ perceptions of the problem matter. In the ongoing of the management process, the actors’ participation in the so-called forums enhances actors’ activity in cooperation. Once cooperation in a CPR management process is established and mature, the actors’ co-participation in these forums ensures the cooperation’s consolidation.

About the Book

xv

Apart from these “evolutionary” dynamics of cooperation, the study shows that the actors’ affiliation to (a) the group of policy implementers and to (b) the same territoriality matters as well. Point (a) reveals the mutual dependency of those realizing the solutions to the CPR problem at stake, while point (b) is of concern in cross-border CPR management situations where actors from different jurisdictions have to work together. A further aspect shown to have influence on cooperation in a CPR problem situation is a supra-national legal guideline that explicitly demands the management of the specific CPR problem. For the present case, this legal guideline is the European Union’s Water Framework Directive (2000/60/EC) that requires the EU member states to ameliorate the ecological and chemical status of the European water bodies. Aside from cognitive factors at the early beginning of resource management, it is mainly institutions that enable the long-lasting establishment of cooperation in CPR problem settings in the Rhine basin. The research project was part of the interdisciplinary research project “CR21I1L_146336 CrossWater—Transboundary Micropollution Regulation in Europe: The Definition of Appropriate Management Scales—An Interdisciplinary Approach” that was funded by the Swiss National Science Foundation (SNF) and the Luxembourg National Research Fund (FNR), 2014–2018.

Contents

1 The Compelling Nature of Water Pollution as a Common-Pool Resource Problem.................................................................................... 1 1.1 The CPR and the Policy Problem of Micro-pollutants in Surface Water................................................................................ 4 1.2 Defining the Key Terms: Management Process, Collaboration, and Cooperation........................................................ 7 1.3 Theoretical Explanations for Cooperation and Methodological Approach.......................................................... 9 1.4 Societal Relevance of the Research.................................................. 12 1.5 Structure of the Book........................................................................ 13 References.................................................................................................. 14 2 Theories and Theoretical Contribution................................................. 19 2.1 Theories of Noncooperation............................................................. 20 2.1.1 The Prisoner’s Dilemma: A Non-cooperative, Nonzero-­Sum Game.............................................................. 21 2.1.2 The Tragedy of the Commons.............................................. 25 2.1.3 The Theory of Collective Inaction........................................ 26 2.2 Theoretical Concepts on Collective Action and Common-­Pool Resources........................................................... 28 2.2.1 A Theoretical Concept of Cooperation’s Core Relationships......................................................................... 29 2.2.2 Preconditions for Collective Action in CPR Settings........... 30 2.2.3 The IAD Framework: An Analytical Tool to Assess Collective Action................................................... 32 2.3 Factors Supporting Cooperation in CPR Settings............................. 33 2.3.1 Recognizing the Environmental Problem............................. 34 2.3.2 The Ecology of Games Framework (EGF): How Forums Matter.............................................................. 37

xvii

xviii

Contents



2.3.3 The Advocacy Coalition Framework (ACF): Actors’ Shared Beliefs.......................................................... 39 2.3.4 Theoretical Relevance of the Research................................. 41 References.................................................................................................. 42 3 Methods and Cases.................................................................................. 47 3.1 A Public Policy Analysis.................................................................. 48 3.1.1 The Policy-Making Process.................................................. 48 3.1.2 The Policy Problem............................................................... 49 3.1.3 The Solutions to a Policy Problem........................................ 52 3.1.4 Policy Networks.................................................................... 54 3.2 The Conceptual Framework Guiding the Research.......................... 55 3.2.1 The Social-Ecological System Framework (SESF) and Its Critique...................................................................... 56 3.2.2 Conceptualizing the Variables............................................... 59 3.3 Case Study Design and Case Studies................................................ 64 3.3.1 The Case Study Design: A Mixed-Method Study................. 65 3.3.2 The Case Studies and Their Selection Criteria..................... 68 3.3.3 The Micro-pollutant Management Process at the Rhine Basin at Basel................................................... 76 3.3.4 The Micro-pollutant Management Process in the Ruhr Region................................................................ 81 3.3.5 The Micro-pollutant Management Process in the Moselle Basin.............................................................. 87 3.3.6 Similarities and Differences Between the Case Studies....... 93 3.4 The Data Collection Process............................................................. 93 3.4.1 Document Analysis............................................................... 97 3.4.2 Actor Identification............................................................... 98 3.4.3 Expert Interviews.................................................................. 100 3.4.4 The Survey............................................................................ 103 3.4.5 Response Rates and Handling Missing Data........................ 108 3.5 The Data Analysis Methods.............................................................. 114 3.5.1 Descriptive Social Network Analysis................................... 114 3.5.2 Exponential Random Graph Models (ERGM)..................... 119 3.5.3 The Qualitative Analysis: A Case Comparison..................... 121 References.................................................................................................. 122 4 Empirical Analysis I: On Cooperation................................................... 133 4.1 The Constituting Elements of Cooperation....................................... 134 4.1.1 Aiming Towards the Same Goal........................................... 135 4.1.2 Coordinating Each Other’s Actions: Actors’ Collaboration............................................................ 138 4.1.3 Exchanging Resources.......................................................... 140 4.1.4 Relating the Three Elements................................................. 144

Contents

xix



4.2 A Network Perspective on Collaboration, the Core of Cooperation.................................................................................. 146 4.2.1 The Macro-level: Reciprocity, Fragmentation, and Components.................................................................... 147 4.2.2 The Meso-level: Factions...................................................... 151 4.2.3 The Micro-level: Core, Important, and Peripheral Actors............................................................ 153 4.3 Actors’ Perceptions of Cooperation.................................................. 163 4.3.1 … in the Basel Case Study.................................................... 163 4.3.2 … in the Ruhr Case Study.................................................... 166 4.3.3 … in the Moselle Case Study................................................ 170 4.4 Qualitative Comparison I: Cooperation at Different Stages............. 173 References.................................................................................................. 177 5 Empirical Analysis II: On the Emergence of Cooperation.................. 179 5.1 The Exponential Random Graph Model........................................... 180 5.2 Problem Perception and Cooperation............................................... 183 5.2.1 Problem Perception as Initial Trigger: The Moselle Case Study....................................................... 185 5.2.2 Knowing the CPR Problem: The Ruhr Case Study.............. 191 5.2.3 Mastering the CPR Problem: The Basel Case Study............ 195 5.2.4 Actors’ Similar Viewpoint on the CPR Problem.................. 198 5.3 Actors’ Participation in Forums and Cooperation............................ 199 5.4 Actors’ Shared Beliefs and Cooperation........................................... 201 5.5 Actors’ Attributes and Cooperation.................................................. 206 References.................................................................................................. 209 6 Empirical Analysis III: On the Consolidation of Cooperation............ 213 6.1 The Forums for Water Quality Management in the Rhine Basin...... 214 6.1.1 The Diversity of Forums Actors Attend................................ 215 6.1.2 The Key Forums for Micro-pollutant Management in the Rhine Basin........................................... 220 6.2 Forums’ Bridging and Bonding Capital............................................ 221 6.2.1 … in the Basel Case.............................................................. 224 6.2.2 … in the Ruhr Case............................................................... 227 6.2.3 … in the Moselle Case.......................................................... 229 6.3 How Forums Reinforce Cooperation in CPR Problem Situations............................................................... 232 6.4 Qualitative Comparison II: The Factor Time and Triggers for Cooperation............................................................ 236 6.4.1 Cooperation in an Evolving CPR Management Process....... 236 6.4.2 Cooperation in an Established CPR Management Process............................................................ 237 6.4.3 Controlling Factors and Cooperation in CPR Situations...... 238 6.4.4 Contextual Factors and Cooperation in CPR Situations....... 240 References.................................................................................................. 243

xx

Contents

7 Conclusion I: Theoretical Insights on Cooperation in CPR Settings........................................................................................ 245 7.1 Factors Contributing to Cooperation in a CPR Problem Setting...... 246 7.2 Explanatory Strength of the Results................................................. 248 7.2.1 Construct Validity................................................................. 248 7.2.2 Internal Validity..................................................................... 249 7.2.3 External Validity................................................................... 250 7.2.4 Reliability.............................................................................. 251 7.3 Discussing the Methods.................................................................... 252 7.4 Contributions to Theory.................................................................... 253 References.................................................................................................. 256 8 Conclusion II: Insights for Practitioners............................................... 259 8.1 Actors in the Center of Attention...................................................... 259 8.2 Cooperation Across Borders............................................................. 260 8.3 Forums as Transmitters of Knowledge and Trust............................. 260 Annex................................................................................................................ 263

About the Author

Laura Mae Jacqueline Herzog is a postdoctoral fellow at the Chair of Resource Management chaired by Prof. Dr. Claudia Pahl-Wostl at the Institute of Environmental Systems Research, University of Osnabrück. She is interested in the different uses and the sustainable management of common-pool resources (CPR), in particular the common-pool resources water and land. In her research, she investigates the interactions of humans with water and land resources, the influences of these interactions on the respective ecological system, and the potential conflicts of interest emerging from such. She is interested in how such influences on nature can be shaped sustainably and how resource conflicts can be moderated and resolved. Taking into account the respective institutional framework, i.e., the laws and regulations that influence water and land management and the state of the respective human-influenced ecosystem, she pursues an interdisciplinary approach, looking at the social-ecological systems she examines from a socioscientific as well as a natural scientific perspective. She currently works as research collaborator in the BiodivERsA and Belmont Forum research project LimnoScenES that investigates the short- and long-term pressures due to climate change and intensifying land and water use on north temperate lakes. The project develops scenarios of future human-freshwater interactions that enable the maintenance of lakes’ biodiversity and ecosystem services. Laura M. J. Herzog received her doctorate from the University of Bern, where she worked as a research assistant at the Chair of Policy Analysis and Environmental Governance chaired by Prof. Dr. Karin Ingold at the Institute of Political Science. This book is her doctoral thesis. Laura M. J. Herzog studied political science at the Otto-Suhr-Institut of the Freie Universität Berlin with a research focus on political ecology, environmental conflicts, and the concepts of human security and Responsibility to Protect (R2P) in postwar settings. In her diploma thesis, she analyzed the local mobilization of social movements in two Peruvian mining regions using social movement theory. She has published in the Policy Studies Journal, Ecology and Society, and Global Environmental Change.

xxi

Abbreviations

A Actor ACF Advocacy Coalition Framework Art. Article AUE BL Amt für Umweltschutz und Energie, Basel Landschaft/Cantonal Office for Environmental Protection and Energy, Basel Country AUE BS Amt für Umwelt und Energie, Basel Stadt/Cantonal Office for the Environment and Energy, Basel City BDEW Bundesverband der Energie- und Wasserwirtschaft/Federation of the Energy and Water Industry BMBF Bundesministerium für Bildung und Forschung/Federal Ministry of Education and Research BMEL Bundesministerium für Ernährung und Landwirtschaft/Federal Ministry of Food and Agriculture BMG Bundesministerium für Gesundheit/Federal Ministry of Health BMJV Bundesministerium der Justiz und für Verbraucherschutz/Federal Ministry of Justice and Consumer Protection BMUB Bundesministerium für Umwelt, Naturschutz und nukleare Sicherheit/Federal Ministry for Environment, Nature Conservation and Nuclear Safety CEO Chief executive officer/Geschäftsführer∗in CPR Common-pool resource CV Control variable/s EGF Ecology of games framework EQS Environmental quality standards EU European Union ERGM Exponential random graph model DV Dependent variable DVGW Deutscher Verein des Gas- und Wasserfaches/German Technical and Scientific Association for Gas and Water DWP Drinking water plant

xxiii

xxiv

Eawag

Abbreviations

Eidgenössische Anstalt für Wasserversorgung, Abwasserreinigung und Gewässerschutz/Swiss Federal Institute of Aquatic Science and Technology FGG Flussgebietsgemeinschaft Rhein/River Basin Commission Rhine FOEN Federal Office for the Environment/Bundesamt für Umwelt (BAFU) FRG Federal Republic of Germany FSVO Federal Food Safety and Veterinary Office/Bundesamt für Lebensmittelsicherheit und Veterinärwesen (BLV) GOF Goodness of fit I Interactions IAD Institutional Analysis and Development Framework ICPR International Commission for the Protection of the Rhine/ Internationale Kommission für den Schutz des Rheins (IKSR) IKSMS Internationale Kommissionen zum Schutz der Mosel und der Saar/ International Commissions for the Protection of the Moselle and the Saar LAWA Bund/Länder-Arbeitsgemeinschaft Wasser/German Working Group on Water Issues of the Federal States and the Federal Government LWG Landeswassergesetz/Federal Water Act IV Independent variable/s GS Governance system MCMC MLE Markov Chain Monte Carlo for Maximum Likelihood Estimation MKULNV Ministerium für Klimaschutz, Umwelt, Landwirtschaft, Natur- und Verbraucherschutz des Landes NRW/Ministry for Climate Protection, Environment, Agriculture, Nature Conservation, and Consumer Protection of NRW MUEEF Ministerium für Umwelt, Energie, Ernährung und Forsten Rheinland-Pfalz/Ministry of the Environment, Energy, Food and Forestry of Rhineland-Palatinate n Number NGO Nongovernmental organization NRW North Rhine-Westphalia O Outcomes OGewV Oberflächengewässerverordnung/Surface Water Ordinance par. Paragraph PFAS Fluorosurfactants/Perfluorierte Tenside PFOS Perfluorooctanesulfonic acid/Perfluoroktansulfonsäure RLP Rhineland-Palatinate RS Resource system RU Resource unit SECO Staatssekretariat für Wirtschaft/State Secretariat for Economic Affairs SES Social-ecological system

Abbreviations

SESF SNA SWG UBA UoA WFD WHG WPA WPO

xxv

Social-ecological system framework Social Network Analysis Saarländisches Wassergesetz/ Water Management Law of the Federal State of Saarland Umweltbundesamt/German Federal Environmental Agency Unit of Analysis European Water Framework Directive (2000/60/EC) Act on Hydrologic Balance/Wasserhaushaltsgesetz Federal Act on the Protection of Waters/Bundesgesetz über den Schutz der Gewässer (GSchG) (SR 814.20) Waters Protection Ordinance/ Gewässerschutzverordnung (GSchV) (SR 814.201)

Chapter 1

The Compelling Nature of Water Pollution as a Common-Pool Resource Problem

Abstract  The first chapter conceptualizes the environmental problem of water pollution as a common-pool resource (CPR) problem and describes why water pollution by micro-pollutants can be considered a wicked policy problem. The chapter further highlights the necessity to overcome the CPR problem of water pollution through actor cooperation and defines the study’s key terms cooperation, collaboration, and management of a CPR problem. The chapter closes with a focus on the case study area—the Rhine basin—and the research’s societal relevance. Keywords  Common-pool resources (CPR) · Collective action · Environmental problems · CPR problems · Micro-pollutants · Rhine basin The focus of this book is on cooperation in a common-pool resource (CPR) problem setting. The study the book is based upon assesses the factors that enhance cooperation among actors who are related to a common-pool resource problem and belong to a, what I call, management process of the CPR problem. A common-pool resource (CPR)—also known as a common—is a natural or humanly created system that gives benefits to its users. The features distinguishing a natural or artificially created resource from mere “nature,” making it a common-­ pool resource, are its social and regulative components: the communities using it and the rules established by these communities with the intention to regulate the handling of the resource (Helfrich et  al. 2010, p.  11). A common consists of a resource stock made out of resource units. The resource units can be overused if the use is not regulated (Ostrom et al. 1994, p. 8). We can find common-pool resources almost everywhere in daily life: the drinking water coming from the nearby river and the groundwater, the public car park where we can park our cars, the forest next to your city with its abundance of species, the lift offered by a person with free space in the car, to name just a few. Commons improve humans’ quality of life, like a lawn in a public park suitable for picnicking. Commons can also be crucial for humans’ lives by providing essential goods. For instance, rivers supply us with water, and rain forests contribute to the global oxygen turnover. The challenge about common-pool resources is the risk that they are overharvested. This is due to their criteria of “exclusion” (Ostrom 1990, p.  32; Ostrom 2000a, p.  148) and “subtractability” (Ostrom 2000a, p.  148; Ostrom et  al. 1994, © Springer Nature Switzerland AG 2020 L. M. J. Herzog, Micro-Pollutant Regulation in the River Rhine, https://doi.org/10.1007/978-3-030-36770-1_1

1

2

1  The Compelling Nature of Water Pollution as a Common-Pool Resource Problem

Table 1.1  The four types of goods (Ostrom 2005, p. 24) Low subtractability High subtractability

Easy exclusion Club goods: Private parks, pay TV Private goods: Food, clothing, cars

Difficult exclusion Public goods: Free TV, national defense Commons: Fish stocks, timber, coal

p. 4, 6). The first refers to the relatively high cost of excluding people from using a common-pool resource; the second indicates the rivalry of the users and the steady and irreversible subtraction of resource units from the resource stock. Common-­ pool resources share the characteristic of difficult exclusion of its users with public goods and the attribute of high subtractability with private goods. Table 1.1 summarizes the four types of goods. Organizing the use of a common involves a dilemma: one person’s use of a CPR is always subtracting a unit from the resource stock, thereby diminishing the CPR’s quantity or quality available for other users. Additionally, the number of users of a CPR cannot be limited, increasing the subtraction of the CPR’s units even more. As long as appropriators of a common keep their consumption pattern, everyone uses the resource freely (Ostrom 2000a, 148).1 This process can reach the point of the resource’s degradation or even exhaustion. In the case of the CPR surface water, such a scenario reads as follows: surface water users take such large amounts of water out of a river2 that other users are left with less water for their own use or without any water at all. The resource has then become a non-excludable, rival CPR. The same holds true when appropriators pollute a resource. Users who pollute surface water reduce the quality of its single resource units and diminish the overall quality of the resource stock. This is another means of reducing the quantity of a CPR, because polluters reduce the quantity of resource units in good condition. A look at today’s environmental problems reveals that many common-pool resources are overused. Rivers are polluted, rainforests are being cleared, the oceans are being overfished, farmland is acidified, micro-plastic deposits are found in soil and oceans, and fossil fuels are being burnt, which releases carbon dioxide in the atmosphere causing climate change (Aktar et al. 2009; Keenan et al. 2015; NASA 2019; Neubauer 30.04.2018; oxfordreference 2018; Plumer 29.10.2013; Schaefer 2015). Many of these commons are essential for humankind. The question of how a CPR problem of over-appropriation can be solved is thus of utmost social and political relevance (Fleishman 2013, p. 34). CPR theory stresses the necessity to coordinate action to achieve joint benefits in a CPR setting (Ostrom 1990, p. 29; Ostrom 1998, p. 3, 6ff) and to overcome a CPR 1  Public goods do not encounter the problem of over-appropriation: access to public goods is available to all (criterion of “difficult exclusion”), but as public goods cannot be diminished in quantity or quality (criterion of “low subtractability”), its use by a large number of appropriators does it no harm; see Hardin (1982, p. 17). 2  For example, for building a dam or irrigating fields.

1  The Compelling Nature of Water Pollution as a Common-Pool Resource Problem

3

problem. A CPR problem is a social dilemma, in which actors take choices independently even though they are interdependent on each other’s action (Ostrom 1998, p. 3), i.e., affect each other with their actions. Collective action is one way out of social dilemmas (ibid., p. 6ff). Collective action is defined as the working together of a number of people who want to achieve a common objective by doing so (Dowding 2018). Collective action is thus synonymous with cooperation, which I define as the coordination of actions and the exchange of resources among two or more actors in order to achieve the same objective, which is beneficial to at least one of the cooperating actors.3 In her dissertation, Rachel Fleishman (2013) analyzes the benefits arising from cooperation. From a broad survey of collaboration literature, she deduces eight key benefits of cooperation (Fleishman 2013, p. 34): 1 . Access to resources and resource exchange. 2. Improved organizational status or legitimacy. 3. Efficiency and productivity gains. 4. Greater flexibility and adaptability. 5. Innovation, knowledge generation, and learning. 6. Achieving coordinated action. 7. Building social capital and social infrastructure. 8. Reducing conflicts. To solve environmental problems, cooperation among actors is needed (Berardo et al. 2015, p. 443). Environmental problems these days can be complex. They may have different causes and affect humans and nature at different levels—regional, national, and even international—and in different intensities (Allen 2013). Actors contributing to these problems may come from different sectors, levels (cf. Ingold and Fischer 2014, p. 88), and jurisdictions (Berardo et al. 2015). Often, a diversity of actors is involved in its solution. Technical, political, and financial resources needed to tackle these problems are fragmented. Actors therefore need to share information and coordinate their actions across levels (Ingold and Fischer 2014, p. 88f.), specifically because (…) stakeholders may have relatively homogeneous views about how a resource should be managed, but lack the knowledge or information needed to discern how others behave or feel about this topic. This (…) hinders their ability to find coordinated responses to common problems. This type of problem is particularly acute in large [social-ecological systems] (…), where resources are often distributed across extensive geographic areas that fall within multiple political jurisdictions. (Berardo et al. 2015, p. 443)

The assumption is that the development and implementation of solutions to environmental problems become easier when actors work together. Cooperation can be interpreted as a prerequisite to overcome social dilemmas and thus CPR problems. The question is how cooperation comes about in a CPR problem setting. Research on cooperation has overlooked this question, focusing mainly on characteristics of

 In the following, I use the terms collective action and cooperation interchangeably.

3

4

1  The Compelling Nature of Water Pollution as a Common-Pool Resource Problem

cooperation and the factors making cooperation successful (Heikkila and Gerlak 2005, p. 584). I thus pose the following research question: Why do actors cooperate in the management of a common-pool resource problem of over-appropriation?

1.1  T  he CPR and the Policy Problem of Micro-pollutants in Surface Water To answer my research question, I focus on the common-pool resource problem of micro-pollutants in the surface water of the river Rhine. My objects of analysis are the actors involved in the management process of this CPR problem and their cooperation pattern. Micro-pollutants are chemical substances that appear in surface water and have become pertinent in the last years. They acquired their name due to their extremely low concentration in the μg/l and ng/1 range in aquatic environments (Kümmerer 2009, p. 2354f.).4 Micro-pollutants are difficult to detect in water bodies (Touraud et al. 2011, p. 437). Until recently, they could not be identified in water resources (Meyer et al. 2011, p. 128), which makes them a rather new issue in environmental politics. Human activities release them to the environment: production processes in industry; pesticide application in agriculture; use of herbicides as plant protectants on facades, roofs, and streets; or the use of pharmaceuticals and detergents with residues entering the water cycle through wastewater treatment plants (ibid., p. 128). The negative impact of micro-pollutants on humans is neither fully understood yet nor proven (Touraud et al. 2011, p. 439). Although some studies state that micro-­ pollutants pose a negligible risk to humans (Cunningham et al. 2009, p. 44; Kümmerer 2009, p. 2360), scientific concern exists that they are carcinogenic and may increase physiological changes in animals (Touraud et  al. 2011, p.  437). Regarding their effects on fauna, studies have proven the negative impacts of micro-­pollutants on ecosystems (Kümmerer 2009, p. 2360; Pal et al. 2010, p. 6063; Touraud et al. 2011, p. 439). There is also the probability that a mixture of micro-­pollutants, each harmless in its particular concentration, can create a joint toxicity (Kümmerer 2009, p.  2359; Touraud et  al. 2011, p.  439). Micro-pollutants involve a certain risk to humans and nature (see Metz and Ingold 2014, p. 1994). Until now, only advanced water treatment plants are able to filter these compounds; but even filters can fail on this task (Jones et al. 2005, p. 164; Rivera-Utrilla et al. 2013, p. 1270). The sources of micro-pollutants are diverse, and their entry paths are difficult to locate. Micro-pollutants enter water bodies, like a river system, through point or

 1  g/L  =  1000  mg/L (milligram per liter); 1  mg/L  =  1000  μg/L (microgram per liter); 1 μg/L = 1000 ng/L (nanogram per liter); see endmemo.com (2017). 1 μg is a millionth of a gram; 1 ng is a billionth of a gram. 1 ng/l equals 1 kg in a water body the size of Lake Biel in Switzerland; see Gälli et al. (2009, p. 33). 4

1.1  The CPR and the Policy Problem of Micro-pollutants in Surface Water

5

diffuse entry paths. Point sources are geographically fixed, like the outlet of a wastewater treatment plant. They are rather easy to determine. In the instance of a diffuse source, micro-pollutants infiltrate the water cycle through a wide area, like the run-­ off of agricultural fields or road surfaces (Lapworth et  al. 2012, p.  289). Micro-­ pollutants in water can spread over long distances, thereby crossing borders and affecting further riparian actors. Micro-pollutants are a typical policy problem. Following Metz and Ingold’s (2014) definition, I define policy problems through their causation, their prevalence, their effects, and their scales. Causation refers to the problem’s causes that can stem from factors or actors. The more diverse causes a problem has, the more complicated its solution becomes—the policy instruments tackling the problem would need to address them all. Prevalence theorizes the frequency at which the problem appears and the amount of factors or actors contributing to it. Effects describe the problem’s negative impacts on society and the environment. Scales represent the levels at which the problem’s effects occur. Instruments addressing the problem and its effects have to consider these levels (cf. Metz and Ingold 2014, p. 1998). Micro-­ pollutants have many different causes, and their occurrence depends on seasonality and weather events—regarding application periods in agriculture and intensified runoff of substances from fields as well as the increased dilution of substances within water in times of heavy rains or inundations. The effects of micro-pollutants are, to a large extent, uncertain, with studies hinting at their negative impact on aquatic systems, invertebrates, and second consumers. When pertinent and bio-­ accumulating, micro-pollutants further “hit” at different levels, starting at the regional level where they are released leading up to the international level when polluted surface water enters territory of another jurisdiction. Micro-pollutants in cross-border, public water bodies are therefore a trans-boundary policy problem in the policy field of environmental politics. They are further a wicked problem in that they are multicausal, continuously evolve, do not fit into one responsible sector, do not have one clear solution, involve various stakeholders with varying perceptions of the problem, and require a behavioral change (Allen 2013, p. 101f.). These aspects make micro-pollutants a highly complex environmental and common-­pool resource problem. They constitute a water quality problem as they intrude into surface and groundwater and diminish water quality. In CPR theory, the phenomenon of resource pollution can be interpreted in two ways: First, a water body containing micro-pollutants can be framed as a CPR problem of over-­ appropriation. Over-appropriation problems concern the allocation of the flow of a CPR, i.e., the resource unit users want to have access to (Ostrom 1990, p. 47). They imply overexploitation of the common by its users, the probability of negative externalities, and strong rivalry between users (Villamayor-Tomas et  al. 2014, p.  364). This means that when a user increases the appropriation, the average return the other users receive from their investments in appropriation is reduced—an externality is created (Ostrom et al. 1994, p. 10f.). In the case of pollution, only the polluters enjoy the benefits of emitting pollutant substances into a common. The benefits from polluting the resource are thus private (cf. Villamayor-Tomas et al. 2014, p. 364). The users do not gain any benefits: they suffer from the reduced quality of the CPR and

6

1  The Compelling Nature of Water Pollution as a Common-Pool Resource Problem

the limited availability of the CPR in good condition. All actors share the costs of pollution (ibid., p. 364). The criterion of subtractability applies. The pollution of a resource creates a conflict between different user groups: the appropriators who use the resource to pollute it and benefit from doing so and the appropriators who pay for the polluters’ attitude as they receive the resource in reduced quality. Second, the problem of a polluted CPR can be interpreted as a provision problem. CPR problems of provision concern the common-pool resource stock, i.e., the resource system in its entirety (Ostrom 1990, p. 47; Ostrom et al. 1994, p. 9). A CPR provision problem refers to behavioral incentives for users to change their consumption pattern in order to give the resource the capacity to recover and reproduce itself. This kind of provision is called demand-side provision. CPR provision problems can also relate to the supply-side of provision. This type of provision problem refers to appropriators’ motivations to contribute to the maintenance of a CPR (Ostrom et al. 1994, p. 12). In contrast to the over-appropriation problem, here users share the benefits of provision, while the costs of providing a good are private (Villamayor-Tomas et al. 2014, p. 364). Letting less micro-pollutants into surface water—avoiding the destruction of a common (Ostrom et al. 1994, p. 9)—renders micro-pollutants a supply-side provision problem. Incentives for changing polluters’ habits relate to micropollutants in terms of a demand-side provision problem. Simply releasing micro-pollutants into surface water creates an over-appropriation problem. Micropollutants are thus a typical and topical example of a common-pool resource problem. Both types of CPR problem are interrelated. As Ostrom et al. (1994, p. 15) state, “the nature of the appropriation problem is affected by how well the provision problem is solved.” The present book considers both CPR problems: the emergence of a user conflict due to over-appropriation through the emission of micro-pollutants and the social processes that (can) reinstall the CPR in good quality, guaranteeing its well-functioning supply-side provision. From a political scientific viewpoint, the question arises how such a wicked problem can be solved. One way is collaborative governance, where those affected by, those responsible for, and those causing a societal problem jointly govern it (Ansell and Gash 2008, p. 544; Emerson et al. 2012, p. 2). The responsibility to regulate the problem is split among the various stakeholders concerned by it. Having actors from different backgrounds participate in the development of and decision on possible solutions enables sharing of knowledge, resources, and specific expertise. These are aspects needed to work out adequate solutions and thus facilitate the problem management process. Resource users who coordinate their actions and behave cooperatively have shown to be more successful in managing their CPR than those who do not show such behavior (Chhatre and Agrawal 2008; Prediger et al. 2011). Cooperation is a societal mechanism that makes governance of societal and environmental problems easier in that it provides the grounds to bridge sectors, levels, and jurisdictions affected by the problems and to share capacities like knowledge, finances, and time. However, Koontz and Thomas (2006, p. 118) state that more research is needed that assesses the link between actors’ collaboration and the environmental outcomes this collaboration achieves. Figure 1.1, the schematic graph of the book’s argument, visualizes this argumentation and takes up the political scientific perspective that cooperation facilitates a

1.2  Defining the Key Terms: Management Process, Collaboration, and Cooperation

Starting point CPR-problem of overappropriation: micro-pollutants in surface water several actors involved

Analysis of … … factors enhancing cooperation among actors participating in the management of the CPR-problem

7

Assumption CPR provision from the supply-side: cooperation facilitates successful management of the problem the provision of the CPR

Fig. 1.1  Schematic graph of book’s argument

successful management of environmental and CPR problems. The first square indicates the starting point of the analysis: the CPR problem of water pollution through micro-pollutants and the different actors involved in and affected by this pollution.5 The second square shows the analysis by which I answer the research question. I assess the factors that influence the emergence of cooperation between actors facing a CPR problem of over-appropriation and participating in its management process. The last square displays the assumption that cooperation facilitates the provision of a common in good quality, which solves the problem of over-appropriation. In the following, I define the main terms of the analysis, the management process of a CPR and actors’ collaboration as well as cooperation.

1.2  D  efining the Key Terms: Management Process, Collaboration, and Cooperation By management process of a CPR problem of over-appropriation, I understand the problem-solving process of corporate actors involved in the CPR problem situation. I rely on Pahl-Wostl’s (2009) definition of resource management to define the management process analyzed in this study. Pahl-Wostl conceptualizes resource management as (…) the activities of analysing and monitoring, developing and implementing measures to keep the state of a resource within desirable bounds. (Pahl-Wostl 2009, p. 355)

I consider actors as participating in the management process if (a) They have common use of the resource. (b) Their daily work deals with the common. (c) They contribute to the problem. (d) They are concerned by it since they handle the issue for administrative reasons.

 These are, among others, user groups, polluters, regulators, and service providers.

5

8

1  The Compelling Nature of Water Pollution as a Common-Pool Resource Problem

The group of actors engaged in the management process of the CPR problem of micro-pollutants is thus broad, comprising common public and private actors like state actors and the industry, user groups, polluters, science, environmental groups, and service providers. Relying on public policy theory, I define the problem-solving process of a CPR problem more precisely as the adoption of guidelines and laws by political bodies or state authorities that aim at the solution to the problem and the implementation of measures that these guidelines and laws produce (Howlett and Giest 2013, p. 17ff; Knill and Tosun 2012, p. 4ff): The management process of a CPR over-appropriation problem is the problem-solving process carried out by corporate actors involved or interested in the CPR problem at stake. The problem-solving process comprises the adoption of political guidelines and laws by responsible institutions and the implementation of measures tackling the CPR problem.

The study focuses on actors’ cooperation within such a management process of a CPR problem. I define cooperation as a “working together” of two or more individual organizations, actor groups, or even enterprises that aim at the same goal (Oxford Living Dictionaries 2018b; Sadoff and Grey 2005, p. 424). Their working together is purposeful. If working together is not intentional, it is collaboration. The difference between cooperation and collaboration is thus the joint aim actors pursue when cooperating and which they do not envisage when collaborating.6 The motivation behind both social phenomena is that actors who unite achieve more benefits than on their own (Huxham 1993, p. 603; O’Leary and Vij 2012, p. 510). The book’s definition of collaboration reads: Collaboration is the working together of two or more actors, thereby producing an outcome the collaborators benefit from.

West et al. (2007) state that cooperation also creates a benefit. They define cooperation as a social behavior and suppose that actors applying this behavior gain a direct benefit. This benefit is high enough to outweigh the costs of performing cooperation (West et al. 2007, p. 416). West et al. (2007) state that this behavior can also provide only an indirect benefit; it would then be only beneficial to the recipient of the action and costly to the actor performing it (Nowak 2006, p. 1560; West et al. 2007, p. 416). Following Sadoff and Grey (2005, p. 424), cooperation among actors exists if the following indicators are given:

6  The Oxford Dictionary defines cooperation as “[t]he action or process of working together to the same end” (Oxford Living Dictionaries (2018b)) and collaboration as “[t]he action of working with someone to produce something” (Oxford Living Dictionaries (2018a)). However, the distinction of the two terms becomes blurry if considering that the produced something—Huxham (1993) calls it collaborative advantage—meets an objective. As Huxham (1993, p.  603) considers: “Collaborative advantage will be achieved when something unusually creative is produced—perhaps an objective is met—that no one organization could have produced on its own and when each organization, through the collaboration, is able to achieve its own objectives better than it could alone. In some cases, it should also be possible to achieve some higher-level ‘meta-objectives’; objectives for society as a whole rather than just for the participating organizations.”

1.3  Theoretical Explanations for Cooperation and Methodological Approach

9

(a) Actors aim at the same goal. (b) Actors jointly regulate and structure their actions; this means they harmonize and coordinate their actions towards the common goal, e.g., by informing one another about planned actions to avoid that they impede each other. (c) Actors exchange resources to achieve their goal, like financial and technical support or knowledge. Information is such a resource. Information can be transformed into knowledge; and actors are keen to acquire knowledge. One can distinguish between political and technical information. Political information informs actors about each others’ strategies and actions and can be a form of power (Leifeld and Schneider 2012, p. 735). Technical information can improve the quality of decision makers’ decisions and increase the legitimacy of these decisions (Leifeld and Schneider 2012, p.  734; Sabatier 1987, p. 650). Technical information also inform about the existence and characteristics of a CPR problem (Henry and Dietz 2011, p. 194). Based on these arguments, I define cooperation as a behavior that is beneficial to another actor (its recipient), while it can be either beneficial or costly to the actor performing it. Actors performing cooperation seek the same goal, exchange resources, and coordinate their actions to reach their common aim.

A fourth element inherent in cooperation is trust. Trust is defined as the risk someone takes who believes in an outcome that depends on others’ actions. Trust is fundamental for cooperation (Araral 2014, p. 14; Giest and Howlett 2014, p. 38; Henry and Dietz 2011, p.  189). A cooperative actor who trusts in other actors’ actions will expect that these others reciprocate his/her actions. If this is the case, the cooperating actors share the costs of the cooperative action. If the actor had no trust in the other actors’ reciprocity, she/he would most likely not start cooperating because the risk of being left alone to settle the costs for the cooperation would seem too high (Henry and Dietz 2011, p. 189). Ostrom stresses the factor trust in her list of variables that increase the likelihood of a group’s self-organization (Ostrom 2000b, p. 40). A certain degree of trust seems necessary for cooperation to form. I assume trust to be an entity that can be observed and analyzed within the context of cooperation. I assess trust among actors by analyzing whether actors reciprocate their actions.

1.3  T  heoretical Explanations for Cooperation and Methodological Approach The cornerstone of this book is the concept of common-pool resources and CPR problems (Ostrom 1990). I draw on several theoretical approaches to answer the research question. I use argumentations from transaction cost theory (Heikkila and Gerlak 2005, p.  585; Taylor and Singleton 1993) and from studies assessing the severeness of an environmental problem (Gerber et  al. 2009; Giest and Howlett

10

1  The Compelling Nature of Water Pollution as a Common-Pool Resource Problem

2014; Lubell et  al. 2002) to check whether actors’ perceptions of the problem’s severeness affect cooperation. I rely on the ecology of games framework (EGF) (Lubell et al. 2012; Lubell et al. 2011; Lubell 2013) to scrutinize the role of forums in the establishment of cooperation in a CPR problem setting. And  I turn to the advocacy coalition framework (ACF) to analyze the effect of actors’ shared beliefs on cooperation (Calanni et al. 2015; Ingold and Fischer 2014; Sabatier et al. 1987; Weible 2005). I derive my empirical case studies of a micro-pollutant management process from the Rhine basin. The Rhine is Europe’s 12th largest river with a length of 1234 km (Kremer 2010, p. 19) and a catchment area of about 185,000 km2 (LANUV 2013). Nine countries are its riparians.7 Figure 1.2 depicts the Rhine catchment area with its nine sub-basins (IKSMS n.d.).8 The Rhine’s waters have been and still are used for a variety of purposes like shipping, electricity generation through hydropower, industrial production processes, or as coolant for power and industrial plants (Kremer 2010, p. 28ff; Ruff et al. 2013, p. 16). More than half of the Rhine basin’s area is used for agriculture (Ruff et al. 2013, p. 17). The river itself is flanked by streets and highways, and a large part of the world’s chemical industry installed itself in its sub-basin Upper Rhine (Kremer 2010, p. 28f.). The Rhine is Europe’s river with the most settlements along its shores (ibid., p.  19, 28f.). The river’s waters are thus under constant pressure: Pollutants from settlements, the industry, and agriculture enter the stream; exhaust fumes from streets and power and industrial plants burden it; thermal pollution and mining activities add to the pressure. Despite the Rhine’s pollution exposure, half of the population living in its catchment area—about 30 million people—receives its drinking water from Rhine surface water (ICPR 2018; Ruff et al. 2013, p. 16). The Rhine catchment area is an appropriate example of an ecological system with a common—surface water—that is used in different ways by various appropriators. The water basin offers regions where the CPR problem of micro-pollutants in surface water occurs and which can serve for case studies. I conduct individual case studies as well as a cross-case comparison for three regions in the Rhine catchment area. Public policy theory provides the concepts with which I determine the elements of the case studies: the policy problem itself (Metz and Ingold 2014); the potential solutions to the problem (Howlett 2005; Vedung 2010); the phases of the policy-making process relevant for the research (Howlett and Giest 2013; Knill and Tosun 2012); and the policy networks that actors who are part of these phases form (Börzel 1998; Kenis and Schneider 1991). To inform my case study selection and structure the case study analysis, I apply the social-ecological system framework (SESF) (McGinnis and Ostrom 2014). Within each case study, I assess whether actors involved in the management process

7  These are Switzerland—where the river originates—,  Austria, Italy, Liechtenstein, Germany, France, Luxembourg, Belgium, and the Netherlands; see Kremer (2010, p. 19, 33). 8  ICPR (2010, p. 3).

1.3  Theoretical Explanations for Cooperation and Methodological Approach

Fig. 1.2  The Rhine catchment area with its nine sub-basins

11

12

1  The Compelling Nature of Water Pollution as a Common-Pool Resource Problem

of the CPR problem micro-pollutants in surface water cooperate with one another. I conceptualize cooperation as the ensemble of actors’ collaboration, their exchange of information, and their sharing of the same goal. I evaluate cooperation and the factors enhancing it through social network analysis (SNA). I conceptualize actors’ coordination of actions as a network of collaboration and actors’ exchange of resources as networks of political and technical information exchange. Applying descriptive SNA, I can reveal the properties of these networks and assess the structure and intensities of actors’ collaboration and information exchange. Comparing the actors’ goals within each case allows me to analyze the actors’ shared objectives. Based on these analytical steps I can determine the states of cooperation in the three case studies. To assess the influence specific factors might have on cooperation, I first apply an inferential network analysis. Based on actor and network characteristics, the analytical technique of the exponential random graph (ERGM) families helps estimate the likelihood for a tie in a network to exist. That means, one can check the conditions under which a tie is likely to exist. In the case of this study, the tie between two actors is their collaboration and their information exchange in the management process of micro-pollutants in Rhine surface water—two of the three elements that I claim define cooperation. Second, I interpret the ERGM results through a descriptive and qualitative study of the explaining factors in each case and a qualitative assessment of their contextual backgrounds. I round off the analysis by comparing each case’s results, which, in turn, gives me insights on the validity of the results of each case study. For the case of the management of micro-pollutants in the Rhine catchment area, the book finds that in young cooperations that have only been around for several years, actors’ perception of the CPR problem triggers their engagement in cooperation. In cooperations that have persisted for a decade or more, it is actors’ joint participation in forums that consolidates the cooperational structure within a CPR management process.

1.4  Societal Relevance of the Research Due to world population growth until the end of this century (Lee 2011), urbanization will increase, leading to massive urban agglomerations with an intensive need for water and sanitation systems (UN 2018; UN Water 2018b). The growing world population also puts pressure on global food production (Carvalho 2017, p. 48; FAO October 2009, p.  2, 8). This process of increased urbanization and agricultural ­production is likely to result in an intensified use and inevitable release of micro-­ pollutants like herbicides, pesticides, and pharmaceuticals into the environment (Carvalho et al. 2014, p. 50f; Owens 19 February 2015) and increases pressure on the resource water (UBA October 2014a; UN Water 2018b). The services of water are essential for humans as well as nature. They include, among others, clean drinking water and water available for hygiene, washing, and

1.5  Structure of the Book

13

cooking; the production of energy and foods like crops and dairy products; and also the maintenance of biodiversity (UN Water 2018a, c). The example of drinking water illustrates the topicality of analyzing a polluted common: drinking water quality and availability is a highly political issue, given that 2 billion people were lacking access to safe drinking water in 2018 and that the demand for water is going to increase by almost one-third by 2050 (cf. WWAP and UN Water 2018, p. v). Drinking water is a prerequisite for a good living standard, and it has to be in a good condition (UN Water 2018d). Micro-pollutants in surface water are a policy problem (cf. Metz and Ingold 2014) of public concern (Friedli 18.06.2017; Liebrich 2./3.09.2017). The different causes for micro-pollutants make the problem specifically complex. Micro-­ pollutants are caused on a regular, constant basis,9 so they are an omnipresent CPR problem. Even though science has not yet fully understood the impact of micro-­ pollutants on humans and the environment (Touraud et al. 2011; UBA December 2014b, p. 3), their effects could become a serious threat in the future (Owens 2015). Moreover, persistent micro-pollutants do not decompose and can bio-accumulate, extending their lifetime in aquatic ecosystems. Released to cross-border water bodies, micro-pollutants affect the aquatic environment at the regional, national, and even international level. Considering the interaction of these characteristics and the amount of polluted commons in the world, the examination of a “polluted common” and the factors leading to its successful management through cooperation is of utmost political relevance.

1.5  Structure of the Book The book consists of eight chapters. The first chapter, the introduction, outlines the scope of the research and the research question, the CPR problem under study, and the theoretical and methodological background of the research as well as its societal relevance. Chapter 2 comprises a literature review of cooperation, based on what I identify the research, and discusses the theoretical strands from which I derive the hypotheses. It further sketches the theoretical relevance of the research. Chapter 3 depicts the methods and the case studies. It starts with a description of the main elements of the public policy process that are part of the research. The chapter then introduces the conceptual framework that structures the analysis and guides the operationalization of the research’s dependent, independent, and control variables: the social-ecological system framework (SESF). The chapter furthermore outlines the case study design, informs about the case study selection criteria, and

9  An exception is agriculture, where there are peaks of pesticide application at certain times of the year.

14

1  The Compelling Nature of Water Pollution as a Common-Pool Resource Problem

presents the three case studies. The chapter finishes with one section on the data collection process and one on the different analytical approaches applied: descriptive Social Network Analysis (SNA); the inferential SNA method of exponential random graph models (ERGM); and the qualitative case comparison approach. Chapter 4 is on cooperation. It examines cooperation within each case study and across them by, first, assessing the constituting elements of cooperation and relating them to each other and, second, analyzing the cases’ collaboration networks descriptively, thereby bringing to light structural parallelisms and differences among them. A section on actors’ interpretation of cooperation in the three management processes compares their statements with the findings of the descriptive Social Network Analysis (SNA). The chapter closes with a comparison of the findings that reveals different intensities and stages of cooperation across the case studies. Chapter 5 presents the ERGM results and discusses the influence the explaining factors have on the emergence of cooperation in the studied CPR problem setting. It does so in light of the case studies’ contextual background. Chapter 6 is on the consolidation of cooperation. It focuses on actors’ forum participation, describing the forums and their functions in more detail, and assesses the patterns of actor-forum-participation through a descriptive network analysis of two-mode networks. The case comparison at the end of the chapter reconciles the theoretical assumptions with the empirical findings and widens the understanding of cooperation’s emergence and consolidation in CPR problem settings. Chapter 7 synthesizes the research findings and discusses the explanatory strength of the results. It critically examines the benefits and limits of the applied methods and concludes by highlighting the contribution to theory on CPR and cooperation. Chapter 8 ends the book with insights for practitioners that tackle water pollution issues in surface water.

References Primary Literature ENDMEMO (2017) online. Available from: http://www.endmemo.com/sconvert/ug_lng_l.php. Accessed 27 Sept 2019 FAO (2009) How to feed the world in 2050. Global agriculture towards 2050. Food and Agriculture Organization of the United Nations (FAO), Rome Gälli R, Schmid-Kleinkemper J, Ort C, Schärer M (2009) Mikroverunreinigungen in den Gewässern. Bewertung und Reduktion der Schadstoffbelastung aus der Siedlungsentwässerung. Bundesamt für Umwelt, Bern ICPR (2010) Our common objective: living waters in the Rhine catchment. Internationally Coordinated Management Plan Part A, Koblenz ICPR (2018) Drinking water [online]. Internationale Kommission zum Schutz des Rheins (IKSR) Available from: https://www.iksr.org/en/topics/uses/drinking-water/?sword_list%5B0%5D=tri nkwasser&cHash=38813ffffb324485d9913685f2864684. Accessed 27 Sept 2019

References

15

IKSMS (n.d.) Das Einzugsgebiet von Mosel und Saar in der Flussgebietseinheit Rhein [online]. Internationale Kommissionen zum Schutze der Mosel und der Saar (IKSMS). Available from: http://www.iksms-cipms.org/servlet/is/20045/. Accessed 25 Sept 2019 LANUV (2013) Gebietsverzeichnis GSK3C [online]. Landesamt für Natur, Umwelt und Verbraucherschutz Nordrhein-Westfalen (LANUV). Available from: https://www.lanuv.nrw. de/fileadmin/lanuv/wasser/pdf/Gebietsverzeichnis%20GSK3C.xls. Accessed 25 Sept 2019 NASA (2019) Global climate change  – vital signs of the planet. Causes [online]. National Aeronautics and Space Administration. Available from: https://climate.nasa.gov/causes/. Accessed 23 Sept 2019 oxfordreference (2018) Tropical rainforest clearance [online]. Available from: http://www.oxfordreference.com/view/10.1093/oi/authority.20110803105837222. Accessed 27 Sept 2019 UBA (2014a) Forschungsprogramm des Umweltbundesamtes 2015–2017. Umweltbundesamt, Dessau-Roβlau, October UBA (2014b) Pharmaceuticals in the environment – the global perspective. Occurrence, effects, and potential cooperative action under SAICM. German Environment Agency, Dessau-Roβlau, December UN (2018) World Urbanization Prospects: The 2018 Revision. key facts [online]. United Nations. Available from: https://population.un.org/wup/Publications/Files/WUP2018-KeyFacts.pdf. Accessed 23 Sept 2019 UN Water (2018a) Water and Ecosystems. Water Facts. Ecosystems [online]. UN Water. Available from: http://www.unwater.org/water-facts/ecosystems/. Accessed 23 Sept 2019 UN Water (2018b) Water Facts. Water and Urbanization [online]. UN Water. Available from: http:// www.unwater.org/water-facts/urbanization/. Accessed 23 Sept 2019 UN Water (2018c) Water facts. water, food and energy [online]. UN Water. Available from: http:// www.unwater.org/water-facts/water-food-and-energy/. Accessed 23 Sept 2019 UN Water (2018d) Water facts. Human Rights. Human Rights to Water and Sanitation [online]. UN Water. Available from: http://www.unwater.org/water-facts/human-rights/. Accessed 23 Sept 2019 WWAP, UN Water (2018) The United Nations world Water development report 2018. Nature-­ Based Solutions for Water. UN Water, Paris

Secondary Literature Aktar MW, Sengupta D, Chowdhury A, Aktar W (2009) Impact of pesticides use in agriculture: their benefits and hazards. Interdiscip Toxicol 2(1):1–12 Allen JH (2013) The wicked problem of chemical policy: opportunities for innovation. J Environ Stud Sci 3(2):101–108 Ansell C, Gash A (2008) Collaborative governance in theory and practice. J Public Adm Res Theory 18(4):543–571 Araral E (2014) Ostrom, Hardin and the commons: a critical appreciation and a revisionist view. Environ Sci Policy 36:11–23 Berardo R, Olivier T, Lavers A (2015) Focusing events and changes in ecologies of policy games: evidence from the Paraná River Delta. Rev Policy Res 32(4):443–464 Börzel TA (1998) Organizing Babylon – on the different conceptions of policy networks. Public Adm Rev 76:253–273 Calanni JC, Siddiki SN, Weible CM, Leach WD (2015) Explaining coordination in collaborative partnerships and clarifying the scope of belief Homophily hypothesis. J Public Adm Res Theory 25(3):901–927 Carvalho FP (2017) Pesticides, environment, and food safety. Food and Energy Security 6(2):48–60 Carvalho RN, Arukwe A, Ait-Aissa S, Bado-Nilles A, Balzamo S, Baun A, Belkin S, Blaha L, Brion F, Conti D, Creusot N, Essig Y, Ferrero VE, Flander-Putrle V, Fürhacker M et al (2014)

16

1  The Compelling Nature of Water Pollution as a Common-Pool Resource Problem

Mixtures of chemical pollutants at European legislation safety concentrations: how safe are they? Toxicol Sci 141(1):218–233 Chhatre A, Agrawal A (2008) Forest commons and local enforcement. Proc Natl Acad Sci 105(36):13286–13291 Cunningham VL, Binks SP, Olson MJ (2009) Human health risk assessment from the presence of human pharmaceuticals in the aquatic environment. Regulatory Toxicology and Pharmacology: RTP 53(1):39–45 Dowding K (2018) Collective action problem [online]. Encyclopaedia Britannica. Available from: https://www.britannica.com/topic/collective-action-problem-1917157. Accessed 27 Sept 2019 Emerson K, Nabatchi T, Balogh S (2012) An integrative framework for collaborative governance. J Public Adm Res Theory 22(1):1–29 Fleishman R (2013) Addressing trans-boundary challenges through collaboration: how organizations “harmonize” actions and decisions across problem landscapes. SURFACE paper 5. Dissertations – ALL Friedli D (2017) Achtung, Pestizide. Die Schweizer Wasserversorger warnen vor steigender Giftbelastung im Grundwasser. NZZ am Sonntag, 18 June, 11 Gerber J-D, Knoepfel P, Nahrath S, Varone F (2009) Institutional resource regimes: towards sustainability through the combination of property-rights theory and policy analysis. Ecol Econ 68:798–809 Giest S, Howlett M (2014) Understanding the pre-conditions of commons governance: the role of network management. Environ Sci Policy 36:37–47 Hardin R (1982) Collective action. A book from resources for the future. The John Hopkins University Press, Baltimore Heikkila T, Gerlak AK (2005) The formation of large-scale collaborative resource management institutions: clarifying the roles of stakeholders, science, and institutions. Policy Stud J 33(4):583–612 Helfrich S, Kuhlen R, Sachs W, Siefkes C (2010) Gemeingüter – Wohlstand durch Teilen. Heinrich-­ Böll-­Stiftung, Berlin Henry AD, Dietz T (2011) Information, networks, and the complexity of trust in commons governance. Int J Commons 5(2):188–212 Howlett M (2005) What is a policy instrument? Tools, mixes, and implementation styles. In: Eliadis P, Hill MM, Howlett M (eds) Designing Government. From instruments to governance. McGill-Queen’s University Press, Montreal, pp 31–50 Howlett, M., Giest, S., 2013. The policy-making process. E.  Araral, Fritzen, S., Howlett, M., Ramesh, M Wu, X., Routledge handbook of public policy. London/New York: Routledge, 17–28 Huxham C (1993) Pursuing collaborative advantage. J Oper Res Soc 44(6):599–611 Ingold K, Fischer M (2014) Drivers of collaboration to mitigate climate change: an illustration of Swiss climate policy over 15 years. Glob Environ Chang 24:88–98 Jones OA, Lester JN, Voulvoulis N (2005) Pharmaceuticals: a threat to drinking water? Trends Biotechnol 23(4):163–167 Keenan RJ, Reams G, Achard F, de Freitas J, Grainger A, Lindquist E (2015) Dynamics of global forest area: results from the FAO global forest resources assessment 2015. For Ecol Manag 352:9–20 Kenis P, Schneider V (1991) Chapter 2: Policy networks and policy analysis: scrutinizing a new analytical toolbox. In: Marin B, Mayntz R (eds) Policy networks. Empirical evidence and theoretical considerations. Campus Verlag/Westview Press, Frankfurt/Main/Boulder, pp 25–59 Knill C, Tosun J (2012) Public policy: a new introduction. Palgrave Macmillan, New York Koontz TM, Thomas CW (2006) What do we know and need to know about the environmental outcomes of collaborative management? Public Adm Rev 66(s1):111–121 Kremer BP (2010) Der Rhein. Von den Alpen bis zur Nordsee. Mercator Verlag, Duisburg Kümmerer K (2009) The presence of pharmaceuticals in the environment due to the human use – present knowledge and future challenges. J Environ Manag 90:2354–2366

References

17

Lapworth DJ, Baran N, Stuart ME, Ward RS (2012) Emerging organic contaminants in groundwater: a review of sources, fate and occurrence. Environ Pollut 163:287–303 Lee R (2011) The outlook for population growth. Science 333(6042):569–573 Leifeld P, Schneider V (2012) Information exchange in policy networks. Am J Polit Sci 56(3):731–744 Liebrich S (2017) Fisch auf Droge. Der steigende Arzneimittelverbrauch belastet die Umwelt. Süddeutsche Zeitung, 2 September, 29 Lubell M (2013) Governing institutional complexity: the ecology of games framework. Policy Stud J 41(3):537–559 Lubell M, Schneider M, Scholz JT, Mete M (2002) Watershed partnerships and the emergence of collective action institutions. Am J Polit Sci 46(1):148–163 Lubell M, Robins G, Wang P (2011) Policy coordination in an ecology of water management games. Southern Illinois University, Carbondale Lubell M, Scholz JT, Berardo R, Robins G (2012) Testing policy theory with statistical models of networks. Policy Stud J 40:351–374 McGinnis MD, Ostrom E (2014) Social-ecological system framework: initial changes and continuing challenges. Ecol Soc 19(2):30 Metz F, Ingold K (2014) Sustainable wastewater management: is it possible to regulate micropollution in the future by learning from the past? A policy analysis. Sustainability 6:1992–2012 Meyer B, Pailler J-Y, Guignard C, Hoffmann L, Krein A (2011) Concentrations of dissolved herbicides and pharmaceuticals in a small river in Luxembourg. Environ Monit Assess 180(1–4):127–146 Neubauer U (2018) Auch unsere Böden sind voller Mikroplastic. Neue Zürcher Zeitung, 30 April Nowak MA (2006) Five rules for the evolution of cooperation. Science 314(5805):1560–1563 O'Leary R, Vij N (2012) Collaborative public management: where have we been and where are we going? Am Rev Public Adm 42(5):507–522 Ostrom E (1990) Governing the commons. The evolution of institutions for collective action. Cambridge University Press, Cambridge Ostrom E (1998) A behavioral approach to the rational choice theory of collective action. Presidential address, American Political Science Association. Am Polit Sci Rev 92(1):1–22 Ostrom E (2000a) Collective action and the evolution of social norms. J Econ Perspect 14(3):137–158 Ostrom E (2000b) The danger of self-evident truths. Political Science and Politics 33(1):33–44 Ostrom E (2005) Understanding institutional diversity. Princeton University Press, Princeton Ostrom E, Gardner R, Walker J (eds) (1994) Rules, games and common pool resources. Michigan University Press, Michigan Owens B (2015) Pharmaceuticals in the environment: a growing problem. Pharmaceuticals in the environment: a growing problem. Pharm J 294(7850) Oxford Living Dictionaries (2018a) British & World English. Collaboration [online]. Available from: https://en.oxforddictionaries.com/definition/collaboration. Accessed 27 Sept 2019 Oxford Living Dictionaries (2018b) British & World English. Cooperation [online]. Available from: https://en.oxforddictionaries.com/definition/cooperation. Accessed 27 Sept 2019 Pahl-Wostl C (2009) A conceptual framework for analysing adaptive capacity and multi-level learning processes in resource governance regimes. Glob Environ Chang 19:354–365 Pal A, Gin Y-HK, Lin Y-CA, Reinhard M (2010) Impacts of emerging organic contaminants on freshwater resources: review of recent occurrences, sources, fates and effects. Sci Total Environ 408:6062–6069 Plumer B (2013) Just how badly are we overfishing the oceans? The Washington Post, 29 October Prediger S, Vollan B, Frölich M (2011) The impact of culture and ecology on cooperation in a common-pool resource experiment. Ecol Econ 70:1599–1608 Rivera-Utrilla J, Sánchez-Polo M, María Ferro-García MÁ, Prados-Joya G (2013) Pharmaceuticals as emerging contaminants and their removal from water. A review. Chemosphere 93:1268–1287

18

1  The Compelling Nature of Water Pollution as a Common-Pool Resource Problem

Ruff M, Singer H, Ruppe S, Mazacek J, Dolf R, Leu C (2013) 20 Jahre Rheinüberwachung. Erfolge und analytische Neuausrichtung in Weil am Rhein. Aqua & Gas (5):16–25 Sabatier PA (1987) Knowledge, policy-oriented learning, and policy change. An advocacy coalition framework. Sci Commun 8(4):649–692 Sabatier PA, Hunter S, McLaughlin S (1987) The devil shift: perceptions and misperceptions of opponents. The Western Political Quarterly 40(3):449–476 Sadoff CW, Grey D (2005) Cooperation on international rivers. Water Int 30(4):420–427 Schaefer A (2015) Mikroplastik in der Umwelt. Infoblatt. oekotoxzentrum, Dübendorf Taylor M, Singleton S (1993) The communal resource: transaction costs and the solution of collective action problems. Polit Soc 21(2):195–214 Touraud E, Roig B, Sumpter JP, Coetsier C (2011) Drug residues and endocrine disruptors in drinking water: risk for humans? Int J Hyg Environ Health 214(6):437–441 Vedung E (2010) Policy instruments: Typologies and theories. In: Bemelmans-Videc M-L, Rist RC, Vedung E (eds) Carrots, sticks & sermons. Policy instruments & their evaluation. Transaction Publishers, New Brunswick/London, pp 21–58 Villamayor-Tomas S, Fleischmann FD, Ibarra IP, Thiel A, van Laerhoven F (2014) From Sandoz to Salmon: conceptualizing resource and institutional dynamics in the Rhine watershed through the SES framework. Int J Commons 8(2):361–395 Weible CM (2005) Beliefs and perceived influence in a natural resource conflict: an advocacy coalition approach to policy networks. Polit Res Q 58(3):461–475 West SA, Griffin AS, Gardner A (2007) Social semantics: altruism, cooperation, mutualism, strong reciprocity and group selection. J Evol Biol 20(2):415–432

Chapter 2

Theories and Theoretical Contribution

Abstract  The chapter discusses different streams of literature that have treated the phenomenon of cooperation. Social action in general and cooperation in particular has been researched on by different disciplines and for several decades. For instance, social capital theory discusses the social norms and actors’ self-interest that shape their actions (Coleman, Am J Sociol 94:S95–S120, 1988); resource dependence theory claims that the social context and interdependencies, understood as power relations and dependencies between organizations, constrain an organization’s behavior and actions (Casciaro and Piskorski, Adm Sci Q 50:167–199, 2005; Pfeffer and Salancik, The external control of organizations: a resource dependence perspective. Stanford University Press, Stanford, 1978); and theory on transaction costs has highlighted the necessity of possessing sufficient resources to meet the costs arising from cooperation (Taylor and Singleton, Polit Soc 21(2):195–214, 1993). In this chapter, I reflect upon the key theoretical concepts of cooperation from sociology and political science that reach into the research domain on common-pool resources. Based on the literature review, I identify a research gap on the topic of cooperation in CPR problem situations. I close this research gap by outlining theoretical explanations for cooperation in the context of CPR problems. I draw these theoretical insights from the ecology of games framework (EGF), the Advocacy Coalition Framework (ACF) and studies on environmental problem perception. Keywords  Cooperation · Collective action · Tragedy of the commons · Institutional Analysis and Development Framework (IAD) · Ecology of games framework (EGF) · Advocacy Coalition Framework (ACF) This chapter discusses different streams of literature that have treated the phenomenon of cooperation. Social action in general and cooperation in particular have been researched on by different disciplines and for several decades. For instance, social capital theory discusses the social norms and actors’ self-interest that shape their actions (Coleman 1988); resource dependence theory claims that the social context and interdependencies, understood as power relations and dependencies between organizations, constrain an organization’s behavior and actions (Casciaro and Piskorski 2005; Pfeffer and Salancik 1978); and theory on transaction costs has

© Springer Nature Switzerland AG 2020 L. M. J. Herzog, Micro-Pollutant Regulation in the River Rhine, https://doi.org/10.1007/978-3-030-36770-1_2

19

20

2  Theories and Theoretical Contribution

highlighted the necessity of possessing sufficient resources to meet the costs arising from cooperation (Taylor and Singleton 1993). In this chapter, I reflect upon the key theoretical concepts of cooperation from sociology and political science that reach into the research domain on common-pool resources. Based on the literature review, I identify a research gap on the topic of cooperation in CPR problem situations. I close this research gap by outlining theoretical explanations for cooperation in the context of CPR problems which, together with the research gap, highlight the theoretical relevance of the study.

2.1  Theories of Noncooperation Game theory treats the aspect of (non-)cooperation among actors by conceptualizing different configurations of situations in which actors’ actions affect the outcome of each other’s actions. A “game” in the logic of game theory is a situation in which the participants each have “(…) at specific times a range of choices of action (…)” (Rapoport and Chammah 1965, p. 13). These choices of action define “(…) an outcome which (…) determines a set of payoffs, one to each participant” (ibid., p. 14). Put simply, players in a game affect each other with their choices of action. A strategy within a game is “a way to play” (Rapoport and Chammah 1965, p. 18), thus the way players choose their actions. Game theory defines different strategies that actors use in accordance with the respective configuration of a game. Game theory distinguishes zero-sum games and nonzero-sum games. In zero-­ sum games, the action of one player always evens out the action of another, i.e., the amount one player is to win from his/her action equals the amount the other player is to lose. The sum of the payoff of both players’ actions equals zero. Actors’ interests in zero-sum games are always opposite (Rapoport and Chammah 1965, p. 14, 248); the situation is thus one of a “pure conflict of interest” (ibid., p.  11). In nonzero-­sum games, the payoff of players’ actions does not come to zero. Here, players’ actions may actually generate a better outcome for all players. Players may therefore favor certain outcomes over others. Players in a nonzero-sum game not only face a confrontation with the other players’ choice of action but also have to settle their own interests when deciding which outcome they would like to achieve (Rapoport and Chammah 1965, p. 11). The theory of nonzero-sum games knows cooperative and non-cooperative games. Their difference lies in the players’ ability to make agreements on which strategy to apply jointly—as is the case in a cooperative game—and players’ inability to do so, which generates a non-cooperative game (Rapoport and Chammah 1965, p.  25). The most prominent paradigm illustrating the latter situation is the prisoner’s dilemma (ibid., p. 11). It explains why humans would tend not to cooperate from a rational choice perspective. The main precondition of non-cooperative games is that the players cannot communicate and are thus incapable of coming to an agreement of how to act in a way that yields a better outcome for all (Ostrom 1998, p. 6).

2.1  Theories of Noncooperation

21

Player A

Table 2.1  The prisoner’s dilemma

Player B

cooperates cooperates defects

defects

3 3

5 0

0 5

1 1

2.1.1  T  he Prisoner’s Dilemma: A Non-cooperative, Nonzero-­Sum Game The prisoner’s dilemma depicts the situation of two players who have the choice between two actions: to cooperate or to defect. They carry out their actions at the same time, and their actions are interdependent. That is, one player’s action affects the other player’s payoff. The underlying assumption is that the players act rationally and cannot communicate (Axelrod 1984, p. 9; Rapoport and Chammah 1965, p. 24f.).1 The players do not know which action to expect from each other and act in their own self-interest. This leads players to prefer acting on their own behalf to cooperating, although they could achieve the “socially optimal outcome” (Poteete et  al. 2010, p. 44) through cooperation. They choose the strategy of defection, which leaves them better off, “(…) no matter what the other player chooses” (Ostrom 1990, p. 5). The situation is depicted in Table 2.1 that visualizes the prisoner’s dilemma and reads as follows: if both players cooperate, they will both be rewarded with 3 points for mutual cooperation; if one player cooperates while the other defects, the one cooperating would receive 0 points while the one betraying—defecting—will receive 5 points. If both defect, both will receive 1 point as punishment for the mutual defection (Axelrod 1984, p. 8f.; Hardin 1982, p. 23f.). The players know the payoffs for the different combinations of their actions, i.e., they have complete information (Ostrom 1990, p. 4; Poteete et al. 2010, p. 45); they just do not know which action the other player will choose because they act simultaneously (Poteete et al. 2010, p. 44). Afraid of not gaining any points for cooperating when the other defects, players tend to choose the “safe” option of gaining at least 1 point for defection instead of cooperating with the risk of not gaining anything at all if the other defects. Defection pays off better, no matter which strategy the other player chooses (Axelrod 1984, p. 9). The example of player B shows that B is better off defecting, no matter whether player A cooperates or defects: in the first scenario in which player A cooperates, player B would receive 5 points instead of 3; in the second scenario, in which player A defects as well, player B would receive 1 point instead of 0. Players are thus always more tempted to defect than to cooperate (Rapoport and Chammah 1965, p. 24f.).

 Acting rationally is understood as acting in one’s own self-interest; see Hardin (1995, p. 46).

1

22

2  Theories and Theoretical Contribution

In a prisoner’s dilemma, the possibly best social outcome for all players—3 points—that can be achieved through collective action is always passed over by the short-term individual benefit each prisoner chooses instead: cooperation is not expected (Ostrom 2010a, p. 156).2 This phenomenon is explained with the so-called Nash equilibrium: when all participants in a social dilemma have chosen a strategy that fulfils their interest best while considering the other participants’ strategies and their effect on their own strategy’s outcome (benefit)—because the strategies are interdependent—then this situation is called Nash equilibrium. It reflects the situation of players whose actions yield the best outcome for themselves considering the other players’ actions and for whom a change in strategy would not improve the outcome unless the other players changed their strategies as well (Ostrom 1998, p. 4; Poteete et al. 2010, p. 43). Translated into real-world contexts, this dilemma situation reflects the problem of actors opting for short-term benefits for themselves which, when summed up, are less beneficial than the outcome that could be achieved if actors cooperated. Such a situation is also called a collective action problem (Ostrom 2010a, p.  155)—the problem is the participants’ incapacity to engage in a collective action. The situation changes if a prisoner’s dilemma is played more than once. By repeated playing, the players gain knowledge of the other players’ past actions (Ostrom 2010a, p. 157) and interaction strategies, and they can react to those. Robert Axelrod (1984) assesses the behavioral strategy that performed best in a series of repeated prisoner’s dilemma games. He organized a tournament of prisoner’s dilemma games where different behavioral strategies in form of algorithms competed against each other to assess the strategy with the best performance (Axelrod 1984, p. 20).3 He found out that The evolution of cooperation requires that individuals have a sufficiently large chance to meet again so that they have a stake in their future interaction. If this is true, cooperation can evolve in three stages. (Axelrod 1984, p. 20)

For cooperation to start, clusters of individuals have to exist that “(…) base their cooperation on reciprocity and have even a small proportion of their interactions 2  In a prisoner’s context, the logic is one of years two persons accused of having committed a crime receive as sentence depending on their testimony. If both cooperate and are silent about the deed, they are both sentenced with the minimum charge of 1  year. If both defect, that is, betray and accuse each other for having committed the crime, they are kept under custody for 2 years. Now, if one prisoner cooperates, that is, remains silent, while the other defects, that is, snitches on the silent comrade, the betrayer goes free, while the one accused—who remained silent—gets 3 years. Defecting is thus the safer bet: you get 2 years of prison if the other defects just like you but 0 years if the other cooperates while you defect; whereas if you cooperate, you get 1  year if the other cooperates as well but 3  years if the other betrays. Defecting while the other is cooperating is always more attractive, as the prisoner gets 0  years when betraying instead of 1  year if she/he cooperates (cf. Ostrom 1990, p. 217; Rapoport and Chammah 1965, p. 24f.). For a depiction of the scenario, see Table 1, Annex I. 3  The winning strategy was “tit-for-tat.” In this strategy, the player starts with cooperation and in the following rounds does what the other player(s) did in the preceding round (Axelrod 1984, p. viii; Poteete et al. 2010, p. 55).

2.1  Theories of Noncooperation

23

with each other” (Axelrod 1984, p. 21). Reciprocity is an individual’s reaction to another person’s action by the same means (Ostrom 1998, p. 10). Axelrod found out that the strategy of reciprocity asserts itself over the other strategies applied in human interactions. Finally, Axelrod detected that if cooperation is built on reciprocity, then strategies that are less cooperative in nature have a hard time intruding actors’ cooperative patterns (Axelrod 1984, p. 21). Axelrod illustrates the importance of reciprocal action for cooperation to start by the interactions of World War I enemy soldiers who were in trench warfare (Axelrod 1984, p. 74ff). During bad weather, in the morning hours, at Christmas, and during the serving of rations, the enemy soldiers would stop the shooting, and short-time truces would be established (Axelrod 1984, pp. 77–79). This was possible as the situation was not a one-time prisoner’s dilemma in which two players play only once, but an iterated prisoner’s game of continuous interaction, i.e., playing, in which the players could observe each other’s playing behavior.4 Based on this, soldiers from the two sides promptly realized when the other side reciprocated their move of not shooting.5 A mutual cooperation of not shooting at each other emerged (Axelrod 1984, p. 77).6 Axelrod thus refutes the assumption of the non-cooperative game theory that actors in a finite social dilemma game apply the so-called “backwards induction” to decide on their strategies. Backwards induction means actors consider which strategies they themselves and the other players might choose in the last round of the game to deduce their own strategies for the game based on these expectations for the last round. As actors would choose not to cooperate in the last round7 and expect others to defect too, there would be no motivation for them not to defect themselves in all the preceding rounds, that is, in the entire game (Ostrom 1998, p. 5; Poteete et al. 2010, p. 54).8 There remains the question, which factors could induce participants of a one-­time prisoner’s dilemma to cooperate. Bohnet and Frey (1999) conducted experiments of prisoner’s dilemma games in which participants would (a) not know each other (condition of anonymity), (b) know each other (condition of mutual identification), and 4  This is because in a trench, warfare enemy soldiers are immobile facing each other for long periods of time (Axelrod 1984, p. 77). 5  These acts of reciprocity were easier to achieve as the two camps had similar needs and activities, like, e.g., eating at similar times of the day, during which each side appreciated not to be attacked (Axelrod 1984, p. 79). 6  This condition was reinforced by harsh retaliation from one side if the other side had defected the situation of mutual agreement and started an attack (Axelrod 1984, p. 79f.). The mutual cooperation came to an end when the high commands of the armies requested the soldiers to raid the other side. Soldiers could not fake raids—which would have been necessary to stay within the mutual cooperation on nonaggression—as either hostages or casualties could testify to their happening (ibid., p. 82f.). 7  Defecting in the last round has no consequences for there is no next round in which the other players could react to a player’s defection. 8  Experiments have shown that actors “(…) do not use backward induction in their decision-making plans (…)” (Ostrom 1998, p. 5).

24

2  Theories and Theoretical Contribution

(c) be able to talk (condition of communication). They showed that knowing the other player as well as being able to talk to one another significantly increases the likelihood of the prisoner’s dilemma participants to cooperate (Bohnet and Frey 1999, p. 49; Ostrom 1998, p. 6). As the experiments were one-time games, players could not “learn” the other players’ strategies—the condition of reciprocity as an incentive to cooperate could be ruled out (Bohnet and Frey 1999, p. 45). Besides this research project, there is an extensive amount of experiments and studies that challenge the assumption about humans not cooperating (Ostrom 1998, p. 4f.). Communication has been proven to increase the likelihood of cooperation in all types of social dilemmas9 (Ostrom 1998, p. 7). It is one of four structural aspects that enhance the likelihood for participants of a one-time prisoner’s dilemma to cooperate and thus opt for the joint benefit for all instead of choosing the short-term individual benefit. The four aspects are (1) the number of participants; (2) the type of benefits—shared by all or individually subtracted; (3) the diversity of the participants; and (4) personal communication (Ostrom 2010a, p. 157). Ostrom explains the strength of communication to foster cooperation with the trust it creates between the participants (Ostrom 2010a, p. 158).10 Trust is defined as an individual’s expectation about other individuals’ actions, which affect the individual’s action. The individual’s action though is chosen before knowing the actions of the others (cf. Dasgupta 1999, p. 330). In a social dilemma, trust is understood as an actor’s expectation that her/his cooperative action will be reciprocated by the others (Ostrom 1998, p. 12).11 As soon as a prisoner’s dilemma is played more than once, other variables emerge that affect actors’ behavior and the likelihood of collective action to rise (Ostrom 2010a, p. 157f.): information about past actions (5); the way participants are connected to each other, e.g., through resource exchange (6); and participants’ option to exit or enter the game freely (7). The assumption of the non-cooperative game theory is that actors in a social dilemma situation—in which actors’ actions are interdependent—would be better off if they cooperated, since cooperation yields a better outcome for the entire group, that actors, however, choose their self-interest over the group interest (Poteete et al. 2010, p. 44). The theory is challenged once the social setting is opened up for structural variables to intervene and alter it: communication between actors and the number and composition of the actors increases the chance of actors cooperating. 9  Social dilemmas comprise public good problems, the free-rider problem, moral hazard, the credible commitment dilemma, generalized social exchange, the tragedy of the commons, and exchanges of threats and violent confrontations (Ostrom 1998, p. 1). 10  Being able to talk in person to an actor of the group enhances the efficacy of communication considerably (Ostrom 1998, p.  7). The reasons why communication enhances cooperation are manifold. Communicating actors can (a) transfer and exchange information, increasing the understanding of the situation and actors’ ability to decide on an adequate joint strategy; (b) exchange their mutual commitment, thereby building and strengthening trust and increasing expectations of the others’ behavior; (c) reinforce norms; and (d) create a group identity (ibid., p. 7). 11  For an extensive discussion on the relevance of trust in establishing social relations, see Coleman (1988).

2.1  Theories of Noncooperation

25

If  actors act repeatedly, inevitably further structural variables interfere that challenge actors’ non-cooperative behavior: actors get to know each other’s actions and thus each other’s behavioral strategies. Based on this information, actors can adjust their own behavior accordingly, for a step that makes cooperative moves possible. Over time, actors also get the chance to build connections, which in turn may influence their willingness to cooperate. Despite these enabling factors, many prisoner’s dilemmas exist in the real world. The fight against climatic change can be viewed as one, as countries are not willing to start reducing their CO2 emissions but hope other countries will do so. Instead of each state contributing its share to solving the problem, with the outcome of halting global temperature rise to the benefit of all states, states defect. That is, they do not contribute their share of action while expecting other countries to contribute. The result is the worst outcome possible for the group of states—an overheating planet with all its expected negative consequences12—because no one cooperates (Ostrom 2010b, p. 551; The Economist 27.09.2007).

2.1.2  The Tragedy of the Commons Gordon (1954) and Hardin (1968) describe the prisoner’s dilemma in the context of the use of commons. In his article on natural resource utilization, the economist H. Scott Gordon depicts the dilemma of humans preferring to go for their own self-­ interest in the fishing industry: In the sea fisheries the natural resource is not private property; hence the rent it may yield is not capable of being appropriated by anyone. The individual fisherman has no legal title to a section of ocean bottom. Each fisherman is more or less free to fish wherever he pleases. The result is a pattern of competition among fishermen which culminates in the dissipation of the rent of the intramarginal grounds. (…) if (…) the individual fishermen are free to fish on whichever ground they please, it is clear that this is not an equilibrium allocation of fishing effort in the sense of connoting stability. (Gordon 1954, p. 130f.)

With the fishermen’ freedom to fish wherever they please and the dissipation of rents on fishing grounds, Gordon states the quintessential characteristics of common-­ pool resources (CPR): the difficult exclusion of users and the subtractability of the  The expected negative effects are, i.a., the rise of the sea level, resulting in the flooding of inhabited islands, coastal areas, and deltas, saltwater intrusion, and damage to infrastructure; the acidification of the oceans and an increase of the sea temperature, affecting the entire maritime ecosystem with uncertain effects on the maritime population and further effects on the global climate, furthermore leading to a reduction of the productivity of fisheries and aquaculture; a growth of heavy weather events, leading to more frequent and more severe droughts, heat waves, inundations, and risks of erosions; water scarcity; a transformation of ecosystems on the terrestrial land from one type to another; loss of biodiversity; smaller net reductions in yields of maize, rice, wheat, and other cereal crops; reductions in projected food availability in Central Europe, the Mediterranean, Southern Africa, the Amazon, and the Sahel; and an increase in heat-related mortality and vector-borne diseases (BMU et al. (2017a, b); IPCC (2018)).

12

26

2  Theories and Theoretical Contribution

resource (Ostrom et al. 1994, p. 7). His example is a prisoner’s dilemma in so far as the fishermen’s actions do not favor the joint benefit of the resource’s stability but mirror their short-term self-interests. Fourteen years later, biologist Garrett Hardin picked up the topic again and described the situation of humans’ overexploitation of natural resources as “tragedy of the commons”: Picture a pasture open to all. It is to be expected that each herdsman will try to keep as many cattle as possible on the commons. (…) the rational herdsman concludes that the only sensible course for him to pursue is to add another animal to his herd. And another; and another. … But this is the conclusion reached by each and every rational herdsman sharing a commons. Therein is the tragedy. Each man is locked into a system that compels him to increase his herd without limit—in a world that is limited. (Hardin 1968, p. 1244)

Hardin’s scenario is a prisoner’s dilemma. Herders do not choose to cooperate, although this would bring the socially optimal outcome: if each herder put less cattle on the meadow, the meadow would not be overused but would continue to provide food for all herders’ cattle. Instead, the herdsmen prefer to gain short-term benefits of their own by letting as much cattle as possible graze. In the long run, their actions deplete the common meadow and cut their cattle short in food supply.13 For Hardin, the interim solution to cope with this tragic problem is the application of “coercive laws” (Hardin 1968, p. 1245) or the turning of commons into private property (ibid., p. 1247).14

2.1.3  The Theory of Collective Inaction Hardin bases his argument on Mancur Olson’s theory of collective action (Olson 1965; Ostrom 2000b, p. 38). Olson argues that the size of a user group matters. The larger a group is, the smaller each group member’s share of contribution to the overall group benefit becomes. The incentive for group members not to spend this share—that is, not to contribute to collective action—rises in large groups as the non-contribution most likely goes unnoticed. As an economist, Olson looks at cooperation within organizations and among firms in an economic environment. His research interest is on actors’ incentives to form collective action—or not to do so. His main arguments cover the problem of free-riding on the provision of a public good and the trivial contribution problem.15  For a detailed interpretation of Hardin’s “tragedy of the commons” as a prisoner’s dilemma game see Ostrom (1990, p. 3f.). 14  In Hardin’s eyes, the tragedy is due to the increasing population growth (Hardin 1968, p. 1243) and can only be solved through low-population density which in turn can only be achieved by “relinquishing the freedom to breed, and that very soon” (ibid., p. 1248). 15  The trivial contribution problem arises out of a group’s large size and the tendency of the actors not to contribute to the group’s greater benefit because this contribution can go unnoticed in a large group. Olson gives the following example to illustrate this case: “Some (…) may argue that the 13

2.1  Theories of Noncooperation

27

The problem of free-riding refers to actors who do not pay for the provision of a public or common good while still benefitting from it as other actors pay for its provision and access to it is free (Olson 1965, p.  21; Ostrom 1990, p.  32). The issue is that [t]hough all of the members of the group (…) have a common interest in obtaining this collective benefit, they have no common interest in paying the cost of providing the collective good. Each would prefer that the others pay the entire cost, and ordinarily would get any benefit provided whether he had borne part of the cost or not. (Olson 1965, p. 21)

The free-riding problem occurs in the provision of public and common goods. The difference lies within the subtractability of commons and the non-subtractability of public goods (Ostrom 1990, p.  23). Free-riders of a common increase its depletion; free-riders of a public good cannot subtract benefits from it as this criterion does not apply to public goods.16 The free-rider problem in public goods situations is called a second-order dilemma. The public good provision problem and the CPR appropriation problem are first-order dilemmas (Okada 2008, p. 172). Olson’s theory of collective action states that collective action in a social dilemma situation is not achievable because actors seek their personal short-term benefit instead of acting for the greater good, that is, the welfare of the group (Olson 1965; Ostrom 2010b, p.  551). Olson’s theoretical considerations about actors’ ability to cooperate thus rather constitute a theory of collective inaction than provide a theory of when actors actually do cooperate (Poteete et al. 2010, p. 47). Furthermore, various studies on empirical common-pool resource uses could show that resource appropriators—the herdsmen in Hardin’s metaphor and the group members in Olson’s example—do not necessarily withdraw from a resource’s provision or overuse a resource until its depletion. They rather establish rules to regulate the resource use (Dietz et al. rational person will (…) support a large organization (…) that works in his interest, because he knows that if he does not, others will not do so either, and then the organization will fail, and he will be without the benefit that the organization could have provided. (…) in a large organization, the loss of one dues payer will not noticeably increase the burden for any other one dues payer, and so a rational person would not believe that if he were to withdraw from an organization he would drive others to do so” (Olson 1965, p. 12). The trivial contribution problem applies to public goods only and refers to the refusal of the actors to contribute to a public good’s provision as the quality of the good neither improves through the individual contributing nor deteriorates through the individual not contributing. Olson argues that the larger the group that provides a good, the less likely a defecting group member will be noticed. This is because the larger the group is, the smaller each member’s contribution becomes and the easier a missing contribution can go unnoticed. As long as the missing contributions are not remarked, the amount the others contribute will not change. The quality of the public good—in Olson’s example the organization’s performance—does not alter to the better or worse if one member does not contribute. Members may thus have the incentive to “save” their share for themselves as they see the good remains intact without their contribution. This logic will counteract itself if too many think and act in this way; see Hardin (2009, p. 62). 16  A prominent example is public transportation. Let us say a city’s subway system is financed by each user buying a ticket. If some users do not buy tickets when using the subway, the large majority of ticket buyers still pays the amount of money needed to provide the public good. The few who cheat remain unnoticed. If the group of cheaters grows, the public good risks under provision or even decay. If nobody paid for using the subway, the subway system could not be maintained.

28

2  Theories and Theoretical Contribution

2003, p. 1907; Ostrom 1990, p. 19f., 58ff). Contrary to Hardin’s scenario where there is no property rights system, reality shows that property rights over natural resources do exist. Experiments of social dilemma configurations related to CPR or public goods could further demonstrate that individuals do not solely choose the non-cooperative strategy (Ostrom 1998, p. 6). Therefore, not all CPR situations are necessarily dilemmas with the appropriators acting in ways that are not rational, from the perspective of the group of appropriators, and thus creating negative outcomes for the group (Ostrom et al. 1994, p. 15)—or for the resource, as in environmental social dilemmas. These findings further support Rapoport and Chammah’s (1965) conclusion that the application of the concept of rationality in nonzero-sum games “makes questionable sense” (Rapoport and Chammah 1965, p. 13).

2.2  T  heoretical Concepts on Collective Action and Common-­Pool Resources A researcher who intensively investigated actors’ behavior in nonzero-sum games in the context of resource use was Elinor Ostrom. She defines social dilemmas as situations “in which individuals make independent choices in an interdependent situation” (Ostrom 1998, p. 3). These social dilemmas create collective action problems (Ostrom 2010a, p. 155), as mentioned above. A common-pool resource (CPR) problem is such a social dilemma. Actors face the challenge of “how to organize to avoid the adverse outcomes of independent action” (Ostrom 1990, p. 29). If these independent actions do not contribute to a joint benefit (Ostrom 1998, p. 3) for all actors, they either only serve the one acting or negatively affect the other actors— because the actions are interdependent. Two problems arise. The CPR over-­ appropriation problem relates to the resource units and is due to their subtractability and the difficult exclusion of users (Ostrom 1990, p. 32, 48)—an actor using the resource unit receives a direct benefit from the use, but harms the other users by reducing the resource stock by one unit. If all users act the same way, the resource is overexploited. The CPR provision problem concerns the resource stock and is twofold. A supply-side provision problem concerns the difficulty of providing and maintaining a common if users act independently, because they will tend to free-­ ride. They thereby undermine the joint provision of the good. A demand-side provision problem refers to the regulation of “withdrawal rates” (ibid., p.  49) for a resource and thus reflects users’ capability and willingness to change their using pattern to allow the resource to regenerate. Collective action helps to overcome negative impacts of single actions and to evade social dilemmas (Ostrom 1998, p. 6ff). It can mitigate the CPR overexploitation problem (Fischer et al. 2004, p. 812). Collective action is defined as “the possibility of benefits from coordinated action” based on mutual interests of different actors (Monge and Contractor 2003, p. 159). If these coordinated actions aim for a joint benefit for a group of actors facing a CPR problem, then this is called cooperation (Ostrom 1998, p. 3f.).

2.2  Theoretical Concepts on Collective Action and Common-Pool Resources

29

2.2.1  A  Theoretical Concept of Cooperation’s Core Relationships As non-cooperative game theory does not hold up with the empirical evidence about individuals starting cooperation in social dilemmas (Ostrom 1998, p.  9), Ostrom claims the need for a “broader theory of human behavior” (Ostrom 2010a, p. 163), a (…) consistent theory to explain why cooperation levels vary so much and why specific configurations of situational conditions increase or decrease cooperation (…). (Ostrom 1998, p. 9)

Ostrom’s baseline for an explanation of collective action in CPR settings is a “theory of boundedly rational, norm-based human behavior” (Ostrom 2010a, p. 156). The main elements she bases her theoretical concept of collective action upon are –– Actors’ reputation “for keeping promises and performing actions with short-­ term costs but long-term net benefits” (cf. Ostrom 1998, p. 12). –– Actors’ trust in each other’s promises about the actions to be taken. –– Actors’ use of reciprocity in their action-taking (Ostrom 1998, p.  12; Ostrom 2000a, p. 142; Ostrom 2010a, p. 163). All three factors—Ostrom calls them “core relationships”—reinforce each other (Ostrom 1998, p. 14). She argues that actors’ reciprocal actions and actors’ trust influence the level of cooperation among the actors. The level of cooperation in turn informs about the net benefits gained from cooperation (Ostrom 2007, p. 202, Fig. 2; Ostrom 2010a, p.  163). All five factors are themselves influenced by exogenous variables (Ostrom 2010a, p. 163). These exogenous structural variables are the factors most frequently identified as having an influence on collective action (ibid., p. 164). They comprise actors’ linkage structure, the number of participants, face-­ to-­face communication, heterogeneity of participants, actors’ freedom to enter or exit the situation, information about past actions, and the characteristic of the resource17 (Ostrom 2010a, p. 163, Fig. 2). Ostrom’s concept of the core relationships influencing collective action is not so much a theory, which explains causal relations among variables, than it is a conceptual framework that helps structuring research on cooperation and collective action. The framework sharpens the focus on certain aspects of a social dilemma whose possible influence on cooperation one can assess.18 Ostrom recognizes that it is not possible (…) to link all of the structural variables (…) in one definite causal model given the large number of variables and that many of them depend for their impact on the value of other variables. (Ostrom 2010a, p. 163)

 The characteristic of the resource refers to whether it is subtractive, i.e., a CPR, or shared, i.e., a public good. For a visualization of the framework, see Fig. 1, Annex II. 18  Ostrom herself calls the theory a “framework” and “specific scenarios of causal direction” at the same time (Ostrom 2010a, p. 164). 17

30

2  Theories and Theoretical Contribution

Thereby, she indirectly admits that her concept cannot be a theory, since theories make distinct assumptions about causal relations among variables. The concept is thus no theory of collective action, but a research approach that suggests which variables to consider when assessing collective action and its causing factors and how these variables may be related to each other. Ostrom’s research concept further stresses the importance of trust among actors for cooperation to emerge. For my research, I claim trust to be a component of cooperation, a structural aspect I am able to measure within actors’ cooperation.

2.2.2  Preconditions for Collective Action in CPR Settings CPR scholars have analyzed the CPR problems of over-appropriation and provision in field studies, assessing under which circumstances and through which factors appropriators are able to solve these problems. They proved Olson’s theory of collective action to be wrong in the context of small-scale environmental social dilemmas (cf. Ostrom 2010b, p. 551). Based on the extensive work of field studies, they derived a set of variables that account for the self-organization of resource user groups and the maintenance of collective action (Agrawal 2001; Araral 2014, p. 14; Baland and Platteau 1996; Ostrom 2000a, p. 148; Ostrom et al. 1994). One set of variables is “associated with an increased likelihood of self-­ organization” (Ostrom 2000b, p. 40) and thus with the emergence of a user group’s self-organization to manage a common. These variables are divided in attributes of a resource and characteristics of the users. The attributes of a resource comprise the feasible improvement of the resource at stake. This implies that the resource is not yet in such a bad condition that joint effort to improve it would be futile. A second attribute is the availability of valid and reliable indicators of the resource system’s condition at a relatively low cost. Two further characteristics concern the relatively predictable flow of resource units and the resource system’s spatial extent that affect the likelihood of collective action. The size of a resource system needs to be sufficiently small so that appropriators are able to develop knowledge of its external boundaries and its microenvironments. The characteristics of the users encompass the salience appropriators attribute to a resource system—in other words, appropriators’ dependence on the resource for a major portion of their livelihoods; a common understanding about the resource’s internal mechanisms and how actors’ actions affect each other and the resource; a sufficiently low discount rate in relation to future benefits gained from the resource; reciprocal relations and trust in each other that promises are kept; autonomy to define access and harvesting rules without an external authority’s influence; and prior organizational experience and local leadership (cf. Ostrom 2000b, p.  39f.; Schlager 2004, p. 151f.). The other set of variables is subsumed under the label of the so-called design principles, reflecting the necessary conditions for the successful maintenance of a CPR (McGinnis 2011, p. 180; Ostrom et al. 1999, p. 279; Schlager 2004, p. 154).

2.2  Theoretical Concepts on Collective Action and Common-Pool Resources

31

The design principles contain the following aspects (cf. McGinnis 2011, p.  180; Ostrom 2000a, p. 149ff): 1 . Clearly defined biophysical and social boundaries of the CPR. 2. Local rules-in-use are in place which restrict the amount, the timing, and the technology used to harvest the resource; which assign the resource use’s benefits according to the users’ inputs; and which account for local conditions. 3. Most affected individuals are able to participate in making and modifying these rules. 4. Monitors are accountable to the users or are users themselves. 5. Graduated sanctions are applied to rule violators. 6. Low-cost, rapid, and local arenas for conflict resolution are available. 7. The right to organize is recognized by the higher-level authority. 8. The governance activities (monitoring, rule enforcement, conflict resolution, resource provision, and appropriation) are organized in multiple layers. Agrawal (2014) criticizes that these different sets of variables have been expanded, mixed together, and summed up under the label of “enabling conditions.” They lack a clear distinction between the variables accounting for the emergence of collective action and those explaining the maintenance of it (Agrawal 2014, p. 88).19 Heikkila and Gerlak (2005) remark that studies on collective action in commonpool resource settings are one-sided, focusing on factors making resource management institutions successful and neglecting the ones driving the formation of these institutions (Heikkila and Gerlak 2005, p. 584). Another deficit of the variables is their applicability to only small-scale CPR settings (Agrawal 2001, p. 1649; Heikkila and Gerlak 2005, p. 586; Kerr 2007, p. 95f.). Kerr (2007, p. 95) argues that the enabling variables have a poor correspondence to watershed characteristics. “Small size, well-defined boundaries, low mobility, possible storage of benefits, predictability” (ibid., p. 96) are not valid characteristics of a watershed, since the resource water is mobile. The characteristics of the users can also only be found in a “watershed no larger than a village” (ibid., p. 96)—a setting most watersheds exceed. The identification of these variables has been a milestone in research on common-­ pool resources.20 However, the lists of variables do not add up to a theory that would provide general theoretical assumptions about the variables’ intensity, specificity, and combination accounting for collective action in a common-pool resource set-

 For a detailed list of the enabling conditions, see Agrawal (2001, p. 1659).  In 2009, Elinor Ostrom as the first woman in history—and together with Oliver E. Williamson from University of California, Berkeley—received the Nobel Prize in Economic Sciences “for her analysis of economic governance, especially the commons” and in particular for characterizing “the rules that promote successful outcomes,” i.e., the design principles, cf. Nobelprize.org (12.10.2009).

19 20

32

2  Theories and Theoretical Contribution

ting. The factors have proved to be important preconditions for collective action to emerge21 and endure.22 They are the (…) conditions that support the emergence and persistence of cooperation as well as those that undermine cooperation (Ostrom et al. 2014, p. 271)

Their specific interplay causing collective action remains contextual. To handle research on collective action in social dilemmas, Elinor Ostrom (2005) together with her colleagues of the Workshop in Political Theory and Policy Analysis at Indiana University conceptualized the Institutional Analysis and Development (IAD) framework (Ostrom 2005). This framework can be understood as a “conceptual map” (Poteete et  al. 2010, p.  57) that guides the assessment of the different variables mentioned above when analyzing collective action in environmental social dilemma situations.

2.2.3  T  he IAD Framework: An Analytical Tool to Assess Collective Action The IAD enables analysis of the “ways in which institutions operate and change over time” (McGinnis 2011, p. 169). Institutions are understood as (…) human-constructed constraints or opportunities within which individual choices take place and which shape the consequences of their choices. (McGinnis 2011, p. 170)

The IAD allows the systematic study of institutional arrangements (Ostrom et  al. 2014, p.  267) to “understand how people use these institutional arrangements to address shared problems and challenges” (ibid., p. 269). The IAD thus structures studies on collective action and the successful management of CPR (ibid., p. 267). The framework conceptualizes the stimulating influence of institutional and noninstitutional factors on actors’ actions (Mayntz and Scharpf 1995, p. 45f.; Ostrom 2005, p. 15ff). They are called contextual factors and comprise (a) the biophysical conditions of the environmental setting, (b) the community the actors come from, and (c) the rules-in-use that exist in the specific context (McGinnis and Ostrom 2014; Ostrom et al. 1994, p. 37ff). The two sets of variables identified to enable and maintain collective action can be located within these contextual factors. The contextual factors influence the framework’s core component, the action situation in which actors engage in interactions, “(…) exchange goods and services, solve problems, dominate one another, or fight (…)” (cf. Poteete et al. 2010, p. 58; Ostrom 2005, p. 13ff, 29). The interactions within the action situation produce outcomes, which in turn can alter the contextual factors and the configurations of the action situation (Ostrom 2005, p.  13f.). The framework illustrates the process of 21 22

 This relates to the characteristics of the resource and the users.  This relates to the design principles.

2.3  Factors Supporting Cooperation in CPR Settings

33

Biophysical Conditions Attributes of Community

Action Situation: actors’ interactions

Outcomes

Rules-in-Use

Fig. 2.1  Simple scheme of the IAD framework Own illustration, based on McGinnis and Ostrom (2014)

biophysical, social, and institutional factors shaping the debate actors have about the use of a resource. The outcomes of this social exchange can be new rules or guidelines that actors have decided upon, actors’ behavioral change, or concrete actions to take. To make assumptions about causal relations between the framework’s elements, one has to apply theory (cf. McGinnis and Ostrom 2014). Figure 2.1, comprises these thoughts, showing a simplified graph of the IAD framework (cf. McGinnis and Ostrom 2014).23 The two sets of variables identified as increasing the likelihood of collective action in social dilemmas and as contributing to its maintenance falsify Olson’s theory on collective action for small-scale social dilemma settings. They also ­challenge the assumption of non-cooperative game theory that actors act purely rational. However, the variables do not constitute a theory of collective action.

2.3  Factors Supporting Cooperation in CPR Settings The literature review has shown common theories on cooperation and collective action in common-pool resource settings to fall short in explaining how cooperation comes about in large-scale common-pool resource settings. The one and only the-

 In the original delineation of the IAD, actors “interactions” are presented as an element (box) of its own that is influenced by and influences the action situation and finally generates the outcomes; it is situated between the action situation and outcomes boxes. I subsume these “interactions” into the action situation itself, as McGinnis and Ostrom et al. (2014), Ostrom (2005), and Poteete et al. (2010, p. 58) highlight actors’ interactions within the action situation. I do not show the “evaluative criteria” Ostrom (2005, p. 13), as for this research, they are of no additional value. For reasons of simplicity, I relinquish the “feedback” arrows presented in the original figure.

23

34

2  Theories and Theoretical Contribution

ory for collective action in CPR problem situations does not exist. As Giest and Howlett (2014) put it more generally: (…) understanding the dynamics of the origins of ‘governance of the commons’ requires going well beyond the self-organizing cooperative structure suggested by Ostrom. (Giest and Howlett 2014, p. 37)

Axelrod (1984) showed that trust and reciprocity are factors essential for cooperation to evolve—two aspects Ostrom (1998, 2010a) took up in her research on collective action in CPR problem settings. The two identified sets of variables—the characteristics of the resource and the resource users and design principles—provide essential explaining variables for an analysis of collective action in CPR problem situations; the IAD framework can be used as an analytical tool for this analysis. However, the characteristics of the resource and the users have deficits concerning large-scale CPR in general and watersheds in particular. They are not suitable to explain the emergence of collective action in a catchment area. I close this research gap by relying on two theories from the field of policy analysis and on empirical studies of cooperation in large-scale CPR problem settings. Ostrom highlighted the influence of the biophysical conditions of a commons, the attributes of a community,24 and the given rules-in-use, understood as the institutional context, on actors’ interactions (Ostrom 2005, p.  15ff). I use these three umbrella terms as point of departure. In the following sub-chapters, I present theoretical arguments from studies that analyze cooperation in environmental governance contexts and that each refer to one of these three overarching factors. I draw on insights of studies investigating cooperation in large-scale CPR settings and theories in the field of policy analysis that consider cooperation to answer the research question: Why do actors cooperate in the management of a common-pool resource problem of over-appropriation? The theoretical arguments guide the formulation of hypotheses and complement the identified research gap in collective action theory. By this, I add valuable observations to the existing literature on collective action in CPR problem situations, and, more specifically, in large-scale CPR contexts. I close the chapter with a discussion of the research’s contribution to and relevance for theories on cooperation in CPR settings.

2.3.1  Recognizing the Environmental Problem Much of the theoretical thoughts on collective action can be reduced to what counts as costs and benefits and how they are distributed among the cooperators (Axelrod and Hamilton 1981; Nowak 2006, p. 1560). Cooperators gain a benefit

 The nature of the actors’ community is described by attributes like generally accepted norms of behavior, the extent of homogeneity of actors’ preferences, and the distribution of resources among actors; oftentimes, culture is used as the overarching term for this group of attributes (Ostrom et al. 1994, p. 45).

24

2.3  Factors Supporting Cooperation in CPR Settings

35

from cooperating, but they also have to pay a cost to do so. Transaction cost theory states that the engagement in cooperation comes along with costs: time, a certain individual effort, and money (Borgatti et al. 2013; Heikkila and Gerlak 2005). Actors thus have to weigh whether the benefit of cooperating outweighs the costs arising from doing so (Heikkila and Gerlak 2005, p. 585).25 In a similar vein, actors have to consider whether the costs of not cooperating could be higher than the ones of cooperating. This can be the case if a harm is to be expected that would be high in its costs and whose intensity—and thus whose produced costs— could be diminished or averted through cooperation with other actors. An example would be an expected flood, which is costly if nothing is done to prevent it and whose prevention would diminish its expected costs. This prevention in turn could only be achieved through joint efforts, that is, the cooperation of many, the cost of which would be less in comparison to the costs arising from the natural harm. Actors’ perception of an environmental problem may thus play a role for actors’ willingness to cooperate: recognizing the severeness of a CPR problem means realizing the costs that might arise from the problem. Cooperating to solve the problem might then be less costly than not cooperating and having to pay for the consequences of the problem. Recognizing a resource problem is thus essential for the beginning of cooperative action (Giest and Howlett 2014; Heikkila and Gerlak 2005; Lubell et  al. 2002). Lubell et  al.’s (2002) assessment of the emergence of 958 watershed partnerships in the USA supports this argument. Their analysis shows that watershed partnerships increase when environmental problems become severe (Lubell et al. 2002, p. 159).26 The logic behind is that the more severe an environmental problem is, the higher the benefit of watershed partnerships becomes (ibid., p. 150). Giest and Howlett (2014, p. 39) emphasize as well that a severe problem serves as a common ground for collaboration. Gerber et al. (2009) argue in the same vein. They introduced a new theoretical concept, the Institutional Resource Regime (IRR). The IRR consists of approaches from institutional economics and the political sciences and analyzes the regulation of heterogeneous, complex, and competitive uses of natural resources (Gerber et al. 2009, p. 799). The concept introduces two categories to evaluate resource regimes: the “extent” and the “coherence” of a regime. Extent refers to the number of the resource’s goods and services that are regulated by a regime.

 Taylor and Singleton (1993) name three different kinds of transaction costs, each referring to a different aspect of solving a collective action problem. Search costs arise from the search for convenient cooperational agreements that yield gains. Bargaining costs incur when actors argue about the distribution of these gains. Monitoring and enforcement costs accrue on the actors’ mutual observation of each other’s behavior; see Taylor and Singleton (1993, p. 196). 26  To measure their concept of “problem severity,” they relied on four different variables: the quality of an ecosystem’s condition, the degradation of water quality due to agriculture and urban runoff, the population pressure in a watershed, and the number of wastewater facilities discharging effluents (Lubell et al. 2002, p. 150f.). 25

36

2  Theories and Theoretical Contribution

Coherence means the degree of coordination among these regulations (ibd., p. 805). The IRR hypothesizes that the greater the threat to stability of a resource, the more it will be perceived as a relevant collective problem to be resolved and the more likely it is that attempts will be made to (…) improve (…) coherence [of new regulations] by (…) coordinat[ing] the actors with regard to their use activities. (Gerber et al. 2009, p. 807)

Studies on adaptation and resilience (Blythe et al. 2014; Schemmel et al. 2016) purposely consider natural threats (e.g., climate change) to assess the vulnerability and adaptability of communities (Adger et al. 2016; Gersonius et al. 2016). Shocks from outside, like natural disasters, and threats within a political system are also examined in policy process theories that investigate the influence of these threats on actors’ actions (Birkland 2011; Nohrstedt and Weible 2010; Sabatier 1993, p. 34). The studies of collective action in CPR settings pay less attention to threats to CPR and how resource users perceive them. Ostrom emphasizes threat only implicitly through the resource and user attributes that enable self-organization (Ostrom 2000b, p. 40): feasible improvement means that appropriators see a chance for the resource’s improvement when taking joint actions. Threat is implied as affecting the resource’s deterioration process. The attribute salience stands for appropriators’ dependence on a CPR for a major portion of their livelihood. It is threat to the resource that challenges the appropriators’ dependence on a resource system. The salience appropriators attribute to the resource system is said to have an influence on cooperation (Ostrom 2000b, p. 40; Schlager 2004, p. 152): if appropriators do not feel or are not dependent on the resource, the CPR problem does not affect them—coordinated action directed towards the solution of the CPR problem is of no interest to them. If actors are dependent on a resource, however, they are affected by the problems relating to the resource, and they are more likely to wish to do something against it—like work together on the problem’s solution. In this research, I test whether the perception of a threat to a common-pool resource relates to actors’ cooperation in a CPR problem setting. I assume that perceiving the seriousness of a CPR problem enhances cooperation among actors facing this problem. My research’s argument is that the more serious actors perceive a CPR problem to be, the more likely they are to engage in cooperation—because they want to reduce the costs arising from the CPR problem. My first hypothesis is: Hypothesis 1a:  The higher actors’ problem perception is, the more likely they are to engage in cooperation. I further test whether actors with a similar problem perception are more likely to cooperate with each other. This assumption is based on the concept of homophily, the tendency of people with similar or identical characteristics to connect with each other (Leifeld and Malang 2014, p. 8; McPherson et al. 2001, p. 416f.). Hypothesis 1b is a variation of hypothesis 1a: Hypothesis 1b:  Actors with a similar problem perception are more likely to engage in cooperation with each other.

2.3  Factors Supporting Cooperation in CPR Settings

37

The next sub-chapter discusses an aspect that reflects the institutional context in which actors meet and decide on or already implement the policies aiming to manage a CPR problem.

2.3.2  T  he Ecology of Games Framework (EGF): How Forums Matter The ecology of games framework (EGF) conceptualizes the relations actors from different sectors entertain with a variety of institutions—the locus where actors make decisions on rules or coordinate the implementation of these rules—in the context of environmental governance. The aim of actors in participating in these so-called forums is to coordinate decision-making on a specific topic across different institutional arrangements.27 Each participation of various actors in one such forum is considered a game; the ensemble of these games constitutes an ecology of games (Lubell et al. 2012, p. 366; Lubell 2013, p. 540). Lubell et al. (2011) regard institutions and actors within an ecology of game setting as complementing each other in structuring and coordinating the games (Lubell et al. 2011, p. 11). The EGF is based on six concepts. Policy issues comprise a collective action problem related to the overuse of a common or the provision of a public good. Policy institutions are the formal and informal norms that structure the way in which actors decide the rules that ought to regulate the given issue. Policy actors have an interest in the rules and their outcomes. They may participate in policy institutions and build relations with each other to access resources like influence and information. Policy actors act boundedly rational with a limited knowledge about the overall setting. Policy subsystems deal with several issues within a geographically determined area and comprise various actors and institutions.28 Policy games comprise the policy institutions and policy actors within a specific policy subsystem. They are the process of actors’ interactions aiming at achieving a specific set of goals. Actors’ behaviors and preferences, the nature of the issue at stake, and the rules established to fix it shape the state of the setting. Time finally refers to the changes that occur to an ecology of games within a policy subsystem over time (Lubell et al. 2011, pp. 5–8; Lubell 2013, pp. 539–42).

 Forums do not necessarily have to be long term. Berardo et al. (2015), for instance, studied actors who came together in short-lived forums to solve the acute environmental problem of forest fire. These forums came up with ad hoc solutions to a concrete issue. 28  The common definition of a policy subsystem in policy process theory is narrower: A policy subsystem has a functional dimension, it thus focuses on a specific topic, and its scope is geographically restricted. In a subsystem, hundreds of actors from different levels of government, science, the media, and interest groups are active; see Sabatier and Weible (2007, p. 192); Weible and Sabatier (2005, p. 181). 27

38

2  Theories and Theoretical Contribution

The EGF’s aim is to understand how actors’ decisions and the set of institutions solve collective action problems (Lubell et  al. 2012, p.  366).29 The framework assumes that learning, the distribution of gains, and cooperation are at the base of governance systems (Lubell 2013, p. 543ff). These three processes help overcome the transaction costs of searching for a cooperational agreement, bargaining about the distribution of benefits, and monitoring actors’ actions. An ecology of games “(…) defines the benefits and transaction costs of solving the underlying policy issues” (ibid., p. 547). Actors play games, that is, participate in institutions (ibid., p. 540), to receive the greatest benefits from cooperation (ibid., p. 547). Cooperation in this regard happens through actors’ participation in forums, generating a gain for them. Fischer and Leifeld (2015) summarize potential gains from actors’ attendance in forums: the pushing forwards of a policy problem; the solution of the policy problem; lobbying for the own position regarding the policy problem; achieving one’s policy goal; increased legitimization of the forum’s political goals; enhanced visibility and reputation of oneself; reducing uncertainty about the given issue; and establishing trust (Fischer and Leifeld 2015, pp. 369–72). Fischer and Leifeld (2015) also give a neat definition of policy forums. They understand policy forums as “(…) organized, stable arrangements situated in the larger network of actors, where resources are exchanged among members” (Fischer and Leifeld 2015, p. 365). Policy forums exist over a certain amount of time or even on a permanent basis; they serve as venues of negotiation and knowledge exchange where a diversity of actor types such as local stakeholders, scientific experts, politicians, state authorities, or interest groups meet. Policy forums address policy-related issues without lobbying for one specific interest; they rather cover a variety of interests and may work on the formulation of rules and measures tackling the policy issue, on their implementation, or on the mere recognition of the policy issue at stake (ibid., p. 365f.). Forums thus influence and shape the work and the interaction of actors who deal with a certain policy problem. Lubell et  al. (2010) study the interplay of actors’ participation in forums and their cooperation. They investigate whether actors’ participation in collaborative institutions—that is, forums specifically designed to coordinate institutional arrangements and the actors participating therein—enhances or reduces the capacity of traditional institutions to produce cooperative behavior (Lubell et al. 2010, p.  288). I do not consider this type of collaborative institution or forum that is ­specific to US environmental governance, but I do examine the relation between actors’ playing of a game—i.e., actors’ participation in a forum—and their likelihood to cooperate with each other. Based on these theoretical considerations, I argue that there is a relation between actors’ participation in forums and their engagement in cooperation. That is because  More precisely, the EGF “(…) intends to produce empirically testable hypotheses about the structure and function of complex adaptive governance systems, analyze the causal processes driving individual behavior and institutional change, and ultimately understand how different types of institutional arrangements are linked to policy outputs and outcomes in order to provide recommendations about how to manage the system” (Lubell 2013, p. 358).

29

2.3  Factors Supporting Cooperation in CPR Settings

39

forums lead actors to get to know each other, which builds trust, a factor that constitutes cooperation. My next hypotheses acknowledge (a) the mere number of forums actors attend and (b) actors’ co-participation in the same forum: Hypothesis 2a:  The more forums an actor participates in, the more likely this actor is to engage in cooperation. Hypothesis 2b:  Actors participating in the same forum(s) are more likely to engage in cooperation with each other. The last sub-chapter focuses on an attribute of the actor community in the CPR problem setting at stake, that is, actors’ beliefs.

2.3.3  T  he Advocacy Coalition Framework (ACF): Actors’ Shared Beliefs Another factor to explain cooperation in a CPR problem setting is the connection between actors based on similar beliefs. The advocacy coalition framework (ACF) by Sabatier (1988) and Sabatier and Jenkins-Smith (1993) claims that actors with similar beliefs group in coalitions and show joint coordination patterns (Sabatier and Weible 2007, p. 197; Weible et al. 2010, p. 524). The ACF is a theoretical framework to analyze policy change in a given political system. It focuses on the interactions of so-called advocacy coalitions within a policy subsystem, their belief systems, and the policy-related learning within and across them (Sabatier 2007, p. 9f.; Sabatier and Weible 2007). The framework distinguishes actors’ beliefs in three levels of belief systems: deep core beliefs, policy core beliefs, and secondary aspects. Deep core beliefs represent ontological and normative perceptions and assumptions actors have about the world and human nature. These fundamental values, which can also be expressed on a left/right-wing continuum, exist within all policy domains. They can comprise actors’ viewpoints on “the nature of man,” the priority actors give to different ultimate values such as freedom, love, or security, or actors’ evaluation of distributive justice (Sabatier 1988, p. 144f.). Policy core beliefs deal with actors’ policy choices and their perception of the policy subsystem within which they act. They include, inter alia, the priority actors give to policy-related values and issues. Secondary aspects are the result of policy core beliefs that have been translated into policy preferences about the specific instruments or programs aiming at the policy goal (Sabatier 1993, p. 30f.; Sabatier 1998, p. 103f.; Sabatier and Weible 2007, p. 194f.). The last group of beliefs is the easiest to alter, while policy core beliefs and especially deep core beliefs are difficult, if not impossible, to change (Sabatier and Weible 2007, p. 194ff; Sabatier 1998, p. 104). Policy core beliefs are said to shape actors’ political behavior (Calanni et  al. 2015, p.  904); through policy core beliefs, one can understand individual behavior in a policy subsystem (Weible 2006, p. 102). Policy core beliefs are the factors that bring allies together and separate opponents (Sabatier 1998, p. 103).

40

2  Theories and Theoretical Contribution

Policy actors orientate themselves towards their policy core beliefs when they interpret evidence about a certain issue or problem (Henry and Dietz 2011, p. 197). They especially rely on their beliefs in situations of uncertainty (Ingold and Fischer 2014, p.  88)—such as presented by micro-pollutants in surface water. Actors constantly compare information they receive: to their own understanding of the world system, to their perception of the causes and severity of problems, and to the policy subsystem they act in. The ACF hypothesizes that similarities in these interpretations are an antecedent for actors to establish relationships (Sabatier et al. 1987, p. 457, 470; Weible 2005, p. 461). Different evaluations of problem situations due to diverging beliefs hinder cooperation, because “two actors with conflicting systems or beliefs will also tend to mistrust one another” (Henry and Dietz 2011, p. 197)30—and trust is an important prerequisite for cooperation.31 Actors identify those actors in the policy subsystem who are similar to themselves through the policy core beliefs. Actors with similar policy core beliefs tend to come together in a so-called advocacy coalition. They engage in a “nontrivial degree of coordinated action in order to translate those beliefs into public policy” (cf. Weible and Sabatier 2005, p. 183). By working together in a coalition, actors try to achieve their policy goals (Fischer and Sciarini 2015; Sabatier and Weible 2007, p. 196). The assumption that actors’ shared beliefs are a driver for actors’ coordination and their working together is called the belief homophily hypothesis (Calanni et al. 2015, p. 903). I base my next hypothesis on the relation between shared beliefs and cooperation (Ingold and Fischer 2014, p.  95; Weible 2005, p. 470). Hypothesis 3:  Actors who share the same belief are more likely to engage in cooperation. By testing this hypothesis, I challenge the finding by Henry et  al. (2010) that actors with divergent beliefs do not tend to collaborate while actors’ sharing of a similar belief does not encourage collaboration either. I further confront the results of the study by Calanni et al. (2015, p. 915ff) proclaiming actors’ shared beliefs are the least important driver for actors’ coordination. I also resume Schlager’s critique (1995, p. 261) that actors’ similar beliefs alone do not account for actors’ coordinated action and that contextual factors have to be considered as well. Instead, I follow the appeal from Weible (2005) and Weible and Sabatier (2005) for more empirical testing of the relation between actors’ similar beliefs and actors’ interactions.

 The misinterpretation of other actors’ intentions is called “devil shift”; see Fischer and Sciarini (2015, p. 66), Fischer et al. (2016), Sabatier et al. (1987), Sabatier and Weible (2007, p. 194), and Weible et al. (2010, p. 524). 31  Cf. Sects. 1.2, 2.1.1, and 2.2.1. 30

2.3  Factors Supporting Cooperation in CPR Settings

41

2.3.4  Theoretical Relevance of the Research The literature review has shown that common theoretical approaches about collective action fall short in explaining the reasons for cooperation in CPR situations. Scholars have identified the variables that enhance actors’ likelihood to self-­organize in CPR situations and the factors maintaining this status of self-organization—but there is no theory connecting these variables to each other and to cooperation in a causal relation. Recognizing the contributions of other theories regarding the explanation of cooperation—namely, studies on environmental threats, the ecology of games framework (EGF), and the advocacy coalition framework (ACF)—I broaden the scope of common research on collective action in large CPR problem settings. Research on CPR problems mainly considers small-scale CPR when assessing how actors cooperate and interact to manage a CPR (Agrawal 2001, p.  1649). Studies on CPR also mostly analyze user groups and their interactions with and use of natural resources. They most often focus on the quantity of and the access to commons. This book goes beyond theoretical assumptions on cooperation in small-scale CPR contexts and analyzes a large-scale CPR. This CPR comprises an area with regional, national, and even international spatial dimensions. Furthermore, the study expands the actors’ group commonly investigated in CPR research and includes state authorities, interest groups, providers, and polluters. Finally, the study specifically targets the quality of a CPR.  The research thus contributes a new research focus on CPR. Analytical frameworks investigating interdependencies and interactions of ecological and social systems tend to focus more on the social than on the ecological system. They bring an imbalance between these two units of analysis (cf. Binder et al. 2013). In this analysis, I look at the ecological and the social system in equal measure (cf. Pelosi et al. 2010) in that I conceptualize the CPR problem through the social-ecological system framework (SESF) and use the SESF’s criteria to categorize micro-pollutants, the resource surface water, micro-pollutants’ influence on it, actors’ relations to the resource, and the regulations and governmental structures in place that tackle the CPR problem. By this, I structure my analytical focus on a CPR problem situation (see Sects. 3.2 and 3.3). However, the explaining factors and the dependent variable I analyze are to be found on the social side of the social-­ ecological system under study. As I apply the SESF, I test its usability and explanatory strength—both aspects need more testing (Thiel 2015). I extend the functions of the SESF by using it for identifying the actors in my case studies (see Sect. 3.4.2). The application of Social Network Analysis (SNA) is a methodological contribution to the research of social-ecological systems. I conceive the management process among actors affected by a CPR problem to consist of networks of collaboration and information exchange. To the best of my knowledge, this is the first time that cooperation is analytically conceptualized as consisting of actors’ aim towards the same goal, their coordination of actions and exchange of resources, as well as their trust in each other.

42

2  Theories and Theoretical Contribution

This study enriches existing definitions of cooperation and encourages studies on cooperation in CPR problem settings for investigating this phenomenon from a social network perspective. In addition, I contribute to network science by focusing on different types of ties—the different relations defining cooperation (cf. Sect. 4.1)—an aspect that deserves more attention in the study of cooperation. I increase the meaningfulness of the results by conducting a cross-case comparison. The combination of quantitative (SNA) and qualitative (contextual case study comparison) methods intensifies the analysis’ explanatory power, while the differentiation of the cases allows me to verify whether results hold true in different contexts. The next chapter outlines the case studies, the data collection process, and the methods of data analysis. Furthermore, I briefly describe the social-ecological system framework’s (SESF) application to the case studies which structures the case study analyses.

References Primary Literature BMU, BMBF, IPCC Deutsche Koordinierungsstelle, UBA (2017a) Kernbotschaften des Fünften Sachstandsberichts des IPCC.  Klimaänderung 2013: Naturwissenschaftliche Grundlagen (Teilbericht 1). IPCC, Deutsche Koordinierungsstelle, Bonn BMU, BMBF, IPCC Deutsche Koordinierungsstelle, UBA (2017b) Kernbotschaften des Fünften Sachstandsberichts des IPCC. Klimaänderung 2014: Folgen, Anpassung und Verwundbarkeit (Teilbericht 2) Kernbotschaften des Fünften Sachstandberichts des IPCC.  IPCC, Deutsche Koordinierungsstelle, Bonn IPCC (2018) Global warming of 1.5°C. summary for policymakers. In: Masson-Delmotte V, Zhai P, Pörtner H-O, Roberts D, Skea J, Shukla PR, Pirani A, Moufouma-Okia W, Péan C, Pidcock R, Connors S, Matthews JBR, Chen Y, Zhou X, Gomis MI, Lonnoy E, Maycock T, Tignor M, Waterfield T (eds) Global Warming of 1.5 °C. An IPCC Special Report on the impacts of global warming of 1.5  °C above pre-industrial levels and related global greenhouse gas emission pathways, in the context of strengthening the global response to the threat of climate change, sustainable development, and efforts to eradicate poverty, Switzerland Nobelprize.org (2009) The Prize in Economic Sciences 2009 [online]. News release, 12 October. Available from: https://www.nobelprize.org/uploads/2018/06/press-7.pdf. Accessed 27 September 2019

Secondary Literature Adger WN, Quinn T, Lorenzoni I, Murphy C (2016) Sharing the pain. Perceptions of fairness affect private and public response to hazards. Ann Am Assoc Geogr 106(5):1079–1096 Agrawal A (2001) Common property institutions and sustainable governance of resources. World Dev 29(10):1649–1672 Agrawal A (2014) Studying the commons, governing common-pool resource outcomes: some concluding thoughts. Environ Sci Policy 36:86–91

References

43

Araral E (2014) Ostrom, Hardin and the commons: a critical appreciation and a revisionist view. Environ Sci Policy 36:11–23 Axelrod R (1984) The evolution of cooperation. Basic Books, Perseus Books Group Axelrod R, Hamilton W (1981) The evolution of cooperation. Science 211(4489):1390–1396 Baland J-M, Platteau J-P (1996) Halting degradation of natural resources. Is there a Role for Rural Communities? [online]. Food and Agriculture Organization of the United Nations (FAO), Rome, Italy. Available from: http://www.fao.org/docrep/x5316e/x5316e00.htm. Accessed 27 Sept 2019 Berardo R, Olivier T, Lavers A (2015) Focusing events and changes in ecologies of policy games: evidence from the Paraná River Delta. Rev Policy Res 32(4):443–464 Binder CR, Hinkel J, Bots PWG, Pahl-Wostl C (2013) Comparison of frameworks for analyzing social-ecological systems. Ecol Soc 18(4):26 Birkland TA (2011) An introduction to the policy process. Theories, concepts, and models of public policy making, 3rd edn. Routledge, London/New York Blythe JL, Murray G, Flaherty M (2014) Strengthening threatened communities through adaptation: insights from coastal Mozambique. Ecol Soc 19(2):6 Bohnet I, Frey BS (1999) The sound of silence in prisoner’s dilemma and dictator games. J Econ Behav Organ 38:43–57 Borgatti SP, Everett MG, Johnson JC (2013) Analyzing social networks. SAGE, London Calanni JC, Siddiki SN, Weible CM, Leach WD (2015) Explaining coordination in collaborative partnerships and clarifying the scope of belief homophily hypothesis. J Public Adm Res Theory 25(3):901–927 Casciaro T, Piskorski M (2005) Power imbalance, mutual dependence, and constraint absorption: a closer look at resource dependence theory. Adm Sci Q 50:167–199 Coleman JS (1988) Social capital in the creation of human capital. Am J Sociol 94:S95–S120 Dasgupta P (1999) Economic progress and the idea of social capital. In: Dasgupta P, Serageldin I (eds) Social capital. A multifaceted perspective, Washington, DC, pp 325–424 Dietz T, Ostrom E, Stern PC (2003) The struggle to govern the commons. Science 302(5652):1907–1912 Fischer M, Leifeld P (2015) Policy forums: why do they exist and what are they used for? Policy Sci 48(3):363–382 Fischer M, Sciarini P (2015) Unpacking reputational power: intended and unintended determinants of the assessment of actors’ power. Soc Networks 42:60–71 Fischer M-E, Irlenbusch B, Sadrieh A (2004) An intergenerational common pool resource experiment. J Environ Econ Manag 48:811–836 Fischer M, Ingold K, Sciarini P, Varone F (2016) Dealing with bad guys: actor- and process-level determinants of the “devil shift” in policy making. J Publ Policy 36(2):309–334 Gerber J-D, Knoepfel P, Nahrath S, Varone F (2009) Institutional resource regimes: towards sustainability through the combination of property-rights theory and policy analysis. Ecol Econ 68:798–809 Gersonius B, van Buuren A, Zethof M, Kelder E (2016) Resilient flood risk strategies: institutional preconditions for implementation. Ecol Soc 21(4):28 Giest S, Howlett M (2014) Understanding the pre-conditions of commons governance: the role of network management. Environ Sci Policy 36:37–47 Gordon HS (1954) The economic theory of a common-property resource: the fishery. J Polit Econ 62(2):124–142 Hardin G (1968) The tragedy of the commons. The population problem has no technical solution; it requires a fundamental extension in morality. Science 162(3859):1243–1248 Hardin R (1982) Collective action. A Book from Resources for the Future. The John Hopkins University Press, Baltimore Hardin R (1995) One for all. The logic of group conflict. Princeton University Press, Princeton Hardin R (2009) How do you know? The economics of ordinary knowledge. Princeton University Press, Princeton

44

2  Theories and Theoretical Contribution

Heikkila T, Gerlak AK (2005) The formation of large-scale collaborative resource management institutions: clarifying the roles of stakeholders, science, and institutions. Policy Stud J 33(4):583–612 Henry AD, Dietz T (2011) Information, networks, and the complexity of trust in commons governance. Int J Commons 5(2):188–212 Henry AD, Lubell M, McCoy M (2010) Belief systems and social capital as drivers of policy network structure: the case of California regional planning. J Public Adm Res Theory 21(3):419–444 Ingold K, Fischer M (2014) Drivers of collaboration to mitigate climate change: an illustration of Swiss climate policy over 15 years. Glob Environ Chang 24:88–98 Kerr J (2007) Watershed management: lessons from common property theory. Int J Commons 1:89–110 Leifeld P, Malang T (2014) National parliamentary coordination after Lisbon: a network approach, Barcelona Lubell M (2013) Governing institutional complexity: the ecology of games framework. Policy Stud J 41(3):537–559 Lubell M, Schneider M, Scholz JT, Mete M (2002) Watershed partnerships and the emergence of collective action institutions. Am J Polit Sci 46(1):148–163 Lubell M, Henry AD, McCoy M (2010) Collaborative institutions in an ecology of games. Am J Polit Sci 54(2):287–300 Lubell M, Robins G, Wang P (2011) Policy coordination in an ecology of water management games. Southern Illinois University, Carbondale Lubell M, Scholz JT, Berardo R, Robins G (2012) Testing policy theory with statistical models of networks. Policy Stud J 40:351–374 Mayntz R, Scharpf FW (eds) (1995) Gesellschaftliche Selbstregelung und politische Steuerung. Campus Verlag, Frankfurt/Main/New York McGinnis MD (2011) An introduction to IAD and the language of the Ostrom workshop: a simple guide to a complex framework. Policy Stud J 39(1):169–183 McGinnis MD, Ostrom E (2014) Social-ecological system framework: initial changes and continuing challenges. Ecol Soc 19(2):30 McPherson M, Smith-Lovin L, Cook JM (2001) Birds of a feather: Homophily in social networks. Annu Rev Sociol 27:415–444 Monge PR, Contractor NS (2003) Theories of communication networks. Oxford University Press, Oxford/New York Nohrstedt D, Weible CM (2010) The logic of policy change after crisis: proximity and subsystem interaction. Risks Hazards Crisis Public Policy 1(2):1–32 Nowak MA (2006) Five rules for the evolution of cooperation. Science 314(5805):1560–1563 Okada A (2008) The second-order dilemma of public goods and capital accumulation. Public Choice 135:165–182 Olson M (1965) The logic of collective action. Public goods and the theory of groups, 2nd edn. Harvard University Press, Cambridge, MA Ostrom E (1990) Governing the commons. The evolution of institutions for collective action. Cambridge University Press, Cambridge Ostrom E (1998) A behavioral approach to the rational choice theory of collective action. Presidential address, American Political Science Association. Am Polit Sci Rev 92(1):1–22 Ostrom E (2000a) Collective action and the evolution of social norms. J Econ Perspect 14(3):137–158 Ostrom E (2000b) The danger of self-evident truths. Political Science and Politics 33(1):33–44 Ostrom E (2005) Understanding institutional diversity. Princeton University Press, Princeton Ostrom E (2007) Collective action theory. In: Boix C, Stokes SC (eds) The Oxford handbook of comparative politics. Oxford University Press, Oxford, pp 186–208 Ostrom E (2010a) Analyzing collective action. Agric Econ 41(s1):155–166 Ostrom E (2010b) Polycentric systems for coping with collective action and global environmental change. Glob Environ Chang 20:550–557

References

45

Ostrom E, Gardner R, Walker J (eds) (1994) Rules, games and common pool resources. Michigan University Press, Michigan Ostrom E, Burger J, Field CB, Norgaard RB, Policansky D (1999) Revisiting the commons: local lessons, global challenges. Science 284(5412):278–282 Ostrom E, Cox M, Schlager E (2014) An assessment of the institutional analysis and development framework and introduction of the social-ecological systems framework. In: Sabatier PA, Weible CM (eds) Theories of the policy process. Westview Press, Boulder, pp 267–306 Pelosi C, Goulard M, Balent G (2010) The spatial scale mismatch between ecological processes and agricultural management: do difficulties come from underlying theoretical frameworks? Agric Ecosyst Environ 139(4):455–462 Pfeffer J, Salancik GR (1978) The external control of organizations: a resource dependence perspective. Stanford University Press, Stanford Poteete AR, Janssen MA, Ostrom E (2010) Working together. Collective action, the commons, and multiple methods in practice. Princeton University Press, Princeton Rapoport A, Chammah AM (1965) Prisoner’s dilemma. A study in conflict and cooperation, 2nd edn. The University of Michigan Press, Ann Arbor Sabatier PA (1988) An advocacy coalition framework of policy change and the role of policy-­ oriented learning therein. Policy Sci 21(2/3):129–168 Sabatier PA (1993) Policy change over a decade or more. In: Sabatier PA, Jenkins-Smith HC (eds) Policy change and learning. An advocacy coalition approach. Westview Press, Boulder, pp 13–39 Sabatier PA (1998) The advocacy coalition framework: revisions and relevance for Europe. J Eur Publ Policy 5(1):98–130 Sabatier PA (2007) The need for better theories. In: Sabatier PA (ed) Theories of the policy process. Westview Press, Boulder, pp 3–17 Sabatier PA, Jenkins-Smith HC (eds) (1993) Policy change and learning. An advocacy coalition approach. Westview Press, Boulder Sabatier PA, Weible CM (2007) The advocacy coalition framework. Innovations and clarifications. In: Sabatier PA (ed) Theories of the policy process. Westview Press, Boulder, pp 189–220 Sabatier PA, Hunter S, McLaughlin S (1987) The devil shift: perceptions and misperceptions of opponents. West Polit Q 40(3):449–476 Schemmel E, Friedlander AM, Andrade P, Keakealani K, Castro LM, Wiggins C, Wilcox BA, Yasutake Y, Kittinger JN (2016) The codevelopment of coastal fisheries monitoring methods to support local management. Ecol Soc 21(4):34 Schlager E (2004) Common-pool resource theory. In: Durant RF, Fiorino DJ, O’Leary R (eds) Environmental governance reconsidered. Challenges, choices, and opportunities. MIT Press, Cambridge, MA, pp 145–175 Taylor M, Singleton S (1993) The communal resource: transaction costs and the solution of collective action problems. Polit Soc 21(2):195–214 The Economist (2007) Playing games with the planet. A version of the “prisoner’s dilemma” may suggest ways to break through the Kyoto impasse. The Economist, 27 September Thiel A (2015) Constitutional state structure and scalar re-organization of natural resource governance: the transformation of polycentric water governance in Spain, Portugal and Germany. Land Use Policy 45:176–188 Weible CM (2005) Beliefs and perceived influence in a natural resource conflict: an advocacy coalition approach to policy networks. Polit Res Q 58(3):461–475 Weible CM (2006) An advocacy coalition framework approach to stakeholder analysis: understanding the political context of California marine protected area policy. J Public Adm Res Theory 17:95–117 Weible CM, Sabatier PA (2005) Comparing policy networks: marine protected areas in California. Policy Stud J 33(2):181–202 Weible CM, Pattison A, Sabatier PA (2010) Harnessing expert-based information for learning and the sustainable management of complex socio-ecological systems. Environ Sci Policy 13:522–534

Chapter 3

Methods and Cases

Abstract  The chapter lays out the methods applied to conduct the research and the case studies the study investigates. The chapter takes King et al.’s (1994) statement as starting point: “a research design is a plan that shows, through a discussion of our model and data, how we expect to use our evidence to make inferences” (King G, Keohane RO, Verba S, Designing social inquiry. Scientific inference in qualitative research. Princeton University Press, Princenton, NJ, 1994, p. 118). I first discuss the concepts of public policy analysis that inform the context of the research’s unit of analysis. The research focuses on actors’ cooperation, the unit of analysis, that happens within the management process of a CPR problem setting. Public policy analysis defines the policy problem the CPR problem represents, constitutionalizes the stage within the policy-making process that the CPR management process is located at, describes the potential solutions to the problem at hand and the way in which these solutions are put in place. The social-ecological system framework (SESF) I introduce thereafter guides the research design and the operationalization of the dependent, independent, and control variables—both conceptually and technically. I further define the case study selection criteria and present the three case studies and the national and regional water laws that apply in each one. In the last two sub-chapters, I describe the data collection process and the data analysis methods of the research, that is, descriptive and inferential Social Network Analysis (SNA) and a most-similar case research design. Keywords  Research design · Public policy analysis · Social-ecological system framework (SESF) · Social Network Analysis (SNA) · Exponential random graph models (ERGM) · EU Water Framework Directive (WFD) · Case study comparison This chapter lays out the methods applied to conduct the research and the case studies the study was investigating. This chapter takes King et al.’s (1994) statement as starting point: “a research design is a plan that shows, through a discussion of our model and data, how we expect to use our evidence to make inferences” (King et al. 1994, p. 118). I first discuss the concepts of public policy analysis that inform the context of the research’s unit of analysis (Sect. 3.1). The research focuses on actors’

© Springer Nature Switzerland AG 2020 L. M. J. Herzog, Micro-Pollutant Regulation in the River Rhine, https://doi.org/10.1007/978-3-030-36770-1_3

47

48

3  Methods and Cases

cooperation, the unit of analysis, that happens within the management process of a CPR problem setting. Public policy analysis defines the policy problem the CPR problem represents, constitutionalizes the stage within the policy-making process that the CPR management process is located at, and describes the potential solutions to the problem at hand and the way in which these solutions are put in place. The social-ecological system framework (SESF) I introduce thereafter guides the research design and the operationalization of the dependent, independent, and control variables—both conceptually and technically (Sect. 3.2). I further define the case study selection criteria and outline the case study design according to which I select the three case studies (Sect. 3.3). In the last two sub-chapters, I describe the data collection process (Sect. 3.4) and the data analysis methods of the research (Sect. 3.5).

3.1  A Public Policy Analysis Public policy analysis focuses on the outputs of a political system, namely, the instruments, strategies, actions, and programs developed to manage societal issues (Knill and Tosun 2012, p. 4). The research discipline scrutinizes the policy-making process that leads to these measures and assesses their content, design, and impact (ibid., p. 1). Certain concepts of public policy analysis are relevant for this book. The policy-making process characterizes the procedure of discussing a problem, developing adequate solutions, implementing them, and evaluating their effects (Knill and Tosun 2012, p.  9; Howlett and Giest 2013, p.  17). The assessed CPR management process in which I expect to find a certain degree of actor cooperation is represented in the implementation of policy instruments that tackle the CPR problem of micro-pollutants in surface water. The policy problem describes the CPR problem that I study from the viewpoint of policy studies. The research’s focus is further on the laws, guidelines, and instruments for tackling the problem of micro-­ pollutants in the Rhine catchment area, the so-called policy. The sub-chapter closes with a definition of policy networks, which include all actors that are involved in the implementation process of a policy (Börzel 1998, p. 260).

3.1.1  The Policy-Making Process The most common description of a policy-making process is that of the policy cycle, which is understood as a sequence of stages. The first stage consists of the identification of a societal or environmental problem and its setting on the political agenda.1

1  Problem identification and agenda-setting can also be defined as two separate stages in the policy process; see Jann and Wegerich (2003, p. 82 ff).

3.1  A Public Policy Analysis

49

In the second, so-called decision-making stage, possible solutions to the problem, the different policies,2 are formulated and selected by decision-makers.3 In the third stage, the adopted policies are implemented. Their performance and effects are evaluated in the last stage. After the evaluation, the cycle starts over again. Within a policy-making process, these stages can overlap and some of them can also be skipped (Knill and Tosun 2012, 9, 12; Howlett and Giest 2013, p. 17). Knill and Tosun (2012, p. 10) suggest using the different stages as the specific analytical foci on the policy process. Two foci are relevant for this research: policy adoption—within the decision-­ making stage—and policy implementation. The first relates to the laws, management plans, and instruments aiming at the regulation of micro-pollutants and having been decided on for the Rhine catchment. The second refers to the realization of the policy instruments. Both make up the CPR management process I analyze in terms of its capacity for cooperation.

3.1.2  The Policy Problem The goal of the policy-making process is to solve a societal problem. The characteristics of a problem are socially constructed (Knill and Tosun 2012, p. 102)—humans try to make sense of what they see and put the different aspects of an observed phenomenon in categories.4 Rochefort and Cobb introduce seven characteristics defining the issues discussed during the agenda-setting. These are causality, severity, incidence, proximity, novelty, and crisis of an issue and the availability of solutions for it (Rochefort and Cobb 1995, pp. 15–23). The characteristics also serve to describe a policy problem (Peters and Hoornbeek 2005, p. 87). Peters and Hoornbeek (2005) borrow from this set when they suggested their own list of problem characteristics. Solubility refers to the degree to which a problem can actually be solved (Peters and Hoornbeek 2005, p. 87ff). Complexity is broken down to political complexity, reflecting the diverse actors and interests that are involved in the problem, and programmatic complexity, representing the problem’s various causes and its scientific dimensions (ibid., p. 90ff). Scale means the problem’s extent and the range of its effects (ibid., p. 93f.). Divisibility relates to the benefits and costs the solution of the

2  Policies can be the pack of measures (legal acts, programs, actions, instruments) of a specific sector; public activities; the target(s) a legal act aims at; or the instruments to achieve a target cf. Knill and Tosun (2012, p. 6). 3  This second stage can also be subdivided into a policy formulation stage, in which actors work out the different policy options and decide on their favorites, and a separate decision-making stage, in which the actors adopt the course of action; see Howlett and Giest (2013, p. 17ff). 4  I will not elaborate on the different ways of interpreting and defining problems, which in turn affects whether they gain public attention and become the object of a policy debate. For a detailed description of this aspect, see Knill and Tosun (2012, pp. 99–102).

50

3  Methods and Cases

problem creates and their distribution among society (ibid., p. 94f.). Monetarization is the translation of the problem into nonmonetary or monetary terms and the discussion of whether money is part of the problem’s solution (ibid., p. 95f.). Scope of activity refers to the amount of activities that need to be regulated to solve the problem, i.e., the problem’s causes (ibid., p. 96f.). Interdependencies finally are an indication of the cross-sectoral character of policy problems, i.e., when a problem concerns not only one, but several policy sectors5 (ibid., p. 98f.). Metz and Ingold, in their study on policy instruments regulating micro-­pollutants, condense this exhaustive list to four concise problem characteristics: causation, prevalence, effects, and scales (Metz and Ingold 2014, p. 1998). Causation refers to the factors and actors that cause the problem. Prevalence reflects the extent of the causes, their levels—from local to global— and whether or not they are seasonal. Effects point out the objects the problem affects. Scale asks at which level the effects occur—from the local to the global. I apply Metz and Ingold’s (2014) characteristics to categorize the research’s CPR problem as a policy problem. The causes of micro-pollutants are manifold. Chemical substances occur in many areas of human life. In the medical sector, humans use pharmaceuticals, hormones, and contrast agents. In the industrial sector, dyes, solvents, plasticizers, and alike are utilized. In the food production sector, pharmaceuticals are given to livestock, and pesticides, insecticides, and herbicides are applied to crops. In households, people use detergents and personal care products containing chemicals (cf. Metz and Ingold 2014, p. 2000; Lapworth et al. 2012).6 These substances enter the water cycle through two paths, the diffuse and the point-source entry path. The substances entering through diffuse sources are those washed away7 from areas that they have been applied to: pesticides and insecticides on fields, herbicides on facades and streets, residuals from dumpsites, and veterinarian pharmaceuticals in factory farming. Point sources, which mean identifiable spots where substances enter the water cycle, are rain water channels, industrial discharges, drainage systems from hospitals, and wastewater treatment plants for household and industrial sewage (Götz et al. 2010, p. 14f.; Lapworth et al. 2012, p. 289). The prevalence of micro-pollutants also depends on the seasons (Metz and Ingold 2014, p. 2003). Pesticides are spread on the fields in spring and fall; rainy weather in fall and winter increases the natural runoff of the substances.8 Geographically speaking, the Rhine catchment area is prone to a high prevalence of micro-pollutants: chemical industry plants in Basel, Mannheim, Ludwigshafen, and Leverkusen contribute to micro-pollutants in the surface water, as do the food industry, the metal and automobile production, the refineries, and biotechnology sites that

5  A policy sector is a topical domain in which the state is active to govern the activities of this domain. 6  For an overview of the different domains in which micro-pollutants appear and their different types therein, see Schluep et al. (2006, pp. 58-63). 7  By precipitation, mainly. 8  Interview N° 22.

3.1  A Public Policy Analysis

51

can be found along the Rhine. Urban agglomerations bring in their share as well (Metz and Ingold 2014, 2002f.). Micro-pollutants have been shown to have effects on the environment, but these need to be studied further (Pal et  al. 2010, p.  6062; Rivera-Utrilla et  al. 2013, p. 1270). The impact of micro-pollutants on humans remains uncertain (Kümmerer 2009, p. 2360). Studies showed that certain substances like diclofenac, ibuprofen, and sulfamethoxazole have ecotoxicological effects on the aquatic environment and fish (Pal et al. 2010, p. 6063). In a recent study, researchers found analgesic diclofenac and the pesticides diurone and diazinon to suppress photosynthesis in algae (Stamm et al. 2017, p. 91). Stamm et al. (2017, p. 91f.) further detected ecotoxicological effects in fish and benthic invertebrates. In the vicinity of wastewater plant dischargers, for instance, freshwater shrimps had a smaller population than elsewhere, which points to the negative impacts substances have either on the fertility of freshwater shrimps or on the mortality of their offspring. Sanderson et al. (2004, p. 36) documented the influence of β-blockers in reducing the fertility of a rainbow trout subtype. A study by Oaks et al. (2004, p. 631) proved the lethal effect of diclofenac on vultures after bioaccumulation of the substance in the food chain.9 Moreover, the metabolites of some substances have stronger bioactive effects than the substance itself, and the potential toxic synergetic effect of substances also needs to be considered (Kümmerer 2009, p. 2356, 2360; Rivera-Utrilla et al. 2013, p. 1270). Micro-pollutants take effect at several scales: at the local level within the regional ecosystem they are released to and at the regional level or even national level when they pass filter systems and appear in urban drinking water (Jones et  al. 2005, p. 164) and national surface and groundwater (Rivera-Utrilla et al. 2013, p. 1270). Yet another scale is reached through bioaccumulation of substances—when they compile and increase in concentration along the food chain (Metz and Ingold 2014, p. 2005; Oaks et al. 2004). Policy problems that have a great prevalence, various causes, and effects and that occur at several scales are especially difficult to solve, because the solutions have to address a variety of causes and involve a diversity of actor groups and their sectors. The solutions also need to approach different types of impacts at different levels. All of this adds to the problem’s complexity and makes it less governable (Lemos and Agrawal 2006, p. 308).10 Complex policy problems are also referred to as “wicked problems” (Allen 2013; Weber and Khademian 2008, p. 336).11 Wicked problems  The vultures had ingested the diclofenac with their food: dead domestic livestock that had been fed the pharmaceutical Oaks et al. (2004, p. 631). 10  Kirschke et  al. (2017) interviewed 65 experts for their study on the complexity of 37 waterrelated problems in Germany and found “(…) that point source problems tend to be complicated, whereas diffuse source problems are rather complicated to complex” (Kirschke et  al. 2017,  p. 545ff). 11  Wicked problems can be characterized as being unstructured, cross-cutting, and relentless. The first characteristic refers to the difficulty of detecting the different causes and effects of the problem as well as to the challenge of finding an adequate solution to the problem. The second hints at the diverse sectors, the various administrative levels and entities, and the different jurisdictions that 9

52

3  Methods and Cases

“(…) have no clear solutions, only temporary and imperfect resolutions (…)” (McGuire 2006, p. 34), and collaborative structures are needed for tackling them with policy instruments (ibid., p. 34).

3.1.3  The Solutions to a Policy Problem Policy problems can be tackled with policies. Policies are: (…) the outputs of a political system, i.e. the decisions, measures, programmes, strategies and courses of action adopted by the government or the legislature. (Knill and Tosun 2012, p. 4)

Hall (1993, p. 278) distinguishes policies in a broad sense as the goals of a political sector, the instruments selected to achieve these goals, and their specific combination. For this research, policies are defined as the instruments for tackling the CPR problem. The analysis of policy instruments and the reasons for adopting them have generated different policy instrument typologies. Lowi (1972) defines a detailed taxonomy of policies, which considers the degree of coercion a policy incorporates and the level the policy targets, the individual or the societal level (cf. Lowi 1972, p. 299f.). Lowi’s typology distinguishes: a) A distributive policy where the state incentivizes an individual to change the behavior by distributing him/her resources via a remotely coercive instrument b) A constituent policy that alters the state institutions, thus working at the societal level with a low level of coercion c) A regulative policy which “(…) specif[ies] conditions and constraints for individual or collective behavior (…)” (Knill and Tosun 2012, p.  16) and is coercive d) A redistributive policy which is coercive as well, shifting resources from one societal group to another, thus acting on the societal level (Knill and Tosun 2012, p. 16; Lowi 1972, p. 300)12 Hood (1986, p.  124f.) classes policy instruments according to their type— whether they are information (nodality), regulation (authority), financial mechanisms (treasure), or of administrative nature (organization)—and according to their objective, whether they observe the addressees’ behavior or whether they aim to change their behavior. Wilson (1989) offers a much simpler typology, distinguish-

are affected by the problem. The last characteristic implies that solutions to the problem have consequences for other policy domains and that the problem cannot be solved completely (Weber and Khademian 2008, p. 336f.). 12  For an overview of the development of the study on policy instruments and their different taxonomies, see Bressers and O’Toole (1998, p. 217f.), Howlett (2005), and Linder and Peters (1989, pp. 39-41).

3.1  A Public Policy Analysis

53

ing policies according to the costs and the benefits they create. Linder and Peters (1989, p. 44) synthesize seven instrument classes that focus on the type of mechanism the instrument represents, differentiating direct provision, subsidy, tax, contract, authority, regulation, and exhortation. Peters (2000, p. 39f.) proposes to assess instruments according to seven qualitative dimensions: their degree of directness regarding their influence, visibility, intensity, degree of automaticity, degree of contingency, degree of coercion, and degree of restriction. Salamon (2001) identifies five criteria to evaluate instruments’ effects.13 Based on these criteria, he derives the following instrument dimensions: degree of coerciveness, directness, automaticity, and visibility (Salamon 2001, pp. 1650–1669). Bressers and O’Toole’s (1998) set of instrument characteristics considers an instrument’s impact on the instrument’s addressees, namely, whether: –– “A policy instrument involves the provision or withdrawal of resources to a target group.” –– The target group has the “freedom of choice to apply the instrument” or not to. –– The instrument addresses bilaterally one target group or multilaterally several ones. –– The instrument incorporates a normative appeal. –– The political system’s reaction to the addressee’s behavior and this addressee’s behavior are proportional. –– The role of policy makers in the implementation process is of specific concern (Bressers and O’Toole 1998, p. 224f.). Howlett (2000, 2005) followed up on the aspect of the instruments’ influence on their addressees. His typology assigns instruments according to their coerciveness executed by the state. Howlett distinguishes substantive policy instruments, which influence the provision and the quality and quantity of societal public goods (Howlett 2000, p.  415), and procedural instruments, which modify the policy-­ making process (ibid., p. 420). The scale of instrument coerciveness ranges from voluntary, thus low coercion by the state, to compulsory, i.e., high coercion (Howlett 2005, p. 38f.). This scale of coercion can be looked at with the threefold classification by Etzioni (1975), which distinguishes three kinds of power. Coercive power means to (threaten to) apply physical punishment, control the satisfaction of physical needs, or be deprived of freedom. Remunerative power reflects the “control over material resources and rewards” (Etzioni 1975, p. 5). Normative power refers to the “manipulation of esteem” and positive values (ibid., p. 6). Understood as a means of control (Vedung 2010, p. 28), the three types of power can be translated into a typology of public policy instruments that distinguishes the state’s degree of coercion. Coercive power becomes the regulations, the “sticks” with which the state sanctions the addressees who do not conform to the instrument. Remunerative power is behind

13  These are effectiveness, efficiency, equity, manageability, and legitimacy with political feasibility (cf. Salamon 2001, pp. 1647–1650).

54

3  Methods and Cases

the economic means, the “carrots” that give or take material resources to incentivize addressees to change their behavior. Information stand for the normative power, the “sermons” that ought to persuade the addressees to change their behavior (ibid., p. 29ff). Policy instruments for tackling the policy problem of micro-pollutants in surface water can also be classified as sticks, carrots, and sermons. Metz and Ingold (2014) claim that in order to find adequate solutions to a policy problem one has to relate them to the specific characteristics of the policy problem. They distinguish source-­ directed and end-of-pipe instruments for the policy problem of micro-pollutants in surface water (Metz and Ingold 2014, p.  2000ff). Source-directed policy instruments address the problem before it arises. In the case of micro-pollutants, these measures can comprise substance bans, product charges, disposal requirements, emission limits, and best environmental practices, for instance. End-of-pipe instruments aim at diminishing the problem once it exists. End-of-pipe measures for micro-pollutants are, i.a., emission charges and limits, disposal requirements, subsidies, and best available techniques (ibid., p. 2006).

3.1.4  Policy Networks The actors who work out and implement these policies form so-called policy networks (Bressers and O’Toole 1998, p. 216). A policy network: (…) includes all actors involved in the formulation and implementation of a policy in a policy sector. They are characterized by predominantly informal interactions between public and private actors with distinctive, but interdependent interests, who strive to solve problems of collective action on a central, non-hierarchical level. (Börzel 1998, p. 260; emphasis in the original in italic)14

To achieve their common interest of solving a collective action problem, actors “exchange resources” (Börzel 1998, p. 254). In this context, Kenis and Schneider (1991) refer to policy resources—such as expertise on the problem’s different characteristics and the required tasks to sort them out—that are dispersed among the state and non-state actors. Policy networks function as vehicles that activate actors’ different capacities and expertise when needed in the policy-making process (Kenis and Schneider 1991, p.  41). One advantage of policy networks is that they can  Bressers and O’Toole (1998) have a narrower definition of policy networks that emphasizes the top-down hierarchy between state and non-state actors in the policy-making process. They describe policy networks as the “(…) pattern of relationship between a governmental authority (…) and the set of actors – the ‘target group’ – toward which the governmental authority’s policy efforts are directed (…)” (Bressers and O’Toole 1998, p. 215). Ansell and Gash’s (2008) definition highlights a decision-making process on equal footing. They understand policy networks as networks of state and non-state corporate actors that engage in informal cooperation to consult and take decisions Ansell and Gash (2008, 547f.). For a detailed description of the development, the different typologies, and the analysis of policy networks, see Börzel (1998), Kenis and Schneider (1991), Pappi and Henning (1998), and Thatcher (1998).

14

3.2  The Conceptual Framework Guiding the Research

55

t­ ransmit valuable policy resources not existent within the state apparatus to the state actors in need of it (Knill and Tosun 2012, p. 203). When analyzing the management process of the CPR problem of micro-­pollutants in the Rhine catchment area, I consider state and non-state actors. The latter are the polluters of the common-pool resource, the actors responsible for the provision of the common, and the ones affected or concerned by the CPR problem. These actors are key to the management process of a CPR problem. Polluters are the ones asked to change their pattern of resource use; service providers15 are affected by the CPR problem but can also be accountable for its maintenance and provision; affected actors can give valuable insights on the nature of the problem and possible solutions to it. Since the research’s definition of the management process is the adoption and implementation stage of policy instruments and since the study focuses on state and non-state actors involved in this stage, I can apply Börzel’s definition of policy networks to my research. I thus analyze policy networks in the management process of a CPR problem, assessing their degree of cooperation and the factors accounting for it. The next sub-chapter introduces the analytical framework with which I align the analysis.

3.2  The Conceptual Framework Guiding the Research The social-ecological system framework (SESF) is the conceptual tool I employ for observing the social-ecological system (SES) under study: the human influence of settlements, agriculture, and pharmaceuticals (the social system), which cause micro-pollutants, on the quality of river surface water (the ecological system); and the social mechanisms, management and cooperation, and products of social interactions, rules and instruments, that ought to ameliorate the uses of the resource surface water and the ecosystem it is a part of. The SESF allows to identify the parts of the SES relevant for the analysis. I situate the discussed elements of policy analysis within the SESF: the adoption and implementation stage of policy instruments; the CPR problem as a policy problem; the potential solution to the problem, i.e., the policy instruments; and the policy network of state and non-state actors. The SESF structures the analytical focus on the environmental problem and on the interactions of the actors that manage it. The framework provides characteristics of the SES that contextualize the dependent, independent, and control variables within this system. The framework’s SES characteristics furthermore guide the case study selection and the actor identification process.

15

 For example, water treatment plant providers.

56

3  Methods and Cases

3.2.1  T  he Social-Ecological System Framework (SESF) and Its Critique The social-ecological system framework (SESF) sharpens the focus on the institutional and noninstitutional factors that shape actors’ interactions. The SESF takes up the IAD’s contextual factors and gives them more weight. It can be interpreted as an amendment to the IAD, which defines the “external” factors influencing the processes of actors’ interactions in the action situation (Cox 2014, p. 312). The SESF allows framing the social and ecological parts of a CPR problem in analytical categories. It consists of the already known action situation and four so-called first-tier components. These components represent the main elements of the social and the ecological system and describe the “external” factors that influence the action situation. Figure 3.1 shows the graph of the SESF and its first-tier components. On one side, representing the ecological system, are the resource systems (RS) and the resource units (RU). The two elements are interlinked, the RUs being part of the respective RS of an analysis. On the other side, the governance systems (GS) and the actors (A) stand for the social system. The connection between them indicates that GS define and set the rules that A behave upon (Basurto et  al. 2013, p. 1367; Ostrom 2009). There can be multiple instances of each of these elements in one social-ecological system (McGinnis and Ostrom 2014). The four elements all influence the action situation. The resource system (RS) and the governance system (GS) set the conditions for the focal action situation; they constitute the context of the SES.

Fig. 3.1  The graph of the SESF and its first-tier components (McGinnis and Ostrom 2014)

3.2  The Conceptual Framework Guiding the Research

57

The RS summarizes the ecological system under study. The GS stands for the policy area in which the environmental problem is addressed; for the regime type in which this policy area exists; for the institutions, that is, the sets of legal rules, that structure the political system and guide its actions (Knill and Tosun 2012, p. 4, 41; McGinnis and Ostrom 2014); and for the organizations that produce these rules. From a political scientific perspective, the governance system of a SES can thus be read as representing the subdiscipline polity. Polity reflects “the institutional structures characterizing a political system” (Knill and Tosun 2012, p. 4). The resource units give the situational input to the action situation and substantiate the environmental problem at stake. The CPR and policy problem lies in the actors’ different types of water use and specifically in the one type that uses surface water to discharge micro-pollutants. The actors participate and interact in the focal action situation. The state and non-state actors that interact in the CPR management process form a policy network. Relying on Henry and Vollan’s (2012) argument that a “commons governance system” is an action arena (Henry and Vollan 2012, p. 131), I claim actors’ CPR management process to be the action situation of the SES under study. The policy network of actors becomes thus part of the action situation. As the CPR management process comprises the adoption and implementation of measures tackling the CPR problem, actors can be interpreted as representing politics, that is, the policy process (Knill and Tosun 2012, p. 4). Actors’ interactions (I) within the action situation (AS) lead to social or ecological performance measures such as sustainability or overharvesting of the given resource or accountability of actors’ performance. These measures are labeled as outcomes (O) in the SESF (McGinnis and Ostrom 2014). Analyzing cooperation within a CPR management process, I define that cooperation is the outcome of actors’ interactions in the action situation. This cooperation facilitates actors’ CPR management process. The CPR management process being the adoption and implementation of policy instruments regarding the CPR problem, the action situation reflects the policy, the outputs of the political system (Knill and Tosun 2012, p. 4). The outcomes (O) produced in an action situation influence the governance system, the resource system, the resource unit, and the actors themselves. The action situation’s outcomes thus feedback to the social ecological system’s elements (McGinnis and Ostrom 2014). Figure 3.2 summarizes these observations in an illustration of the SESF’s application to the present research. The framework offers a variety of so-called second-tier variables of its main elements. These variables describe the characteristics of the first-tier components and allow the researcher to interpret the object of study in the social-ecological system at stake (McGinnis and Ostrom 2014). To avoid confusion of the different uses of the term “variable,” I call these second-tier variables “second-tier characteristics” throughout the research.16

16

 For a list of the SESF’s second-tier characteristics, see Table 2 in Annex III.

58

3  Methods and Cases

RS: Rhine sub-catchment provides the ecological context

GS: the policy subsystem with its institutions (legal rules)—polity

I Interactions in the policy network lead to O Outcome cooperation AS: Actors’ CPR management process, the adoption of political guidelines and laws & implementation of policy instruments—policy RU: type of surface water use creates the CPR & policy problem

A: state & non-state actors form a policy network—politics

Fig. 3.2  Illustration of the SESF’s application to the present research

Regarding the research’s focus on diminished water quality, the SESF is the appropriate analytical tool, as it has been: designed primarily for the study of common pool resources problems [and as it] provides useful insights for the analysis of the provision of environmental services presenting public goods characteristics such as the restoration/maintenance of water quality. (Amblard 2012, p. 9)

I select the SESF because of the balanced relevancy it attributes to social institutions and ecological factors for studying actors’ interactions in a social-ecological system. I consider the polluted common surface water and the social actors involved in the problem to be part of the same system, a “social-ecological system” (SES). The term SES itself stresses the connectedness of the two spheres (cf. Folke et al. 2005, p. 444). Through the SES analysis along the SES framework, I can define the dynamics and features of the environmental problem of micro-pollutants in surface water and the resource system this environmental problem is placed in. I can further name the different components that form the governance system delivering the conditions for the actors’ action situation, the characteristics of the actors that take part in the CPR problem setting, and the characteristics of the action situation that constitute cooperation. The SES analysis serves the purpose of determining: –– The spatial extent of the social system and the environmental problem forming the case study regions’ spatial area –– The jurisdictions and policy instruments of the case studies regarding the environmental problem –– The actors and their potential characteristics

3.2  The Conceptual Framework Guiding the Research

59

–– The constituting parts and localization of the dependent and independent variables within the social-ecological system under study The social-ecological system framework is a helpful analytical tool to structure research on social-ecological systems and to define the different parts of an SES under study. However, the framework has several shortcomings. A first flaw of the SESF lies in the vagueness of the meaning of several second-­ tier characteristics. The framework’s first-tier elements are well described and comprehensible, some of its second-tier variables lack precision in their meaning and purpose. “Equilibrium properties” (McGinnis and Ostrom 2014), for example, a characteristic of the resource system (RS6), could mean anything between the balance of an ecosystem and the chemical composition of a natural resource—two juxtaposed poles on a scale that ranges from the micro to the macro level. “Distinctive characteristics” is a similarly vague feature of the resource unit (RU6) that could signify literally anything about a natural resource. “Knowledge of SES/mental models” belonging to the element Actor (A7) is yet another characteristic that can be interpreted in any way that would seem to suit the specific context best. The framework offers the freedom to apply the characteristics in the manner most convenient for the respective analysis. However, this freedom reduces the comparability of studies that apply the SESF, and it impedes reliability checks. The same applies to the feedback mechanisms that lead from the action situation to the first-tier variables. They lack precision. Furthermore, it is not defined how an analysis of several of the first-tier elements should be carried out. As soon as there are two or more of each of the four first-tier elements, their second-tier characteristics double (or triple) as well. This could lead to a confusing rather than illuminating analysis and description of the SES. There is also no guidance on how to apply the SESF in a longitudinal study. The framework is a tool for structuring analyses; it cannot provide insights on relational dependencies and influences within a SES. Nevertheless, it could form the basis for grounded theory. If researchers identified a certain pattern between specific elements across a variety of similar SES, they could make theoretical assumptions based on their observations. The second-tier characteristics I selected for the analysis are precisely defined and justified. I extend the framework’s field of application by using it to identify the case studies and the actors in the case study regions.

3.2.2  Conceptualizing the Variables To show how the SESF can be complemented with theory, I actively assign the independent, the dependent and the control variables to the appropriate parts of the framework, that is, I conceptualize them within the context of the social-ecological system under study.

60

3  Methods and Cases

3.2.2.1  The Dependent Variable Cooperation From the perspective of the SESF, I understand cooperation as the outcome (O) of actors’ interactions (I) (McGinnis and Ostrom 2014) in the specific action situation “management process of the CPR problem micro-pollutants in surface water.” Actors’ coordination of actions and their resource exchange happen within this management process. I define actors’ joint goal as actors’ similar understanding of the management process goal. Based on the characteristics and pattern of actors’ interactions and based on actors’ goals, I can claim the existence or absence of cooperation within such a management process. Originally, the SESF’s most important types of interactions (I) and outcomes (O) were the process of resource extraction and resource maintenance (McGinnis and Ostrom 2014). Today, the SESF lists ten second-tier characteristics under the element interactions (ibid.). I consider five as relevant for the research.17 I interpret the characteristic harvesting (I1) as using surface water. It reflects the CPR problem itself: the use of surface water for different purposes, one of them being resource pollution, leads to a CPR over-appropriation problem and to possible conflicts between users. I choose information sharing (I2) as it a) leads to learning and knowledge generation, which is one of the benefits of cooperation (Fleishman 2013), and b) represents a type of resource exchange, which is one of the constituting indicators of cooperation (Sadoff and Grey 2005). Conflicts (I4) reflect the CPR problem, the possible conflict(s) between users. The characteristic of self-­organizing activities (I7) hints at actors’ impetus to coordinate their actions. I define this coordination as actors’ collaboration on the topic of micro-pollutants. Collaboration is defined as actors’ working together that the actors benefit from. Since collaboration is similar to—although not the same as—cooperation in that it is a working together that actors benefit from (Huxham 1993, p.  603; O’Leary and Vij 2012, p.  510; Oxford Living Dictionaries 2018a; Oxford Living Dictionaries 2018b; West et al. 2007, p. 416), I use actors’ collaboration as a proxy for actors’ cooperation in the inferential network analysis. I interpret evaluative activities (I10) as referring to actors’ overarching goal, representing the third indicator of cooperation. The SESF also distinguishes the different outcomes of the interactions: the social performance measure (O1), the ecological performance measure (O2), and externalities to other social-ecological frameworks (O3) (McGinnis and Ostrom 2014). I capture the research’s specific outcome cooperation under the second-tier characteristic social performance measure. This argumentation is supported by the IAD’s conceptualization of outcomes: apart from new rules or guidelines that actors have decided upon, the IAD also defines actors’ behavioral change as one possible outcome of an action situation. Claiming that cooperation evolves among actors who participate in an action situation implies that cooperation has not existed before. 17  The ten 2nd-tier characteristics are I1 harvesting, I2 information sharing, I3 deliberation processes, I4 conflicts, I5 investment activities, I6 lobbying activities, I7 self-organizing activities, I8 networking activities, I9 monitoring activities, and I10 evaluative activities; see McGinnis and Ostrom (2014); see also Table 2 in Annex III.

3.2  The Conceptual Framework Guiding the Research

61

Table 3.1  The research’s action situation with its 1st-tier and 2nd-tier characteristics Action situation 1st tier 2nd tier

Management process of the CPR problem of micro-pollutants in Rhine surface water Interactions (I) Outcome (O) = DV I 1 Harvesting Using surface O 1 Social performance Cooperation water measure I 2 Information Exchanging sharing resources I 4 Conflicts Having a user conflict I 7 Self-organizing Collaborating activities I 10 Evaluative Sharing the same activities goal

Cooperation can thus be considered a sign of behavioral change—from noncooperation to cooperation—and subsequently be defined as an outcome of the action situation. Table 3.1 summarizes the studied action situation “management process of micro-pollutants in Rhine surface water” with its different types of actors’ interactions and with the dependent variable cooperation being an outcome of the three interactions that constitute cooperation conceptually: sharing the same goal, exchanging resources, and collaborating. 3.2.2.2  The Independent Variables The independent variables are actors’ problem perception (IV 1a), actors’ similar problem perception (IV 1b), actors’ participation in forums (IV 2a), actors’ co-­ participation in forums (IV 2b), and actors’ shared belief (IV 3). Independent variables 1a and 1b reflect actors’ perception of the CPR problem micro-pollutants. IV 1a comprises each actor’s perception of the environmental issue as rather problematic or less urgent; independent variable 1b reflects whether two actors have a similar perception of the environmental problem, that is, whether they perceive micro-pollutants as problematic in the same way. I categorize these two independent variables within the social-ecological system under study as follows: the research’s resource systems (RS) are the Rhine sub-­ catchment areas that serve as case studies. Their characteristics productivity of system (RS5) and predictability of system dynamics (RS7) refer to the CPR problem (McGinnis and Ostrom 2014). The first could be interpreted as the production or regeneration rate of a system’s resource stock. I construe it as surface water’s natural regeneration rate, i.e., the natural process of micro-pollutant decomposition after a certain amount of time. Fleischmann et al. (2014) depict RS5 as resource pollution, stating that “for pollutants, productivity can be interpreted as the rate at which pollutants are released into the environment” (Fleischmann et al. 2014, p. 447). I include this perspective into my interpretation of RS5 productivity of system, as the

62

3  Methods and Cases

regeneration rate also tells how much micro-pollutants remain in the surface water. Characteristic RS7 complements this CPR problem description. Predictability of system dynamics can be read as the distribution and reach of micro-pollutants in the sub-catchment area. These two SESF characteristics thus inform the intensity of the CPR problem at stake—which in turn relates to actors’ perception thereof (IVs 1a and 1b). I define the research’s resource unit (RU) as type of surface water use. Several of the RU’s characteristics objectify the pollutants’ threat to the resource surface water. Mobility (RU1) adds to RS7 in that it represents micro-pollutants’ mobility in surface water, which is informed by their spatial and temporal distribution in water (RU7). Growth and replacement rate (RU2) inform the productivity of system (RS5), while interaction among RUs (RU3) can be interpreted as the disturbance of one type of water use, i.e., drinking water generation, by another, i.e., the contamination of water with micro-pollutants. Distinctive characteristics (RU6) can reflect the different concentrations of micro-pollutants in surface water. The characteristic importance of resource (A8) comprehends IV 1a and b since a threat to a resource weighs more when an actor depends on this resource. Actors’ participation and actors’ co-participation in forums (IVs 2a & b) describe the number of forums actors participate in with regard to micro-pollutant management and the number of forums two actors attend together. Within the SESF, the two variables find a match in the “actor” characteristic norms/social capital (A6) if we consider forums a social structure in which actors act and that is part of their social capital (Coleman 1988, 98). Actors sharing the same belief (IV 3) mean that two actors have the same opinion about essential societal values. The characteristic norms (A6) refer to actors’ belief systems, since beliefs can be understood as norms that actors adhere to. 3.2.2.3  The Control Variables When testing for significant relations between variables, it is crucial to check the influence of intervening variables. In this analysis, I include the following control variables. Actor has a regulatory power (CV 1) is the first control variable. Actors who have no decision-making power but want to influence the decision-making process provide actors with decision-making power—like state actors—with policy goods. Non-decision-makers offer expertise and political information to decision-makers in order to gain access to the decision-making processes (Beyers and Braun 2014, p.  95f.) and to influence decision-makers’ policy preferences (Stokman and Zeggelink 1996, p. 80). Non-decision-makers try to obtain the resources necessary to access the venues in which decisions and regulations are made (Baumgartner and Leech 2001, p. 1197; Beyers and Braun 2014, p. 93f.). As Ingold and Leifeld found in their study on influence reputation in different policy networks: (…) decision-makers find one another particularly important (a homophily effect) but (…) decision-makers are also judged to be more influential than nondecision-makers by nondecision-­makers. (Ingold and Leifeld 2014, p. 17)

3.2  The Conceptual Framework Guiding the Research

63

Formal decision-making power thus attracts actors who try to influence the policy-­making process (Beyers and Braun 2014, p. 93; Ingold and Leifeld 2014). Decision-­makers—defined here as actors with regulatory power—tend to receive ties from those actors who try to influence the decision-making process (Ingold and Fischer 2014, p. 89). I employ the dummy control variable actor with regulatory power to control whether these actors are sought out more for collaboration than actors without regulatory power. An actor who implements policy measures (CV 2)—that is, who implements instruments—serves as control variable opposed to the regulatory power. Being an implementer does not apply to all non-decision-makers: actors in the management process might be neither decision makers nor implementers—such as consumer organizations, NGOs, or certain associations of the water supply industry. The second control variable checks whether actors who implement measures regarding a CPR problem play a specific role in activating collaboration in a CPR problem setting. Actors’ positions within the management process as regulators and/or implementers are reflected in actor’s socio-economic attributes (A2) and the rules-in-use (GS6). Socioeconomic attributes can be interpreted precisely as an actor’s role as a regulator or implementer, while rules-in-use represent the institutions that assign these particular roles and related tasks to actors. I also control for actors’ importance in the management process. Reputation (CV 3) reveals whether an actor who is part of a defined social system is seen by the other actors of this particular social system as fulfilling an important role for the system, as, e.g., influencing the outcomes of a policy process (Abu-Laban 1965, p.  35f.; Knoke 1998, p. 508; Scott 2000, p. 56). An actor’s reputation comprises the beliefs other actors hold about his/her role, capacity, and duties (Carpenter 2010, p. 45).18 Studies have shown that actors who entertain links, be they collaborational or communicational, are thereby enabled to rate each other as important (Fischer and Sciarini 2013, p. 10; Heaney 2014, p. 78f.). Controlling for actors’ reputation in this study sheds light on how actors’ importance as perceived by others relates to their cooperational behavior. Scott and Christopoulos (2018, p. 14), for instance, showed in their study on bank policy networks that well-reputed organizations in a leadership position are more likely to receive collaboration ties. An SESF characteristic capturing this aspect is norms/social capital (A6) if we interpret the importance that is attributed to actors as one element of their social capital (cf. Coleman 1988, p. 104f.). One factor that could also account for cooperation in CPR settings is appropriators’ sensitivity regarding the resource’s quality: actors’ pollution-sensitive water use (CV 4). Elinor Ostrom lists “dependence on the resource system” as one of the crucial variables that account for an increased likelihood of users’ self-organization (Ostrom 2000b, p. 40). This self-organization is assumed to lead to long-term cooperation among the appropriators. If appropriators are independent of the resource,

 Carpenter (2010, p. 45ff) distinguishes performative reputation, moral reputation, technical reputation, and legal-procedural reputation that organizations can be attributed with.

18

64

3  Methods and Cases

the CPR problem does not affect them, and they are indifferent towards coordinated action to solve it (Schlager 2004, p. 151f.). The assumption is thus that if an actor is dependent on the CPR, this actor is affected by its problem. The actor is likely to cooperate. I apply this aspect of resource dependence by considering those actors of the management process of micro-pollutants who have a pollution-sensitive water use. They depend on good water quality for their activities (e.g., drinking water or fishing), and they are directly affected by the CPR problem of micro-pollutants in surface water. I suppose that these actors tend to connect with each other to work on a solution of the CPR problem. For this control variable, the SESF “actor” characteristic importance of resource (A8) applies: it reflects actors’ dependence on the resource surface water and transmits the resource’s diminished usefulness for actors once it is polluted. The control variable actors’ territoriality (CV 5) is based on the homophily effect. This term means “that a contact between similar people occurs at a higher rate than among dissimilar people” (McPherson et al. 2001, p. 416). Two of the case studies lie in cross-border regions, with actors coming from different nation states. The effect I measure with this control variable is whether actors with the same nationality tend to collaborate more with each other than with actors of another nationality. The SESF characteristic fitting this control variable is actor’s location (A4). I also control for network effects (CV 6), as structures inherent in the actor network may apply for cooperation. Networks enable or constrain interaction between actors, thereby shaping and influencing actors’ exchange pattern and the evolution of their cooperation (Henry and Vollan 2012, p.  131).19 The network effects are, however, not reflected by any of the SESF 2nd-tier characteristics. Table 3.2 summarizes the dependent, independent, and control variables and their characterization by the SESF. The next chapter defines the case study design, presents the case study selection criteria, and presents the three case studies.

3.3  Case Study Design and Case Studies To answer the research question, I use a case study method. It is the “preferred strategy (…) when “why” questions are being posed (…) and when the focus is on a contemporary phenomenon within some real-life context” (Yin 2003, p. 1). Gerring (2004) defines a case study as: an intensive study of a single unit for the purpose of understanding a larger class of (similar) units. A unit connotes a spatially bounded phenomenon (…) observed at a single point in time (…). (Gerring 2004, p. 342; emphasis in the original in italic)

The unit one studies thus represents the phenomenon a researcher analyzes (cf. Gerring 2004, p. 344); or, as Bennet (2004) puts it, the unit is an “instance[s] of a 19

 For a detailed description of the considered network effects, see Sects. 3.4.4.3 and 3.5.2.

3.3  Case Study Design and Case Studies

65

Table 3.2  The study’s dependent, independent, and control variables

N° DV a b c IV 1 a & b

Variable Aiming at the same goal Coordinating actions Exchanging resources Actors’ CPR problem perception & actors’ similar CPR problem perception

2 a & b Actors’ participation & co-participation in one or more forum(s) 3 Actors sharing the same belief CV 1 Actor having regulatory power 2 Actor being an implementer 3 4 5 6.1-­6.6

Actors’ reputation Actors’ pollution-sensitive water use Actors’ territoriality Network effects

Social-ecological system framework 2nd-tier characteristics I 10 Evaluative activities I 7 Self-organizing activities I 2 Information sharing RS5 Productivity of system RS7 Predictability of system dynamics RU1 Mobility RU2 Growth and replacement rate RU3 Interaction among RUs RU6 Distinctive characteristics RU7 Spatial and temporal distribution A 8 Importance of resource A 6 Norms/social capital

A 2 Socio-economic attributes GS 6 Rules-in-use A 6 Norms/social capital A 8 Importance of resource A 4 Location –

class of events of interest to the investigator” (Bennet 2004, p. 20f.). For this study, this unit or class of an event—the case study—is actors’ management process of the CPR problem micro-pollutants in surface water. In the following, I present the case study design which is followed by the case study selection criteria and the introduction of the case studies themselves.

3.3.1  The Case Study Design: A Mixed-Method Study Case studies allow “to measure in a case the indicators that best represent the theoretical concept we intend to measure” (Bennet 2004, p. 34). I conducted a survey to collect the data needed for the analysis (see Sect. 3.4.4). As Yin states, “for some questions, a choice among [research] strategies might actually exist” (Yin 2003, p. 7), doing a “survey within a case study” being such a choice (ibid., p. 9). Within the cases, I analyze two units of analysis (UoA) (Yin 2003, p. 42f.).

66

3  Methods and Cases

Each unit of analysis represents a different analytical perspective on the dependent variable cooperation. On the network level, I assess the three constituting elements of cooperation (UoA 1): actors’ same goal; actors’ coordination of actions; and actors’ resource exchange. I describe actors’ coordination more thoroughly by breaking open the overall network perspective, focusing on the macro-, meso-, and micro-level of the case studies’ collaboration networks which stand for actors’ coordination. On the dyadic level, I examine the factors related to the existence of a dyad between actors in the collaboration networks (UoA 2). I call this relational tie shared by two network actors a collaboration tie. On this level, I consider factors that enhance the likelihood that such a tie between two actors exists. The inferential network analysis technique of exponential random graph models (ERGM) provides the test method. The two units of analysis in each case study make the research design an embedded case study design (Cox 2015; Yin 2003, p. 42f.). I compare the cases’ units of analysis on a cross-case level: I contrast the intensity and development stage of cooperation in the three case studies and I discuss the ERGM results and the explanatory factors across the cases. The cross-case analytical level reflects a methodological technique: the comparison of the analyses’ results and the reasoning about their causes across the case studies enables me to juxtapose differences and similarities of the case studies and their networks. This analytical procedure allows to interpret the ERGMs and understand from a broader perspective under which circumstances cooperation in a CPR problem situation evolves. In a mixed-method study that combines a single-unit with an across-unit analysis (cf. Gerring 2004, p. 344), I statistically analyze the two units of analysis within each case, and I qualitatively compare the units of analyses across the cases (Gerring 2004, p. 343). My case study method is a hierarchical across- and within-unit analysis, because I (a) do not consider time and (b) assume to find variation within and across the units (ibid., p. 343). Through the single-unit analyses, I am able to assess and understand the units of analysis in depth; the across-unit analysis allows me to check the validity of the inferences found in each of the single units (Gerring 2004, p. 347f.). The case study design is furthermore multiple, since I examine three case studies of CPR problem management in sub-catchment areas of the Rhine basin. The study is covariational in that I assume a causal relationship between the independent variables and the dependent variable (Gerring 2004, p.  342f.). The three single case studies can “elucidate[…] causal mechanisms” (ibid., p.  349) between the ­independent variables and the dependent variable. A causal mechanism exists if a correlation between two variables is plausible (cf. ibid., p. 348), that is, if the connection between the two is meaningful. The study is covariational in that I expect variation on the observations of the independent variables across the cases (cf. Anckar 2008, p.  395). The different observations “can sustain causal inference” (King et al. 1994, p. 208). I expect variations of the independent variables, as the cases are similar but not identical (cf. Gerring 2004, p. 351). Figure 3.3 is inspired by Yin’s depiction of the basic types of case study designs. It illustrates the research’s multiple, embedded, and covariational mixed-method case study design.

3.3  Case Study Design and Case Studies

67

Context: CPR problem of micro-pollutants in surface water in the Rhine catchment area

SINGLE-UNIT ANALYSIS—IN EACH CASE STUDY Case study 1: Mgmt. process Case study 2: Mgmt. process Case study 3: Mgmt. process in the Rhine basin at Basel in the Ruhr basin in the Moselle basin Units of analysis:

Units of analysis:

Units of analysis:

UoA 1 Network level— descriptive SNA: cooperation’s constituting elements

UoA 1 Network level— descriptive SNA: cooperation’s constituting elements

UoA 1 Network level— descriptive SNA: cooperation’s constituting elements

UoA 2 Dyadic level— ERGM: collaboration tie between two actors

UoA 2 Dyadic level— ERGM: collaboration tie between two actors

UoA 2 Dyadic level— ERGM: collaboration tie between two actors

ACROSS-UNIT ANALYSIS—CASE COMPARISON Comparison I: Cooperation across the case studies Comparison II: Factors enhancing cooperation across the case [own depiction, inspired by Yin (2003, p. 40)]

Fig. 3.3  Illustration of the book’s case study method design [own depiction, inspired by Yin (2003, p. 40)]

The inferential network analysis applied on the dyadic level does not enable me to assess causal effects between the explaining and the dependent variables. However, the technique allows to make suggestions about possible correlations between the variables. The contextual qualitative in-depth analysis of each case and the cases’ subsequent comparison furthermore enable me to reveal the variables’ potential causal mechanisms. As Gerring states, single case studies can identify causal effects, but cross-case evidence can also reveal causal mechanisms (Gerring 2004, p. 349). The cross-case variation on the independent variables, which I expect, is necessary for assessing causal effects through a case comparison (ibid., p. 348). By comparing the cases, I can generate general statements about the social phenomenon analyzed (Garrick et  al. 2013, p.  7; Kannonier-Finster 1998, p.  57). Generalization of the findings is important as it enables the researcher to make “inferences that go beyond the particular observations observed” (King et al. 1994, p. 8). It also allows judgments about “which phenomena are “more” or “less” alike in degree (…) or in kind (…)” (ibid., p. 5).

68

3  Methods and Cases

To ensure the inference of causal relations between the independent variables and the dependent variable, I apply the constant effect and the unit homogeneity assumption to my cases. Unit homogeneity (cf. King et al. 1994, p. 92) is the expectation that “all units with the same value of the explanatory variables have the same expected value of the dependent variable” (ibid., p. 91). It is the basis for comparative case studies, in which the researcher assumes that differences in the dependent variables’ values are due to the differences in the independent variables’ values (cf. ibid., p. 93). This criterion is difficult to accomplish in social science research (cf. King et al. 1994, p. 95) the reason for why one needs to be aware of the uncertainty about causal inference within this research. The constant effect assumption is a weaker version of unit homogeneity and states that an observed causal effect is constant across cases (ibid., p. 92f.). A further criterion is the replication logic. A selected case should either foresee differing results for predictable reasons or predict similar results. The first condition refers to a theoretical replication, the second to a literal replication (cf. Yin 2003, p. 47). I use the cases for a variation of the theoretical replication: I assume a certain variance on the independent variables and on the dependent variable—with differing results—across the cases. The comparison across the cases gives me insights on the independent variables’ effects on the dependent variable. I do justice to cases’ conditional independence by ensuring that the “values of the explanatory variables are not caused by the dependent variables” (King et al. 1994, p. 94). The question of generalizability of the results needs special consideration when selecting the case studies. It is essential to avoid interconnectedness of cases to allow for generalizability when comparing the cases. I ensure the independence of the cases by selecting sub-catchment areas of the Rhine basin that are hydrologically independent and geographically distant. This guarantees that actors are disconnected across the case studies and ensures that the case studies with their variables are separate from each other.

3.3.2  The Case Studies and Their Selection Criteria The Rhine catchment area is the empiric context in which I analyze the management process of the CPR problem micro-pollutants and actors’ cooperation therein. I choose the case studies based on the following selection criteria derived from the SESF’s elements. Each case study needs to: a. Lie in a sub-catchment of the Rhine (RS) b. Have a considerable concentration of micro-pollutants in the region’s river surface water and different forms of water use (RU) c. Have instruments or guidelines concerning micro-pollutants in place (GS) d. Have a group of distinctive actors (A) working on the management (AS) of the CPR problem I further consider certain chemical substances that have to appear in the case study resource system. Four groups of chemical compounds that cause micro-­

3.3  Case Study Design and Case Studies

69

Table 3.3  List of indicator compounds Substance Indicator group compound Pharmaceuticals Carbamazepine Sulfamethoxazole Herbicides Isoproturon S-Metolachlor

Biocides

Terbuthylazine Diurone

Carbendazim

Mecoprop

Scope of application Humans: anti-epileptics Humans: antibiotics Agriculture: applied to wheat, barley, and rye Agriculture: applied to corn, turnips, and beans Agriculture: applied to corn Settlements: facade protection Agriculture: applied to fruits and grapevines Settlements: facade protection Agriculture: applied to fruits and grapevines Settlements: applied to lawn Industry: material protection Agriculture: cereal protection

EU WFD priority substance – – yes – – yes





pollution can be distinguished. These groups represent four different types of use and application and therefore four different sources of chemical compounds: (1) pharmaceuticals, used by humans to treat diseases or chronic illness and entering the water circle mainly from households and through wastewater treatment plants; (2) herbicides, used in agriculture and applied to crops to protect them from rot and pest; (3) biocides, used in urban settings to avoid mold and moss at facades and which in some cases can also be applied to crops; and (4) chemicals used in production processes. For my research, I focus on the first three substance groups. Within these groups, I define so-called indicator compounds20 as listed in Table  3.3. Indicator compounds serve as representatives for the vast number of existing substances and are selected according to their: a. High application rate b. Presence in surface water in considerable concentration c. Persistence in drinking water recycling processes, such as riverbank filtration d. General persistence in surface water e. Estimated toxicity f. Prominence to public authorities g. Biochemical composition similar to other substances h. Possible applicability in other domains  I developed the selection criteria for the indicator compounds in close collaboration with the research group from the Swiss Federal Institute of Aquatic Science and Technology, Eawag, Switzerland, which was part of the research project “CrossWater—Transboundary Micropollution Regulation in Europe: The Definition of Appropriate Management Scales—An Interdisciplinary Approach.”

20

70

3  Methods and Cases

For the group of pharmaceuticals, the indicator substances are sulfamethoxazole, which is used in antibiotics, and carbamazepine, a substance used in anti-­ epileptics. Representatives for the group of herbicides are isoproturon, s-metolachlor, and terbuthylazine. Isoproturon is applied to wheat, barley, and rye and is listed as priority substance by the EU Water Framework Directive (WFD) (EU 2008). S-Metolachlor is used on corn, turnips, and beans. Terbuthylazine, a substance also applied to corn, substituted the EU-wide forbidden compound atrazine and is continuously increasing in consumption. The group of biocides is represented by diurone, carbendazim, and mecoprop. Diurone and carbendazim are used for facade protection and in agriculture, where they are applied to fruits and in wine-growing. Diurone is listed as a priority substance under the EU WFD as well (EU 2008). Mecoprop is employed on lawn in urban settlements and as material protection in industry. It can also be used as herbicide for cereal protection (Abegglen and Siegrist 2012, p.  19; Braun and Gälli 2014, p.  28; Wittmer et  al. 2014, p.  55, 59). The Swiss Federal Office for the Environment (FOEN) defined mecoprop as official indicator substance for herbicides like isoproturon, diurone, and carbendazim, as all these substances have a similar entry pattern (Abegglen and Siegrist 2012, p. 177). 3.3.2.1  The Three Case Studies The case studies deliver the paradigmatic conditions on which I test my theoretical arguments. They all comprise: –– A management process of micro-pollutants in a Rhine sub-catchment area—the case –– In which I expect to find a certain degree of cooperation—the first unit of analysis –– Upon which I test my hypotheses—the second unit of analysis Two cases comprise the management process of micro-pollutants in catchment areas of two Rhine tributaries, while the third management process takes place in a spatial area that lies within a Rhine sub-catchment area. In all three selected regions, at least one compound of each of the substance groups mentioned above has been detected (see also Table  3.8). The case studies are the management processes of micro-pollutants in the following regions: 1. The catchment area of the river Rhine in the urban area of the canton Basel City, Switzerland, belonging to the hydrological sub-catchment High Rhine 2. The catchment area of the river Ruhr in the German federal state North Rhine-­ Westphalia (NRW) 3. The catchment area of the Moselle on the territory of Luxembourg and the two German federal states Rhineland-Palatinate and Saarland Each case study region meets the criteria deduced from the SESF. Micropollutants have been detected in the river surface water and/or in the catchment

3.3  Case Study Design and Case Studies

71

area’s groundwater in each case study. In all cases, surface water is used for different pollution-­ sensitive water uses, namely, drinking water and fishing. Actors engaged in the CPR problem come from different sectors in each region (for more details on the actors, see Sect. 3.4.2). The governance systems of the three case studies comprise the policy area, the jurisdictional area in which the legal acts and instruments apply, and the political system in the case study regions, furthermore the rule-making organizations and the repertoire of norms and strategies in place, that is, the instruments. The case study regions are located on Swiss, German, and Luxembourgian territory. I therefore take into account the legislations regarding the management of micro-pollutants in the European Union (EU), in Luxembourg, Germany, and Switzerland, and their realization at the case studies’ regional levels. As Switzerland is not a member of the EU, it is not bound to EU law, while the Grand Duchy of Luxembourg and the Federal Republic of Germany (FRG) are (EU 2019a, b, c). The case studies are similar in that they belong to the same territorial region, Central Europe; they represent management processes of the CPR over-­appropriation problem of water pollution by micro-pollutants; and they comprise heterogeneous actor groups. They differ in that the management processes are at different development stages. The regulations regarding the treatment of micro-pollutants diverge across the cases. For instance, in the case of Basel, instruments have been worked out and are already implemented. In the Ruhr case, state authorities have developed an action plan to tackle the CPR problem and have started implementing measures. In the Moselle case, state authorities are starting to discuss and work out measures. The cases follow a sequential logic regarding the state of instrument adoption and implementation. The characteristic GS6, rules-in-use, accordingly varies across the cases. As the GS sets the conditions for the action situation—the management process of the CPR problem— I assume all cases to be in different “phases” of the management process. Consequently, I presume some variation of the independent and the dependent variables across the cases. Assuring variation on both types of variables allows me to infer meaningful findings across differing cases (King et al. 1994, p. 93). Figure 3.4 visualizes the SESF’s application for the three case studies in a simplified manner. 3.3.2.2  The SESF’s Characteristics Informing the Case Studies Each case study’s resource system is a sub-catchment of the Rhine basin. The resource system (RS) has nine second-tier characteristics (McGinnis and Ostrom 2014). I consider six of them, which sketch the CPR problem’s extent.21

 The nine 2nd-tier characteristics are RS1 sector, RS2 clarity of system boundaries, RS3 size of resource system, RS4 human-constructed facilities, RS5 productivity of system, RS6 equilibrium properties, RS7 predictability of system dynamics, RS8 storage characteristics, and RS9 location; see McGinnis and Ostrom (2014).

21

72

3  Methods and Cases

Larger eco-system: the river Rhine catchment area Case study 1: mgmt. process in the Rhine basin at Basel Rhine basin at Basel

Policy subsystem

Case study 2: mgmt. process in the Ruhr basin

Management of CPR problem

Ruhr basin

Water use

Policy subsystem

Actors

Water use

Management of CPR problem Actors

Case study 3: mgmt. process in the Moselle basin Policy subsystem

Moselle basin

Water use

Management of CPR problem Actors

Fig. 3.4  The SESF’s application for the three case studies

The characteristics sector (RS1) and clarity of system boundaries (RS2) set the analysis’ prerequisites. The sector defines the study’s CPR, i.e., surface water. Clarity of system boundaries is given: the system boundaries are the boundaries of the respective Rhine sub-catchment under study. Human-constructed facilities (RS4) are relevant for this research because they indicate the technologies and plants that eliminate micro-pollutants. They also hint at the facilities and systems (i.e., agriculture) that let the substances enter surface water. They stand for the CPR problem’s causes and problem-solving actors.22 The characteristic location (RS9) describes the geographical and hydrological region where the resource system under study is located. Productivity of system (RS5) and predictability of system dynamics (RS7) as discussed in Sect. 3.2.2.2 reflect the behavior of micro-­pollutants in surface water, that is, their concentration, their distribution within the element, and their decomposing rate. The two characteristics thus stand for the common-pool resource problem: the first representing the regeneration rate of the resource system, the second shows the distribution and reach of micro-pollutants in the sub-­catchment  The problem-solving actors meant here are drinking water providers that are able to filter the substances. This is but one possible solution to the CPR problem, referring to the CPR maintenance and thus supply-side provision. More solutions, applied at different levels and in distinctive sectors, are needed to solve the CPR problem of micro-pollutants; see Metz and Ingold (2014).

22

3.3  Case Study Design and Case Studies

73

Table 3.4  The resource system and its 2nd-tier characteristics Resource system RS1 Sector RS2 Clarity of system boundaries RS4 Human-constructed facilities

RS5 Productivity of system

RS7 Predictability of system dynamics RS9 Location

River Rhine sub-catchment area Surface water Sub-catchment boundaries  Technologies and plants eliminating micro-pollutants  Systems and facilities causing or discharging micro-pollutants  Amount of micro-pollutants in surface water  Natural processes decomposing micro-pollutants (regeneration rate) Distribution and reach of micro-pollutants in the sub-catchment Sub-catchment’s geographical and hydrological location

area. Table 3.4 summarizes the resource system’s second-tier characteristics chosen for analysis and their contribution to the research. I interpret the research’s resource unit (RU) as type of surface water use. The present study knows three types of resource use. In the case of pollution, surface water is used to discharge micro-pollutants into it. Purposes to release micro-­ pollutants into the environment are diverse. The first and main type of water use is caused in the three domains of agriculture, urban settlements, and the consumption of medicals. The second type is drinking water use. Here, appropriators take the resource surface water from the catchment area (RS) to process it to drinking water. The third type is surface water use for fishing—as fish are a species that is likely affected by micro-pollutants (AWWR and Ruhrverband 2016, p. 96). The following resource unit’s characteristics inform the research’s social-­ ecological system: Resource unit mobility (RU1) indicates the surface water flow. Based on Fleischmann et al.’s (2014) interpretation of mobility as “the spatial extent of spread of a pollutant” (Fleischmann et al. 2014, p. 447), I argue that, in this study, resource unit mobility represents the rate at which surface water contaminated with micro-pollutants spreads within surface water and flows downstream. The resource unit’s characteristic mobility stands for micro-pollutant movement within the surface water of the respective sub-catchment area under study. Growth and replacement rate (RU2) represents the continuous replacement of surface water with new surface water—i.e., no decrease of surface water. This is necessary to guarantee a continuous provision of drinking water and a habitat for fish. Regarding pollution, I refer again to Fleischmann et  al. (2014), who define “renewability [replacement rate] as the residence time of a pollutant” (Fleischmann et  al. 2014, p.  447). In this vein, replacement rate indicates the time a micro-­ pollutant needs to dissolve. This aspect points to the concentration and persistence of micro-pollutants and thus to the severity and intensity of threat they pose to the environment. Interaction among resource units (RU3) is relevant as it discusses the exchange of surface water between the different types of water use. RU3 highlights the confrontation of different types of surface water use due to micro-pollutants in surface water. The characteristic describes the CPR over-appropriation problem.

74

3  Methods and Cases

Table 3.5  The resource units and their 2nd-tier characteristics Type of surface water use Resource unit RU1 Mobility RU2 Growth and replacement rate RU3 Interaction among resource units RU6 Distinctive characteristics RU7 Spatial and temporal distribution

Drinking water generation Pollution through micro-pollutants and fishing Flow of surface water contaminated with micro-pollutants Persistence of micro-pollutants Replacement of surface water Flow of micro-pollutants into other area of surface water use Concentration of micro-pollutants in surface water Up- and downstream asymmetries

Infiltration of surface water by micro-pollutants Degree of surface water quality

Distinctive characteristics (RU6) is a vague term. I apply it to distinguish the different degrees of surface water quality and, in consequence, the different concentrations of micro-pollutants in surface water. RU6 hints at the CPR problem’s intensity. Similar to RU3, spatial and temporal distribution (RU7) is inherent to the resource surface water. Surface water runs from one point in space to another in a certain amount of time. It is one-directional. This upstream-downstream asymmetry makes the CPR asymmetric: something happening to the surface water upstream inevitably has consequences for the appropriators downstream. For micro-pollutants, spatial and temporal distribution is a crucial characteristic as it refers to the pace and extent at which micro-pollutants spread. Table  3.5 lists the different resource units and their second-tier characteristics and summarizes their interpretation for the analysis. The governance system (GS) comprises the policy subsystem with its legal rules within which the CPR management process takes place. The GS represents thus the formal policy structure with the legal acts, instruments, and guidelines related to micro-pollutants. For the second-tier characteristics of the GS, I draw on the “alternative list of second-tier properties for governance systems” by McGinnis and Ostrom (2014, Table 2).23 Policy area (GS1) defines the policy field water policy. Geographic scale of governance system (GS2) indicates the jurisdictional region where the legal acts, instruments, and guidelines regarding the management of micro-pollutants apply. GS2 refers to the governance system’s geographical outreach. The characteristic regime type (GS4) informs about the political system that exists in the respective sub-­ catchment area under study. GS4 points to the internal, administrative structure of the respective state. Rule-making organizations (GS5) hint at the actors concerned with the CPR problem that may decide on the rules and instruments aiming to tackle  I ignore the characteristic population (GS3) because the SESF element “actors” and its characteristics cover this aspect in detail (McGinnis and Ostrom 2014). I further do not consider the three characteristics property-rights systems (GS7), historical continuity (GS10), and network structure (GS9).

23

3.3  Case Study Design and Case Studies

75

Table 3.6  The governance system and its 2nd-tier characteristics Governance system GS1 Policy area GS2 Geographic scale of GS GS4 Regime type GS5 Rule-making organizations GS6 Rules-in-use

GS8 Repertoire of norms and strategies

The policy subsystem with its institutions: legal acts, instruments and guidelines related to micro-pollutants Water Policy Jurisdictional area in which rules-in-use apply The political system of the respective government/s in the sub-­ catchment area Public sector organizations; private sector organizations; nongovernmental organizations; community-based organizations Legal acts and instruments concerning micro-pollutants; inform about the regulators and implementers among the actors; indicate supranational guidelines (Directive 60/2000/EC) The policy instruments

the CPR problem. These actors comprise public sector organizations, like state institutions and public wastewater treatment plant operators; private sector organizations, like the industry causing micro-pollutants; nongovernmental organizations, like interest groups or environmental associations; and community-based organizations, like local drinking water providers. The characteristic GS5 is therefore a crucial asset for the identification of the actors of a SES.  It has to be considered, however, that depending on the context of the studied case, not every type of ­organization labeled as rule-making organization by McGinnis and Ostrom (2014) necessarily has the right to make rules. Rules-in-use (GS6) consists of the legal acts and instruments concerning micro-pollutants. They build the core for the study’s analysis of the governance system. The characteristic also indicates whether an actor is a regulator—deciding on an instrument—or an implementer, realizing an instrument, and thus informs control variables 1 and 2. Furthermore, rules-in-use can indicate guidelines issued by a supranational governance system. For this research, the EU’s Water Framework Directive (Directive 60/2000/EC) is a supranational guideline on micro-pollutants. The repertoire of norms and strategies (GS8) reflects the policy instruments tackling micro-pollutants. Table 3.6 gives an overview of the GS characteristics relevant for the analysis. The research focus is on cooperation among collective actors who are affected by the CPR problem of micro-pollutants in Rhine surface water and participate in its management process. I take into consideration actors who cause the CPR problem, the ones affected by it, actors assigned to regulate environmental problems, those who research on the issue, and actors who lobby and inform about the resource in general and the CPR problem specifically. They are collective actors, i.e., they are an institution, a firm, or an organization. The SES framework’s element actor (A) has nine characteristics, out of which seven relate to the research. The number of relevant actors (A1) is simple and useful. It indicates the number of actors involved in the management process of the CPR problem. Socio-economic attributes of actors (A2) illustrates actors’ diverse

76

3  Methods and Cases

Table 3.7  Actors and their 2nd-tier characteristics Actors A1 N° of relevant actors A2 Socioeconomic attributes A4 Location A5 Leadership/ entrepreneurship A6 Norms/social capital A8 Importance of resource A9 Technologies available

Corporate actors related to the issue of micro-pollutants in surface water The relevant actors in the given CPR problem setting Actors’ characteristics, i.e., the actors’ types and roles Actors’ spatial location in the sub-catchment area Actors’ centrality within network Actors’ participation in forums and actors’ beliefs and actors’ reputation Actors’ dependence on the resource surface water Technological solutions to the CPR problems (e.g., filtering techniques)

characteristics. In the present case, A2 hints at the different actor types and the roles they are assigned to by the governance system. The characteristic location (A4) demonstrates actors’ geographical position in each case study region and indicates actors’ territoriality, a control variable. Leadership/entrepreneurship (A5) is an attribute I consider in assessing actors’ positions in the collaboration networks—an analysis that could reveal leaders or entrepreneurs. As already discussed, the characteristic norms/social capital (A6) informs the independent variables actors’ participation and actors’ co-participation in forums (IVs 2a & b) and actors’ same belief (IV 3) as well as the control variable reputation (CV 3) (see Table  3.2, Sect. 3.2.2.3). The characteristic importance of resource (A8) finally applies to CV 4 pollution-sensitive water use. Technologies available (A9) covers the aspect of potential solutions to the CPR problem—like technical improvements of water filter techniques. Table 3.7 lists the actors’ characteristics relevant for the analysis. The SESF’s application clarified the relevant parts of the social-ecological system under study. The resource system’s and the resource units’ characteristics revealed the dynamics between the resource surface water and its pollutant. The structured focus on the attributes of the governance system showed in which policy area the study takes place and hints at the case studies’ different jurisdictions, their laws, and instruments. The actors’ specification further facilitates the actor identification process (see Sect. 3.4.2). The following three chapters describe the case studies along the SESF’s elements, while the last sub-chapter summarizes the case study description in a general overview board.

3.3.3  T  he Micro-pollutant Management Process at the Rhine Basin at Basel Canton Basel City, that is, the city of Basel and the municipalities of Riehen and Bettingen, lies within the hydrological Rhine sub-catchment area High Rhine. The sub-basin High Rhine comprises the catchment area from Lake Constance to Basel

3.3  Case Study Design and Case Studies

77

Fig. 3.5  Map of the Basel case study region, the city and the canton Basel City (left), and Switzerland’s location within Europe (right)

(Kremer 2010, p.  56). The case study is localized within the city’s and the two municipalities’ boundaries. Figure 3.5 shows the case study’s territorial boundaries, canton Basel City with the city of Basel and the two communities Riehen and Bettingen, and Switzerland’s location within Europe.24 3.3.3.1  The Different Water Uses The Rhine surface water in Basel is processed into drinking water. The Rhine surface water at this geographical spot shows considerable concentrations of micro-­ pollutants. They are caused by the effluents from the pharmaceutical and chemical industry in Basel (Braun and Gälli 2014, p. 14) and the urban drainage. In 2006, the public learned about chemical residues from a waste disposal site close to Basel that had entered the groundwater basin. The reaction from the media and the general public was immediate (Rechsteiner 2006), as that very groundwater basin is one of the drinking water sources for Basel City and its agglomeration. People in this region are thus sensitized for drinking water issues. Two service providers are in charge of the drinking water supply in Basel: the drinking water treatment plants Lange Erlen, operated by IWB, and the drinking water operator Hardwasser AG (WWB). Both rely on good surface water quality as they gain drinking water through riverbank filtration from surface water (WWB) and artificial groundwater recharge (IWB).25 In addition to its use as drinking water source, Rhine surface water in the city of Basel is also used for fishing (KFVBS 2016).  For the map of the city of Basel, see wikimedia (2018a); for the map of Switzerland’s location within Europe, see wikimedia (2018b). 25  Interviews N° 1 and 2. Riverbank filtration is a natural treatment process to filter surface water. The surface water, mostly from a river system, is abstracted close to the surface water (e.g., a 24

78

3  Methods and Cases

Micro-pollutants are generally found in Swiss rivers. Most of them belong to the Rhine catchment area (Gälli et al. 2009, p. 46f.) which makes up 68% of the country’s surface—an area in which about 80% of the Swiss population live (Ruff et al. 2013, p. 17). Persistent substances that enter Swiss rivers connecting to the Rhine inevitably show up in the surface water in Basel.26 For instance, in 2013 more than 100 tons of organic micro-pollutants were found in the Rhine surface water at Basel (ibid., p. 24). Data on concentrations of indicator compounds detected in Swiss surface water thus apply to Rhine surface water at Basel. Along the Rhine and the Aare, which is a tributary to the Rhine, there are 221 dischargers from the chemical and food processing industry that potentially discharge micro-pollutants into the rivers’ surface water (Braun and Gälli 2014, p. 12).27 The indicator compound sulfamethoxazole was detected in 34 of 66 measurements of Swiss surface water in 2010; the pharmaceutical carbamazepine was identified 112 times in 509 measurements. Both substances are mainly discharged from treatment plants which are not yet able to decompose the substances efficiently (Abegglen and Siegrist 2012, p. 36f., 187f.). Carbamazepine does not decompose itself biologically, which means it has a high persistence (ibid., p. 70). The Swiss Federal Office for the Environment (FOEN) declared the two substances sulfamethoxazole and carbamazepine as indicator compounds for antibiotics and anti-­ epileptics, respectively, as they spread widely in considerable concentrations throughout Switzerland (ibid., p. 177). The same applies to the biocide mecoprop (Abegglen and Siegrist 2012, p. 177; Wittmer et al. 2014, p. 59). Mecoprop, as well as the herbicide isoproturon, are also often found in discharge from treatment plants (Abegglen and Siegrist 2012, p. 188; Gälli et  al. 2009, p.  47), with mecoprop being especially difficult to decompose (Abegglen and Siegrist 2012, p. 36). The herbicides terbuthylazine and s-­metolachlor, as well as the biocides diurone and carbendazim, are widely used in Switzerland (Gälli et al. 2009, p. 47; Wittmer et al. 2014, p. 58). While diurone has a very high persistence (Gälli et al. 2009, p. 71), terbuthylazine and s-metolachlor have decom-

riverbank) and infiltrated into the subsurface. As the water flows through the soil and sediments, a combination of biological, physical, and chemical processes occur, filtering the water and improving its quality; see Hülshoff et  al. (2009) and Sharma and Amy (2009). Artificial groundwater recharge is a hydrologic process where surface water percolates through the unsaturated ground and soil downward to groundwater. Through its travel through the vadose zone, the infiltrated water is purified; see Asano and Cotruvo (2004) and Greskowiak et al. (2005). 26  About 70% of the Swiss territory drains into the Rhine. Substances entering Swiss water bodies usually end up in the Rhine. One interviewee stated: “Etwa 70% der Fläche der Schweiz werden in den Rhein entwässert. Das, was in Schweizer Gewässer gelangt, endet meistens im Rhein.” English translation: “About 70% of the Swiss territory drains into the Rhine. Most of the times, anything that enters Swiss water bodies, ends up in the Rhine” (Interview N° 6). 27  Out of these 221 dischargers, 172 are indirect and 49 are direct dischargers. Direct dischargers emit industrial wastewater—treated or untreated—into water bodies; indirect dischargers emit wastewater, treated or untreated, into the public canal network (Braun and Gälli 2014, p. 6). For a map showing the enterprises that directly discharge into Swiss water bodies, see Braun and Gälli (2014, p. 14).

3.3  Case Study Design and Case Studies

79

position products that exceed concentration rates of 0.1  μg/l in surface water (Wittmer et  al. 2014, p.  66). The persistent compounds mentioned above pass through River bank filtration. These facts turn the indicator compounds into a serious challenge for actors responsible for water politics in the region and for those depending on good surface water quality. 3.3.3.2  The Laws and Instruments Swiss politics support research on advanced techniques extracting micro-pollutants from the water cycle. Swiss WWTP providers started to upgrade their plants. The upgrading of WWTP is only one measure within the strategy of the Swiss Federal Office for the Environment (FOEN) to reduce concentration of micro-pollutants (Gälli et al. 2009, p. 58). Measures at the source, such as material bans, restrictions, and voluntary measures, have also been agreed upon. Water Policy Regarding Micro-pollutants at the National Level The Swiss confederation regulates its water policy through its Waters Protection Ordinance (WPO) and its Water Protection Act (WPA) (Bundesversammlung der Schweizerischen Eidgenossenschaft 1991; Der Schweizerische Bundesrat 1998). The WPO regulates the examination of the degree of pollution of wastewater (Der Schweizerische Bundesrat 1998, Art. 3) and the discharge of wastewater (ibid., Art. 3, 6–9). Article 47 states that in case of water deterioration, authorities have to “assess the type and extent of the pollution” (ibid., Art. 47(a)) and its cause (ibid., Art. 47(b)) and come up with potential measures (ibid., Art. 47(c&d)). The act explicitly considers the two international agreements regarding the Rhine river: the Convention on the International Commission for Protection of the Rhine against Pollution and the Convention on the Protection of the Rhine against Chemical Pollution (ibid., Art. 51). Article 52a backs up the general consideration of the Rhine’s potential pollution and decrees the “elimination of organic trace substances at wastewater plants.” Costs for necessary measures are compensated (Der Schweizerische Bundesrat 1998, Art. 52a(1)). The ordinance also decrees compensatory payments for measures in agriculture that prevent the entrance of substances into the water cycle (ibid., Art. 54). The ordinance thus introduces point source and source-directed measures. Furthermore, the WPO purports to achieve a water quality that assures that substances detected in Swiss water bodies do not accumulate, are not harmful to the aquatic ecosystem, and occur in concentrations in which they would naturally be present (ibid., Annex 1(1)). In Annex 3.2, the WPO lists the parameters of substances that need to be met; in Annex 3.1.2(8), the ordinance defines the WWTP that must assure a removal efficiency of 80% for organic substances. The WPA specifies the regulations already outlined by the WPO. For instance, it regulates the financing of the renovation and upgrading of WWTP towards an elimination of micro-pollutants by introducing a wastewater charge (Bundesversammlung der Schweizerischen Eidgenossenschaft 1991, Art. 60 & 61a).

80

3  Methods and Cases

Water Policy Regarding Micro-pollutants in the Cantons Basel City and Basel Country At the regional level, the WPO is enforced by the two cantons, canton Basel City and canton Basel Country (Bundesversammlung der Schweizerischen Eidgenossenschaft 1991, Art. 45(1)). The Cantonal Office for the Environment and Energy, Basel City (AUEBS), and the Cantonal Office for Environmental Protection and Energy, Basel Country (AUEBL), are the region’s regulatory authorities supervising the implementation of the WPO’s and WPA’s measures.28 For the implementation of measures, canton Basel City works closely together with the FOEN. Whenever problems arise in the enforcement process, the canton contacts the FOEN to have the issue solved at the national level.29 Canton Basel Country plays a role for water management within canton Basel City as both cantons arrange most of their policies jointly.30 On the side of the addressees of legislation, the two drinking water providers in Basel, IWB and WWB, have installed activated charcoal in their treatment system, which sufficiently eliminates micro-pollutants.31 The city’s main WWTP operator, ProRheno AG, will upgrade its plant until 2023 to eliminate micro-pollutants as efficiently as possible (Kanton and Kanton 2013; ProRheno AG 2016).32 With concentrations of micro-pollutants regularly found in the region’s surface water, with regulatory measurements installed, pollution-sensitive water uses at  Interview N° 6. For the list of interviews, see Table 3 in Annex IV.  Interview N° 3. 30  Both cantonal state actors, AUEBS and AUEBL, confirmed that they work closely together on the issue of micro-pollutants. AUEBS stated: “The neighboring canton Basel-Country is very relevant. It surely is the most important actor. Because a lot of what we measure comes from the Canton Basel Country or from the Canton Aargau. Basel Country is one of the most important partners we have, also later on during the carrying out [of measures].” Original quote: “Der Nachbarkanton Basel-Landschaft ist sehr relevant. Der ist sicher der wichtigste Akteur. Weil sehr vieles, was wir messen eigentlich aus dem Kanton BL, und manchmal aus dem Kanton Aargau, kommt. Kanton BL ist einer der wichtigsten Partner, die wir haben, dann auch nachher im Vollzug.” (Interview N° 3). AUEBL reckoned: “We have a lot to do with AUE BS.” Original quote: “Mit dem AUE BS haben wir viel zu tun” (Interview N° 6). 31  Interviews N° 1 and 2: In the case of IWB, after the treatment with activated charcoal, the two indicator substances sulfamethoxazole (antibiotic) and carbamazepine (anti-epileptic) have been detected 8 and 48 times, respectively, between 2010 and 2015, each time below the limiting value. Both do not appear in drinking water. In 2014, four other substances could be detected below the limiting value after the activated charcoal treatment: the contrast agents diatrizoate and iopamidol, the sugar substitute acesulfame, and the chemical EDTA (Interview N° 2). The same substances, again below the limiting value, were also found in the water after the activated charcoal treatment at WWB in 2014 (Interview N° 1). 32  The upgrading includes the so-called 4th treatment stage which is an advanced process engineering method to eliminate micro-pollutants. The two common practices are using ozone to filter micro-pollutants or applying activated carbon. Both treatment techniques necessitate further water treatment afterwards. Even these elaborated filtering techniques cannot hold back all substances. For instance, the complexing agent EDTA and radiocontrast agents like iomeprol stay in the water, cf. Hillenbrand et al. (2015, p. 13) and Interview N° 4. 28 29

3.3  Case Study Design and Case Studies

81

stake, and actors working on the CPR problem, the case Basel is suitable for a case study. Marginally, it is a cross-border case study, in that actors from France and the German federal state Baden-Wurttemberg are part of the wider management process. Since Switzerland does not belong to the European Union, the borders between Germany and Switzerland and between France and Switzerland are not internal EU borders; the case is not an EU cross-border case.

3.3.4  T  he Micro-pollutant Management Process in the Ruhr Region The second case is the management process of micro-pollutants in the catchment area of the river Ruhr in North Rhine-Westphalia (NRW), Germany. With a longitude of about 221 kilometers and a catchment area covering 4485 km2, the Ruhr flows through NRW’s industrial region, the so-called Ruhrgebiet. It meets the Rhine off Duisburg (LANUV 2013; Ruhr-Guide 2018). The main sources of micro-­ pollutants in this region are agriculture; production processes, such as electroplating and paper manufacture; and settlements.33 Figure 3.6 shows a map of the Ruhr case study region (commons.wikimedia 2016).34 3.3.4.1  The Different Water Uses Drinking water in NRW is mainly gained from artificial groundwater recharge, a process in which surface water is added to groundwater, and from groundwater and spring water (LANUV 2019; LWL 2018). The Ruhr alone accounts for 20% of the region’s drinking water (MKULNV 2009, p. 16), providing 4.6 million people with drinking water. The Rhine provides 9% of drinking water in this region (ibid., p. 16). In the Ruhr region, 26 drinking water providers process surface water into drinking water (AWWR 2018). The waters of the Ruhr are also used for fishing (FV NRW 2018; RFG 2018). In 2006, researchers from the University of Bonn analyzed the waters of the Rhine and detected high concentrations of perfluorooctanesulfonic acid (PFOS) in the Ruhr35 posing a serious threat to the region’s drinking water source (AWWR and Ruhrverband 2016, p. 110f.).36 The general public and politicians were alarmed. As  Interview N° 9.  The geographical outlines of the catchment area of the Ruhr are also referred to as “the galloping elephant” (Interview N° 10). 35  Interview N° 10. Perfluorooctanesulfonic acid (PFOS) together with perfluorooctanoic acid (PFOA) belongs to the chemical class of fluorosurfactants (PFAS). They are used for textile impregnation and coating paper, in fire-fighting foams, and in production processes of electroplating AWWR and Ruhrverband (2016, p. 110). 36  Interview N° 10. 33 34

82

3  Methods and Cases

Fig. 3.6  Map of the Ruhr case study region

a result, state authorities started the political program “Reine Ruhr,” meaning as much as “Clean Ruhr.” The program suggests suitable measures to manage PFOS in the Ruhr. It also raises awareness for other kinds of micro-pollutants in the river as further studies detected several other kinds of micro-pollutants in the Ruhr surface water (MKULNV 2012). The pharmaceutical indicator compound sulfamethoxazole was detected above the threshold value of 0.1 μg/l in the periods between 2004 and 2006 and between 2008 and 2009 (MKULNV 2012, p.  49; RWTH Aachen and IWW 2008, p.  68). Official authorities warn that toxic impacts from this compound cannot be ruled out (MKULNV 2012, p. 51)—especially, because concentrations of this substance are significantly increasing. A similar pattern can be observed for carbamazepine, which has been detected in concentrations up to 0.31  μg/l (RWTH Aachen and IWW 2008, p. 68, 127) between 2004 and 2006 and in concentrations still above the precautionary value of 0.1 μg/l in 2008 and 2009 (MKULNV 2012, p. 49). Its average concentration in the Ruhr catchment area is 2.8 μg/l (RWTH Aachen and IWW 2008, p. 99). The substance has also been found in the discharge of sewage treatment plants (ibid., p.  70), in raw waters used for drinking water and in drinking water itself (MKULNV 2012, p. 11). The herbicides isoproturon, terbuthylazine, and metolachlor were detected in concentrations above the threshold value of 0.1 μg/l. The two biocides mecoprop and diurone also exceeded the precautionary value, although less often than the other indicator substances. Micro-pollutants are thus an omnipresent environmental issue in the Ruhr region and pose a threat to the quality of regional drinking water.

3.3  Case Study Design and Case Studies

83

3.3.4.2  The Laws and Instruments The region of the Ruhr case study lies entirely on German territory, within the German federal state North Rhine-Westphalia (NRW). In German water politics, the German federal states have the right to adapt regulations deviating from the national law (Berger 2017, p. 6). The FRG is a member state of the European Union (EU) and has to translate EU law into domestic law.37 In 2000, the European Union commissioned the Directive 2000/60/EC, known as the Water Framework Directive (WFD), to provide a framework for a European water policy (EU 2000, par. 10 & 18). The WFD demands from the EU member states to achieve the maintenance and improvement “of the aquatic environment in the Community” (ibid., par. 19). One aspect that has to be tackled in order to achieve this goal is micro-pollutants. For its national and regional water policy in general and the management of micro-­ pollutants in particular, the FRG has to comply with the WFD. The European Union Water Framework Directive (WFD) The WFD’s aim is to “achieve the objective of at least good water status” (EU 2000, par. 26) with the “(…) ultimate aim (…) to achieve the elimination of priority hazardous substances” (ibid., par. 27) through the “progressive reduction of emissions of hazardous substances to water“ (ibid., par. 22). The EU defines hazardous substances as “substances or groups of substances that are toxic, persistent and liable to bio-accumulate, and other substances or groups of substances which give rise to an equivalent level of concern” (ibid., Art. 2(29)). The WFD thus describes the main concerns about micro-pollutants: their potential toxicity and persistence and their possible accumulation in water. To eliminate the pollution caused by “the direct or indirect introduction (…) of substances (…) into the (…) water (…) which may be harmful to human health or the quality of aquatic ecosystems (…)” (EU 2000, 7, Art.2(33)), EU member states are required to adopt measures (ibid., Par. 45, Art. 11(2-4)). These measures shall be: based on the precautionary principle and on the principles that preventive action should be taken, environmental damage should, as a priority, be rectified at source and that the polluter should pay. (EU 2000, par. 11)

The directive is thus explicit in its recommendation of which type of measures member states should prefer: preventive and source-directed measures that ought to be financed by polluters themselves. The directive expects member states to ensure water services are covering their costs by 2010 (EU 2000, Art. 9(1)). The WFD stipulates to apply best available techniques or to meet emission limit values in the case of point sources, i.e., emissions, and to use best environmental practices in the case of diffuse sources of micro-pollutants (cf. ibid., Art. 10(2a-c)). The directive further asks member states to work out river basin management plans (ibid., 16, 37

 See EU (07.03.2019); EU (22.07.2019).

84

3  Methods and Cases

Art. 13; and Annex VII), to monitor their water bodies’ status (ibid., par. 36), and to set emission limit values and environmental quality standards (ibid., par. 40).38 Environmental quality standards (EQS) refer to the “concentration of a particular pollutant or group of pollutants in water (…) which should not be exceeded in order to protect human health and the environment” (EU 2000, p.  7, Art. 2(35)). In Directive 2000/60/EC, the European Parliament and the Council stipulate to “(…) agree on the substances to be considered for action as a priority (…)” (ibid., par. 43). The European Union published a first list of 33 priority substances in 2001 (Decision N° 2455/2001/EC, EU 2001). This list was revised in 2008 and 2013, and EQS for the now 45 substances were laid down accordingly (EU 2008, Annex I; EU 2013, Annex II).39 Since the European Parliament and the European Council had realized that the member states would not be able to “(…) achiev[e] good surface water chemical status by laying down EQS for priority substances and certain other pollutants (…)” (EU 2013, par. 35), they determined that the EU itself “(…) may adopt measures, in accordance with the principle of subsidiarity (…)” (ibid., par. 35), the objective of which is to keep the protection level of surface water even throughout the EU (cf. ibid., par. 35). The EU further initiated a roadmap on a strategic approach towards pharmaceuticals in the environment. Its goal is to understand the uncertainties and knowledge gaps about the impact of pharmaceuticals on the environment. The roadmap aims at showing ways to address this concern while assuring a continuous treatment of humans and animals with pharmaceuticals (EC 2017, p. 2). After the period of consulting member states and experts on the topic ended, the initiative  started to be worked out, beginning in the first quarter of 2018 (ibid., p. 2). The first deadline for the EU member states to “(…) bring into force the necessary laws, regulations and administrative provisions necessary to comply with th[e] Directive (…)” was 22 December, 2003 (EU 2000, Art. 24). This deadline was extended several times.40 Member states should consider the revised EQS for existing priority substances in their river basin management plans covering the period from 2015 to 2021 (EU 2013, par. 9).41 The EU member state federal republic of Germany (FRG) is obliged to implement the WFD in its domestic law.  Emission limit values are “the mass, expressed in terms of certain specific parameters, concentration and/or level of an emission, which may not be exceeded during any one or more periods of time” EU (2000, p. 8), Art. 2, Par. 40. 39  The amendment of the WFD in 2013 (2013/39/EU) further recorded that the list of priority substances should be reviewed “four years after the date of entry into force of this Directive and at least every six years thereafter” EU (2013, p. 6), Art. 1(1). 40  The first extension of the deadline for member states to “bring into force the laws, regulations and administrative provisions necessary to comply with (…) the Directive” was set to July 13, 2010, in Directive 2008/105/EC; see EU (2008), Art. 13(1). A second prolongation until 14 September 2015 was set in Directive 2013/39/EU, which is an amendment to the WFD; see EU (2013), Art. 3(1). The deadline was again postponed to 20 May 2016 in Directive 2014/101/EU; see EU (2014), Art. 2(1). 41  In 2018, the “Commission shall (…) verify that emissions, discharges and losses as reflected in the inventory are making progress towards compliance with the reduction or cessation objectives 38

3.3  Case Study Design and Case Studies

85

Water Policy Regarding Micro-pollutants in the Federal Republic of Germany The federal republic of Germany transposes the WFD in its act on hydrologic balance Wasserhaushaltsgesetz (WHG) (Berger 2017, 4f.; BMJV 2009, par. 80(1), 82(3,4), 83(2)) and in its surface water ordinance Oberflächengewässerverordnung (OGewV) (BMJV 2016). The WHG demands the production of risk maps as required by the WFD (BMJV 2009, par. 80(1)), the implementation of the measures in compliance with the WFD (ibid., par. 82(3,4)), and the working out of river basin management plans in accordance with the WFD (ibid., par. 83). The OGewV states the measures more precisely: the monitoring of anthropogenic pollution of German surface waters (BMJV 2016, par. 4(1)) and of emissions of priority substances (ibid., par. 4(2)); the application of the standards outlined in the WFD when monitoring quality elements (ibid., par. 9(1)); the monitoring of the WFD priority substances (ibid., par. 11); and the assurance of cost recovery for water services (ibid., par. 16(2)). Based on the Directives 2000/60/EC, 2008/105/EC, and 2013/39/EU, the EQS in Germany have been adapted accordingly (Hillenbrand et al. 2015, p. 4). The German Federal Environmental Agency (UBA) claimed three biocides, five pharmaceuticals, and four further substances as relevant for Germany42 (ibid., p. 6). These twelve substances were added to the list of priority substances (ibid., p. 4). In a study on potential measures to reduce and avoid the entrance of micro-pollutants into water bodies in 2014, the UBA recommends potential measures at the source, measures to reduce pharmaceuticals, and measure mixes as proposed by the FRG (ibid., p. 20f., 25). Various state authorities work intensively on the topic. The German Working Group on water issues of the Federal States and the Federal Government (LAWA) published a report on the existence and significance of micro-pollutants in German surface waters (LAWA 2016). The Ministry for the Environment, Nature Conservation and Nuclear Safety (BMUB) and the UBA further initiated a stakeholder dialogue intended to provide the base for a nationwide strategy to protect water bodies from micro-pollutants. The participants came from the industry, state agencies responsible for water service provision, the civil society, and the water sector. Their recommendations for reducing the entry of biocides, pesticides, cosmetic products, industrial chemicals, and pharmaceuticals were published in June 2017 (BMU 2016; BMU et al. 2017). The federal states can decide how to implement the WHG at the federal level. That is, they are allowed to develop regulations that deviate from state law (WHG) if these regulations are issued after the federal legislation (Berger 2017, p. 6). The WHG deliberately leaves space for the federal authorities to substantiate instruc-

laid down in (…) Directive 2000/60/EC” EU (2008), Art. 5(5) to assess whether the member states have achieved the WFD objective. 42  The substances are terbutryn, triclosan, and TBT (biocides); diclofenac, ibuprofen, metoprolol, sulfamethoxazole, and iomeprol (pharmaceuticals); and PAK, nonylphenol, PFOS, and HBCDD; see Hillenbrand et al. (2015, p. 6).

86

3  Methods and Cases

tions and regulations within the regional context. However, the federal states must not develop own regulations that concern the emission of substances (ibid., p. 7). Water Policy Regarding Micro-pollutants in the Ruhr Region In North Rhine-Westphalia, the national and European regulations are realized through the Landeswassergesetz (LWG), the Federal Water Act (MKULNV 2018). After high concentration of PFAS had been detected in 2006 in the Ruhr and in the region’s drinking water (AWWR and Ruhrverband 2016, p. 109), state authorities and water service providers started considering micro-pollutants.43 The Ministry for Climate Protection, Environment, Agriculture, Nature and Consumer Protection in NRW (MKULNV) worked out the abovementioned strategy plan for rivers and water bodies in NRW, called “Programm Reine Ruhr”(MKULNV 2012).44 The program targets the improvement of drinking water and water body quality in NRW by outlining the status quo of micro-pollutants in surface water in NRW. It introduces a water monitoring system (MKULNV 2012, p. 19ff) and suggests an evaluation scheme for the substances. Furthermore, the program presents common water treatment practices and planned improvements—for instance, wastewater treatment plants will be upgraded in the coming years to be able to filter persistent compounds such as carbamazepine (RWTH Aachen and IWW 2008, p. 70). Its overall goal is to keep concentrations of micro-pollutants below 0.1 μg/l (ibid., p. 17). To comply with the EU WFD, the MKULNV also published a river basin management plan for the rivers of federal state. The second management plan covers the years 2016–2021 and lays down the requirement to reduce the pollution of groundwater and surface water, the monitoring plan of the waters and nature reserves, the environmental objectives to achieve, and the measures planned to be taken (MKULNV 2015). The regional government takes the so-called multibarriers approach to reduce micro-pollutants. This approach combines three types of measures: measures taken at the place of pollution; treatment of wastewater; and processing of drinking water.45

 PFAS—fluorosurfactants—being the chemical class that perfluorooctanesulfonic acid (PFOS) and perfluorooctanoic acid (PFOA) belong to (AWWR and Ruhrverband 2016, p. 110). 44  English translation: “Clean Ruhr Program”. Although it is named after the river Ruhr, the program relates to the management of micro-pollutants in NRW in general (Interview N° 10). 45  The multibarriers approach—“Multibarrierenansatz” in German—works in several steps that form a cycle: 43

1 ) 2) 3) 4) 5) 6)

Water quality monitoring in water bodies Early diagnosis and risk evaluation of micro-pollutants entering the water cycle Preparation of an action plan including a value-costs analysis and an impact assessment Taking measures (following the polluter pays principle and the principle of proportionality) Risk communication and ongoing research Repeating step 1, monitoring water quality (Interview N° 15)

3.3  Case Study Design and Case Studies

87

Actors from the water service sector and the municipalities further published a memorandum for the protection of waters against micro-pollutants (agw et al. 2014). Its objective is to inform municipalities, the water service provision sector, and water associations about the measures taken so far to manage micro-pollutants in surface water in NRW.  It aims at opening a discussion between state actors and service providers about future measures (ibid.). Furthermore, the Federal Ministry of Education and Research (BMBF) financed a research program46 for the Ruhr, which investigated drinking water treatment, microbiological analyses, early warning systems, and the socioeconomic acceptance of measures (BMBF 2019). The case study of the Ruhr catchment area fulfills the selection criteria: concentrations of micro-pollutants have been detected in Ruhr surface water; pollution-­ sensitive surface water uses exist; regulatory measures are worked out and being put in place; and regional actors have started working on the issue.

3.3.5  T  he Micro-pollutant Management Process in the Moselle Basin The Moselle is the second biggest tributary of the Rhine, and its catchment area forms one of the Rhine’s nine sub-basins. It originates in the French Vosges region and joins the Rhine in Koblenz after 520  km and after having passed through Luxembourg and the German federal state Rhineland-Palatinate. Its catchment area furthermore comprises Belgian territory and parts of the German federal state Saarland (ICPR 2019; IKSMS n.d.-a; UNECE 2011, p. 320). The case study area comprises the Moselle catchment area on Luxembourgian and on German territory.47 Figure 3.7 depicts the map of the Moselle catchment area and the case study region within it.48 3.3.5.1  The Different Water Uses …in Luxembourg In Luxembourg, drinking water is provided by regional unions—so-called syndicats des eaux—that are headed by the Luxembourgian Union of the Waters of the Esch-­ sur-­Sûre Dam (SEBES).49 SEBES is in charge of the drinking water reservoir at  The research program is called “Projekt Sichere Ruhr” (English: project safe Ruhr) and ran from January 2012 until December 2015; see also BMBF et al. (n.d.); juraforum (2012). 47  The Moselle catchment area accounts for 97.2% of the Luxembourgian territory. The remaining 2.8% are covered by the Maas catchment area (see MDDI 2014, p. 29f.). 48  Map by Andreas Moser, Swiss Federal Institute of Aquatic Science and Technology, Switzerland (Eawag), Dübendorf, Switzerland, June 2015. 49  Original name: Syndicat des Eaux du Barrage d’Esch-sur-Sûre. 46

88

3  Methods and Cases

Fig. 3.7  Map of the Moselle case study region

Esch-sur-Sûre in the northwest of the country, which is fed by the Moselle’s tributary Sauer. The regional drinking water service providers receive their water from the reservoir and from groundwater (Administration de la Gestion de l’Eau 2016; MDDI 2014, p. 19; SEBES n.d.). The Moselle’s surface water is thus not used for drinking water purposes in Luxembourg. Nevertheless, compounds entering Luxembourgian groundwater or the Sauer, which flows into the Moselle (UNECE 2011, p.  320), contribute to the Moselle’s burden with micro-pollutants as it continues its way through Germany and to the Rhine. Furthermore, the rivers of the Luxembourgian parts of the Moselle catchment area are used for fishing purposes (FLPS 2019).

3.3  Case Study Design and Case Studies

89

Studies on micro-pollutants in Luxembourg in the 2000s revealed high concentrations of sulfamethoxazole, diurone, and metolachlor—among a range of other substances—in the surface water of the rivers Mess and Pétrusse in Southwestern Luxembourg (Krein et al. 2013, p. 287; Meyer et al. 2011, p. 133). In 2014, residues of pesticides were found in more than 70% of the country’s measurement sections for groundwater (Luxemburger Wort 2014b). The herbicides s-metolachlor, metazachlor, and benta-zone are especially pertinent in Luxembourgian groundwater (MDDI 2014, p. 117). An acute incident of micro-pollutants in the Sauer alerted officials to the threat to their drinking water: due to an agricultural accident in Belgium in fall 2014, about 7.5  l of the herbicide metazachlor entered the river Sauer and the Luxembourgian main drinking water reservoir (Luxemburger Wort 2014a). Studies following up this incident revealed the concentration of metazachlor to have risen up to 2 μg/l (MDDI 2014, p. 95)—while the EU limiting value for pesticides and their metabolites is 100 ng/l (i.e., 0.1 μg/l). The water from the barrage no longer fulfilled EU standards for drinking water quality (SES 9. 2014). The studies also revealed heavy pollution of surface and groundwater with s-metolachlor. This led to a ban of the compound in Luxembourg in February 2015 (MDDI 2015a, p. 153). The herbicides isoproturon and terbuthylazine as well as the biocide diurone are also frequently detected in groundwater (MDDI 2014, p. 125, 155). In 2009, isoproturon was measured in concentrations above the EU limiting value of 0.1 μg/l (ibid., p. 125). …in Germany On German territory, the Moselle’s catchment area covers 2% of the federal state Saarland and about a third of the federal state Rhineland-Palatinate (RLP) (IKSMS n.d.-c, p. 17). In RLP, surface water has only a minor significance for the production of drinking water (Breitenfeld 2007, p. 172); however, the Moselle’s surface water is used for fishing (bfv-trier 2019; FV Saar 2010). State authorities in RLP face particular problems with the pesticide isoproturon in the Moselle surface water. The substance is applied in fall, and its main runoff occurs with heavy rains. High concentration levels led to the temporal chemical degradation of the Moselle and the detection of the substance in the Rhine. Banned since 2017, isoproturon will disappear in the future. Pesticides, however, remain an issue in the region. In 2016, residues of pesticides exceeded the limiting value in 45 water bodies in RLP. Metolachlor is only one of the substances traced in water bodies in RLP, but also in the Rhine, as happened in 2016.50 Another substance detected above the limiting value is the analgesic diclofenac. A common painkiller, diclofenac enters the water cycle through the sanitation system. To regulate it, some of the

50

 Interview N° 19.

90

3  Methods and Cases

region’s WWTP are being upgraded to the 4th treatment stage.51 In addition, state authorities in RLP monitor water samples of selected WWTP to identify the substances’ dischargers. In a second step, they request the dischargers to reduce their use of the detected substances.52 In Saarland, the University launched a series of research projects to examine river water quality in the federal state. Increased concentrations of micro-pollutants are not reported (Universität Saarland 2019). 3.3.5.2  The Laws and Instruments The Moselle case study region extends across the Moselle catchment area on the territory of the two German federal states Rhineland-Palatinate and Saarland and the Grand Duchy of Luxembourg. The Moselle forms the natural and national border between Luxembourg and Germany. The two countries share the responsibility for the river’s waters (IKSMS n.d.-d, p. 17; UNECE 2011, p. 24f.), and the case covers two jurisdictions: that of the FRG and of Luxembourg. Water Policy Regarding Micro-pollutants in Rhineland-Palatinate and Saarland Rhineland-Palatinate (RLP) has its own Landeswassergesetz (LWG) that translates national into state law (Ministerium der Justiz, RLP 2015). The Ministry for the Environment, Energy, Food and Forest in RLP developed its river basin management plans for the periods of 2010–2015 and 2016–2021 (MUEEF 2010, 2015). The second plan outlines the measures intended for tackling the problem of micro-­ pollutants (MUEEF 2015, p. 202f., 207, 211, 216). Measures in RLP are mainly source-directed and consider the cost-benefit ratio. Rather than pushing for costly end-of-pipe solutions, reasonable measures to tackle the problem at its source are preferred.53 State authorities in RLP are waiting for the EU guidelines on micro-­ pollutants before taking further measures themselves. Meanwhile, they participate in working out a nationwide strategy regarding micro-pollutants in Germany.54 Water use in Saarland is regulated by the Saarländische Wassergesetz (SWG), the state’s water management law (Ministerium der Justiz, Saarland 2004). To comply with the WFD at federal state level, the Ministry for the Environment and Consumer Protection Saarland (MUV) worked out a second river basin management plan for the years 2016–2021. It envisages technical and administrative measures for holding enterprises responsible that emit micro-pollutants (MUV 2015, p. 76). In 2016, the ministry further launched a study on the elimination of micro-­ pollutants through the 4th treatment stage in communal WWTP (MUV 2016).

 Telephone call N° 3.  Interview N° 19. 53  Telephone call N° 3. 54  Interview N° 19. 51 52

3.3  Case Study Design and Case Studies

91

Water Policy Regarding Micro-pollutants in Luxembourg As an EU member state, Luxembourg has to comply with the EU WFD, which has accordingly been integrated into Luxembourgian law (law “A217”).55 Article 5(3) of the law requires priority substances to be reduced and priority hazardous substances to be eliminated; Art. 21 stipulates a monitoring program to evaluate the water bodies’ quality. The law demands that the causes of point and diffuse pollution apply the best available techniques or best environmental practices (Art. 27). Agricultural substances considered pollutants are to be limited or banned (Art. 26(3)). The deadline for agreement on environmental quality norms for the priority substances was 22 December 2009 (Art. 34). In 2016, however, the Luxembourgian Ministry of Sustainable Development and Infrastructure had neither determined the type of micro-pollutants for measures to focus on yet, nor had it decided on limiting values for micro-pollutants.56 The first river basin management plan, issued in 2009, did not explicitly refer to micro-pollutants.57 The second river basin management plan for the Rhine and the Maas on Luxembourgian ground covers the years 2015–2021 (MDDI 2015b). Here, the Luxembourgian government sets itself the goal to develop and implement regulations for micro-pollutants. The authorities plan voluntary measures, among others, such as renunciation and reduced consumption, and bans of certain compounds (MDDI 2015b). The plan focuses primarily on water monitoring and pilot projects at wastewater treatment plants, i.e., feasibility studies.58 Once limiting parameters for micro-pollutants will be fixed, wastewater treatment plant operators will have to implement the 4th treatment stage.59 For the moment, the generally outdated Luxembourgian wastewater treatment plants are being upgraded (MDDI 2014, p. 77).60 Basic measures to reduce micro-substances, e.g., voluntary measurements  Interview N° 16, Le Gouvernement du Grand-Duché de Luxembourg 2008. The law “A217: Protection et gestion des eaux, loi du 19 décembre 2008 relative à l’eau modifiant” was amended in July 2017. Other than before, agricultural substances considered pollutants are supposed to be permanently limited, not only temporarily; see Le Gouvernement du Grand-Duché de Luxembourg (2017, Art. 9(1)). 56  Interview N° 20. 57  The reduction of pesticide use is suggested; the plan also contains a brief discussion of the country’s WWTPs as point sources of certain substances and agricultural use as diffuse source of substances; see MDDI (2009). 58  Interview N° 16. A monitoring measure implemented in Luxembourg is the separation of wastewater in old people’s homes, hospitals, and psychiatric wards in order to discern the types of substances and their concentration in the sewage water (Telephone call N° 1). 59  Interview N° 20. 60  Luxembourg is lagging 3–4 years behind regarding the standard of wastewater treatment plants in the EU; see Telephone call N° 1; MDDI (2014, p. 77). Luxembourgian WWTPs that process sewage of a population above 100.000 have to be upgraded. This primarily concerns WWTPs in the country’s south, where half of the country’s population lives. WWTPs will be upgraded so as to facilitate the inclusion of a 4th treatment stage at a later point in time, as this stage is not included now. The first river basin management plan reported 58 wastewater treatment plants to require being either built or upgraded; see MDDI (2009, p. 15). 55

92

3  Methods and Cases

or charges on products, are not yet in place. Meanwhile, research on efficient elimination processes of pharmaceuticals (LIST 2018a), on the reduction of pesticide use (LIST 2018c), and on mixture toxicity of herbicides in rivers (LIST 2018b) is being conducted. The Luxembourgian government is awaiting the measures regarding micro-­ pollutants outlined by the EU before taking decisions on measures.61 The country currently lacks a national strategy for the regulation of micro-pollutants62; ­developing of solutions takes time, and the responsible institutions are not sufficiently staffed to tackle the various tasks simultaneously.63 Water Policy Coordination at the Level of the IKSMS At the supranational level, the commissions for the protection of the Moselle and the Saar (IKSMS) coordinate the river basin management plans for the two rivers’ catchment areas across the different jurisdictions—namely, those of France, Luxembourg, Wallonia in Belgium, and Rhineland-Palatinate, Saarland, and North Rhine-Westphalia in Germany (IKSMS n.d.-d, p. 10). To do so, the IKSMS published river basin management plans for the international area of the Moselle and Saar river basins for the periods between 2010 and 2015 and 2016 and 2021 (ibid., p. 10). The plan for the years 2016–2021 juxtaposes the load of substances detected in the different countries’ river waters (ibid., p.  19ff); the countries’ application of the EQS (according to 2008/105/EC or 2013/39/EU); the chemical and ecological good water status across the different regions (ibid., p. 36ff); and the regions’ monitoring stations (ibid., p.  42ff). It further lists the substances relevant to the Moselle and Saar river basins (ibid., p. 50ff); the different national threshold values (ibid., p. 55); the environmental goals and the degree of reaching them (ibid., p. 58ff); the different types of water use across the jurisdictional areas (ibid., p. 68ff); and the measures taken by the countries (ibid., p. 78ff). In this way, the plan serves as orientation for the responsible state authorities in the six jurisdictional territories to improve the coordination of their actions and measures. The micro-pollutant management process in the Moselle basin is a cross-border case study comprising the Moselle and its tributary Sauer on the Luxembourgian side and the Moselle’s section flowing from Luxembourg through German territory

One interviewee stated Luxembourg’s lagging behind bluntly: “Au niveau européen, principalement aussi encore au Luxembourg, même si on est aussi un pays apparemment même très riche, on a toujours des problèmes pour garantir au niveau de l’épuration des eaux usées le standard actuel, disons, c’est à dire élimination des éléments nutritifs, nitrates et tout ça.” English translation: “At the European level, in Luxembourg, even though it is an apparently quite rich country, we still have troubles to guarante the current standard of sewage disposal, that is, the elimination of nutrients, nitrates and all that” (Interview N° 17). 61  Interview N° 16. 62  Telephone call N° 1; Interview N° 17. 63  Telephone call N° 1.

3.4  The Data Collection Process

93

into the Rhine. Micro-pollutants are an issue in both territories. Surface water of the Moselle is only partially used for drinking water—in Luxembourg, drinking water is obtained from its tributary Sauer; in RLP, the share of riverbank filtrate in drinking water is low, but the resource is needed for fishing. The case study region lies within EU borders, as both Luxembourg and Germany are EU members. Due to the river’s geographical location between the two countries, Luxembourg and Germany are jointly responsible for the management of the Moselle waters. Regulations to tackle the environmental problems are being discussed and developed in both countries. On the cross-border level, the international commissions for the protection of the Moselle and the Saar (IKSMS) contribute to an appropriate management of the river water across the neighboring nation states according to the EU Water Framework Directive. The commissions acknowledge the existence of polluting substances in the rivers’ waters and the problems resulting thereof (IKSMS n.d.-b).

3.3.6  Similarities and Differences Between the Case Studies Table 3.8 summarizes the similarities and differences between the case studies, listing the SESF characteristics and how they apply to each case study.

3.4  The Data Collection Process For a case study, the researcher investigates several sources of evidence: primary and secondary documents, observations of the event, and “interviews of the persons involved in the events” (Yin 2003, p. 8). To obtain detailed information on the context of the “event” under study, I assessed primary documents. To know who is involved in the event, I identified each case study’s corporate actors. To observe the events, I conducted a survey. I tested the survey’s questionnaire through expert interviews. These interviews further provided background information on the context of the management process in each case study region. The survey was the tool to gather the network data (Borgatti et al. 2013, p. 44ff). The survey’s response rates differ across the case studies. A closer look at the non-respondents reveals which actors did not reply the questionnaire. The timeframe of the analysis is the same in all case studies. Actors were asked to indicate the current state of the art of the micro-pollutant management process in their region and to also reflect the last few years. The time of data gathering and analysis was the year 2016.

Causing m-p: households WWTP settlements chem. industry agriculture Eliminating m-p:a WWTP and activated charcoal treatment Pharmaceuticals: carbamazepine sulfamethoxazole Herbicides: isoproturon s-metolachlor terbuthylazine Biocides: mecoprop

RS4 Human-­constructed facilities

RS5 Productivity of system; RU1 Mobility; RU2 Growth and replacement rate; RU3 Interaction among RUs; RU6 Distinctive characteristics; RU7 Spatial; and temporal distributionb

Rhine basin on the territory of the canton Basel City

RS2 Clarity of system boundaries

RS1 Sector

Basel Rhine surface water

Table 3.8  Overview of the case studies’ SESF characteristics

Pharmaceuticals: carbamazepine sulfamethoxazole Herbicides: isoproturon metolachlor terbuthylazine Biocides: diurone and mecoprop Others PFAS

Pharmaceuticals: diclofenac Herbicides: isoproturon metolachlor

Eliminating m-p: WWTP

Eliminating m-p: WWTP

Causing m-p: households WWTP settlements agriculture

Moselle basin’s hydrological boundaries on the territory of RLP and Saarland Causing m-p: households WWTP settlements winegrowing

Moselle Germany Moselle surface water

Ruhr basin’s hydrological boundaries

Ruhr Ruhr surface water

Pharmaceuticals: sulfamethoxazole Herbicides:c metolachlor metazachlor Biocides: diurone

Luxembourg Moselle and Sauer surface water Moselle basin’s hydrological boundaries on Luxembourgian territory Causing m-p: households WWTP settlements agriculture

94 3  Methods and Cases

GS6 Rules-in-use

GS4 Regime typee GS5 Rule-­making organi-zations

Regional

EU National

AUEBS, AUEBL WPA WPO

The Federal Assembly

Cantons Basel City and Basel Country Directorial system — The Federal Council,

Moselle-Saar German water policy

Within Rhine sub-catchment

(continued)

Luxembourgian water policy Federal state North Federal states Rhineland-­ Grand Duchy of Rhine-­Westphalia Palatinate and Saarland Luxembourg Parliamentary democracy Parliamentary democracy Parliamentary democracy European Parliament & European Council The Federal Government, Grand Duke of Luxembourg, The Bundesrat (upper house of parliament), Council of State, The Bundestag (lower house of parliament), Chamber of Deputies BMUB, BMEL MKULNV MUEEF, MUV – EU WFD EU WFD EU WFD WHG & OGewV WHG and OGewV Laws LWG LWG and SWG A217 and A690 River basin mgmt. plans River basin mgmt. plan River basin mgmt. and strategy plan IKSMS river basin management plans

GS2 Geographic scale of GS

GS1 Policy area

RS9 Locationd

Seasonal peaks: Application on vineyards and fields; precipitation Within Rhine subWithin Rhine catchment High Rhine sub-catchment German Lower Rhine Swiss water policy German water policy

RS7 Predictability of system dynamics

3.4  The Data Collection Process 95

Ruhr Monitoring “Multi-barrier approach” Feasibility studies and upgrade to 4th treatment stage

Basel Monitoring Research Bans Compensation for preventive mea-sures in agriculture WWTP and DWPf upgrading

Moselle Germany Monitoring Source-directed measures Feasibility studies on 4th treatment stage Luxembourg Monitoring Feasibility studies on sewage treatment Research on reduction of pharmaceutical residues

a

Note: wastewater treatment plants are being upgraded to a 4th treatment stage that is capable of filtering micro-pollutants. There are still substances that pass even those advanced filters and enter the water cycle b The substances listed here are the ones detected in concentrations above the limiting value in the case study regions throughout the last years c Even more herbicides were found in Luxembourgian groundwater: the herbicides bentazone, isoproturon, metazachlor, s-metolachlor, and terbuthylazine and the biocide diurone d See ICPR (2019) e Der Bundesrat (2017); Le Gouvernement du Grand-Duché de Luxembourg (2019); Schüttemeyer (2007) f DWP drinking water plant

GS8 Repertoire of norms and strategies

Table 3.8 (continued)

96 3  Methods and Cases

3.4  The Data Collection Process

97

3.4.1  Document Analysis The starting point for the document analysis was the relevant legislation for each of the case studies regarding water governance in general and micro-pollutants in particular. I next searched for actors’ statements referring to the laws and relating to the topic of micro-pollutants.64 This second step revealed: a) Actors involved in the CPR problem of micro-pollutants b) The variety of aspects that make up the CPR problem’s complexity65 c) Actors’ positioning towards the issue66 I also identified a range of reports and studies provided by state officials, scientific institutions, water associations, and service providers as well as press releases on the topic of micro-pollutants for each case. I scrutinized the relevant European legislation for the two European territory cases. The European directive 2000/60/EC— the Water Framework Directive (WFD)—treats the issue of micro-pollutants and requests river basin management plans from the EU member states (EU 2000)). I therefore examined for the river basin management plans drafted by the state officials in the pertinent case study regions. The primary data used for the analysis comprised: –– Laws regarding the environmental problem in the case study regions –– Amendments to laws and the official statements of interest groups –– Reports and studies on the issue from the case study regions’ official authorities, scientific institutions, NGOs, water associations, and service providers –– Regional river basin management plans The documents provided information for the case study regions on: • • • •

Strategies of action and measures regarding micro-pollutant management Causation, prevalence, and extent of the environmental problem Actors involved in the management process Potential conflict lines regarding micro-pollutant management

The information on actors obtained from these documents served as a basis for the actor identification process.

 For example, by considering the evaluation of statements to the amendment of the WPA by the State Office for the Environment in the case of Basel, see BAFU (2012). 65  For instance, actors would comment on the advantages and disadvantages of the different filter techniques; on the continuing water pollution through different sources, see Cercl’Eau (2010); and the uncertainty about micro-pollutants’ effects, and the undervalued relevancy of agriculture, see Rheinaubund (2012). 66  For instance, the Basel chamber of commerce (HKBB) opposed the amendment of the Water Protection Act, while the Swiss Farmers’ Union (SBV) approved it; see HKBB (2012); SBV (2012)—both actors belonging to the actor group “polluter.” 64

98

3  Methods and Cases

3.4.2  Actor Identification This book analyzes cooperation between actors engaged in the water policy sector in general and the management process of micro-pollutants in particular for the three case study regions. Mayntz and Scharpf (1995) highlight the importance to include every actor relevant in the pertinent field under study when analyzing actor constellations within a policy sector (Mayntz and Scharpf 1995, p. 44). Following this logic, I consider all actors within the case study regions who have a stake in the CPR problem setting. The analysis focuses on a variety of actor groups, reaching from political actors (e.g., state departments and regional state authorities) to political associations (e.g., interest groups and NGOs), resource users, polluters (e.g., wastewater treatment plants and agricultural actors), and service providers. I previously identified the actors in the three case study regions based on two concepts: the SES framework and the reputational approach (Knoke 1998). The first concept enabled me to find the relevant actors through a guided but broad search, covering various sectors. The second confirmed that I had identified the right actors and that all key actors had been included. 3.4.2.1  Based on the SESF The SESF serves to identify the actors involved in a CPR problem. With regard to the resource unit different types of surface water use, I searched for the surface water users in each case study. I distinguished between direct and indirect users based on the three criteria: –– Intensity of surface water use: actors use the resource surface water on a daily base for their work or their living—direct use. –– Sensitivity of surface water use: the surface water’s quality is essential for the type of use—direct use. –– Importance of surface water: actors are dependent on surface water for their work; they work on the topic surface water quality—indirect use. This approach helped me identify a range of environmental NGOs and scientific actors (indirect users); service providers, such as water works (direct users); consumer interest groups (direct users); and polluters, like wastewater treatment plant operators and pharmaceutical firms (direct users).67 To find the users, I searched reports, studies, and official statements for indications of resource users. Next, I searched for the resource users via the Internet, browsing the homepages of water-­

 In the instance of the pharmaceutical firms, use is understood as the disposal of chemical sewage into surface water, thereby creating the CPR problem of micro-pollutants in surface water. WWTP operators are polluters in that they release substances they were not able to filter into the water cycle.

67

3.4  The Data Collection Process

99

related institutions in the case study regions for information on further regional resource users. I based my identification of actors related to the governance system on McGinnis and Ostrom’s (2014) categorization of rules-in-use. They distinguish operational-­ choice, collective-choice, and constitutional-choice rules (McGinnis and Ostrom 2014). Constitutional-choice rules determine who is responsible for formulating collective-choice rules. Collective-choice rules are used by appropriators, state officials, and authorities to derive policies that guide the management of a CPR.  Operational rules—like policy measures—are implemented (Ostrom 1990, p. 52). Based on the management plans and, again, on the reports and studies of each case study and on the conceptual distinction of different rules-in-use, I identified two types of GS-related actors:68 –– Actors responsible for defining policy instruments and working out measures at the collective-choice level –– Actors responsible for the implementation of instruments and measures and addressees of measures and instruments at the operational-choice level69 3.4.2.2  Through the Reputational Approach This first phase of the actor identification process resulted in a preliminary actor list for each case study. I had the lists evaluated by the experts interviewed70 to check whether important actors were missing, following the logic of the reputational approach for assessing the “important players” within a policy network. Reputation can be defined as actors’ attribution of social status (cf. Knoke 1998, p. 509); it is a “group judgment” of an actor based on the other actors’ perception of his/her attributes and attitudes (ibid., p. 509). The reputational approach is a “sociometric technique” (Abu-Laban 1965, p.  35) originally used to identify local leaders in communities (ibid., p. 35f.). Experts with knowledge about the social group under study are asked to select and grade from a list of actors those they consider influential or to create a list themselves of the influential actors of the respective group (Abu-Laban 1965, p. 35f.; Scott 2000, p. 56). Influence reputation is understood as the impression members of a policy network have about each member’s capacity to influence the outcomes of a policy process (cf. Knoke 1998, p. 508). In my research, I refer to the influence reputation when asking experts to judge the actors on the list regarding their importance in the management process of micro-pollutants.  Since the research focus is on the adoption and implementation of instruments and thus on the collective-choice and operational-choice levels as defined by McGinnis and Ostrom (2014), I did not include the actors responsible for making laws at the constitutional-choice level. 69  To name just a few instruments: guidelines on the amount of micro-pollutants permitted in surface water; the order for users to stay within the recommended limits or to reduce micro-pollutants in surface water; or the declaration at which concentration micro-pollutants should be considered a serious threat; see Metz and Ingold (2014). 70  For the list of interviews, see Table 3 in Annex IV. 68

100

3  Methods and Cases

Furthermore, I requested the experts to tell whether any important actors were missing. Through this approach, I ensured the identification of the relevant actors within the management process of micro-pollutants in each case study region. The two procedures produced 51 relevant collective actors in the Basel case, 39 in the Ruhr case, and 44 in the Moselle case. Actors comprise public and private entities from different sectors, such as the industry, civil society, water associations, science, and the state. To ensure I do not miss any actors relevant to the management process, I kept the boundaries of the actor sample broader and further included actors that the experts were undecided about regarding their importance. With this more extensive actor sample, the actor lists in the questionnaire comprise 60 actors in the Basel case, 55 in the Ruhr case, and 54 in the Moselle case. In each organization, I contacted the person highest in rank responsible for the topic of micro-pollutants.71 As would become apparent throughout the data collection process, the additional actors I considered in the wider actor sample neither proved to be important based on the judgment by the other actors, nor did they regard themselves as relevant: several of them answered via e-mail that they had no stake in the issue of micro-­ pollutants. For the analysis, I thus take the sample of relevant actors into account that amount to 51, 39, and 44 actors for the Basel, Ruhr, and Moselle case, respectively. Figure 3.8 visualizes the distribution of the different actor groups across the cases with the entire actor samples and the total number of actors. The share of the different types of actors is relatively even across the cases except for the actor-type NGO and science. In the Ruhr case, there is only one NGO, making up only 2.6% of the actor sample, while in the Basel case, five actors are NGOs, constituting 9.8% of the case’s entire actor sample. In the Moselle case, scientific actors amount to only 9.1% (four actors), and in the Basel and Ruhr cases, they have a share of 17.6% (nine actors) and 23.1% (nine actors), respectively. Luxembourgian actors (59.1%) outnumber German actors (38.6%) in terms of actors’ territoriality in the Moselle case study region—see Fig.  3.9 that depicts actors’ territoriality across the case studies, considering the entire actor samples.72

3.4.3  Expert Interviews I interviewed experts in each case study region in order to: a) Pretest the questionnaire b) Validate the actor list c) Confirm I had identified the right laws regarding micro-pollutant management d) Collect information on the contextual background of each case study 71 72

 For the cases’ actor lists, see Tables 4, 5 and 6 in Annex V.  See Tables 7, 8 and 9 in Annex VI for the statistics.

3.4  The Data Collection Process

100% 90%

1

101

2

3

9

4 9

80% 70%

7

60%

5

50%

6

40%

5

30% 8

20% 10% 0%

11

10

5 6

4 2

5 1

9

6 7

Basel National Pol. Actor Service Provider

6

3

Ruhr Regional Pol. Actor Polluter

Moselle NGO Science

Water Association Consumer Organiz.

Fig. 3.8  Distribution of actor types across the case studies; entire actor samples, total number of actors

I chose the experts based on three criteria. They had to be in a leading position within their department or organization, and they had to work on the issue of micro-­ pollutants. The organization they worked for had to have an important role within the management process of micro-pollutants. I interviewed CEOs and heads of office with a thorough understanding of the regional water quality policy regarding micro-pollutants and the actors involved in it. I conducted two preliminary interviews with two drinking water providers in Basel to gain first insights about the field in fall 2015. These interviews were guided by certain questions, but generally open-ended in nature (Yin 2003, p. 90). In spring 2016, I interviewed six experts for the Basel case, seven for the Ruhr case, and five for the Moselle case. They were focused interviews (cf. ibid., p. 90): In the beginning of each interview, I asked the expert to check the actor list and tell me whether actors important for the respective management process of micro-pollutants were missing. The experts would also tell me if I had actors on the list that were irrelevant to the management process. Based on this information, I edited the actor lists of the case studies. I further asked about the current legislation on micro-pollutant management to ensure I had identified the right laws. After these first checks, I guided

102

3  Methods and Cases

100%

0,00% 0,00%

5,90% 0,00%

2,30%

9,80%

90%

7,80%

80%

59,10%

70% 60% 97,40%

50% 76,50%

40% 30%

38,60%

20% 10%

0,00% 2,60%

0%

Basel Swiss

French

German

0,00%

Ruhr

Moselle

Luxembourgian

International

Fig. 3.9  Actors’ territoriality across the case studies; entire actor samples

the experts through the questionnaire, making a note when there were questions about the questionnaire. After the interviews, the questionnaire needed only a slight revision to be improved. Apart from these information related to the survey, experts would also provide me with information on the case study’s background. I conducted two more interviews with experts working in two water associations, one of which treats cooperation among Rhine riparian states, and the other one monitors the Rhine’s surface water quality. In these last two open-ended conversations, I posed general questions about the issue of micro-pollutants in Rhine surface water and its management by the Rhine neighboring states. The interviews were either recorded and then transcribed or documented by written notes. The transcripts and notes serve as data for the case studies’ contextual analysis.

3.4  The Data Collection Process

103

3.4.4  The Survey A survey served to collect the network data. I developed the survey questionnaire in line with my hypotheses73 and under consideration of the dependent variable. The questionnaire comprises thus questions about actors’ collaborative connections, about their exchange of information exchange, about their aims regarding the management process, and questions concerning the explaining factors for cooperation as outlined in the hypotheses and the control variables. As the cases’ actor lists were after their final alteration, the survey’s questionnaire has a closed-ended format (Borgatti et al. 2013, p. 47). I sent the questionnaire to individuals in key positions regarding the topic of micro-pollutants within the identified organizations. Key positions comprised CEOs, heads of department, heads of working-groups, and professors. By this, I ensured that the persons who received the questionnaire were able to respond to its questions on behalf of their organization. The questionnaire was sent together with an introductory letter via e-mail and post. Data was collected for all three cases throughout the following periods: –– Basel case study: 18 April 2016–24 August 2016 –– Ruhr case study: 5 September 2016–6 February 2017 –– Moselle case study: 5 September 2016–17 February 2017 During the data collection period, I sent up to five reminders via mail to all those actors who had not yet answered the questionnaire. I telephoned all actors who after the fifth mail-reminder had not yet reacted. In the following, I depict how the data was collected through a survey. For a detailed description of the data preparation process, see Annex VIII. 3.4.4.1  Collecting Data on the Dependent Variable The dependent variable cooperation consists of the four elements already introduced: a) Actors aiming towards the same goal b) Coordination of each other’s actions c) Actors’ resource exchange d) Trust (Henry and Dietz 2011; Ostrom 1998; Ostrom 2000a; Ostrom 2000b; Sadoff and Grey 2005) The first three elements are assessed through survey questions, the last one through an analysis of actor interactions’ reciprocity (Cranmer et al. 2017, p. 241).

 See Annex VII for the entire questionnaire. Figure VII shows the German version of the questionnaire for the Basel case study. This questionnaire is representative for the other two questionnaires that were sent to the actors in the other two case study areas and adapted accordingly (i.e., the actor lists and the names of the rivers differ).

73

104

3  Methods and Cases

Cooperation defined as intended and well-directed interaction can be understood as a network, because, as Jakobi (2009) and Scharpf (1993) argue, an “established pattern of interaction between different actors that are interested in a common subject matter” (cf. Jakobi 2009, p. 4) can be defined as a network (Jakobi 2009, p. 4; Scharpf 1993, p. 72). I thus conceptualize two of the constituting elements of cooperation as social networks: Coordinating actions (b) is operationalized as actors’ collaboration network. To gather the network data on actors’ collaboration (b), I asked the closed-ended question: With which actors has your organization been closely collaborating within the management process of micro-pollutants throughout the last years? (Question N° 5)

I defined close collaboration in the CPR management process as discussing new findings on the issue, working out possible courses of action regarding the management of micro-pollutants, exchanging viewpoints on the topic, and accomplishing joint projects regarding micro-pollutants.74 The definition was stated right below the question. A roster of the actor list followed in which the actors could mark with a cross the organizations they collaborate with. The answers were then transposed into a matrix (cf. Borgatti et al. 2013, p. 63f.). Resource exchange (c) is translated into actors’ exchange of information. I choose the resource information over finances or technical support (cf. Sadoff and Grey 2005, p.  424) and distinguish between political and technical information exchange. I do so since actors seek both types of information to learn about each other’s strategies and to acquire knowledge about the policy problem at stake and inform their decisions regarding this policy problem (Henry and Dietz 2011; Leifeld and Schneider 2012; Sabatier 1987). Actors’ information exchange is operationalized as one actor network of technical information exchange and another actor network of political information exchange. Ties in these networks represent the exchange of information between actors. Data on these networks was collected through Question N° 6 about actors’ information exchange: To whom of the actors did your organization send technical and political information and from whom of the actors did your organization receive technical and political information in the last years?

Definitions of the two information types were added to the question. Technical information is defined as information about the newest technical standard regarding the treatment of micro-pollutants; about the newest insights on micro-pollutants’ impact on aquatic organisms, animals, and humans; and about measuring data and analytics. Political information is described as information concerning political matters and enabling the actor to participate in political debates, to negotiate positions, and to take actions. This question was posed in a multigrid format: the actor list was set in front of a series of columns, each column indicating one of the ques-

 Conceptually, collaboration is defined as the working together of two or more actors, which produces an outcome the collaborators benefit from (cf. Sect. 1.2). The difference to cooperation is that the actors do not aim at a common goal and do not exchange resources.

74

3.4  The Data Collection Process

105

tions (cf. Borgatti et al. 2013, p. 50). As was the case for actors’ collaboration network, the answers to these questions were transposed into matrices representing networks of sent and received political and technical information. Aiming at the same goal (a) is operationalized as a similarity index constructed out of five objectives actors are asked about regarding the goal of micro-pollutant management. Question N° 8 asked the actors to indicate their agreement on the five following goals regarding the management process of micro-pollutants in the case’s catchment area that represent extremes of what the management of micro-pollutants should aim at. The answers could be given on a Likert scale from 1, do not agree at all, to 4, fully agree: –– Measures should focus on the source of pollution. –– Measures should be end-of-pipe. –– As long as the effects of micro-pollutants are not yet fully understood, preventive measures should be applied (precautionary principle). –– As long as the effects of micro-pollutants are not yet fully understood, no measures should be taken. –– Measures should aim at cleaning surface water virtually completely from micro-pollutants. Actors’ evaluation of these five objectives reflects their attitude towards the goal that a management process of micro-pollutants should attain. Based on the actors’ answers about their agreement to these goals, I constructed a goal similarity index. The index value ranges between 0 and 1 for each pair of actors and indicates how similar each pair of actors evaluates the goal of the management process. 0 means two actors do not at all aim at the same goal within the management process; 1 represents that both actors agree entirely on the management process’s goal—that is, they evaluated the five different objectives in the same way. The fourth element of cooperation, trust (d), is determined through an analysis of the reciprocity of actors’ collaboration ties. The data used to determine actors’ trust is thus the data of actors’ collaboration network. The method to assess reciprocity within the collaboration network is descriptive and inferential Social Network Analysis (SNA)—I present both methods in Sect. 3.5. 3.4.4.2  Collecting Data on the Independent Variables To obtain data for the independent variables actors’ problem perception (IV 1a) and actors’ similar problem perception (IV 1b), I asked actors in how far they perceive seven different consequences of micro-pollutants’ occurrence in river surface water as problematic (Question N° 9). The potential effects of micro-pollutants were deduced from the literature on micro-pollutants’ impact on ecosystems and humans (Bundschuh and Schulz 2011; Bunzel et al. 2013; Carvalho et al. 2014; Chèvre et al. 2006) and validated through the expert interviews. The seven “effects” comprise: 1) The uncertain interaction effects of micro-pollutants, i.e., the so-called cocktail effect of different merging micro-pollutants, that possibly generate a toxic effect

106

3  Methods and Cases

2) The negative effects micro-pollutants could have on aquatic organisms 3 ) The negative effects micro-pollutants could have on secondary consumers 4) Micro-pollutants’ possible carcinogenic effect 5) Raised levels of hormones in surface water 6) The persistence of certain micro-pollutants 7) Micro-pollutants’ uncertain effects on drinking water security The scale on which actors could evaluate these seven effects as a problem ranged from 1, no problem, to 4, severe problem. The values actors attributed to the different “effects” were aggregated into a perception similarity index. Its value ranges from 0—no similar perception of a pair of actors—to 1, complete similar perception of two actors. For the independent variables 2a and 2b, actors’ participation in forums and actors’ co-participation in forums, data was collected through two open-ended questions. Question N° 11a asked actors about their participation in national or international platforms, like working groups, water body commissions, or scientific platforms; Question N° 11b asked them about their international cooperation regarding micro-pollutants, i.e., their working together with actors from other countries (Lubell et al. 2010, p. 295; cf. Lubell et al. 2011, p. 18). The answers were transformed into a numerical variable for IV 2a which comprises the number of forums actors participate in and into a co-participation matrix for IV 2b, which reflects the number of forums participated in by two actors. Data on IV 3, actors’ same belief, is based on Question N° 2, with five keywords for reflecting three deep core beliefs. They represent normative and ontological axioms that define an actor’s philosophy (Sabatier 1988, p.  144f.; Sabatier 1998, p. 103). Economic efficiency and free enterprise economy/competition stand for the deep core belief representing actors’ positive attitude towards economy; the population’s safety and social justice refer to actors’ deep core belief in supporting societal standards; ecological compatibility reveals actors’ deep core belief about the importance to protect the environment. Regarding environmental policy, actors’ economy-­ friendly deep core beliefs that favor free market over state intervention are mostly opposed to environmental deep core beliefs that put the protection of the environment center stage and prefer the state to intervene for that (Ingold et  al. 2017, p. 447). Actors were asked to assign the priority (low, medium, high) that these five key criteria have for them in the management process of micro-pollutants. Out of actors’ answers, I constructed a belief similarity index ranging from 0 to 1—the closer to 1 the indicator value is, the more similar two actors’ beliefs are. 3.4.4.3  Collecting Data on the Control Variables All control variables (CVs) are dummy variables. The control variable regulatory actor (CV 1) distinguishes actors with the competency to design policy instruments and who actively shape the decision-making process. They act at the collective-­

3.4  The Data Collection Process

107

choice level, where policies are developed and decided upon (Ostrom 1990, p. 52). This categorization is based on information about actors’ tasks and roles within the respective management process of micro-pollutants, which I obtained from official documents. The assignment of actors’ attribute being an implementer (CV 2) was obtained in the same way. These are the actors who act at the operational-choice level, translating policy instruments into practice (McGinnis 2011, p. 173). Actors’ reputation (CV 3) indicates the actors judged important in the management process by more than a third of the actors. This information is covered by Question N° 4, which listed the actors in a multigrid format and asked the interviewees to indicate the actors they consider especially important in the management process in the second column and to mark the three actors they consider most important among all actors with a cross in the third column. Actors’ attribute pollution-sensitive water use (CV 4) was obtained through an analysis of each actor’s type of water use—based on official documents and information from the actors’ website revealing the actor’s activities and whether these activities can be considered sensitive to surface water quality. Actors’ territoriality (CV 5) reflects the nationality on behalf of which an actor acts. The control variables of the network effects (CVs 6.1–6.6) are inherent in the inferential network analysis technique (ERGM). They comprise the network’s density, which corresponds to the edges term that serves as baseline for all network effects and represents the networks’ number of edges; the network vertices’ in- and out-degrees, which are geometrically weighted and tested through the network terms gwidegree and gwodegree; the network ties’ reciprocity, controlled for with the mutual term; and the geometrically weighted edgewise shared partners, a term (gwesp) that controls for the configuration of two actors who share a tie also sharing a joint partner via an edge each (Hunter 2007; Hunter and Handcock 2006).75 Table 3.9 lists the variables and their operationalization. The questionnaire comprises further questions that provide information on each actor’s contextual background. Question N° 1 calls up the tasks of water management the actors work on. It asks actors to prioritize these tasks in relation to the task of micro-pollutant management. Question N° 3 investigates the geographical area actors cover with their work. Question N° 7 demands actors’ agreement or disagreement with the other actors regarding the management process of micro-pollutants. Question N° 12 asks actors to evaluate ten statements about the influence national frontiers can have on cooperation. As the study was part of the research project “CrossWater  – Transboundary Micropollution Regulation in Europe: The Definition of Appropriate Management Scales – An Interdisciplinary Approach”76, one question of the questionnaire relates to the project’s focus and research question. The project investigated the following

 Section 3.5.2 outlines the analytical method.  The research project was funded by the Swiss National Science Foundation (SNF) and the Luxembourg National Research Fund (FNR), Grant N°: CR21I1L_146336.

75 76

108

3  Methods and Cases

Table 3.9  Operationalization of the dependent, independent, and control variables N° a. b. c.

DV’s elements Aim towards the same goal Coordination of actions Exchange of resources

N° IVs 1a Problem perception and b

2a Participation in forum(s) and b

3 N° 1 2 3 4 5

6

Operationalization Actors’ goal similarity index (range of 0 to 1) Directed binary matrix of actors’ collaboration Directed binary matrices of actors’ technical and political information exchange Operationalization a) Interval variable: actors’ perception of the CPR problem, mean value (scale from 1 to 4) b) Actors’ perception similarity index (range of 0 to 1) a) Interval variable: N° of forums actors participate in b) Weighted, undirected, co-participation matrix (N° of forums two actors participate in) Actors’ belief similarity index (range of 0 to 1) Operationalization Dummy variable

Two actors share similar belief CVs Actor being a regulator Actor being an implementer Actors’ reputation Actors’ pollution-sensitive water use Actors’ territoriality (Swiss/ non-Swiss; German/ Luxembourgian) Network effects ERGM terms: edges, mutual, gwidegree, gwodegree, gwesp

research question: how can the potential mismatch between physical and jurisdictional/political areas be identified and understood; and what conclusions can be drawn for the development of future micropollution management and regulation? Question N° 10 of the questionnaire thus asks about actors’ level of agreement to 15 different policy instrument types tackling micro-pollutants.

3.4.5  Response Rates and Handling Missing Data The actors’ response rates range between 66.6% in the Ruhr case study and 72.5% in the Basel case study. In the case of Basel, I identified and contacted 51 actors out of which 37 gave valid answers.77 In the Ruhr case, 26 out of 39 actors responded, which results in a response rate of 66.6%. In the Moselle region, 31 out of the 44  Even though more than 37 actors answered, not all returned questionnaires could be used for the analysis, as some of them were incomplete.

77

3.4  The Data Collection Process

109

Table 3.10  Response rate statistics for each case study sample N° of actors contacted N° of responses Response rate in % Judged important by 40% of the actors (N°/%) N° of important actors who answered (N°/%)

Basel 51 37 72.5% 5/10% 5/100%

Ruhr 39 26 66.6% 7/18% 5/71.4%

Moselle 44 31 70.5% 7/16% 5/71.4%

identified actors answered the questionnaire properly, leading to a response rate of 70.5%. The cases’ response rates thus meet the standard needed for sound Social Network Analysis and are higher than the actor samples’ response rates of several SNA studies (Lubell et al. 2010, p. 294; Lubell et al. 2014). 3.4.5.1  The Respondents A closer look at who exactly responded the questionnaires and who did not reveals the following picture (see also Table 3.10): as the actors were asked to evaluate their most important peers within the management process, I was able to assess actors’ attributed importance. I did so by calculating actors’ in-degree centrality: I checked for actors’ incoming ties based on the data on reputation (Question N° 4). The number of an actor’s incoming ties tells how many of the other actors judge this actor as important. The benchmark by how many actors someone must be judged as ­important to consider him/her important for this analysis was set at 40% of each case’s actors.78 Table 3.10 shows the number and share of actors evaluated as important by this criterion for each case study and the number and share of these important actors who answered the survey. Note that I considered the entire list of all actors contacted for the evaluation of the important actors in each case. The reputational judgment by those who did not answer the survey is thus missing. The identification of the actors judged important by 40% or more based on the total actor list is informative for three reasons: a) The actors answering the questionnaire have a stake in the management process of micro-pollutants and thus know its important actors. b) The analysis shows that in the Basel case study, 10% of the actors are deemed important by 40% of the other actors, while this share is even higher in the Moselle (16%) and the Ruhr case studies (18%). c) All important actors answered the survey in the Basel case study; in the Ruhr and Moselle case studies, their share is 71.4%.

 In the Basel case, 40% of 51 actors amounts to 20.4 actors—I considered all actors with an indegree of 20 and above; in the Ruhr case, 40% of 39 actors amounts to 15.6 actors, I considered all actors with an in-degree of 15 and above; in the Moselle case, 40% of 44 actors is 17.6 actors, I considered all actors with an in-degree of 17 and above.

78

110

3  Methods and Cases

100%

1

2

4

3

90% 80% 70% 60% 50% 40%

3 7

7

8 5

5 3

5 5

5

30% 20% 10% 0%

2 1

2 1 2

8

8 2

2

3

Basel National Pol. Actor Service Provider

Ruhr Regional Pol. Actor Polluter

Moselle NGO Science

Water Association Consumer Organiz.

Fig. 3.10  Distribution of actor types across the case studies; actor samples of the respondents

The response rates of the cases are satisfactory, and the vast majority of each case study’s key actors responded the survey, ensuring this study to cover and grasp the right actors for each case. Since not all of the contacted actors answered the survey, the different backgrounds of the actors are not even among the cases. Figure 3.10 shows the distribution of the different actor groups for the actor samples of the respondents in entire numbers.79 The highest discrepancy is within the group of regional political actors. While in the Ruhr case study only two actors (7.7%) belong to this group, eight actors (25.8%) in the Moselle and six actors (16.2%) in the Basel case studies are regional political actors. In the group of scientific actors, we see a similar difference across the cases: in the Moselle case study, their share is lowest with 9.7% (three actors), while they are the largest group in the Ruhr case study, amounting to 26.9% of all actors (seven actors). The Basel case study ranges in between, with 16.2% (six actors). Service providers are also unevenly represented across the case studies: in the Moselle region, only three actors (9.7%) belong to this group, while it is double that value in the Ruhr case study. Basel, again, lies in between with five actors (13.5%) from this group. For the other actor type groups, one can state a somewhat similar distribution across the cases. There are two exceptions for the Basel case

79

 See Table 8 in Annex VI for the percentages.

3.4  The Data Collection Process

111

study, in which the share of NGOs and water associations is higher than in the other two case studies.80 Overall, each actor group is represented in all case studies. Of the three case studies, Basel has the most regular distribution of actor types. In the following, I turn the focus to those actors who did not respond the survey to see what kind of actors and whether relevant ones are missing from the study. 3.4.5.2  The Non-respondents Basel Case Study The 14 non-respondents of the Basel case study are two regional state actors from Germany and France each, as well as one regional state actor from Switzerland.81 Furthermore, three national Swiss state actors did not answer the survey—one of which did partially answer the questionnaire but missed out on questions essential for the analysis. An explanation for these three actors for not—or only partially— answering the survey could be that they felt “too far” away from the regional level on which the survey focused. Three polluters from the chemical industry and agriculture did not respond, as well as two Swiss scientific institutions and a German water association. The non-respondents were judged important by between 14% and 28% of the actors with the majority being deemed relevant by 22–24% of the 39 actors who answered the questionnaire. These numbers show that a) all identified actors are at least to some extent important for the management process and that b) the non-respondents are not the key actors in the micro-pollutant management process in the Basel region. Ruhr Case Study In the Ruhr case study, two highly important regional state actors did not answer the survey: the district administration of Arnsberg, responsible for the implementation of the WFD in the Ruhr region and judged as important by 46%, and the State Office for Nature, the Environment and Consumer Protection in NRW (LANUV) deemed important for the management process by 51% of the actors. I interviewed a representative of the LANUV, but the person did not answer the entire questionnaire. The two other regional state actors (Landtag NRW; RVR) and the one national state actor (BMUB) who did not respond are by far less important for the management process. One of them, the RVR, responded that they were incapable of answering the questionnaire, as the organization has no competences regarding the topic. Two

 In the Basel case study, five actors are NGOs (13.5%), whereas in the Ruhr and Moselle case studies, only one actor is a NGO (3.9% and 3.2% of the actors, respectively). 81  For the cases’ lists of non-respondents, see Tables 10, 11 and 12 in Annex IX. 80

112

3  Methods and Cases

polluters, from the chemical industry and agriculture, did not answer the survey as well as two scientific institutions, which are—after all—thought to be of importance by 20% of the surveyed actors. Furthermore, three water associations and one service provider are non-respondents. Three of them are considered important by 28% of the actors. One of the water associations (DVGW) called after having received the questionnaire, explaining why it could not respond the survey and offering a range of valuable insights on the topic.82 The actor sample of the Ruhr case study is thus missing two key actors of the region’s management process of micro-­pollutants: the district administration of Arnsberg and the LANUV. Two of the contacted actors further explained that they were incapable of answering the questionnaire in a way that would satisfy its purpose. Moselle Case Study In the Moselle case study, three Luxembourgian state actors—two of which considered important by 50% and 34%, respectively—did not respond to the questionnaire. The water protection division of the Luxembourgian administration for water management83 was not capable of answering all questions, but willing to discuss the questionnaire, presenting their viewpoint on the topic.84 The national ministry of health and the national management of health did not provide any answers for why they did not answer the questionnaire. Three polluters did not answer the survey either—one being from the agricultural and two from the settlement sector. Two Luxembourgian regional service providers did not respond, as well as one German scientific institution. Department VI—Regional and Environmental Sciences—of the University of Trier clarified that most of the questions did not apply to the department’s area of work. The department was not involved in the management process so far and thus unable to evaluate the questionnaire.85 On the German side, an environmental NGO, a regional state actor responsible for consumer and health protection and the regional state association of the DVGW did not answer the survey. Finally, the international water association coordinating the cross-border river basin management of the Moselle and Saar (IKSMS) did not respond either. The Moselle case study sample is thus missing three relevant actors of the management process: the Luxembourgian administration for water management, the Luxembourgian Ministry of Health, and the IKSMS.  At the same time, two actors explained why they were incapable to answer. The number of important actors who did not respond is considerably low. In the case of the Ruhr, I can rely on the interview with the LANUV to add this actor’s

 Telephone call N° 4.  Original name: L’administration de la gestion de l’eau du Grand-Duché de Luxembourg. 84  Telephone call N° 1. 85  e-mail N° 1. 82 83

3.4  The Data Collection Process

113

Table 3.11  List of imputed values Variable and its N° of values DV: Aim 5

Case Ruhr Moselle

IV 1: Problem perception

7

Basel

IV 3: Deep core beliefs

5

Moselle Ruhr Moselle

Actors RWTH Aach WLV Hospitals.LUX FLPS MWVLW.RLP SGD CITYWeil Eawag Hospitals.LUX MKULNV ULC

N°/% of imputed values 1/20% 3/60% 1/20% 1/20% 1/20% 3/60% 3/43% 1/14% 1/14% 5/100% 3/60%

viewpoint to the qualitative analysis. In the Moselle case, insights from the phone call with the Luxembourgian administration for water management can enrich the case’s qualitative assessment. Apart from the non-respondents, some actors did not answer certain questions completely. To keep these actors in the cases’ actor sample, I imputed the missing values. 3.4.5.3  Handling Missing Data Returned questionnaires that did not answer the network questions on collaboration and information exchange—Questions N° 5 and 6—were excluded from the sample. For those actors who answered questions concerning their attributes—the independent variables—only partially, I imputed the missing values using the mice package in R (RStudio Team 2016).86 The algorithm uses Gibbs sampling and imputes the missing values of a variable in consideration of that variable’s values of the other actors. The algorithm thus “imputes an incomplete column (the target column) by generating ‘plausible’ synthetic values given other columns in the data” (RStudio Team 2016, mice::mice). Table 3.11 lists the actors who did not answer all questions entirely and the number of the variable values that were imputed. The table shows that values were imputed for three different variables and ten actors. The MKULNV is the only actor that did not answer an entire question—Question N° 2 on deep core beliefs. For this actor, I had to impute all of this variable’s values. The other actors did not answer single parts of questions. Their answers were therefore close to completeness.

86

 “mice” stands for Multivariate Imputation by Chained Equations.

114

3  Methods and Cases

3.5  The Data Analysis Methods The study’s methodological approach follows Ostrom’s logic of the interplay between framework, theory, and model (cf. Ostrom 2005, p. 27f.): the SES framework delivers the structure into which I fit the object of study. Theory-based hypotheses emphasize the factors relevant for the analysis, the independent variables. The methods Social Network Analysis (SNA) and the case comparison were used as models for analyzing cooperation and its triggers in the management process of a CPR problem. Actors’ information exchange and collaboration networks function as the social system I assess (cf. Wasserman and Faust 1994, p. 93). I analyze the cases’ collaboration and information exchange networks’ properties through descriptive SNA, thereby gaining insights into actors’ collaboration and information exchange patterns as well as into the relation between these two types of interactions. To test the hypotheses—to test the independent variables’ influence on the dependent variable—I apply the inferential SNA technique of the exponential random graph model (ERGM) family. The analytical technique reveals the factors that enhance the likelihood of a tie’s existence in the collaboration networks under study. I interpret the models’ results by analyzing the independent variables’ distinctness across the cases and comparing them as well as the contexts of the case studies. The following sub-chapters outline descriptive SNA, explain the ERGM technique, and discuss the technique of qualitative case comparison.

3.5.1  Descriptive Social Network Analysis Social Network Analysis (SNA) evolved from sociological, anthropological, and social psychological efforts in the 1930s. It aimed at studying human social and psychological behavior and at picturing and understanding social groups and their structures and structure inherent patterns (Scott 2000, p. 7ff; Wasserman and Faust 1994, p. 10ff). Interest in the theoretical reasoning behind certain network structures like cliques, transitivity, or social position gained momentum and pushed the evolution of this method (Wasserman and Faust 1994, p. 13f.). In parallel, mathematical approaches from probability and statistical theory and algebraic models were integrated into the “new” practice (Scott 2000, p.  33; Wasserman and Faust 1994, p.  15f.). SNA provides a set of methods to assess social structures (Scott 2000, p. 38): it comprises the basic concepts constituting social networks;87 a variety of methodological techniques to assess network structures and actors’ positions within them;88 and statistical methods to assess relations among different network struc “These concepts are: actor, relational tie, dyad, triad, subgroup, group, relation and network”; see Wasserman and Faust (1994, p. 17). 88  Like cliques, triads, block models, or components; actors’ eigenvector, betweenness, or reach centrality; a network’s density, connectivity, or connectedness. 87

3.5  The Data Analysis Methods

115

tures and between nodes’ attributes and network relations. I use SNA as analytical tool to assess actors’ cooperation in a CPR problem setting and to identify its triggers. A social network functions as a model of the social system under study (Wasserman and Faust 1994, p. 93). A social network can be understood as a form of coordination of interactions. Its core comprises the trustful cooperation of autonomous, but interdependent, actors who work together for a certain amount of time to achieve their particular goals. While working together, they pay regard to the interests of the other actors (Weyer 2011, p. 49). As an element of social organization, a network facilitates coordination and cooperation for actors’ mutual benefit (Crona et al. 2011, p. 57; Provan and Kenis 2008, p. 233). The notion of cooperation is thus already inherent in the definition of a social network. A policy network can be described as: a heterogeneous set of persons or organizations, linked by one or more relationships into an enduring social structure with the potential to influence public policy decisions of interest to the network’s members. (Knoke 1998, p. 508)

The analysis of network data allows highlighting certain features of a network. One can “zoom in” on a specific phenomenon within the network and assess possible explanations for its development. Social Network Analysis fits the analysis of a social-ecological system, because: SNA has been applied to social-ecological systems to study how network structure and variation in connectivity influence network performance and the behavior of individual actors. (Bergsten et al. 2014, p. 4)

SNA helps reveal the structures and dynamics of the social-ecological system’s action situation, that is, cooperation within the management process of a CPR problem. SNA enables me to comprehend actors’ interaction pattern and reveal the degree of intentional interaction among actors. Through SNA, I can identify social actors’ information exchange patterns, their connectedness among each other, and their structural positions within the management process. A Social Network Analysis of actors’ information exchange as well as collaborative interactions reveals the degree of actors’ cooperation. Certain network characteristics will be relevant for the descriptive SNA. 3.5.1.1  A Graph, an Edge, and Vertices A social network consists of nodes, the actors, and the links between these nodes, the actors’ relations.89 The set of vertices, nodes, and the set of edges, links, that make up the network are called graph (Borgatti et al. 2013, p. 11f.). A graph is thus a network’s mathematical representation (ibid., p.  11). The relation between two

 Nodes can also be called points or vertices while ties between actors are also referred to as edges or lines, see Wasserman and Faust (1994, p. 95).

89

116

3  Methods and Cases

vertices is called a dyad: it is the “pair of actors and the (possible) tie(s) between them” (Wasserman and Faust 1994, p. 18). Two actors that have a line between them are adjacent to each other (Borgatti et al. 2013, p. 12; Scott 2000, p. 67; Wasserman and Faust 1994, p.  95). Actors’ relations can be of different nature: they can be resource exchange, communication, and information exchange, shared points of view on a certain topic, the mutual perception of each other, a biological relationship, or a behavioral interaction (Schneider 2014, p.  274; Wasserman and Faust 1994, p. 18). They can further be structurally different. Relations can be undirected when the tie’s direction towards one or the other actor is not relevant. For instance, in the case of two actors who are married, the line’s direction is trivial as they are both bound to each other by marriage. Ties between nodes are directed when they express the direction of a relation, e.g., whether an actor likes another actor. Directed ties are called arcs (Wasserman and Faust 1994, p. 72). Ties can also be weighted to indicate the relation’s intensity, e.g., to indicate how often actors attended a certain event together (Scott 2000, p. 65). The types of relations among nodes—i.a., friendship, kinship, colleagues, neighbors, etc.—are recorded in relational data (ibid., p. 3). Nodes, on the other hand, have attributes. These are the nodes’ characteristics or properties, which make them distinct from or similar to each other. Attributes are treated like variables that possess different values and are stored as attribute data (ibid., p. 2).90 Walks, Trails, and Paths In a network, the connections among actors have different forms. A walk is the simplest way in which adjacent actors are connected. A walk is a sequence of lines among nodes, which starts and ends with a node. Ties and nodes in this sequence can be included more than once (Scott 2000, p. 68; Wasserman and Faust 1994, p. 105). A trail is the connection between nodes in a network in which a node may be included twice or more times, but in which all lines are only passed once (Wasserman and Faust 1994, p. 107). A path is the most restrict type of a walk in which all nodes and the connecting ties between them might be included only once (Scott 2000, p.  68). A path between nodes makes them “reachable” (Wasserman and Faust 1994, p. 107). In a directed network, the lines of such a sequence have to show into the same direction to make it a path (Scott 2000, p. 68). The connection among actors exclusively through paths creates so-called network components.

 A node attribute could be, for example, a person’s age, a country’s GDP, or an organization’s ideology.

90

3.5  The Data Analysis Methods

117

Components A component is defined as a maximal set of nodes in which every node can reach every other by some path. The ‘maximal’ part means that if you can add a node to the set without violating the condition that everyone can reach everyone, you must do so. (Borgatti et al. 2013, p. 16)

A component is thus a subgraph in which all nodes are reachable (Scott 2000, p. 101). There is furthermore no path between a node of this subgraph and any node outside this subgraph (cf. Scott 2000, p. 101; Wasserman and Faust 1994, p. 109). For directed networks, one distinguishes weak and strong components. Weak components do not consider the direction of ties, while strong components do (Borgatti et al. 2013, p. 16; Scott 2000, p. 103f.). A component analysis shows how cohesive a network is. As actors within a component are all reachable among each other, a network with several components is less dense than a network consisting of one large component and several smaller components or isolates (Scott 2000, p. 104).91 A graph with only one component is connected; a graph comprising more than one component is disconnected (Wasserman and Faust 1994, p. 109). Factions To find out whether certain types of actors form subgroups within a network, one can analyze factions within a network. Factions are cohesive groups of nodes whose number is predetermined (cf. Borgatti et al. 2013, p. 191). To become part of a faction, actors have to be connected by adjacency or through a path (Wasserman and Faust 1994, p. 290). Faction analysis is an explorative method in which the researcher decides into how many factions to split the network. The number of groups actors are assigned to is thus defined before the analysis is carried out. The algorithm fits each actor—i.e., node—into one group only. If an actor has paths or is adjacent to one or more actors in several factions, the algorithm still forces this actor into one group only. To guarantee the faction partition is robust, the analysis has to be repeated several times to check whether actors are constantly allocated into the same faction (Borgatti et al. 2013, p. 192).

91  Isolates are nodes that have no connections; see Borgatti et al. (2013, p. 14). The actors within the large component are all connected through paths—the majority of the network’s actors reaches out to each other; while in a network with several components, actors are reachable only to their peers within their component—the components themselves are not connected through paths; the network is looser.

118

3  Methods and Cases

Density Density is the share of ties existent within a network compared to the amount of ties theoretically possible within this network. The measure represents “the probability that a tie exists between any pair of randomly chosen nodes” (Borgatti et al. 2013, p. 150). In a directed graph, the density’s value reaches from 0 to 1: at a value of 0 no edges would be present; the value of 1 indicates that all arcs are reciprocated (Wasserman and Faust 1994, p. 129). Degree Centrality Centrality measures focus at actors’ positions within a network from different conceptual perspectives.92 For my analysis, I use degree centrality, which measures the amount of ties a node has (Borgatti et al. 2013, p. 165; Scott 2000, p. 83; Wasserman and Faust 1994, p. 101, 178). In a directed graph, this measure distinguishes outand in-degree. A node’s in-degree, dI(ni), is the number of nodes that have a link to ni—one could also say the number of nodes that are adjacent to ni. The in-degree is thus the number of arcs a node receives (Wasserman and Faust 1994, p. 126). A node’s out-degree, dO(ni), reflects the number of nodes ni is adjacent from. It is the number of arcs ni sends to other nodes (ibid., p. 126). Degree centrality is interpreted in terms of the network’s type of ties. In a network of information exchange, an actor with a high degree centrality receives and distributes a high amount of information and could be called an “information hub.” An actor with a high degree centrality of ties representing friendships possesses many friends and is thus popular, while an actor with a low degree centrality of this kind of ties has few friends and is socially more isolated (cf. Borgatti et al. 2013, p. 165f.). However, if ties represented the social relation “does not like,” actors with a high in-degree centrality would be less popular than the ones with a low in-degree centrality. Actors’ degree centrality is dependent on network size. To compare actors’ degree centralities across networks of different sizes, the degree has to be normalized by dividing the degree, d(ni), by all connections potentially available to the actor, i.e., the network’s number of actors, g, minus one—the actor him-/herself (cf. Wasserman and Faust 1994, p. 178f.): d ( ni )

g -1

 Two prominent centrality measures are, for instance, closeness centrality and betweenness centrality. The former considers a node’s (ni) distance to the other nodes and is the sum of ni’s shortest paths to all nodes ni is connected to; see Borgatti et al. (2013, p. 173) and Wasserman and Faust (1994, p. 183ff). The latter represents a node’s capacity to connect other nodes of the network. It measures the proportion of the shortest paths of all pairs of nodes passing through the focal node one is estimating the measure for and sums these proportions up, resulting in a single value for the focal node (Borgatti et al. 2013, p. 174f.); (Wasserman and Faust 1994, 187ff). For an extensive examination of centrality measures, see Chapter 10  in Borgatti et  al. (2013) and Chapter 5  in Wasserman and Faust (1994). 92

3.5  The Data Analysis Methods

119

Core-Periphery The core-periphery structure divides a network into two groups, revealing a network’s peripheral nodes, which are only linked to core nodes, and the core nodes, which are connected with each other and the peripheral nodes (Borgatti et al. 2013, p. 161). A network’s partition into a core and a periphery reveals one block of actors that are in the core, connected with each other; one block of actors that are in the periphery and only connected among themselves; one block of the peripheral actors that have ties to the core actors; and a last block of the core actors that send ties to the peripheral actors.93 A network’s core-periphery structure is one of nodes that are densely connected within the core and a periphery of nodes that are sparsely connected to the core and to one another. A core-periphery analysis casts light on the actors that are well connected, forming a network’s center, and the type of actors that are farther away from the center of interaction. In the case of actors’ collaboration, this analysis reveals the actors actively engaged in collaboration and the ones less integrated into collaboration. The descriptive SNA techniques serve to assess collaboration and information exchange among actors in the management process of micro-pollutants in the Rhine sub-catchments. The case comparison of which allows to draw some general assumptions about cooperation in a CPR problem setting in a central-European context. The inferential SNA technique enables me to test the hypotheses.

3.5.2  Exponential Random Graph Models (ERGM) At the dyadic analytical level, I look at ties between actors in the collaboration network. More precisely, I focus on the factors that influence the likelihood of such ties to exist. The operationalization of the dependent variable at the dyadic level is the existence of a collaboration tie between two actors. To assess the independent variables’ influence on ties in the collaboration network, an exponential random graph model (ERGM) is estimated. An ERGM is a “tie-based model[…] for understanding how network ties arise” (Berardo 2014, p. 207). The model sees “network structures as possible realizations of stochastic network processes” (ibid., p. 207). ERGM techniques can test the influence of various endogenous network effects as well as exogenous effects on the structure of a network. Endogenous network effects are, inter alia, triangles, cycles, or clustering, whereas exogenous network effects comprise edge covariates like node attributes (Hunter et al. 2008; Leifeld and

 The partition reveals a blocked adjacency matrix of four blocks. The upper left matrix quadrant contains the connections of the core nodes; the lower right quadrant the interactions of the peripheral nodes. The two remaining quadrants show the links from the core nodes to the peripheral nodes and the connections reaching from the peripheral to the core nodes (see Borgatti et al. 2013, p. 225).

93

120

3  Methods and Cases

Schneider 2012, p. 737). 94 The model not only estimates the effects of the covariates on network ties while simultaneously controlling for the influence of endogenous network effects on network ties; it also projects parameters that describe the forms of dependence existing in relational data (Cranmer and Desmarais 2010, p. 67). The ERGM thus takes network dependencies into consideration when estimating the causes for tie creation in a network. The model has the advantage that it “represents a complete and proper probability model of the entire network” (Cranmer et  al. 2017, p.  241). The model is estimated via Markov Chain Monte Carlo Maximum Likelihood Estimation (MCMC MLE) and computed using the ERGM package for R (R Core Team 2016) that comes with the stat-net suite of packages (Handcock et al. 2008; Handcock et al. 2016). The model uses the parameters of the original network to compute thousands of random permutations of the original network.95 Keeping the original network features, such as the number of nodes and edges, it generates random combinations of these parameters and in the process assesses the influence of the factors included into the model’s function in order to test these factors’ effect on the network ties. After having run-through of a thousand different combinations of the factors and network features, the model discerns the moment at which an effect occurs not merely randomly anymore, but keeps appearing in the configuration at a rate indicating that the effect is not happening by pure chance. Based on this recognition, the model determines which of the factors account for the existence of a tie not by chance, but significantly by intent. An ERGM thus calculates and generates possible configurations of a given network stochastically. The ERGM then determines the dominant configuration that appears in a higher frequency in the observed network—the original one—than in all the randomly generated ones (Berardo 2014, p. 208). For my research, I determine the dominant configuration of ties that indicate collaboration. To see what the existence of these ties depends upon—in other words to understand which independent variables account for a tie’s existence—I include endogenous network effects as well as actor attributes, the so-called covariates, into the model. They represent the independent and control variables. By this, I check whether these attributes correlate with the existence of a tie or not. The model’s results show the endogenous and exogenous network effects and covariates that account for the likelihood of a tie to exist in the collaboration network as “the existence of a tie depends […] on particular attributes of the node that creates (or receives) the tie” (Berardo 2014, p. 209). I control for the following endogenous network effects: the network’s density (edges), the networks vertices’ in- and out-degree (gwidegree and gwodegree), the network ties’ reciprocity (mutual), and the geometrically weighted edgewise shared partners (gwesp) (Hunter 2007; Hunter and Handcock 2006).

94 95

 Node attributes can be, for example, actors’ age, sex, or profession.  The number of random permutations depends on the number set by the analyst.

3.5  The Data Analysis Methods

121

Running an ERGM with directed networks provides information both on the independent variable’s effect on the out-going and on the in-coming ties of actors. This distinction is an added value allowing to discern the actors’ engagement within the network, i.e., whether actors in the collaboration networks are more likely to send ties to other actors due to the independent variables tested. This behavior is described as actors’ activity in the network. It is further possible to see whether actors are being more or less selected for collaboration by other actors by way of actors’ specific attributes, i.e., the independent variables. Such a selection by others is called an actor’s popularity. Whether actors are more likely to share a tie due to their similarity of a covariate is indicated with a homophily term. The opposite effect—actors’ absolute difference in a covariate’s value that decreases their likelihood to share a tie—is expressed in the ERGM’s heterophily term. The general stochastic logic underlying ERGMs is logistical regression. ERGMs do not infer causality, but only correlations between covariates and network ties. However, as some studies have shown, ERGMs can be interpreted as providing insights on causality (Ingold and Leifeld 2014, p.  14).96 In my study, I treat the ERGM results as indications of non-directional correlations between the independent variables and the dependent variable.

3.5.3  The Qualitative Analysis: A Case Comparison To interpret the ERGM results, I assess the independent and the control variables qualitatively and compare their distinctness across the three case studies. Furthermore, I evaluate the case studies’ contextual factors and the insights from the expert interviews. The qualitative analysis of each case study and the subsequent comparison of the three cases elucidate the background upon which I can interpret the results of the quantitative analysis. The detailed case study analyses enable me to reject or accept the working hypotheses and to answer the research question. The case comparison further allows me to draw general conclusions about the social phenomenon under study (Gerring 2004). The case study comparison design is Mill’s method of difference or rather a “most-similar case” research design (Bennet 2004, p. 31). The most-similar case research design states that the independent variables of the cases are all identical except for one that differs across the cases. The dependent variable differs as well (Bennet and Elman 2008, p.  506; Gerring 2001, p.  210). In a most-similar case research design the: intensive study of the (…) cases will reveal one – or at most several – factors that differ across these cases. The differing factors (…) are the putative causes. (Gerring 2007, p. 131)

 In the specific study, the researchers examined the evolution of collaboration throughout time, which enabled them to deduce a causal direction.

96

122

3  Methods and Cases

Table 3.12  Most-similar case study research design Case studies 1 2 3

IVs A, B, C, D A, B, C, E A, B, C, F

DV—cooperation in micro-pollutant management process D—trigger for cooperation of intensity 1 E—trigger for cooperation of intensity 2 F—trigger for cooperation of intensity 3

Claiming that different intensities of cooperation across the case studies signify a difference of the dependent variables across the cases, the most-similar case research design allows identifying the factors that account for actors’ cooperation in each case. Table 3.12 depicts the concept of the research’s most-similar case research design as applied to the present study.97 Mill’s method only works well under specific circumstances that are rarely to be found in the “real world.” I thus apply the method in a rather general way (cf. Bennet 2004, p. 32). The reasons to conduct a case study comparison for this research are twofold. A case comparison allows finding the different factors that explain different intensities of cooperation in a CPR problem situation within the same resource system. By identifying differences and similarities between the case studies, I can draw a “broader picture” of the factors that are most likely to influence cooperation in the management process of a CPR problem, such as micro-pollutants in surface water. Having presented the analytical approach, the case study design, and the case studies as well as data collection methods and data analysis methods, the next chapter presents the results of the analysis of cooperation in each case study.

References Sources e-mail N° 1. Department Raum- und Umweltwissenschaften, field of Analytische und Ökologische Chemie, University of Trier, 10 November 2016 Telephone call N° 1. Division “Protection des Eaux” of the Administration de la Gestion de l’Eau, Ministry for Durable Development and Infrastructure, Luxembourg, 22 November 2016 Telephone call N° 3. Division “Innovation, Umwelt & Energie”, Industrie- und Handelskammer Rheinland-Pfalz, 12 December 2016 Telephone call N° 4. DVGW NRW, 21 December 2016

97

 The Table is adapted from Bennet (2004, p. 31).

References

123

Primary Literature Administration de la Gestion de l’Eau (2016) Distribution et responsabilités [online]. Grand-­ Duché de Luxembourg; Administration de la Gestion de l’Eau. Available from: https://eau. public.lu/eau_potable/production_distribution_responsabilites/Distribution-et-responsabilites/ index.html. Accessed 25 Sept 2019 agw, BWK NRW DWA NRW Städtetag NRW, Städte- und Gemeindebund NRW (2014) Memorandum für einen Schutz der Gewässer vor Spurenstoffen der agw, des BWK-­ Landesverbandes NRW, des DWA-Landesverbandes NRW, des Städte- und Gemeindebundes NRW und des Städtetages NRW. [online]. Available from: https://bwk-nrw.de/wp-content/ uploads/2018/10/agw_memorandum_2014.pdf. Accessed 27 Sept 2019 AWWR (2018) Trinkwassergewinnung an der Ruhr. Über 230 Mio. m3 Trinkwasser pro Jahr für die Menschen im Einzugsgebiet [online]. Arbeitsgemeinschaft der Wasserwerke an der Ruhr (AWWR). Available from: https://www.awwr.de/trink-wassergewinnung/. Accessed 25 Sept 2019 AWWR, Ruhrverband (2016) Ruhrgütebericht 2016. Arbeitsgemeinschaft der Wasserwerke an der Ruhr (AWWR)/Ruhrverband, Essen BAFU (2012) Verursachergerechte Finanzierung der Elimination von Spurenstoffen im Abwasser Änderung des Gewässerschutzgesetzes (GSchG). Auswertung der Vernehmlassung von April-­ August 2012. Bundesamt für Umwelt, Bern bfv-trier (2019) Bezirks-Fischerei-Verband Trier 1922 e.V. (BFV-Trier) [online]. Available from: https://www.bfv-trier.de/. Accessed 25 Sept 2019 BMBF (2019) Risikomanagement von neuen Schadstoffen und Krankheitserregern im Wasserkreislauf (RiSKWa) [online]. Available from: http://www.riskwa.de/de_1294.html. Accessed 25 Sept 2019 BMBF, RiSKWa, FONA (n.d.) Sichere Ruhr. Das Projekt Sichere Ruhr  - Was machen wir? [online]. Available from: https://sichere-ruhr.de/. Accessed 25 Sept 2019 BMJV (2009) Gesetz zur Ordnung des Wasserhaushalts (Wasserhaushaltsgesetz). WHG. 31 July. Available from: http://www.gesetze-im-internet.de/whg_2009/WHG.pdf. Accessed 25 Sept 2019 BMJV (2016) Verordnung zum Schutz der Oberflächengewässer (Oberflächengewässerverordnung). OGewV. 20 June. Available from: http://www.gesetze-im-internet.de/ogewv_2016/OGewV. pdf. Accessed 25 Sept 2019 BMU (2016) Spurenstoffstrategie des Bundes [online]. Bundesministerium für Umwelt, Naturschutz, Bau und Reaktorsicherheit (BMU). Available from: https://www.bmub.bund.de/ pressemitteilung/spurenstoffstrategie-des-bundes/. Accessed 23 Sept 2019 BMU, UBA, Frauenhofer Institut Karlsruhe (2017) Empfehlungen des Stakeholder-Dialogs “Spurenstoffstrategie des Bundes” an die Politik zur Reduktion von Spurenstoffeinträgen in die Gewässer. Policy Paper: Bundesministerium für Umwelt, Naturschutz, Bau und Reaktorsicherheit (BMU) Braun C, Gälli R (2014) Mikroverunreinigungen aus Industrie und Gewerbe. Erste Grundlagenerhebungen mittels Umfragen bei den Kantonen zu vorhandenen Informationen. Bern, BMG Engineering AG, Schlieren Bundesversammlung der Schweizerischen Eidgenossenschaft (1991) Bundesgesetz über den Schutz der Gewässer (Gewässerschutzgesetz). GSchG. 24 January. Available from: https:// www.admin.ch/opc/de/classified-compilation/19910022/201701010000/814.20.pdf. Accessed 25 Sept 2019 Cercl’Eau (2010) Massnahme zur Verringerung der Mikroverunreinigungen in den Gewässern zum Schutze des Ökosystems und des Trinkwassers. Änderung der Gewässerschutzverordnung (GSchV; SR 814.201). Stellungnahme des Cercl’Eau, 28 April commons.wikimedia (2016) Map of ruhr [online]. Available from: https://commons.wikimedia. org/wiki/Ruhr#/media/File:Verlaufskarte_Ruhr2.png. Accessed 25 Sept 2019

124

3  Methods and Cases

Der Bundesrat (2017) Das Portal der Schweizer Regierung. Kollegium seit 1848 [online]. Available from: https://www.admin.ch/gov/de/start/bundesrat/geschichte-des-bundesrats/schweizerregierung-zusammensetzung-seit-1848.html. Accessed 25 Sept 2019 Der Schweizerische Bundesrat (1998) Gewässerschutzverordnung. GSchV. 28 October. Available from: https://www.admin.ch/opc/de/classified-compilation/19983281/201801010000/814.201. pdf. Accessed 25 Sept 2019 EC (2017) Communication on a strategic approach to pharmaceuticals in the environment. Roadmap: European Commission (EC). Ares(2017)2210630, 28 April EU (2000) DIRECTIVE 2000/60/EC OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL of 23 October 2000 establishing a framework for Community action in the field of water policy EU (2001) DECISION No 2455/2001/EC OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL of 20 November 2001 establishing the list of priority substances in the field of water policy and amending Directive 2000/60/EC EU (2008) DIRECTIVE 2008/105/EC OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL of 16 December 2008 on environmental quality standards in the field of water policy, amending and subsequently repealing Council Directives 82/176/EEC, 83/513/EEC, 84/156/EEC, 84/491/EEC, 86/280/EEC and amending Directive 2000/60/EC of the European Parliament and of the Council EU (2013) DIRECTIVE 2013/39/EU OF THE EUROPEAN PARLIAMENT AND THE COUNCIL of 12 August 2013 amending Directives 2000/60/EC and 2008/105/EC as regards priority substances in the field of water policy EU (2014) COMMISSION DIRECTIVE 2014/101/EU of 30 October 2014 amending Directive 2000/60/EC of the European Parliament and of the Council establishing a framework for Community action in the field of water policy EU (2019a) EU member countries in brief [online]. European Union. Available from: https:// europa.eu/european-union/about-eu/countries/member-countries_en. Accessed 25 Sept 2019 EU (2019b) EU law. Regulations, Directives and other acts [online]. European Union. Available from: https://europa.eu/european-union/eu-law/legal-acts_en. Accessed 25 Sept 2019 EU (2019c) About the EU. Countries [online]. Available from: https://europa.eu/european-union/ about-eu/countries_en. Accessed 25 Sept 2019 FLPS (2019) Fédération Luxembourgeoise des Pécheurs Sportifs [online]. Fédération Luxembourgeoise des Pécheurs Sportifs a.s.b.l. (FLPS). Available from: http://www.flps.lu/. Accessed 25 Sept 2019 FV NRW (2018) Fischereiverband Nordrhein-Westfalen e.V. [online]. Fischereiverband Nordrhein-­ Westfalen e.V. Available from: https://www.fischereiverband-nrw.de/. Accessed 25 Sept 2019 FV Saar (2010) Fischereiverband Saar, Körperschaft des öffentlichen Rechts [online]. Fischereiverband Saar KöR. Available from: https://www.fischereiverband-saar.de/. Accessed 25 Sept 2019 HKBB (2012) Änderung des Gewässerschutzgesetzes. Mikroverunreinigungen: Spezialfinanzierung für ARA-Ausbau. Stellungnahme. Basel, 30 August ICPR (2019) Moselle-Saar. Moselle [online]. Internationale Kommission zum Schutz des Rheins (IKSR/ICPR). Available from: https://www.iksr.org/en/topics/rhine/sub-basins/mosellesaar/?sword_list%5B0%5D=mosel&cHash=38ddeb3d59cdb226b67f1d3f92215535. Accessed 25 Sept 2019 IKSMS (n.d.-a) Das Einzugsgebiet von Mosel und Saar in der Flussgebietseinheit Rhein [online]. Internationale Kommissionen zum Schutze der Mosel und der Saar (IKSMS). Available from: http://www.iksms-cipms.org/servlet/is/20045/. Accessed 25 Sept 2019 IKSMS (n.d.-b) Die IKSMS  - Abstimmungsplattform zur Umsetzung der europäischen Wasserrahmenrichtlinie [online]. Internationale Kommissionen zum Schutze der Mosel und der Saar (IKSMS). Available from: http://www.iksms-cipms.org/servlet/is/20052/. Accessed 25 Sept 2019

References

125

IKSMS (n.d.-c) Richtlinie 2000/60/EG Internationale Flussgebietseinheit RHEIN Internationales Bearbeitungsgebiet “Mosel-Saar”. Bestandsaufnahme (Teil B). Konz: Internationale Komissionen zum Schutze der Mosel und der Saar (IKSMS) IKSMS (n.d.-d) Richtlinie 2000/60/EG Internationale Flussgebietseinheit RHEIN Internationales Bearbeitungsgebiet Mosel-Saar (Teil B). Bewirtschaftungsplan 2016-2021. Konz: Internationale Komissionen zum Schutze der Mosel und der Saar (IKSMS) juraforum (2012) UDE: Neues Projekt “Sichere Ruhr”  - Alles im Fluss? [online]. Einbock GmbH.  Available from: https://www.juraforum.de/wissenschaft/ude-neues-projekt-sichereruhr-alles-im-fluss-417646. Accessed 25 Sept 2019 Kanton BS, Kanton BL (2013) Erweiterung der ARA Basel: Projektierungskredit beantragt. Press release. Basel, Liestal: Kantone BS und BL, 27 August. KFVBS (2016) Kantonaler Fischereiverband Basel-Stadt (KFVBS) [online]. Kantonaler Fischereiverband Basel-Stadt (KFVBS). Available from: http://www.basler-fischerei.ch/ web7old/. Accessed 24 Sept 2019 LANUV (2013) Gebietsverzeichnis GSK3C [online]. Landesamt für Natur, Umwelt und Verbraucherschutz Nordrhein-Westfalen (LANUV). Available from: https://www.lanuv.nrw. de/fileadmin/lanuv/wasser/pdf/Gebietsverzeichnis%20GSK3C.xls. Accessed 25 Sept 2019 LANUV (2019) Wasserversorgung. Trinkwasserressourcen - Herkunft des Trinkwassers [online]. Landesamt für Natur, Umwelt und Verbraucherschutz Nordrhein-Westfalen (LANUV). Available from: https://www.lanuv.nrw.de/umwelt/wasser/wasserversorgungtrinkwasser/trinkwasserressourcen-herkunft-des-trinkwassers. Accessed 25 Sept 2019 LAWA (2016) Mikroschadstoffe in Gewässern: Bund/Länder-Arbeitsgemeinschaft Wasser (LAWA). Beschlossen auf der 151. LAWA-VV am 17./18. März 2016 in Stuttgart Le Gouvernement du Grand-Duché de Luxembourg (2017) Loi du 20 juillet 2017 modifiant la loi modifiée du 19 décembre 2008 relative à l’eau. A690. Available from: http://legilux.public.lu/ eli/etat/leg/loi/2017/07/20/a690/jo. Accessed 25 Sept 2015 Le Gouvernement du Grand-Duché de Luxembourg (2019) The official portal of the Grand Duchy of Luxembourg. Politics and institutions [online]. Le Gouvernement du Grand-Duché de Luxembourg. Available from: http://luxembourg.public.lu/en/le-grand-duche-se-presente/ luxembourg-tour-horizon/politique-et-institutions/index.html. Accessed 25 Sept 2015 LIST (2018a) noPills. No PILLS in waters! [online]. Luxembourg Institute of Science and Technology (LIST). Available from: https://www.list.lu/en/research/project/nopills/. Accessed 27 Sept 2019 LIST (2018b) RivHerb. Exploring mixture toxicity of herbicides in rivers and its treatment by ecotoxicological models [online]. Luxembourg Institute of Science and Technology (LIST). Available from: https://www.list.lu/en/research/project/rivherb/. Accessed 27 Sept 2019 LIST (2018c) Sentinelle III. Warning and advisory platform for the main pests and diseases in the major crops in Luxembourg [online]. Luxembourg Institute of Science and Technology (LIST). Available from: https://www.list.lu/en/research/project/sentinelle3/. Accessed 26 February 2018 Luxemburger Wort (2014a) Nach Verunreinigungsvorfall in Belgien: Chemische Substanz in der Sauer ist Rapsherbizid. Meazachlor könnte im schlimmsten Fall Folgen für Trinkwasseraufbereitung am Obersauer-Stausee haben. News release, Luxemburg Stadt, 29 Sept. Available from: http://www.wort.lu/de/lokales/nach-verunreinigungsvorfall-in-belgienchemische-substanz-in-der-sauer-ist-rapsherbizid-542959ceb9b398870806dfbf. Accessed 25 Sept 2019 Luxemburger Wort (2014b) Grundbelastung des Wassers: Pestizide im Trinkwasser  - und jetzt?! Der “Mouvement écologique” fordert konkretes Angehen der offensichtlichen und vernachlässigten Probleme. Luxemburg Stadt, 20 October. Available from: http:// www.wort.lu/de/lokales/grundbelastung-­des-wassers-pestizide-im-trinkwasser-und-jetzt54450924b9b398870807c3e5. Accessed 25 Sept 2019

126

3  Methods and Cases

LWL (2018) Öffentliche Trinkwasserversorgung Westfalen. Wasserförderung [online]. Landschaftsverband Westfalen-Lippe (LWL). Available from: https://www.lwl.org/LWL/ Kultur/Westfalen_Regional/Wirtschaft/Trinkwasser/#anker-5. Accessed 25 Sept 2019 MDDI (2009) Bewirtschaftungsplan für das Groβherzogtum Luxemburg. Umsetzung der EG-Wasserrahmenrichtlinie. Le Gouvernement du Grand-Duché de Luxembourg; Ministère du Développement durable et des infrastructures (MDDI); Administration de la Gestion de l’Eau, Esch-sur-Alzette, December MDDI (2014) Bericht zur Bestandsaufnahme für Luxemburg. Umsetzung der europäischen Wasserrahmenrichtlinie (2000/60/EG). Le Gouvernement du Grand-Duché de Luxembourg; Ministère du Développement durable et des infrastructures (MDDI); Administration de la Gestion de l’Eau, Esch-sur-Alzette, October MDDI (2015a) Entwurf des Bewirtschaftungsplans für die luxemburgischen Anteile an den internationalen Flussgebietseinheiten Rhein und Maas (2015-2021). Umsetzung der europäischen Wasserrahmenrichtlinie (2000/60/EG). Le Gouvernement du Grand-Duché de Luxembourg; Ministère du Développement durable et des infrastructures (MDDI); Administration de la Gestion de l’Eau, Esch-sur-Alzette, February MDDI (2015b) Bewirtschaftungsplan für die luxemburgischen Anteile an den internationalen Flussgebietseinheiten Rhein und Maas (2015-2021). Umsetzung der europäischen Wasserrahmenrichtlinie (2000/60/EG). Le Gouvernement du Grand-Duché de Luxembourg; Ministère du Développement durable et des infrastructures (MDDI); Administration de la Gestion de l’Eau, Esch-sur-Alzette, December Ministerium der Justiz, RLP (2015) Landeswassergesetz (LWG) Vom 14. Juli 2015. LWG Ministerium der Justiz, Saarland (2004) Gesetz Nr. 714  - Saarländisches Wassergesetz (SWG) Vom 28. Juni 1960. in der Fassung der Bekanntmachung vom 30. Juli 2004. zuletzt geändert durch das Gesetz vom 3. Dezember 2013. SWG MKULNV (2009) Trinkwasserbericht Nordrhein-Westfalen. Düsseldorf, February MKULNV (2012) Programm Reine Ruhr. zur Strategie einer nachhaltigen Verbesserung der Gewässer- und Trinkwasserqualität in Nordrhein-Westfalen. Available from: http://www.taz. de/fileadmin/static/pdf/ProgrammReineRuhrBerichtFINAL.pdf. Accessed 25 Sept 2019 MKULNV (2015) Bewirtschaftungsplan 2016-2021 für die nordrhein-westfälischen Anteile von Rhein, Weser, Ems und Maas. Ministerium für Umwelt und Naturschutz, Landwirtschaft und Verbraucherschutz des Landes Nordrhein-Westfalen MKULNV (2018) Bekanntmachung der Neufassung des Wassergesetzes für das Land Nordrhein-­ Westfalen (Landeswassergesetz) Vom 25. Juni 1995. LWG. 19 January MUEEF (2010) Zusammenfassung der Beiträge des Landes Rheinland-Pfalz zum Bewirtschaftungsplan und der Maβnahmenprogramme für den internationalen Bewirtschaftungsplan Rhein. Ministerium für Umwelt, Forsten und Verbraucherschutz, Rheinland-Pfalz; Abteilung Wasserwirtschaft, Mainz, 22 March MUEEF (2015) Rheinland-Pfälzischer Bewirtschaftungsplan 2016-2021. Ministerium für Umwelt, Landwirtschaft, Ernährung, Weinbau und Forsten, Rheinland-Pfalz; Abteilung Wasserwirtschaft, Mainz, 22 December MUV (2015) 2. Bewirtschaftungsplan Saarland. nach Artikel 13 der Richtlinien 2000/60/EG des Europäischen Parlaments und des Rates vom 23. Oktober 2000. Ministerium für Umwelt und Verbraucherschutz Saarland MUV (2016) Studie zur Reduktion von Mikroverunreinigungen in Gewässern: Umweltminister Jost überreicht rund 222.500 Euro an EVS. Ministerium für Umwelt und Verbraucherschutz Saarland, 21 December ProRheno AG (2016) Erweiterung ARA Basel [online]. Accessed 25 Sept 2019 Rechsteiner R (2006) An die Geschäftsprüfungskommissionen von Basel-Landschaft und von Basel-Stadt. Hardwasser AG RFG (2018) Die Ruhrfischereigenossenschaft (RFG) [online]. Ruhrfischereigenossenschaft (RFG). Available from: http://www.ruhrfischereigenossenschaft.de/. Accessed 25 Sept 2019

References

127

Rheinaubund (2012) Stellungnahme zur Änderung des Gewässerschutzgesetzes (Verursachergerechte Finanzierung der Elimination von Spurenstoffen im Abwasser). Schaffhausen, 30 August Ruhr-Guide (2018) Die Ruhr [online]. Available from: http://www.ruhr-guide.de/freizeit/seen-­ und-fluesse/die-ruhr/14123,0,0.html. Accessed 23 Sept 2019 RWTH Aachen, IWW (2008) Abschlussbericht zu den Forschungsvorhaben:. “Senkung des Anteils organischer Spurenstoffe in der Ruhr durch zusätzliche Behandlungsstufen auf kommunalen Kläranlagen  - Gütebetrachtungen” und “Senkung des Anteils organischer Spurenstoffe in der Ruhr durch zusätzliche Behandlungsstufen auf kommunalen Kläranlagen Kostenbetrachtungen”. Mühlheim an der Ruhr SBV (2012) Vernehmlassung zur Änderung des Gewässerschutzgesetzes (Elimination von Spurenstoffen). Brugg SEBES (n.d.) La Distribution. Le réseau d’adduction du SEBES [online]. Syndicat des Eaux du Barrage d’Esch-sur-Sûre (SEBES). Available from: https://sebes.lu/syndicat-eaux-barrage-­ desch-sure/reseau-du-sebes/. Accessed 25 Sept 2019 SES (2014) Präsenz von Pestiziden im Trinkwasser. Pressekonferenz vom 9. Oktober 2014 über die momentane Situation des Trinkwassers. Press release, 9 October. Available from: http:// www.ses-eau.lu/online/www/navMain/13/40/402/contentContainer1/1097/984/DEU/communique%20SES.pdf. Accessed 25 Sept 2019 UNECE (2011) Second Assessment of Transboundary Rivers, Lakes and Groundwaters [online]. United Nations Economic Commission for Europe (UNECE). Available from: http://www. unece.org/fileadmin/DAM/env/water/publications/assessment/English/ECE_Second_ Assessment_En.pdf. Accessed 25 Sept 2019 Universität Saarland (2019) Gewässer-Monitoring. Arbeitsgruppe der Universität des Saarlandes [online]. Universität des Saarlandes, Anorganische und Analytische Chemie. Available from: http://www.gewaesser-monitoring.de/?Projekte. Accessed 25 Sept 2019 wikimedia (2018a) Karte_Gemeinden_des_Kantons_Basel_Stadt_2007 [online]. Available from: https://upload.wikimedia.org/wikipedia/commons/7/77/Karte_Gemeinden_des_Kantons_ Basel_Stadt_2007.png. Accessed 25 Sept 2019 wikimedia (2018b) Switzerland_in_Europe [online]. Available from: https://upload.wikimedia. org/wikipedia/commons/a/ad/Switzerland_in_Europe.svg. Accessed 25 Sept 2019

Secondary Literature Abegglen, C., and Siegrist, H. (2012) Mikroverunreinigungen aus kommunalem Abwasser. Verfahren zur weitergehenden Elimination auf Kläranlagen. Bern. Abu-Laban B (1965) The reputational approach in the study of community power: a critical evaluation. Pac Sociol Rev 8(1):35–42 Allen JH (2013) The wicked problem of chemical policy: opportunities for innovation. J Environ Stud Sci 3(2):101–108 Amblard L (2012) The potential of collective action for the control of nonpoint source pollution in European rural areas. Conference “Design and Dynamics of Institutions for Collective Action”, Utrecht University, 29 November until 1 December Anckar C (2008) On the applicability of the most similar systems design and the most different systems design in comparative research. Int J Soc Res Methodol 11(5):389–401 Ansell C, Gash A (2008) Collaborative governance in theory and practice. J Public Adm Res Theory 18(4):543–571 Asano T, Cotruvo JA (2004) Groundwater recharge with reclaimed municipal wastewater: health and regulatory considerations. Water Res 38:1941–1951 Basurto X, Gelcich S, Ostrom E (2013) The social-ecological system framework as a knowledge classificatory system for benthic small-scale fisheries. Glob Environ Chang 23(6):1366–1380

128

3  Methods and Cases

Baumgartner FR, Leech BL (2001) Interest niches and policy bandwagons: patterns of interest group involvement in national politics. J Polit 63(4):1191–1213 Bennet A (2004) Case study methods: design, use, and comparative advantages. In: Sprinz DF, Wolinsky-Nahmias Y (eds) Models, numbers, and cases: methods for studying international relations. University of Michigan Press, Michigan, pp 19–55 Bennet A, Elman C (2008) Case study methods. In: Reus-Smit C, Snidal D (eds) The Oxford handbook of international relations. Oxford University Press, Oxford/New York, pp 499–517 Berardo R (2014) Bridging and bonding capital in two-mode collaboration networks. Policy Stud J 42(2):197–225 Berger H (2017) Die Länderkompetenz im Wasserrecht. Zeitschrift für Landesverfassungsrecht und Landesverwaltungsrecht 2(1):4–11 Bergsten A, Galafassi D, Bodin Ö (2014) The problem of spatial fit (in social-ecological systems): detecting spatial mismatches between ecological connectivity and land management in an urban region. Ecol Soc 19(6) Beyers J, Braun C (2014) Ties that count: explaining interest group access to policymakers. J Publ Policy 34:93–121 Borgatti SP, Everett MG, Johnson JC (2013) Analyzing social networks. Sage, London Börzel TA (1998) Organizing Babylon – on the different conceptions of policy networks. Public Adm Rev 76:253–273 Breitenfeld J (2007) Woher kommt unser Trinkwasser? Statistische Monatshefte Rheinland-Pfalz 3:170–175 Bressers HTA, O’Toole LJ (1998) The selection of policy instruments: a network-based perspective. J Publ Policy 18(3):213–239 Bundschuh M, Schulz R (2011) Ozonation of secondary treated wastewater reduces ecotoxicity to Gammarus fossarum (Crustacea; Amphipoda):. Are loads of (micro)pollutants responsible? Water Res 45:3999–4007 Bunzel K, Kattwinkel M, Liess M (2013) Effects of organic pollutants from wastewater treatment plants on aquatic invertebrate communities. Water Res 47(2):597–606 Carpenter D (2010) Reputation and power: organizational image and pharmaceutical regulation at the FDA. Princeton University Press, Princenton, NJ Carvalho RN, Arukwe A, Ait-Aissa S, Bado-Nilles A, Balzamo S, Baun A, Belkin S et al (2014) Mixtures of chemical pollutants at European legislation safety concentrations: how safe are they? Toxicol Sci 141(1):218–233 Chèvre N, Loepfe C, Singer H, Stamm C, Fenner K, Escher BI (2006) Including mixtures in the determination of water quality criteria for herbicides in surface water. Environ Sci Technol 40(2):426–435 Coleman JS (1988) Social capital in the creation of human capital. Am J Sociol 94:S95–S120 Cox M (2014) Applying a social-ecological system framework to the study of the Taos valley irrigation system. Hum Ecol 42:311–324 Cox M (2015) A basic guide for empirical environmental social science. Ecol Soc 20(1):63 Cranmer SJ, Desmarais BA (2010) Inferential network analysis with exponential random graph models. Polit Anal 19(66-86) Cranmer SJ, Leifeld P, McClurg SD, Rolfe M (2017) Navigating the range of statistical tools for inferential network analysis. Am J Polit Sci 61(1):237–251 Crona BI, Ernstson H, Prell C, Reed M, Hubacek K (2011) Combining social network approaches with social theories to improve understanding of natural resource governance. In: Bodin Ö, Prell C (eds) Social networks and natural resource management. uncovering the social fabric of environmental governance. Cambridge University Press, Cambridge/New York, pp 44–71 Etzioni A (1975) A comparative analysis of complex organizations. The Free Press, New York Fischer M, Sciarini P (2013) Unpacking reputational power: actors’ attributes and relations. Swiss Federal Institute of Aquatic Science and Technology (EAWAG), Dübendorf Fleischmann FD, Ban NC, Evans LS, Epstein G, Garcia-Lopez G, Villamayor-Tomas S (2014) Governing large-scale social-ecological systems: Lessons from five cases. Int J Commons 8(2):428–456

References

129

Fleishman R (2013) Addressing trans-boundary challenges through collaboration: how organizations “Harmonize” actions and decisions across problem landscapes. SURFACE Paper 5. Dissertations – ALL Folke C, Hahn T, Olsson P, Norberg J (2005) Adaptive governance of social-ecological systems. Annu Rev Environ Resour 30(1):441–473 Gälli R, Schmid-Kleinkemper J, Ort C, Schärer M (2009) Mikroverunreinigungen in den Gewässern. Bewertung und Reduktion der Schadstoffbelastung aus der Siedlungsentwässerung. Bundesamt für Umwelt, Bern Garrick D, de Stefano L, Fung F, Pittock J, Schlager E, New M, Connell D (2013) Managing hydroclimatic risks in federal rivers: a diagnostic assessment. Philos Trans R Soc 371 Gerring J (2001) Social science methodology. A critical approach. Cambridge University Press, Cambridge Gerring J (2004) What is a case study and what is it good for? Am Polit Sci Rev 98(2):341–355 Gerring J (2007) Case study research. Principles and practices. Cambridge University Press, Cambridge Götz C, Kase R, Hollender J (2010) Mikroverunreinigungen. Beurteilungskonzept für organische Spurenstoffe aus kommunalem Abwasser. Studie im Auftrag des Bundesamtes für Umwelt (BAFU), Eawag (ed.), Dübendorf, Switzerland Greskowiak J, Prommer H, Massmann G, Jonston CD, Nützmann G, Pekdeger A (2005) The impact of variably saturated conditions on hydrogeochemical changes during artificial recharge of groundwater. Appl Geochem 20:1409–1426 Hall PA (1993) Policy paradigms, social learning, and the state: the case of economic policymaking in Britain. Comp Polit 25(3):275–296 Handcock MS, Hunter DR, Butts CT, Goodreau SM, Morris M (2008) statnet: software tools for the representation, visualization, analysis and simulation of network data. J Stat Softw 24(1) Handcock MS, Hunter DR, Butts CT, Goodreau SM, Krivitsky PN, Bender-deMoll S, Morris M (2016) statnet: Software tools for the statistical analysis of network data: the statnet project. CRAN.R-project.org/package=statnet. Available from: http://www.statnet.org. Accessed 27 Sept 2019. Heaney MT (2014) Multiplex networks and interest group influence reputation: an exponential random graph model. Soc Networks 36:66–81 Henry AD, Dietz T (2011) Information, networks, and the complexity of trust in commons governance. Int J Commons 5(2):188–212 Henry AD, Vollan B (2012) Risk, networks, and ecological explanations for the emergence of cooperation in commons governance. Ration Mark Moral 59(3):130–147 Hillenbrand T, Tettenborn F, Menger-Krug E, Marscheider-Weidemann F, Fuchs S (2015) Maβnahmen zur Verminderung des Eintrages von Mikroschadstoffen in die Gewässer. Kurzbericht 86/2014. Umweltbundesamt, Dessau-Roβlau Hood C (1986) The tools of government: public policy and politics. N.J.  Chatham House Publishers, Chatham Howlett M (2000) Managing the “hollow state”: procedural policy instruments and modern governance. Can Public Adm 43(4):412–431 Howlett M (2005) What is a policy instrument? Tools, mixes, and implementation styles. In: Eliadis P, Hill MM, Howlett M (eds) Designing government. From instruments to governance. McGill-Queen’s University Press, Montreal/Ithaca, pp 31–50 Howlett M, Giest S (2013) The policy-making process. In: Araral E Jr et al (eds) Routledge handbook of public policy. Routledge, London/New York, pp 17–28 Hülshoff I, Greskowiak J, Grützmacher G (2009) Analysis of the vulnerability of bank filtration systems to climate change by comparing their effectiveness under varying environmental conditions. TECHNEAU, 3 December Hunter DR (2007) Curved exponential family models for social networks. Soc Networks 29(2):216–230 Hunter DR, Handcock MS (2006) Inference in curved exponential family models for networks. J Comput Graph Stat 15(3):565–583

130

3  Methods and Cases

Hunter DR, Handcock MS, Butts CT, Goodreau SM, Morris M (2008) ergm: a package to fit, simulate and diagnose exponential-family models for networks. J Stat Softw 24(3) Huxham C (1993) Pursuing collaborative advantage. J Oper Res Soc 44(6):599–611 Ingold K, Fischer M (2014) Drivers of collaboration to mitigate climate change: An illustration of Swiss climate policy over 15 years. Glob Environ Chang 24:88–98 Ingold K, Leifeld P (2014) Structural and institutional determinants of influence reputation: a comparison of collaborative and adversarial policy networks in decision making and implementation. J Public Adm Res Theory 26(1):1–18 Ingold K, Fischer M, Cairney P (2017) Drivers for policy agreement in nascent subsystems: an application of the advocacy coalition framework to fracking policy in Switzerland and the UK. Policy Stud J 45(3):442–463 Jakobi AP (2009) Policy networks in comparative politics and international relations: perspectives, typologies and functions. Bremen: Universität Bremen, Collaborative Research Center 597 “Transformations of the State”. Jann W, Wegerich K (2003) Phasenmodelle und Politikprozesse: der Policy Cycle. In: Schubert K, Bandelow NC (eds) Lehrbuch der Politikfeldanalyse. De Gruyter, München, pp 71–104 Jones OA, Lester JN, Voulvoulis N (2005) Pharmaceuticals: a threat to drinking water? Trends Biotechnol 23(4):163–167 Kannonier-Finster W (1998) Methodologische Aspekte soziologischer Fallstudien. In: Kannonier-­ Finster W (ed) Waltraud. Exemplarische Erkenntnis. Studien-Verlag, Innsbruck, pp 35–64 Kenis P, Schneider V (1991) Chapter 2. Policy networks and policy analysis: scrutinizing a new analytical toolbox. In: Marin B, Mayntz R (eds) Policy networks. Empirical evidence and theoretical considerations. Campus Verlag/Westview Press, Frankfurt/Main/Boulder, CO, pp 25–59 King G, Keohane RO, Verba S (1994) Designing social inquiry. scientific inference in qualitative research. Princeton University Press, Princenton, NJ Kirschke S, Borchardt D, Newig J (2017) Mapping complexity in environmental governance: a comparative analysis of 37 priority issues in German water management. Environ Policy Gov 27:534–559 Knill C, Tosun J (2012) Public policy: a new introduction. Palgrave Macmillan, New York Knoke D (1998) Who steals my purse steals trash. The structure of organizational influence reputation. J Theor Polit 10(4):507–530 Krein A, Keßler S, Meyer B, Pailler J-Y, Guignard C, Hoffmann L (2013) Concentrations and loads of dissolved xenobiotics and hormones in two small river catchments of different land use in Luxembourg. Hydrol Process 27(2):284–296 Kremer BP (2010) Der Rhein. Von den Alpen bis zur Nordsee. Mercator Verlag, Duisburg Kümmerer K (2009) The presence of pharmaceuticals in the environment due to the human use – present knowledge and future challenges. J Environ Manag 90:2354–2366 Lapworth DJ, Baran N, Stuart ME, Ward RS (2012) Emerging organic contaminants in groundwater: a review of sources, fate and occurrence. Environ Pollut 163:287–303 Leifeld P, Schneider V (2012) Information exchange in policy networks. Am J Polit Sci 56(3):731–744 Lemos MC, Agrawal A (2006) Environmental governance. Annu Rev Environ Resour 31:297–325 Linder SH, Peters BG (1989) Instruments of government: perceptions and contexts. J Publ Policy 9(1):35–58 Lowi TJ (1972) Four systems of policy, politics, and choice. Public Adm Rev 32(4):298–310 Lubell M, Henry AD, McCoy M (2010) Collaborative institutions in an ecology of games. Am J Polit Sci 54(2):287–300 Lubell M, Robins G, Wang P (2011) Policy coordination in an ecology of water management games. Southern Illinois University Carbondale. Lubell M, Robins G, Wang P (2014) Network structure and institutional complexity in an ecology of water management games. Ecol Soc 19(4):23 Mayntz R, Scharpf FW (eds) (1995) Gesellschaftliche Selbstregelung und politische Steuerung. Campus Verlag, Frankfurt/Main/New York

References

131

McGinnis MD (2011) An introduction to IAD and the language of the Ostrom workshop: a simple guide to a complex framework. Policy Stud J 39(1):169–183 McGinnis MD, Ostrom E (2014) Social-ecological system framework: initial changes and continuing challenges. Ecol Soc 19(2):30 McGuire M (2006) Collaborative public management: assessing what we know and how we know it. Public Adm Rev:33–43 McPherson M, Smith-Lovin L, Cook JM (2001) Birds of a feather: Homophily in social networks. Annu Rev Sociol 27:415–444 Metz F, Ingold K (2014) Sustainable wastewater management: is it possible to regulate micropollution in the future by learning from the past? A policy analysis. Sustainability 6:1992–2012 Meyer B, Pailler J-Y, Guignard C, Hoffmann L, Krein A (2011) Concentrations of dissolved herbicides and pharmaceuticals in a small river in Luxembourg. Environ Monit Assess 180(1-4):127–146 O’Leary R, Vij N (2012) Collaborative public management: where have we been and where are we going? Am Rev Public Adm 42(5):507–522 Oaks LJ, Gilbert M, Virani MZ, Watson RT, Meteyer CU, Rideout BA, Shivaprasad HL et al (2004) Diclofenac residues as the cause of vulture population decline in Pakistan. Nature 427:630–633 Ostrom E (1990) Governing the commons. The evolution of institutions for collective action. Cambridge University Press, Cambridge Ostrom E (1998) A behavioral approach to the rational choice theory of collective action. Presidential address, American Political Science Association. Am Polit Sci Rev 92(1):1–22 Ostrom E (2000a) Collective action and the evolution of social norms. J Econ Perspect 14(3):137–158 Ostrom E (2000b) The danger of self-evident truths. Political Science and Politics 33(1):33–44 Ostrom E (2005) Understanding institutional diversity. Princeton University Press, Princenton, NJ Ostrom E (2009) A general framework for analyzing sustainability of social-ecological systems. Science 325:419–422 Oxford Living Dictionaries (2018a) British & World English. collaboration [online]. Available from: https://en.oxforddictionaries.com/definition/collaboration. Accessed 18 April 2018 Oxford Living Dictionaries (2018b) British & World English. cooperation [online]. Available from: https://en.oxforddictionaries.com/definition/cooperation. Accessed 18 April 2018. Pal A, Gin Y-HK, Lin Y-CA, Reinhard M (2010) Impacts of emerging organic contaminants on freshwater resources: review of recent occurences, sources, fates and effects. Sci Total Environ 408:6062–6069 Pappi FU, Henning C (1998) Policy networks: more than a metaphor? J Theor Polit 10(4):553–575 Peters BG (2000) Policy instruments and public management: bridging the gaps. J Public Adm Res Theory 10:35–47 Peters BG, Hoornbeek JA (2005) The problem of policy problems. In: Eliadis P, Hill MM, Howlett M (eds) Designing government. From instruments to governance. McGill-Queen’s University Press, Montreal/Ithaca, pp 77–105 Provan KG, Kenis P (2008) Modes of network governance: structure, management, and effectiveness. J Public Adm Res Theory 18(2):229–252 R Core Team (2016) R: a language and environment for statistical computing. {R} {F}oundation for {S}tatistical {C}omputing, Vienna, Austria Rivera-Utrilla J, Sánchez-Polo MJ, Ferro-García MÁ, Prados-Joya G (2013) Pharmaceuticals as emerging contaminants and their removal from water: a review. Chemosphere 93:1268–1287 Rochefort DA, Cobb RW (1995) Problem definition: an emerging perspective. In: Rochefort DA, Cobb RW (eds) The Politics of problem definition: shaping the policy agenda. University Press of Kansas, Lawrence, Kansas RStudio Team (2016) RStudio: Integrated Development Environment for R (Version 1.0136). Available from: https://rstudio.com/. Accessed 25 Sept 2019. Boston, MA. Ruff M, Singer H, Ruppe S, Mazacek J, Dolf R, Leu C (2013) 20 Jahre Rheinüberwachung. Erfolge und analytische Neuausrichtung in Weil am Rhein. Aqua Gas (5):16–25

132

3  Methods and Cases

Sabatier PA (1987) Knowledge, policy-oriented learning, and policy change. An advocacy coalition framework. Sci Commun 8(4):649–692 Sabatier PA (1988) An advocacy coalition framework of policy change and the role of policy-­ oriented learning therein. Policy Sci 21(2/3):129–168 Sabatier PA (1998) The advocacy coalition framework: revisions and relevance for Europe. J Eur Publ Policy 5(1):98–130 Sadoff CW, Grey D (2005) Cooperation on international rivers. Water Int 30(4):420–427 Salamon LM (2001) The new governance and the tools of public action: an introduction. Fordham Urban Law J 28(5):1611–1674 Sanderson H, Brain RA, Johnson DJ, Wilson CJ, Solomon KR (2004) Toxicity classification and evaluation of four pharmaceuticals classes: antibiotics, antineoplastics, cardiovascular, and sex hormones. Toxicology 203:27–40 Scharpf FW (1993) Positive und negative Koordination in Verhandlungssystemen. In: Héritier A (ed) Policy-Analyse: Kritik und Neuorientierung. PVS Sonderheft 24. Westdeutscher Verlag, Opladen, pp 57–83 Schlager E (2004) Common-pool resource theory. In: Durant RF, Fiorino DJ, O’Leary R (eds) Environmental governance reconsidered. Challenges, choices, and opportunities. MIT Press, Cambridge, MA, pp 145–175 Schluep M, Thomann M, Häner A, Gälli R, Stucki G (2006) Organische Mikroverunreinigungen und Nährstoffhaushalt. Eine Standortbestimmung für die Siedlungswasserwirtschaft. Umwelt-­ Wissen 0614. Bundesamt für Umwelt, Bern Schneider V (2014) Akteurskonstellationen und Netzwerke in der Politikentwicklung. In: Schubert K, Bandelow NC (eds) Lehrbuch der Politikfeldanalyse. De Gruyter, München, pp 259–287 Schüttemeyer S (2007) Parlamentarische Demokratie. Bundeszentrale für Politische Bildung, Bonn Scott J (2000) Social network analysis. Sage, London Scott J, Christopoulos D (2018) Reputational leadership and preferences similarity: Explaining organisational collaboration in bank policy networks. Eur J Polit Res 57:518–538 Sharma SK, Amy G (2009) Bank filtration – a sustainable water treatment technology for developing countries. 34th WEDC international conference, Addis Ababa, Ethiopia. Stamm C, Burdon F, Fischer F, Kienle C, Munz N, Altermatt F, Behra R et  al (2017) Einfluss von Mikroverunreinigungen. Lebensgemeinschaften in Fliessgewässern - Ergebnisse aus dem Projekt ECOIMPACT. Aqua Gas (6):90–95 Stokman FN, Zeggelink E (1996) Is politics power or policy oriented? A comparative analysis of dynamic access models in policy networks. J Math Sociol 21(1-2):77–111 Thatcher M (1998) The development of policy network analyses. From modest origins to overarching frameworks. J Theor Polit 10(4):389–416 Vedung E (2010) Policy instruments: typologies and theories. In: Bemelmans-Videc M-L, Rist RC, Vedung E (eds) Carrots, sticks & sermons. Policy instruments & their evaluation. Transaction Publishers, New Brunswick/London, pp 21–58 Wasserman S, Faust K (1994) Social network analysis: methods and applications. Cambridge University Press, Cambridge Weber EP, Khademian AM (2008) Wicked problems, knowledge challenges, and collaborative capacity builders in network settings. Public Adm Rev 68(2):334–349 West SA, Griffin AS, Gardner A (2007) Social semantics: altruism, cooperation, mutualism, strong reciprocity and group selection. J Evol Biol 20(2):415–432 Weyer J (2011) Zum Stand der Netzwerkforschung in den Sozialwissenschaften. In: Weyer J (ed) Soziale Netzwerke. Konzepte und Methoden der sozialwissenschaftlichen Netzwerkforschung. Oldenbourg Verlag, München, pp 39–69 Wilson JQ (1989) Bureaucracy: what government agencies do and why they do it. Basic, Boulder, CO Wittmer I, Stamm C, Singer H, Junghans M (2014) Mikroverunreinigungen. Beurteilungskonzept für Mikroverunreinigungen aus diffusen Einträgen. Studie im Auftrag des Bundesamtes für Umwelt (BAFU), Dübendorf, Switzerland: Eawag Yin RK (2003) Case study research: design and methods. Sage, Thousand Oaks, CA

Chapter 4

Empirical Analysis I: On Cooperation

Abstract  The chapter analyzes the constituting elements of cooperation in each case study and relates them to each other to understand their interplay and the cooperation’s complexity. The chapter further assesses the case studies’ collaboration networks more thoroughly, analyzing their cohesion, their degree of fragmentation, their components and factions, and their core and peripheral actors. This descriptive SNA of specific features of the actor collaboration networks in the context of water quality management shows how collaboration—as a proxy for cooperation—can be conceptualized and understood in network terms. I complete the analysis of cooperation by focusing on the case study actors’ viewpoints of cooperation in the three case study regions in the Rhine catchment area. The chapter closes with a case comparison: I compare the insights of the analyses—that is, the constituting elements of cooperation and the collaboration networks’ specific features—across the case studies to draw conclusions about the nature of cooperation in each case study. Keywords  Cooperation · Descriptive Social Network Analysis (SNA) · Reciprocity · Cohesion · Fragmentation · Components · Factions This chapter captures the dependent variable cooperation. I first analyze the constituting elements of cooperation in each case study—the first unit of analysis—and relate them to each other in a second step to understand their interplay and the cooperation’s complexity. I further assess the case studies’ collaboration networks more thoroughly, analyzing their cohesion, their degree of fragmentation, their components and factions, and their core and peripheral actors. This descriptive SNA of specific features of the actor collaboration networks shows how collaboration—as a proxy for cooperation1—can be conceptualized and understood in network terms. I complete the analysis of cooperation by focusing on the case study actors’ viewpoints of cooperation in the three case study regions.

1  Collaboration serves as a proxy for cooperation because since it is conceptually close to cooperation and I consider it to lie at the very heart of cooperation.

© Springer Nature Switzerland AG 2020 L. M. J. Herzog, Micro-Pollutant Regulation in the River Rhine, https://doi.org/10.1007/978-3-030-36770-1_4

133

134

4  Empirical Analysis I: On Cooperation

Context: CPR problem of micro-pollutants in surface water in the Rhine catchment area

SINGLE-UNIT ANALYSIS— IN EACH CASE STUDY Case study 1: Mgmt. process Case study 2: Mgmt. process Case study 3: Mgmt. process in the Rhine basin at Basel in the Moselle basin in the Ruhr basin Units of analysis:

Units of analysis:

Units of analysis:

UoA 1 Network level— descriptive SNA: cooperation’s constituting elements

UoA 1 Network level— descriptive SNA: cooperation’s constituting elements

UoA 1 Network level— descriptive SNA: cooperation’s constituting elements

UoA 2 Dyadic level— ERGM: collaboration tie between two actors

UoA 2 Dyadic level— ERGM: collaboration tie between two actors

UoA 2 Dyadic level— ERGM: collaboration tie between two actors

ACROSS-UNIT ANALYSIS— CASE COMPARISON Comparison I: Cooperation across the case studies Comparison II: Factors enhancing cooperation across the case studies

Fig. 4.1  Illustration of the book’s case study method design; analytical part 1

The chapter closes with the first part of the across-unit analysis, the case comparison I. I compare the insights of the analyses—that is, the constituting elements of cooperation and the collaboration networks’ specific features—across the case studies to draw conclusions about the nature of cooperation in each case study. Figure 4.1 depicts the case study method design once more, highlighting the analytical parts that will be assessed in this chapter.

4.1  The Constituting Elements of Cooperation I conceptualize cooperation as consisting of three elements, two of which I operationalized as networks: a network of actors’ collaboration, a network of actors’ political information exchange, and one of actors’ technical information exchange. Actors’ aim at the same goal is descriptively evaluated. The following sub-chapters assess each element of cooperation.

4.1  The Constituting Elements of Cooperation

135

Table 4.1  Share of actors agreeing on goals of micro-pollutant management in the same way Management process’ trajectory Acting upon the source of pollution

Interpretation Acting end-of-pipe

4 3 2 1 4 3 2 1

Interpretation Precautionary principle

Interpretation No actions

Interpretation Zero concentration through measures

Interpretation

4 3 2 1 4 3 2 1 4 3 2 1

Basel Ruhr Moselle 81.1% 88.5% 87.1% 16.2% 7.7% 12.9% 2.7% 3.8% 0% 0% 0% 0% Majority agrees on goal 29.7% 23.1% 3.2% 37.8% 23.1% 32.2% 27% 30.7% 58.1% 5.5% 23.1% 6.5% Majority Torn Majority does agrees not agree 45.9% 34.6% 41.9% 40.5% 50.0% 51.6% 10.8% 15.4% 6.5% 2.7% 0% 0% Majority agrees on goal 2.7% 0% 3.2% 13.5% 15.4% 12.9% 37.8% 65.4% 48.4% 45.9% 19.2% 35.5% Majority does not agree on goal 40.5% 7.7% 32.2% 40.5% 57.7% 45.2% 10.8% 26.9% 19.4% 8.1% 7.7% 3.2% Majority Small torn Majority agrees agrees

1 = do not agree with goal; 4 = totally agree with goal

4.1.1  Aiming Towards the Same Goal Question 8 of the questionnaire asked actors to evaluate in how far they agree on five different directions the micro-pollutant management process in their catchment area should take (cf. Sect. 3.4.4.1). The scale of answers was “do not agree at all with goal” (1), “do somewhat not agree with goal” (2), “do somewhat agree with goal” (3), and “totally agree with goal” (4). Table 4.1 shows the share of actors that agree on the routes of the management process in the same way. The majority of actors in all three case studies totally agree on the implementation of source-directed measures and the elimination of micro-pollutants at the source as a goal of the management process. The goal of establishing end-of-pipe measures to reduce micro-pollutants in surface water is more controversial. In the

136

4  Empirical Analysis I: On Cooperation

Basel case study, almost 30% of the actors totally agree on the goal and about 38% agree somewhat on it. Still, a share of 27% somewhat does not agree on it, and two actors, amounting to about 5%, do not at all agree on the goal. This discrepancy of agreement on end-of-pipe measures is even stronger in the Ruhr case study. Here, actors’ different evaluations of the goal are almost evened out among all four categories. In the Moselle case study, the tendency is clearly towards nonagreement with about 58% somewhat not and 6.5% not at all agreeing on end-of-pipe measures. Only one actor (3%) fully agrees on this goal. Preventive measures are welcome in all case studies. In the Basel region, 86.4% of the actors somewhat or totally agree on this goal. In the Ruhr region, this group amounts to 84.6%, and in the Moselle region, it is even higher with 93.5%. The goal not to implement any measures at all receives only little support. In all three case studies, around 13–15% of the actors somewhat agree on this objective. Most of the actors do somewhat not agree on this goal—in the Ruhr case their share is the highest with 65.4%—or do not agree at all on it; here, their share is the highest in the Basel case study with 45.9%. Achieving zero concentration of micro-pollutants in surface water is a goal that for some of the actors is “not realistic”2 or simply “not possible.”3 However, 81% of the actors of the Basel and 77.4% of the Moselle case study totally and somewhat agree on this objective. In the Ruhr case study, actors are rather torn regarding this goal: about 58% agree somewhat on it, and about 28% do somewhat not agree on it. Two actors each (i.e., 7.7%) totally agree or do not agree at all. This detailed picture of actors’ share of agreement on the five ways the management process could take shows that in the Basel and the Moselle case studies, actors tend to concur on their agreement—or nonagreement—on the five ways. They follow the same aims. In the Ruhr case, actors tend to differ in their evaluation of measures for tackling micro-pollutants. They do not necessarily have the same objective when managing micro-pollutants. I disclose details about the Ruhr case to outline possible conflict lines among actors that might explain actors’ different aims. Conflict Lines Regarding the Management of Micro-pollutants in the Ruhr Case In the Ruhr region, municipal services and waterworks have been the actors that drew the attention to micro-pollutants and set action taking in motion.4 The regional government’s strategy to tackle the CPR problem is the “multi-barrier approach.” Although the approach combines source-directed measures, the procession of drinking water, and wastewater treatment, its main focus remains on end-of-pipe mea-

 Interviews N° 5, 7, and 9.  Interview N° 10 and ROCHE (comment in the questionnaire). 4  Interview N° 13. When researcher detected high concentrations of PFOS in the Rhine and the Ruhr in 2006, pressure on waterworks to clean the water increased immediately. Waterworks in the Ruhr region in turn raised awareness for the issue at the political level; see AWWR and Ruhrverband (2016, p. 109ff). 2 3

137

4.1  The Constituting Elements of Cooperation 4

3

2

1

100% 80% 60% 40% 20% 0%

naonal polit. actors

regional polit. actors

1

0%

0%

2

0%

0%

3

50%

50%

4

50%

50%

NGOs

consumer organizaons

service providers

50%

0%

50%

80%

0%

100%

50%

20%

40%

50%

0%

0%

0%

0%

0%

0%

0%

0%

polluters

water associaons

0%

0%

29%

60%

14% 57%

science

Fig. 4.2  Actors’ attitude towards end-of-pipe measures in the Ruhr case study

sures.5 Within this category, the upgrading of waterworks,6 and wastewater treatment plants to the so-called fourth treatment stage are promoted.7 Meanwhile, the cost recovery of this second measure has not been regulated at the time of data gathering. Costs are divided among the members of the wastewater and water service associations.8 Actors causing concentration of micro-pollutants in the Ruhr, as agriculture and hospitals, are not called to account as much as needed,9 even though agriculture is considered by far the biggest contributor of micro-pollutants in the region.10 The Ruhr case study shows a divide between end-of-pipe-solutions, which focus mainly on the upgrade of wastewater treatment plants, and source-directed measures, which address the causes and thus bring the polluters to book. A contentious issue in the Ruhr case is further the financing of water examination by waterworks. Expenditures on these costly water quality checks are not covered by the water charge (Wasserpreis). Waterworks have to pay for them themselves.11 The line of conflict between actors regarding their attitude towards end-of-pipe measures becomes apparent when looking at how the different actor types evaluated this class of policy instruments. Figure 4.2 presents actors’ attitude towards end-of-­ pipe measures in the Ruhr case study with the evaluations of “totally agree” (4),

 Interview N° 15.  Interview N° 9. 7  Interview N° 13. 8  Telephone call N° 2. 9  Interview N° 13. 10  Interview N° 9. 11  Interview N° 13. 5 6

138

4  Empirical Analysis I: On Cooperation

“do somewhat agree with goal” (3), do somewhat not agree with goal” (2), and “do not agree at all with goal” (1). Actors’ attitude towards end-of-pipe measures is divided between actor groups. Political and scientific actors are in favor of them; service providers and actors from the civil society are against them. Polluters, to whom these measures mainly apply, are a little more against than in favor of them. The case study’s two water associations, both from the service provision sector, have different opinions on this measure. The lines of conflict regarding the attitude towards treatment plant upgrades also become apparent in the faction analysis (see Sect. 4.2.2).

4.1.2  C  oordinating Each Other’s Actions: Actors’ Collaboration The second element constituting cooperation is actors’ joint coordination of their actions within the management process. It implies that actors harmonize and complement their actions for regulating the environmental problem. Such coordination can take different forms: the exchange of each other’s viewpoints on the topic at stake, the joint working out of solutions, or the joint carrying out of actions. Coordination of actions is needed to ensure that actions do not undermine each other’s goals, but support and enhance one another. Data on actors’ collaboration regarding the joint alignment and complementation of actions—actions’ coordination—was gathered through question N° 5 of the questionnaire. Actors indicated with whom they had been, and still are, collaborating on the topic of micro-­ pollutants. Based on the answers, I constructed a collaboration matrix for each case, in which a 1 in a cell indicates collaboration between two actors while a 0 indicates no collaboration. These collaboration matrices served as database for the cases’ collaboration networks. The collaboration networks’ main statistics12 give a first impression of the similarities and differences between collaboration in the three cases (Table 4.2). The Basel case study has the largest collaboration network with 37 actors, followed by the Moselle case study with 31 actors. The Ruhr case study offers the Table 4.2  Network statistics of the case studies’ collaboration networks Statistics Nodes n Ties Density Average degree Standard deviation

12

Basel 37 276 20.7% 7.459 0,405

 Network statistics were calculated in UCINET.

Ruhr 26 160 24.6% 6.154 0,431

Moselle 31 216 23.2% 6.968 0,422

4.1  The Constituting Elements of Cooperation

139

Table 4.3  Collaboration networks’ cores and peripheries

n/% Density

Basel Core 14/37.8% 50.0%

Periphery 23/62.2% 7.1%

Ruhr Core 8/30.8% 83.9%

Periphery 18/69.2% 10.8%

Moselle Core 12/38.7% 61.4%

Periphery 19/61.3% 10.2%

Calculated in UCINET with the following measures: fitness measure, CORR; N° of iterations, 50; population size, 100; output partition, CorePartition

smallest network with 26 actors. As the number of actors varies across the networks, so does the number of network ties. With 276 ties, the Basel case study network has the most connections among actors; the network of the Moselle case study has 60 ties less, i.e., 216; the network in the Ruhr case study has 160 links among its actors. One possibility to compare the networks is in terms of their density. Its advantage “over the simple number of ties (…) is that it adjusts for the number of nodes in the network, making density figures comparable across groups of different sizes” (Borgatti et al. 2013, p. 151). However, it might be easier to reach out to each other in small networks than in large ones with an abundance of actors: “(…) densities are almost always lower in large networks than in small networks” (ibid., p. 151). The study’s data confirms this assumption: the largest collaboration network has the lowest density (20.7%), while the smallest network shows the highest density (24.6%). The Moselle case study network lies in between with a density of 23.2%. The average degree is the average number of ties an actor possesses within a network (cf. Borgatti et  al. 2013, p.  162). It is highest in the Basel (7.5), almost equally strong in the Moselle (7), and a bit lower in the Ruhr case study (6.2). The networks’ standard deviation is about the same in all case studies (0,41; 0,43; and 0,42) and indicates that the error rate of estimating a tie between two nodes in the network is at 40%. These basic statistics do not tell us anything about the distribution of ties within the networks. I therefore analyze the collaboration networks’ structural patterns in more detail in Sect. 4.2. A first insight into the networks’ structural specificities is their division into core and periphery. The networks’ cores reveal the centers of intensive collaboration in the three case study regions. Table 4.3 shows the numbers of actors in the collaboration networks’ cores and peripheries and the densities within the two groups. In the Basel and the Moselle case studies, 38% and 39% of the actors build the networks’ cores. The cores’ densities indicate that these actors are well connected among each other: they are high in both case studies—50% in the Basel and 61% in the Moselle case study. The core of the Ruhr case study collaboration network is smaller (31% of the actors) and much denser (84%). The actors in the core of each collaboration network are of special interest: they are the central actors that seem to “have the say” in the respective management process. The networks’ peripheral actors are more sparsely connected. In the Basel case study, the 62% of actors connect at a rate of 7.1% compared to all connections possible between them. In the Moselle case study, 61% are peripheral actors with a tie

140

4  Empirical Analysis I: On Cooperation

density of 10.2% among them. The density among the 69% of actors in the Ruhr case study collaboration network is in the same range (10.8%). Overall, network statistics show dense webs of collaborative interactions in all three cases. Actors’ backgrounds and positions within the management process are discussed in more detail in the following chapters to better understand the actor networks’ structural patterns.

4.1.3  Exchanging Resources The third element that complements cooperation is resource exchange. For this study, resources are defined as information regarding the environmental problem at stake. In N° 6 of the questionnaire, I distinguished between political and technical information. I asked the actors to indicate from whom they had received and to whom they had given such information. The information exchange networks’ statistics show that political information flow among actors is half the size of the technical information flow in all three cases (see Table 4.4). The political information exchange networks’ connections among actors and densities of ties are about half the size of the ones in the technical information exchange networks. As a consequence, actors’ average degree is half as big in the political information exchange networks as it is in the technical information exchange networks. For instance, the received and given political information in the Moselle case study account for 90 and 91 links of information flow among the case’s actors. The densities of these ties amount to 9.7% and 9.8%. Actors’ average degree in these two networks is 2.9. In the networks of received and given technical information exchange, there are about twice as many informational linkages among the actors (179 and 161) with densities and actors’ average degrees consequentially twice as big as in the political information exchange network (19.2% and 17.3% and 5.8 and 5.2). Technical information exchange is about twice as high as political information exchange in all three case studies. The networks also show differences regarding whether information is given or received: the networks of given information exchange have less ties and smaller

Table 4.4  Information exchange network statistics

Received political information Given political information Received technical information Given technical information

Basel Ruhr Density / N° of ties / average degree 9.4% / 125 / 3.38 10.8% / 70 / 2.69 7.7% / 103 / 2.78 7.4% / 48 / 1.85 19.6% / 261 / 7.05 22.9% / 149 / 5.73 14.6% / 194 / 5.24 21.7% / 141 / 5.42

Moselle 9.7% / 90 / 2.9 9.8% / 91 / 2.94 19.2% / 179 / 5.77 17.3% / 161 / 5.19

4.1  The Constituting Elements of Cooperation

141

densities than the networks of received information exchange.13 The discrepancy between receiving and handing out information is highest for the technical information exchange networks in the Basel case study—261 connections in the received information exchange network versus 194 ties in the given information exchange network. As data on the given information exchange networks is likely to be more reliable,14 I focus on the given information exchange networks for the continuing analysis. The densities of the two types of information exchange networks are similar across the case studies: densities of the received political information exchange networks range from 9.4% to 10.8% and of the given political information exchange networks from 7.4% to 9.8% across the case studies. Densities of the received technical information exchange network range from 19.2% to 22.9%. The densities in the given technical information exchange networks vary more across the cases, from 14.6% in the Basel up to 21.7% in the Ruhr case study. The actors’ centrality measures reveal who exchanges most information and thus the actors that possess and distribute most information in the case study regions. To compare actors’ centralities across the case studies, I use actors’ normalized outand in-degree centrality values, which express the measure as proportion of the connections theoretically possible (Hanneman and Riddle 2005, Chapter 7). For each case study, I focus on the 20% of actors with the highest degree centralities in the two information exchange networks (see Table 4.5). Basel Case Study In the Basel case study, the actors sending and receiving most political information differ slightly from the ones exchanging most technical information. The Federal Office for the Environment (FOEN) is by far the most central actor in the political exchange network. It is still the fourth central actor in the technical information exchange network. This is probably due to the fact that the FOEN serves not only as ministry but also as agency that accumulates and distributes knowledge about environmental topics. As both agency and Ministry for the Environment (BAFU 1 June 2016, p. 7), it is responsible for the implementation of measures regarding micro-pollutants and for the provision of information regarding the quality of Swiss waters. The FOEN supervises the monitoring process of the Swiss water bodies, informs cantonal authorities when limiting values are surpassed (Der Schweizerische Bundesrat 1998, Art. 49),15 and researches on the topic’s relevance in Switzerland (e.g., Braun et al. 2015).

 With the exception of the Moselle case’s political information exchange networks, where the given political information exchange network has one more tie than the received political information exchange network. 14  People have a better overview on whom they pass information to than on the multitude of sources they receive information from. 15  Interview N° 7. 13

142

4  Empirical Analysis I: On Cooperation

Table 4.5  The 20% of actors with highest normalized degree centralities in the given information exchange networks

1 2 3 4 5 6 7

Basel, 20% = 7.4 actors Political information FOEN_W National state actor IAWR Water association AUEBL Regional state actor AWBR Water association WWF NGO IWB Service provider Roche Polluter Ruhr, 20% = 5.2 actors Political information ARW Water association MKULNV Regional state actor RV Polluter

Out In 44.4% 25%

Technical information IWB Service provider 27.8% 11.1% AUEBL Regional state actor 25% 11.1% Eawag Science

25%

22.2% 11.1% FOEN_W

22.2% 36.1%

16.7% 16.7% WWTP Birs 19.4% 11.1% AUEBS

National state actor Polluter Regional state actor Science

Out In 44.4% 33.3% 36.1% 33.3% 41.7%

44.4% 8.3% 13.9% 38.9%

16.7% 8.3%

TZWK

Out 32%

In 12%

Out 72%

In 60%

4%

40%

Technical information MKULNV Regional state actor RV Polluter

44%

60%

32%

8%

RWW

56%

32%

4 AWWR

12%

24%

AWWR

44%

36%

5

16%

12%

Gelsen.Plc

40%

36%

Out 40%

In 3.3%

Out 50%

In 23.3%

1 2 3

1 2 3 4 5 6

Water association UBA National state actor Moselle, 20% = 6.2 actors Political information MinDev. National state LUX actor Aluseau Water association OffNat. National state LUX actor MinAgri. National state LUX actor MUEEF. Regional state RLP actor SEBES Service provider

Service provider Water association Service provider

Technical information MinDev. National state LUX actor 40% 16.7% OffNat.LUX National state actor 43.3% 10% MUEEF.RLP Regional state actor 26.7% 3.3% LIST Science 13.3% 3.3%

SEBES

16.7% 20%

SES & UNI. LUX

30.6% 16.7%

53.3% 13.3% 43.3% 20% 20%

43.3%

Service 26.7% 26.7% provider Service prov. & 26.7% 23.3% science

Network statistics were calculated in UCINET; Freeman centrality. When actors had the same degree centrality, I chose the one with the higher out-degree centrality for the last position, as I consider giving information a stronger indicator for power than receiving information The networks are the former directed given political and technical information exchange networks respectively

4.1  The Constituting Elements of Cooperation

143

The most central actors in the political exchange network comprise two of the three most important state actors, actors from the drinking water service sector,16 the civil society, and the industry. The most active actors exchanging technical information are scientific actors, the case study’s three main state actors, one of the city’s drinking water providers, and one of its wastewater treatment operators. In the Basel management process, the three main state actors hold key positions in general information distribution. Regarding the technical information exchange, scientific actors and service provider figure prominently, while water associations circulate political information to many other actors. Ruhr Case Study The Ruhr case study has about the same actors in central positions in both information exchange networks. The Ministry for Climate Protection, Environment, Agriculture, Nature and Consumer Protection NRW (MKULNV) is very central in both—in the political information exchange network, however, its out-degree centrality is low. This is due to the fact that those actors the MKULNV mainly sends political information to are non-respondents who do not show in the case study actor sample. Besides this regional state actor, the regional nonprofit water management company Ruhrverband17 is central in the management process regarding information exchange, as are drinking water service providers18 and the German Federal Environmental Agency. Moselle Case Study In the Moselle catchment area in Luxembourg and Germany, the Luxembourgian Ministry of Sustainable Development and Infrastructure (MinDev.LUX); the Luxembourgian State Office of the Environment (OffNat.LUX); the Ministry for the Environment, Energy, Food and Forest in RLP (MUEEF.RLP); and the drinking water provider SEBES19 are all central in both types of information exchange networks. Political information is further passed on by the Luxembourgian association for sewage treatment (Aluseau) and the Luxembourgian Ministry of Agriculture, Viticulture and Consumer Protection (MinAgri.LUX). Like in Basel, two scientific institutions and a drinking water provider are important regarding technical information.

 IAWR and AWBR being umbrella organizations of the waterworks in the entire Rhine basin and in the region of Lake Constance. 17  The Ruhrverband is an association of mainly wastewater treatment plant operators and other water service providers in the Ruhr region. 18  ARW being the association of drinking water providers along the Rhine between Mannheim and the German-Dutch border; AWWR being the association of drinking water providers in the Ruhr basin. 19  Syndicat des eaux du barrage d’Esch-sur-Sûre, meaning in English Union of the Waters of the Esch-sur-Sûre Dam. 16

144

4  Empirical Analysis I: On Cooperation

Throughout the case studies, state actors are the main distributors and receivers of political and technical information. Apart from this actor group, political ­information is mainly passed on by drinking water providers; and technical information is mainly passed on by scientific institutions and polluters.

4.1.4  Relating the Three Elements Data has shown that in all case studies, the three elements constituting cooperation are present, although in the Ruhr case study, we observe an absence of a clear common aim among the actors. I consider collaboration to be the essential part of cooperation as precisely the established social interactions among actors build the very base for cooperation to manifest itself. By relating actors’ collaboration to actors’ joint aims and actors’ information exchange, I highlight the symbiotic characteristic of the three elements. The elements’ statistical correlation further affirms the existence of cooperation in the two case studies of the Basel and the Moselle region and to a lesser degree in the Ruhr region. If actors working together on a specific topic aim towards a common goal, then their collaboration is intended and can be called cooperation (cf. Sect. 1.2). Therefore, I relate actors’ aims with actors’ collaboration. Welch’s two-sample t-test of each case’s similar goal index and the collaboration matrix tests whether two populations have the same mean (Ruxton 2006). In my case, the two populations are the group of actors who collaborate (group 1) and the group of actors who do not collaborate (group 0). The similar goal index reaches from 0, actors have no agreement on the five aims, to 1, actors agree on the five aims in the same way. Based on the definition of cooperation, I assume that actors who collaborate agree on average on the same goal, i.e., have a higher similar goal index value than those actors who do not collaborate. Welch’s two-sample t-test proves this assumption (see Table 4.6): throughout all case studies, the mean of the similar goal index is higher in group 1 than in group 0. The relation is statistically significant in the Basel and in the Moselle case study.20 The fact that the relation between agreeing on the same goal and collaborating is not significant for the Ruhr case substantiates that actors in this case study region do not cooperate to the full extent as in the other two cases.

Table 4.6  Results of Welch’s two-sample t-test of similar goal index and collaboration matrices Statistics p-value Index’ mean in

20

Basel 0.00001 Group 1 0.8245

Group 0 0.7957

Ruhr 0.1491 Group 1 0.8269

 For the entire statistic, see Table 13, Annex X.

Group 0 0.8147

Moselle 0.00002 Group 1 0.8581

Group 0 0.8334

4.1  The Constituting Elements of Cooperation

145

Fig. 4.3  Boxplot of t-test of actors’ similar goal index and actors’ collaboration

The boxplot graph in Fig. 4.3 visualizes the finding. It shows the t-test of actors’ similar goal index and actors’ collaboration: bars indicate the index’s median in each population; the dots within the plots indicate the mean values. The graph ­highlights the tendency of collaborating actors to have the same opinion on the five aims examined in the survey. To see whether actors’ collaboration and actors’ information exchange patterns are related, I correlate the cases’ collaboration and information exchange networks using Pearson’s Chi-squared test (Plackett 1983). The assumption that the two correlated matrices are independent, the null hypothesis, can be rejected because p-­values are significant.21 There is a significant, positive correlation between each case study collaboration network and its political as well as technical information exchange network. The likelihood that actors who collaborate also exchange information is very high—the two constituting elements of cooperation are strongly related in all three case studies. Table 4.7 shows the test results.

 Political and technical information exchange networks are undirected, i.e., they have been symmetrized. 21

146

4  Empirical Analysis I: On Cooperation

Table 4.7 Chi2 test of information exchange and collaboration networks Basel Ruhr Political information exchange network and Collaboration network X2 330.78 122.24 Df 1 1 p < 0.000 < 0.000 Technical information exchange network and collaboration network X2 475.97 258.17 Df 1 1 p < 0.000 < 0.000

Moselle 215.33 1 < 0.000 435.94 1 < 0.000

Analyzing the constituting elements of cooperation has shown that: (a) Actors have similar attitudes towards the pathways the management processes in the Basel and the Moselle region should take and thus aim towards a common goal; in the Ruhr case study, there are lines of conflict regarding the goal of the management process. (b) Actors are densely connected with each other in terms of collaboration within the management process in all three case study regions. (c) Actors exchange more technical than political information among each other on the topic of micro-pollutants. (d) Actors who collaborate also tend to exchange information with each other and strive for the same goal; the latter, however, does not apply to the Ruhr case study in the same intensity as in the other two cases. In two case study regions—the Basel and the Moselle case studies—all three elements constituting cooperation in the management process of micro-pollutants are given. In the Ruhr case study, not all actors in this case study focus on the same goal regarding the management process. Actors’ cooperation in this micro-pollutant management process is thus less intense than in the Basel and Moselle case studies. The following focus on the specific shapes and patterns of each case’s collaboration network will reveal the particularities of actor collaboration in each case —the center of cooperation.

4.2  A  Network Perspective on Collaboration, the Core of Cooperation To grasp the collaboration networks’ specificities, I examine different network characteristics and compare them across the case studies—see Table 4.8.22 I do this from a macro-level, at which I assess the networks’ reciprocity, fragmentation, and their  The network statistics were calculated with UCINET. For the entire lists of the network statistics; see Tables 13–21, Annex X. The number of mutual ties was calculated in R.

22

4.2  A Network Perspective on Collaboration, the Core of Cooperation

147

Table 4.8  Network statistics of the collaboration networks—in % Mutual ties Reciprocity Fragmentation Connectedness Normed mean of reach centrality

Basel 80 0.2899 0.08 0.92 0.54

Ruhr 47 0.2938 0.149 0.851 0.54

Moselle 64 0.2963 0.063 0.937 0.55

components; from a meso-level, at which I examine the networks’ factions; and from a micro-level, at which I analyze the peripheral and core actors as well as the important and the central actors (cf. Wasserman and Faust 1994, p. 25f.).

4.2.1  T  he Macro-level: Reciprocity, Fragmentation, and Components A crucial condition for collaboration to function is reciprocity. Collaboration as the working together of different actors requires that both sides engage in action taking. The calculated reciprocity for the three collaboration networks shows that in all three case studies, about 30% of the networks’ ties are reciprocated or, in other words, 30% of actors’ collaborative actions are responded. To acquire an understanding of the networks’ cohesion, two measures are useful: fragmentation and connectedness. They are two opposite ways to look at a network’s coherence. Fragmentation is “the proportion of pairs of nodes that cannot reach each other” (Analytictech n.d.). In the Moselle case study, this share is the smallest: 6.3% of actor pairs do not reach out to each other—through other actors— in the collaboration network. This share is only slightly higher in the Basel case study (8%). In the Ruhr case study, about 15% of possible actor couples do not connect by any means. The collaboration network of the Ruhr case study has the highest fragmentation compared to the other two networks. Generally, the Ruhr collaboration network’s fragmentation is nevertheless low. Connectedness is the other side of the “medal of cohesion” and defined as the proportion of pairs of nodes that can reach each other by a path of any length – in other words, the proportion of pairs of nodes that are located in the same component. (Borgatti et al. 2013, p. 154)

Connectedness thus equals 1 minus fragmentation (cf. Borgatti et  al. 2013, p. 154). This opposite focus on the collaboration networks highlights their cohesion even more: 92% and 94% of actor pairs are connected by network ties in the Basel and the Moselle case study; in the Ruhr case study, this measure amounts to 85% (see fourth row in Table 4.8).

148

4  Empirical Analysis I: On Cooperation

Table 4.9  Collaboration networks’ strong components N° 1 2 3 4

Basel Size 34 1 1 1

Share 0.919 0.027 0.027 0.027

Ruhr Size 23 1 1 1

Share 0.885 0.038 0.038 0.038

Moselle Size 29 1 1

Share 0.935 0.032 0.032

Reach centrality is another way to assess how connected actors of a network are. This measure sheds lights on the paths between the actors and indicates how many actors an actor reaches out to in 1, 2, 3, or more steps—a step understood as one tie between two actors (Hanneman and Riddle 2005, Chapter 10). The share of actors an actor reaches out to on average with only one step is even in the three networks: it is 54% in the Basel and Ruhr case studies and 55% in the Moselle case study. This means that on average actors in all three collaboration networks connect with more than half of the respective network’s actors by just one tie. This measure stresses the actors’ strong connectedness in all three case study regions. Cohesion can also be conceptualized as the number and the size of a network’s components (cf. Borgatti et al. 2013, p. 153 f.). A component is a subgraph of a network in which all pairs of nodes are connected through a path (cf. Sect. 3.5.1). Network components thus mirror network connectedness in another way. The component analysis for the three case studies’ collaboration networks produces four strong components for the networks of the Basel and the Ruhr case studies and three strong components in the Moselle case study network.23 As has already become apparent through the networks’ connectedness measures, all three networks consist of one large component and few components made out of one actor each (see Table 4.9). If these isolates were added to the large component, they would not be reached by all of the main component’s actors. The main components comprise between 89% (Ruhr case study) and up to 94% (Moselle case study) of the networks’ actors. They literally visualize the networks’ strong cohesion, because “(…) the bigger the main component (in terms of nodes), the greater the global cohesion of the network” (cf. Borgatti et al. 2013, p. 153). Weak components are only found in the Ruhr case study, where one actor is completely disconnected from the rest of the network. This single actor accounts for one weak component; the other actors comprise the network’s second weak component.24 The networks’ graphs (Figs.  4.4, 4.5 and 4.6) visually depict the network components.

23 24

 For the component analysis’ statistics, see Table 17, Annex X.  For the Ruhr case study’s weak component analysis statistics, see Table 18, Annex X.

4.2  A Network Perspective on Collaboration, the Core of Cooperation

149

Fig. 4.4  Network graph of the Basel case study collaboration network; single strong components: CITYWeil; HKBB and KI.SSV.SGV Legend: Square nodes represent Swiss actors; circles represent non-Swiss actors; colors represent actor types: green = NGO; orange = consumer group; turquoise = science; dark blue = water association; light blue  =  service provider; violet  =  polluter; dark pink  =  national state actor; Bordeaux red = regional state actor The graph was visualized with NetDraw in UCINET (Borgatti et al. 2002)

Given the big size of the networks’ main components, the networks’ component ratio is accordingly low. The component ratio is the number of components, c, minus 1 divided by the number of nodes in the network, n, minus 1:



c -1 n - 1

If every actor is an isolate, the ratio equals the value of 1. The value 0 indicates that the network possesses only one component. Subtracting the component ratio from 1 produces the network’s cohesion measure (Borgatti et al. 2013, p. 153). For the collaboration networks of the three case studies, the component ratios and the cohesion measures amount to the following values25 (Table 4.10): The macro-level network analysis proved that collaboration among actors is strong in all case study regions. In the three collaboration networks, 30% of actors’ connections are reciprocated. Fragmentation is low in all networks, and actors’ con-

25

 For the calculation of the component ratios, see Fig. 3, Annex X.

Fig. 4.5  Network graph of the Ruhr case study collaboration network; single strong components: Fish.RUHR, Paper.NRW, and UBA Legend: Square nodes represent German actors; circle represents single Swiss actor; colors represent actor types: green = NGO; orange = consumer group; turquoise = science; dark blue = water association; light blue  =  service provider; violet  =  polluter; dark pink  =  national state actor; Bordeaux red = regional state actor The graph was visualized with NetDraw in UCINET

Fig. 4.6  Network graph of the Moselle case study collaboration network; single strong components: StGB.RLP and ULC Legend: Square nodes represent Luxembourgian actors; circles represent German actors; colors represent actor types: green = NGO; orange = consumer group; turquoise = science; dark blue = water association; light blue = service provider; violet = polluter; dark pink = national state actor; Bordeaux red = regional state actor The graph was visualized with NetDraw in UCINET

151

4.2  A Network Perspective on Collaboration, the Core of Cooperation Table 4.10  Collaboration networks’ component ratios & cohesion measure Component ratio Cohesion

Basel 0.083 0.917

Ruhr 0.12 0.88

Moselle 0.067 0.933

nectedness is high. Actors’ average reach centrality lies at about 55% in each case study, and all three networks comprise one large component consisting of 89% (Ruhr), 92% (Basel), and 94% (Moselle) of each network’s actors. Given all these indices, cohesion is high in all three networks (0.92 in the Basel; 0.88 in the Ruhr; and 0.93 in the Moselle case study). A look at the networks’ meso-level reveals that the network pattern splits up actors into different subgroups, each with particularly intense collaboration.

4.2.2  The Meso-level: Factions In order to identify such potential groups within the networks, I conducted a faction analysis.26 I ran the algorithm several times to make sure no actor belongs to more than one faction of the network and that actors were always assigned to the same groups.27 I determined the factions’ number for each case study collaboration network after having run a series of different numbers of faction partitions. I compared them regarding their final proportion correctness, their validity in actor assignment, and their value in meaningfulness.28 The analysis revealed three factions for the Ruhr and four for the Basel and the Moselle case studies.29

 The analysis was conducted using the factor analytic program of UCINET.  I tested six times for each case whether actors would be assigned to the same factions. All actors were always assigned to the same faction, cf. Borgatti et al. (2013, p. 192). 28  I ran algorithms on two, three, and four faction partitions for each network and compared the results’ final proportion correctness. In a second step, I ran the algorithms on the same number of partitions for each network another five times to assure actors were always placed in the same faction. This was not the case, for instance, when splitting the Basel case study’s collaboration network into three factions. I further qualitatively compared actors’ constellation within the factions. If a further added faction did not reveal a group of newly combined actors but rather split an already existing faction into two smaller factions with increasing low densities, I decided on the former number of factions. 29  For an overview of the cases’ factions, see Tables 19, 20 and 21, Annex X. 26 27

152

4  Empirical Analysis I: On Cooperation

Basel Case Study In the Basel case study, the collaboration network is grouped into four factions: three heavily dense factions and one extremely sparse faction (see Table 4.11). I call the first and largest faction of 12 actors service providers & science faction. The majority of scientific actors, two regional state actors in charge of water examinations, Basel’s drinking water providers, and a national and an international water association, both related to service provision, fall into this category. The second faction of ten actors comprises the region’s polluters & main political actors. The third faction consists of nine actors: NGOs and a consumer organization, again a national and an international water association, the association of Swiss cantonal experts on aquatic biology and chemistry, and the key national state actor of the case study region. This faction is labeled the civil society faction. The last and least dense faction is made up of one Swiss and two German regional state actors, a French NGO, a Swiss industrial association, and a Swiss service provider in charge of wells. As for its internationality as well as for the loose connections among its actors and towards the other factions it is titled FrenchGerman-Swiss peripheral faction. Ruhr Case Study In the Ruhr case study, the scientific and regional state actors are grouped together with the regional association of wastewater treatment plants, a large regional drinking water provider, and the regional NGO working on the issue, forming the science & regional state actors faction. The service provision faction is composed of eight actors: two umbrella associations of the Ruhr and Rhine waterworks, regional service providers, one polluter, and one scientific actor. The third faction comprises three agricultural and industrial polluters, two consumer organizations—i.e., fishing associations—and the case’s two national state actors. They are more loosely connected and titled peripheral polluters & national state actors (Table 4.12).

Table 4.11  Faction analysis of the Basel case study—final proportion correctness: 79.4% Faction 1 2 3 4

n 12 10 9 6

Density 53% 61% 51% 3%

Title Service providers & science faction Polluters & main political actors faction Civil society faction French-German-Swiss peripheral faction

Table 4.12  Faction analysis of the Ruhr case study—final proportion correctness: 73.8% Faction 1 2 3

n 11 8 7

Density 59% 52% 12%

Title Science & regional state actors faction Service provision faction Peripheral polluters & national state actors

4.2  A Network Perspective on Collaboration, the Core of Cooperation

153

Table 4.13  Faction analysis of the Moselle case study—final proportion correctness: 81.3% Faction 1 2 3 4

n 12 7 7 5

Density 67% 60% 57% 5%

Title Luxembourgian faction RLP faction Saarland-Luxembourg faction Luxembourgian consumer & polluter faction

Moselle Case Study In the Moselle case study, actors are grouped in the following four factions: all Luxembourgian state actors, several Luxembourgian polluters and service providers as well as the only NGO, and one scientific actor form the largest faction with 12 actors, the Luxembourgian faction. The second and third factions both have seven actors. One consists of regional state actors, a service provider, a water association, and a polluter—all from the German federal state Rhineland-Palatinate (RLP). It is thus called the RLP faction. The other holds regional state actors and a consumer organization from the German federal state Saarland, polluters from Saarland and Luxembourg, and scientific actors from RLP and Luxembourg. Its title is Saarland-­ Luxembourg faction. The smallest and least dense faction with five actors is the Luxembourgian consumer & polluter faction, comprising two consumer organizations and two polluters from Luxembourg and one regional state actor from RLP (Table 4.13). Although the case studies’ collaboration networks are strongly cohesive, the faction analysis has shown that actors group differently in cohesive subgroups across the cases. In the case of the Moselle region, actors form factions along their territoriality. In the Ruhr case study, there is a divide between sectors: service providers are on the one, scientific, and regional state actors on the other side. This structural separation underlines the line of conflict between service providers and scientific and state actors regarding the implementation of end-of-pipe measures. The Basel case shows a division roughly according to specialization: strong collaboration can be observed among actors from the civil society, among service providers with scientific actors, and among polluters with regional state actors. The next chapter focuses on the actors themselves and assesses the core and central actors of the collaboration networks. The analysis reveals whether the central actors are also considered important for the management process.

4.2.3  The Micro-level: Core, Important, and Peripheral Actors Section 4.1.2 already touched upon the concept of a network’s core and periphery. To know the actors of a network’s core means to know the key actors of the network. At the same time, it is important to know the actors lingering at the periphery of a network. What if, for instance, a highly expert and engaged actor that is crucial for the social system is situated far from the core and from the central actors of the

154

4  Empirical Analysis I: On Cooperation

network? I first examine the collaboration network’s core actors and assess their degree centrality. I then check whether the actors in the networks’ core are also those said to be important in the management process of micro-pollutants. The chapter ends with an assessment of the collaboration networks’ most peripheral actors. 4.2.3.1  The Collaboration Networks’ Cores The network statistics show that core actors come from different sectors, all of which are essential to resolve the CPR problem of micro-pollutants. The data further show that actors in the core are deemed important by the network’s actors—the share of actors evaluating the core actors as important ranges from 33% up to 83%. Basel Case Study In the case of Basel, 14 actors (37.8% of the network’s actors) are in the core. Table 4.14 lists the core actors’ degree centralities: the sum of their incoming and outgoing collaboration ties with other actors (degree); their normalized in- and out-­ Table 4.14  Core actors in the Basel case study collaboration network and their degree centrality & reputational measures

1 2 3 4 5 6 7 8 9 10 11 12 13 14

Actor FOEN_W

Type National state actor Eawag Science AUEBL Regional state actor AUEBS Regional state actor TZWK Science WWTP Rhein Polluter IWB Service provider SVGW Water association LABBL Regional state actor WWB Service provider WWTP Basel Polluter Polluter WWTP ChemBasel ICPR Water association VSA Water association

Degree 39

Normalized in-degree 0,639

Normalized out-degree 0,444

Reputation in-degree 30/83.3%

32 30

0,583 0,389

0,306 0,444

28/77.8% 20/55.6%

28

0,389

0,389

23/63.9%

26 21 21

0,389 0,250 0,472

0,333 0,333 0,111

18/50.0% 17/47.2% 19/52.8%

20

0,306

0,250

15/41.7%

20

0,222

0,333

15/41.7%

20

0,278

0,278

12/33.3%

19 19

0,222 0,222

0,306 0,306

16/44.4% 15/41.7%

18

0,250

0,250

18/50.0%

18

0,417

0,083

23/63.9%

4.2  A Network Perspective on Collaboration, the Core of Cooperation

155

degree measures in the collaboration network; and their in-degree of reputation. Reputation reflects how many actors deemed the core actors important for the ­management process of micro-pollutants. The value is rendered in number of ties and percentage of actors. State actors are relevant as they put the environmental issue on the political agenda and work out the measures aiming at the problem’s solution. The water division of the Swiss Federal Office for the Environment (FOEN_W) is the network’s actor with the highest reputation and the highest degree centrality: 83% of the actors find this actor important for the management process, and 64% have collaborative ties to him. FOEN_W itself entertains collaborative ties to 44% of the actors. The two main regional state actors working on the issue of micro-pollutants in the Basel region, the cantonal offices for the protection of the environment (AUEBL and AUEBS), have incoming collaboration ties from 39% of the network’s actors and connect with 44% and 39% of them. They are engaged in the collaboration network as strongly as the national state actor. They are judged important by 56% and 64% of the actors. The cantonal laboratory of the Canton Basel Country (LABBL) is a regional state actor in charge of examining the quality of the cantonal water bodies. The laboratory is considered important by 42% of the actors and is less active, reaching out to 33% of the actors, and less popular—sought out for collaboration by 22% of the actors—than the two other regional state actors. Scientific actors inform state actors, service providers, and resource users about the possibilities and the feasibility of technical solutions to the CPR problem. The two scientific experts on the topic,30 the Swiss Federal Institute of Aquatic Science and Technology (Eawag) and the technology center on water in Karlsruhe (TZWK), are said to be important by 78% and 50% of the actors, and they are central in that they collaborate with 31% and 33% of the actors and contacted for collaboration by 58% and 39%. Service providers and wastewater treatment plant (WWTP) operators are water users and polluters and addressees of the rules developed to handle the CPR problem. They implement the specific measures regarding micro-pollutants’ regulation. The two drinking water providers of Basel, IWB and WWB, are considered important by 53% and 33% and contacted for collaboration by 47% and 28% of the actors. The more “popular” IWB engages less in collaboration with others (11%) than WWB (28%). The three WWTP operators in Basel are also in the network core. Between 42% and 47% of the actors see them as important for the management process. They are reached by 22% to 25% of the actors while they contact 31% to 33% of them. Water associations, finally, function as informing and connecting actors on the topic. The most important supra-governmental institution regarding the regulation and protection of the waters and ecosystems of the Rhine catchment area, the International Commission for the Protection of the Rhine (ICPR), is in the network’s core. So are the Swiss Gas and Water Industry Association (SVGW) and the Swiss Water Association (VSA). The three associations are important—50%, 42%,

30

 Interview N° 3.

156

4  Empirical Analysis I: On Cooperation

Table 4.15  Core actors in the Ruhr case study collaboration network and their degree centrality & reputational measures

1 2 3 4 5 6 7 8

Actor MKULNV RV AWWR IWW Gelsen.plc DA.Duess RWW ARW

Type Regional state actor Polluter Water association Science Service provider Regional state actor Service provider Water association

Degree 37 27 22 22 19 17 15 15

Normalized in-degree 0,840 0,520 0,440 0,480 0,440 0,360 0,240 0,120

Normalizedout-­ degree 0,640 0,560 0,440 0,400 0,320 0,320 0,360 0,480

Reputation in-degree 24/96% 23/92% 16/64% 14/56% 13/52% 12/48% 8/32% 7/28%

and 64% of the actors think so—and engage in collaboration with 25% (ICPR and SVGW) and 8% (VSA) of the actors while they are contacted to collaborate by 25%, 31%, and 42% of the actors. Ruhr Case Study The core of the collaboration network in the management process in the Ruhr region comprises eight actors, i.e., 30.8% of the actors, and has a similar mix of actors as the Basel case—Table 4.15 summarizes this case study’s statistics. The principal regional state actor, the Ministry for Climate Protection, Environment, Agriculture, Nature and Consumer Protection of the German federal state NRW (MKULNV), has by far the highest degree centrality and reaches out to 64%, while it is contacted for collaboration by 84% of the network’s actors. Except for one, all actors agree that the MKULNV is important for the management process of micro-pollutants in the Ruhr region (96%). The other regional state actor, the district authority Düsseldorf (DA.Duess), is less active (32%) and less contacted (36%) than the MKULNV, but still considered important by about 50% of the actors.31 The IWW Water Centre is the only scientific actor in the network’s core. More than half of the actors say the IWW is important for the management process, and it is quite engaged in collaboration, contacting 40% and reached by 48% of the network’s actors. Most of the core actors are service providers, i.e., drinking water providers. Gelsenwasser AG is more important (52%) than the RhenishWestphalian Waterworks, RWW (32%), and also contacted by more actors (44%) than RWW (24%). The two are similarly active in the collaboration network (32% and 36%).  A third regional state actor needs to be mentioned: the State Office for Nature, the Environment and Consumer Protection NRW (LANUV.NRW). The LANUV.NRW takes regular measurements of the water bodies in NRW, examines the samples, and produces maps and reports of the water bodies’ conditions (Interview N° 12). As this actor did not answer all of the questionnaire’s questions, it could not be included in the analysis.

31

4.2  A Network Perspective on Collaboration, the Core of Cooperation

157

Table 4.16  Core actors in the Moselle case study collaboration network and their degree centrality & reputational measures

1 2 3 4 5 6 7 8 9 10 11 12

Actor MinDev.LUX MUEFF.RLP OffNat.LUX Aluseau LIST TU Kais MinAgri.LUX UNI.LUX SIDEN SIDEST SEBES SES

Type National state actor Regional state actor National state actor Water association Science Science National state actor Science Polluter Polluter Service provider Service provider

Degree 29 26 26 25 23 23 19 21 16 17 18 19

Normalized in-degree 0,300 0,400 0,200 0,400 0,533 0,367 0,300 0,333 0,400 0,300 0,333 0,267

Normalized out-degree 0,667 0,467 0,667 0,433 0,233 0,400 0,333 0,367 0,133 0,267 0,267 0,367

Reputation in-degree 16/53.3% 17/56.7% 15/50.0% 14/46.7% 17/56.7% 16/53.3% 15/50.0% 14/46.7% 14/46.7% 13/43.3% 11/36.7% 8/26.7%

Both water associations in the core belong to the service provision sector as well. The Association of Waterworks Ruhr (AWWR) is more important (64% to 28%) and contacted more (44% to 12%) than the Association of Waterworks Rhine (ARW). This seems a logical consequence as the AWWR is the waterworks’ umbrella organization within the Ruhr region while the ARW acts on behalf of the waterworks within the Rhine catchment area extending between Mannheim and the German-Dutch border (ARW 2018). The two associations engage equally strong in the collaboration network, contacting 44% and 48% of the network’s actors. The single WWTP operator in the network’s core is at the same time the case study’s second most important actor (92%). The Ruhrverband is a key actor32 for the management process and contacting and contacted by more than half of the collaboration network’s actors. Moselle Case Study In the case of micro-pollutant management in the Moselle catchment area, the composition of the collaboration network’s 12 core actors (38.7% of the actors) is again similar to the preceding cases. The vast majority of the core’s actors are Luxembourgian—only two are German (see Table 4.16). Three state actors from Luxembourg and one from the German federal state Rhineland-Palatinate (RLP) are in the core. The Ministry for the Environment in RLP (MUEEF.RLP) is considered the most important of the core’s state actors

32

 Interviews N° 9 and 12.

158

4  Empirical Analysis I: On Cooperation

Table 4.17 Chi2 test of the cases’ reputation and collaboration networks X2 df p

Basel 420.11 1 0.000

Ruhr 164.77 1 0.000

Moselle 295.3 1 0.000

(57%), followed by the Luxembourgian Ministry for Development (MinDev.LUX, 53%). Half of the network’s actors judge the Luxembourgian National Office for the Environment (OffNat.LUX) and the Luxembourgian Ministry for Agriculture (MinAgri.LUX) important. The last one is the least engaged in collaboration (contacting 33% of the actors), while the other three reach out to 47% (MUEFF.RLP) and 67% of the network’s actors. The state actors are contacted by between 20% and 40% of the actors. The Luxembourg Institute of Science and Technology (LIST), the Technical University of Kaiserslautern (TUKais), and the University of Luxembourg (UNI. LUX) are considered important by between 47% and 57% of the actors. These scientific actors collaborate with between 23% and 40% and are contacted for collaboration by 33% to 53% of the actors. Two Luxembourgian service providers are part of the core: the Union of the Waters of the Esch-sur-Sûre Dam (SEBES) and the Union of the Waters of South Koerich (SES). They are considered important by 37% and 27% of the actors, and they contact 27% and 37% of the actors while themselves being contacted by 33% and 27%. The water association Aluseau is the Luxembourgian association of WWTPs and considered important by almost half of the actors. Aluseau reaches out to and is contacted by around 40% of the actors. The two WWTP operators (SIDEN and SIDEST) are less active in the collaboration network (contacting 13% and 27%) than they are popular (reaching 40% and 30% of the actors). More than 40% of the actors consider them important for the management process. The three collaboration networks’ cores have about the same size and comprise actors from the same different sectors, which need to be involved when tackling the environmental problem of micro-pollutants. All networks’ cores cover only polluters from settlements and lack those from the agricultural and industrial sectors. The Moselle case core actors are furthermore mainly from Luxembourg—making up 83% of the core actors. All core actors of the case studies are considered important by their peers. This is not surprising, as it is conceivable that actors perceive those actors as important with whom they collaborate or, vice versa, that actors collaborate with those that they find important. A Pearson’s Chi-squared test of the ­collaboration and reputation networks shows a clear correlation of the two networks for all case studies (Table 4.17) and supports this assumption.33

 The strong correlation between the actors’ collaboration and reputation networks further supports the decision of having tested for actors’ reputation in the ERGM via a binary variable and not

33

4.2  A Network Perspective on Collaboration, the Core of Cooperation

159

Table 4.18  The case studies’ 30% most important actors N° 1 2 3 4 5

Basel Actor FOEN_W Eawag AUEBS VSA AUEBL

Reput. 83.3% 77.8% 63.9% 63.9% 55.6%

Core Yes Yes Yes Yes Yes

6 7 8 9

IWB ICPR TZWK WWTP Rhein

52.8% 50.0% 50.0% 47.2%

Yes Yes Yes Yes

10 WWTP Basel 44.4% Yes 11 LABBL 41.7% Yes

Ruhr Actor MKULNV RV AWWR UBA IWW

Reput. 96.0% 92.0% 64.0% 64.0% 56.0%

Core Yes Yes Yes No Yes

CompCent.NRW 52.0% No Gelsen.plc 52.0% Yes DA.Duess 48.0% Yes

Moselle Actor MUEFF.RLP MUV.SAAR LIST CoA.LUX MinDev. LUX OffNat.RLP TUKais OffNat.LUX MinAgri. LUX CoA.RLP

Reput. 56.7% 56.7% 56.7% 56.7% 53.3%

Core Yes No Yes No Yes

53.3% 53.3% 50.0% 50.0%

No Yes Yes Yes

50.0% No

Various studies on reputation within policy networks revealed strong correlations between actors’ reputation and their positions within the actor network (Knoke 1998, p. 511; Laumann and Pappi 1976, p. 98)—a relation confirmed once more by this study. The possibility remains, though, that some important actors are not in the core. Table  4.18 lists the 30% of each case’s most important actors and highlights the ones who are not in the collaboration networks’ core.34 4.2.3.2  The Collaboration Networks’ Most Important Actors In the Basel case study, the 30% most important actors are all in the collaboration network’s core (for the case studies’ 30% most important actors, see Table 4.18). In the case of the management process in the Ruhr region, two of the eight most important actors are not in the collaboration network’s core. The German Federal Environmental Agency (UBA) is considered important by 64%. As Germany’s “main environmental protection agency” (UBA 2015), the UBA conducts research on environmental issues of national interest. The UBA gathers data on surface water quality and assesses environmental threats and risks for aquatic systems. It reports to the

via the actors’ reputation matrix. The reputation matrix would not have given a valid explanation for the existence of a tie in the actors’ collaboration network since it is too similar to the collaboration matrix. 34  In the Basel case, 30% of the actors are 11.1 actors; in the Ruhr case, 30% correspond to 7.8 actors; and in the Moselle case, 30% amount to 9.3 actors.

160

4  Empirical Analysis I: On Cooperation

federal government, which uses the information in their decision-making processes (UBA October 2014, p. 3). As an advisory national state actor, the UBA is sought out for collaboration by 28% of the Ruhr case study actors but is itself not active in the collaboration network—i.e., it has no outgoing ties. The UBA is furthermore a single strong component, meaning that it cannot reach all actors of the large component directly or via paths in keeping with the ties’ directions. The scientific actor “Competence Center Micro-pollutants NRW“(CompCent.NRW) is the other important actor that is not in the core. The competence center is a consortium of four companies, which was founded by the MKULNV and researches on industrial engineering and operations for the reduction of micro-pollutants. It reaches out to 24% of the actors to collaborate and is contacted by 28%; 52% of the actors reckon it important. The Moselle case actors have a low normalized in-degree reputation compared to the other two cases—none of the Moselle actors exceed 57%, while the other two cases have actors with a reputation of up to 83% and 96%. Actors’ territoriality offers an explanation: the Moselle case consists of 41.9% German actors and 58.1% Luxembourgian actors. As shown in the faction analysis, the Moselle collaboration network comprises a strictly Luxembourgian and an entirely German faction. Actors in this case study thus seem to collaborate more within each country than across the countries. German actors probably know their German colleagues better than the Luxembourgians and vice versa; this may lead to actors’ bias towards the own nationality when evaluating the most important actors in the management process. Four important actors are not part of the collaboration network’s core, two of which are state actors from the German federal states RLP and Saarland. The other two are the chambers of agriculture in Luxembourg and RLP. They are all considered important by half or more than half of the collaboration network’s actors. The Ministry for the Environment and Consumer Protection in the federal state Saarland (MUV.SAAR) is not as active (contacting 26.7% of the actors) and popular (contacted by 20% of the actors) in the collaboration network. The State Office of the Environment in RLP (OffNat.RLP) is sought out by more actors (33%), but itself contacts only 10% of the actors. The Luxembourgian chamber of agriculture is slightly more active than its German counterpart (16.7% to 13.3%) and receives more collaborative ties than the German agricultural chamber (33% to 16.7%). Actors in the collaboration networks’ cores of all three case studies come from different domains, representing state actors, science, service providers, WWTP, and water associations. Polluters from the industry and the agricultural sector are missing in the networks’ cores. The actors central in collaboration in each management process thus cover almost all sectors needed to overcome the CPR problem at stake: the regulators and decision-makers responsible for developing the policy solution, i.e., the measures; the scientific experts informing on the extent and severeness of the problem as well as on the feasibility and costs of possible solutions; the addressees of the measures, i.e., the resource users and the polluters from the settlements; and associations of the water sector, which function as information channels and hubs of knowledge exchange and acquaintance (Fischer and Leifeld 2015).

161

4.2  A Network Perspective on Collaboration, the Core of Cooperation

Core actors are predominantly deemed important for the management process— although it remains unclear whether they are considered important because of their central position within collaboration or whether they are central because they are important. The Ruhr case study’s nucleus of collaboration lacks two important actors, one from the scientific sector and the other from the state side, affiliated to science. In the Moselle case study, the collaboration core has a majority of Luxembourgian actors and lacks four important actors: two from the agricultural sector and two German state authorities at the regional level. 4.2.3.3  The Collaboration Networks’ Most Peripheral Actors On the other side of the spectrum are the peripheral actors. The focus here is on the 15% least central ones of each case study collaboration network (see Tables 4.19, 4.20, and 4.21 for each case).35 Table 4.19  The 15% least central actors of the Basel case study collaboration network

1 2 3 4 5 6

Actor KI.SSV. SGV CITYWeil

Type Regional state actor Service provider APRONA NGO AdmLoerr Regional state actor HKBB Polluter SBrV Service provider

Normalized Normalized Degree in-degree out-degree 2 0 0,056

Reputation in-degree 2/5.6%

Single strong component Yes

3

0

0,083

3/8.3%

Yes

4 4

0,056 0,056

0,056 0,056

2/5.6% 6/16.7%

No No

4 5

0,111 0,111

0 0,028

7/19.4% 7/19.4%

Yes No

Table 4.20  The 15% least central actors of the Ruhr case study collaboration network

1 2 3 4

Actor Fish. RUHR Paper. NRW VKU. NRW Fish. NRW

Type Consumer organization Polluter Service provider Consumer organization

Normalized Degree in-degree 0 0

Normalized out-degree 0

Reputation in-degree 5/20.0%

Single strong component Yes

1

0

0,040

4/16.0%

Yes

4

0,040

0,120

4/16.0%

No

5

0,160

0,040

5/20.0%

No

 In the Basel case, 15% of the actors amount to 5.5 actors; in the Ruhr case, 15% equals 3.9 actors; and in the Moselle case, 15% correspond to 4.65 actors.

35

162

4  Empirical Analysis I: On Cooperation

Table 4.21  The 15% least central actors of the Moselle case study collaboration network

1 2 3 4 5

Actor ULC

Type Consumer organization SIVEC Polluter LDEW Water association StGB. Regional state RLP actor Fish. Consumer SAAR organization

Normalized Degree in-degree 2 0

Normalized out-degree 0,067

Reputation in-degree 3/10.0%

Single strong component Yes

4 4

0,100 0,033

0,033 0,100

9/30.0% 3/13.3%

No No

4

0,1330

0

3/10.0%

Yes

5

0,067

0,100

4/13.3%

No

In the Basel case study, these are the Swiss association of local authorities (KI. SSV.SGV); the waterworks of the German city of Weil am Rhein (CITYWeil); a French environmental NGO (APRONA); the district administration of the German city of Lörrach (ADMLoerr); the Basel chamber of commerce (HKBB); and the Swiss association of well craftsmen (SBrV). The first three actors are not considered important; neither are the last three; only 17% and 19% of the actors see them as important. The two most peripheral actors—KI.SSV.SGV and CITYWeil—and HKBB are single strong components, this highlighting their peripheral positions even more. The Ruhr case study collaboration network’s most peripheral actor is the industrial federation of the paper industry in NRW (Paper.NRW), which at the same time is a single strong component. The network’s second single strong component cannot even be claimed peripheral: the Cooperative Association of Fishing in the Ruhr (Fish.RUHR) is not connected to any actor of the network at all and is therefore an isolate. Still, the association is evaluated as being important by 20% of the actors. The association of municipal enterprises in NRW (VKU.NRW) and Paper.NRW is not as important—only 16% judge them so. The case’s second consumer organization, the Fishing Union of NRW, is also among the most peripheral actors. Like its counterpart in the Ruhr region, 20% think the Fishing Union is of importance for the management process. A consumer organization is also the most peripheral actor in the Moselle case study: the Luxembourgian Union of Consumers (ULC) is a single strong component and is not considered important for the management process of micro-pollutants. Further peripheral actors are a wastewater treatment plant operator in Southern Luxembourg (SIVEC); the regional Association of the Supply Industry in Hesse and RLP (LDEW); the Association of Communities and Cities in RLP (StGB.RLP)— the case’s other single strong component; and the Fishery Association Saar (Fish. SAAR), yet another consumer organization. Except for the WWTP, which is considered important by 30% of the actors, none of the least central actors are considered important for the management process of micro-pollutants. Two aspects of the actors at the collaboration networks’ edges are common across the cases: they are not considered important; and, except for one, all single

4.3  Actors’ Perceptions of Cooperation

163

strong components of the collaboration networks are part of the group of the least central actors. In the Moselle and Ruhr case studies, the peripheral actors are, among others, from the consumer sector—which could be an indication that this sector is neither important for nor integrated into the management process. In the Basel case study, two German state actors from the regional level and a French NGO are peripheral. This might point to the fact that actors across the Swiss border are not so relevant for the handling of micro-pollutants in the Basel region.

4.3  Actors’ Perceptions of Cooperation Proceeding from the network level to the actor level depicts the impression that actors in the management processes themselves have about cooperation. The following sub-chapters shed light on their viewpoints, which demonstrate a more detailed picture of actor cooperation in the three case studies. I support the statements with insights from the descriptive SNA of the cases’ collaboration networks.

4.3.1  … in the Basel Case Study The interviewed actors from the micro-pollutant management process in the Basel region commented on cooperation from different angles. A regional state actor assessed cooperation between the different state levels regarding the development and implementation of measures as follows: Das [BAFU] ist ein sehr wichtiger Partner für uns im Vollzug. Wir halten Rücksprache [bezüglich] wie sollen wir es [die Umsetzung der Massnahmen] machen. Sämtliche Probleme, die wir im Vollzug haben, werden immer mit dem BAFU diskutiert; Revisionen, aber auch Anträge. So dass wir aktiv ans BAFU gelangen und sagen ‘wir haben ein Problem in Basel und wir müssen eigentlich das Problem auf nationaler Ebene gelöst bekommen.’ Und bringen uns auch so aktiv in die ganze Gesetzgebung mit ein.36 (Interview N° 3)

From the viewpoint of a regional service provider,37 cooperation with higherlevel authorities is only indirect, which at times is laborious. The Federal Office for the Environment (FOEN_W) contacts the regional state actors for information on

 English translation: “Regarding the implementation, the FOEN is a very important actor. We confer how to do it. Any problem we have in the implementation process, we discuss with the FOEN; revisions, but also petitions. We actively approach the FOEN, telling them ‘We’ve got a problem at Basel that actually needs to be solved at the national level’. By this, we participate in the legislation process.” The Federal Office for the Environment (FOEN) provides so-called Wegleitungen guidelines for the implementation of measures. The FOEN also sends the announcements (Vernehmlassungen) that the Cantonal state actors can comment on (Interview N° 3). 37  Interview N° 4. 36

164

4  Empirical Analysis I: On Cooperation

the state of the regional implementation process. However, the regional state actors do not always have the requested answers: Im März hat eine Sitzung bezüglich Mikroorganismen stattgefunden, nur zwischen BAFU und AUE[BS]. Sie wollten nicht die ProRheno dabei. Und dann hatten sie viele Fragen, die das AUE nicht wusste, und sind zu uns gekommen. Wenn wir alle an einem Tisch gesessen wären… aber sie wollen das so.38 (Interview N° 4)

The Social Network Analysis of the Basel case’s collaboration network does reflect the intense collaboration the FOEN entertains with most of the management process’s actors: this actor’s degree centrality—that is, the in-degree and out-degree of this actors’ collaboration ties put together—is 39 and the highest of all the case study’s actors. However, regarding the contacts to regional state actors, the FOEN “only” collaborates with the regional state actor AUEBS.39 The FOEN shares a collaboration tie to the ProRheno AG, the WWTP operator not invited to the round table in question. The statement thus shows that even if ties exist, it is also a question at which occasions they come into action. Furthermore, cooperation between service providers on the regional level could be better, as one interviewee stated. Actors engaged in the management of water seem to be poorly coordinated.40 This could be because water service providers in Switzerland act mainly at the local level and local communities are relatively autonomous in their work. The data confirms this statement in that the case’s service providers are in different factions (cf. Table 19, Annex X). Fig. 4 in Annex X depicts the graph of collaboration ties between service providers and regional state actors.41 Here one sees that the drinking water provider Hardwasser AG (WWB) connects to the service providers IWB, WWR, and SBrV who themselves do not share collaboration ties among each other but only with WWB. The other drinking water provider in Basel, IWB, is the only one collaborating with the German counterpart Waterworks Südliches Markgräflerland (CITYWeil). The case’s sixth service provider (KI) does not collaborate with any of the other service providers. According to one service provider, intensification of the communal-cantonal cooperation would improve the coordination of actions among the disconnected local service providers. According to this actor, the cooperation of local entities with cantonal and national authorities seems to be less intense in the Swiss context than compared to the according cooperation in Germany or the Netherlands.42 The collaboration ties between service providers and regional state actors (Fig. 4, Annex X) show that cantonal state actors do function as a bridging connection between  English translation: “In March, there was a meeting regarding micro-pollutants, only between the FOEN and the AUE(BS). They didn’t want to have ProRheno there. And then they had a lot of questions which AUE couldn’t answer. And so they came to us. If we all had been sitting at the table… but that’s how they want it.” 39  The FOEN collaborates with the two cantonal laboratories of Basel City and Basel Country (LABBL & LABBS), which are run by the cantons but coded as scientific actors. 40  Interview N° 3. 41  The graph was visualized with NetDraw in UCINET. 42  Interview N° 2. 38

4.3  Actors’ Perceptions of Cooperation

165

certain service providers, although not in an immediate way: the service provider KI collaborates with the cantonal state actor AIBBL who in turn collaborates with AUEBL and AUEBS.  The Cantonal Office for Environmental Protection and Energy, Basel Country (AUEBL) is the cantonal state actor with most collaboration ties to the further service providers, three altogether. This actor is in a gateway position connecting service providers and regional state actors. The other key cantonal actor, the Cantonal Office for the Environment and Energy, Basel City (AUEBS), “only” collaborates with one service provider, the drinking water operator IWB.  Overall, the graph rather shows a divide between Cantonal/regional state actors on the one side and service providers on the other than an intertwined cooperation pattern between the two actor groups. Communal-cantonal cooperation in the micro-pollutant management process thus could still be intensified. Regarding alarming in case of a drastic pollution incident, cooperation among the relevant actors works well. State authorities, the cantonal labors, drinking water providers, WWTP operators and the region’s chemical industry of the Cantons Basel City and Basel Country, the neighboring Canton Aargau, and the German federal state Baden-Wurttemberg form a so-called “circle of trust.”43 Established before 2006, the circle of trust functions as an alarming system and has been especially active since 2012, when even lower concentrations of substances were detected due to the new screening method. If one of the circle’s members detects a substance in a concentration higher than permitted, this actor contacts the other members of the circle. This circle of trust cannot be retraced in the case’s collaboration network since not all the circle’s members are also part of the actor sample. Nevertheless, the collaborative connection from the German service provider and the German state actor to the cantonal state actor in Basel City (AUEBS) can be seen in Fig. 4 in Annex X; and the connection between polluters—WWTP operators and the chemical industry in Basel—and cantonal state actors becomes apparent in the case’s polluters & main political actors faction (faction 2).44 Through this communication channel, the causes of the pollution can be identified and acute measures can be taken faster. The technique works so well that industrial actors became active themselves and report when they release a substance in high concentrations.45 Cooperation between state authorities and polluters is thus considered to be good: Wir behandeln alle [industriellen Einleiter, egal ob groß oder klein] gleich. Also ich arbeite sehr gerne mit der Chemie zusammen. Das ist nicht der ‘Feind’, überhaupt nicht. Die machen viel mehr Positives als Negatives. (…) wenn sie mal erwischt werden [beim Verschmutzen] , dann sind sie sehr kooperativ. Sie machen auch viel mehr als sie müssten. Das ist eigentlich eine gute Zusammenarbeit.46 (Interview N° 3)  Interviews N° 3, 5, and 8. The German name is “Vertrauenskreis.”  See Table 4.11, Sect. 4.2.2, and Table 19, Annex X. 45  Interviews N° 3 and 5. 46  English translation: “We treat all [industrial discharger, big and small] alike. I like working with the chemical industry. They are not ‘the enemy’, not at all. They do much more positive than negative things. (…) if they are caught [polluting], they are very cooperative. They also do much more than they have to. It’s actually a good cooperation.” An example of an additional, positive action is the voluntary remediation of old landfills by the company Novartis, which costs 200 million Swiss francs (Interview N° 3). 43 44

166

4  Empirical Analysis I: On Cooperation

Cooperation between actors from across the border is less intense between Swiss actors and their French counterparts and more intense with the German partners. Swiss interviewees said that it is difficult to discover the responsible authorities on the French side.47 The case’s French-German-Swiss peripheral faction (faction 4), in which the case’s six most peripheral actors are loosely connected, confirms this statement.48 The contact to German actors is more intense and regular, which is also due to the circle of trust. For instance, the Ministry for the Environment in Canton Basel City (AUEBS) receives technical information from its counterpart in the federal state Baden-Württemberg and from the German scientific actor TZWK.49 Cross-border contacts outside the management process of micro-pollutants in the Basel region exist because of an inter-regional project financed by the European Union.50 The regional state actors AUEBS and AUEBL work in a so-called interreg project that monitors micro-pollutants in groundwater in the Upper Rhine. The project lasted from 2016 until the end of 2018, when the actors AUEBS and AUEBL worked with state actors from France and the German federal states Hesse, Baden-­ Württemberg, and Rhineland-Palatinate (interreg Oberrhein 2018a).51

4.3.2  … in the Ruhr Case Study In North Rhine-Westphalia (NRW), two cooperative systems work on the improvement of the regional water bodies’ quality: the “Kooperation zwischen Wasserwirtschaft und Landwirtschaft,” a cooperation between drinking water providers and farmers,52 and the “Rohwasserdatenbank Pflanzenschutz”,53 which monitors water quality of wells and works out measures to improve their water quality. The Kooperation zwischen Wasserwirtschaft und Landwirtschaft was established between farmers and drinking water providers in 1992. Back then, drinking water providers in NRW did not charge a water withdrawal fee (Wasserentnahmeentgelt) that could have been used for financing measures to reduce pesticides. Due to the lack of such a financing, the Association of Water Works along the Ruhr (AWWR) and the chamber of agriculture Westfalen-Lippe (WLV) formed the Kooperation  Interviews N° 3 and 5.  The six actors are a French NGO, one German state actor and one German service provider, and a polluter, a service provider, and a regional state actor from Switzerland. See Tables 4.11 and 4.19 and Table 19, Annex X. 49  The German counterparts are the Landratsamt Lörrach (the district administration of the city of Lörrach) and the LUBW, the Landesanstalt für Umwelt, Messungen, und Naturschutz BadenWürttemberg (the State Institute for the Environment, Measurements, and Nature Conservation Baden-Württemberg), Interview N° 3. 50  Interviews N° 3 and 4. 51  Interviews N° 3 and 6. 52  English translation: “Cooperation between water management and agriculture.” 53  English translation: “Pest control raw water database.” 47 48

4.3  Actors’ Perceptions of Cooperation

167

zwischen Wasserwirtschaft und Landwirtschaft to develop measures that reduce the use of pesticides. Today, about 100 of these cooperations exist with 800 farmers participating. In 2004 the water withdrawal fee was introduced (five cents per cubic meter water), charged by drinking water operators and used for financing the measures and the fees for the advisors (AWWR and Ruhrverband 2015, blurb).54 The objective of the cooperation is to advise the agricultural actors on how to protect water bodies and soil from pesticides. The chamber of agriculture functions as mediator between the drinking water providers and the local farmers. Employees of the chamber advise the local actors on developing measures. The farmers’ acceptance of the measures is enhanced by the fact that the advisors come from the agricultural sector. Besides the continuous local engagement, twice a year the chamber of agriculture and the drinking water operators also meet with the farmer’s union and the district authorities.55 This cooperational tool is only partially visible in the case’s collaboration network: the initiators AWWR and WLV are not directly connected through collaboration ties—the chamber of agriculture NRW (CoA.NRW) functions as bridging actor between them. While the service providers entertain collaboration ties to almost every other service provider and the two water associations of the drinking water sector, collaboration with agricultural actors concentrates on the chamber of agriculture NRW only. This actor is in a gateway position between service providers and drinking water provider associations on the one hand and agricultural actors on the other hand. The cooperation tool applied throughout the entire federal state NRW is only partly reflected in Fig. 5 (Annex X) for the Ruhr region. Complementary to these local cooperations, the national tool Rohwasserdatenbank Pflanzenschutz was initiated in 2010 by the German association of municipal enterprises (VKU), the Federation of the Energy and Water Industry (BDEW), the German Technical and Scientific Association for Gas and Water (DVGW), and the water service provider Gelsenwasser AG. The tool’s task is to measure the water quality of wells and to find local solutions when a well does not fulfill this standard. Local “caretakers” investigate the causes for the water pollution and—in a dialogue with responsible actors, drinking water operators, and regional authorities—work out local measures to fix the problem. Most caretakers are members of the industrial association “Industrieverband Agrar”56 or come from municipal services or other local associations. For instance, the service provider Gelsenwasser AG was caretaker for the chemical enterprise BASF in Leverkusen and managed to have BASF discontinue the use of a certain substance.57 The cooperational tool Rohwasserdatenbank Pflanzenschutz cannot be retraced in the data since the key actors BDEW, DVGW, and BASF are not part of the case’s actor sample.

 The remaining money goes into a federal fund and into financing the implementation of the WFD (Interview N° 9). 55  Interview N° 9. 56  IVA, Industrieverband Agrar. 57  Interview N° 9. 54

168

4  Empirical Analysis I: On Cooperation

Apart from these two established cooperational tools for micro-pollutant management in NRW, a strong bilateral cooperation exists between the Association of Water Works along the Ruhr (AWWR) and the water management company Ruhrverband (RV). They are in close contact and constant technical exchange, publishing the annual “Ruhrgütebericht,” a report on the chemical and ecological state of the Ruhr.58 Data show that the two actors reciprocate their collaboration ties and that the AWWR receives and gives technical information from and to the RV while the RV receives and gives political information from and to the AWWR. The two actors also work on solutions to diminish the entry of radiocontrast agents.59 The Ruhrverband further participated in several other research projects on micro-­ pollutants. For instance, the RWW and the Ruhrverband tested activated charcoal filter techniques for drinking water treatment.60 One actor remarked: “Without the cooperation of the large water associations of the service sector and the ministry, nothing much would be happening.”61 By this, the actor specifically referred to the Ruhrverband. Cooperation between scientific actors is also strong, with the NRW Competence Center micro-pollutants entertaining contacts to all universities in NRW working on micro-pollutants. The center cooperates especially close with the University of Duisburg-Essen in the development of new process engineering.62 This strong collaboration among scientific actors is also shown in the data. Fig. 6 (Annex X) depicts the graph of the case’s scientific actors and their collaboration ties. The NRW Competence Center micro-pollutants (CompCent.NRW) sends collaboration ties to four of the case’s six scientific actors and receives collaboration ties from three of those. However, it does not entertain a collaborative connection to the University of Bochum. The other scientific actors are also well connected among each other, with the water research institute IWW receiving five and sending three collaboration ties, the Rhenish-Westphalian Technical University Aachen (RWTH Aach) sending and receiving four, the University of Duisburg (UNI.Duis) sending four and receiving three, Eawag sending three and receiving two, the University of Bochum (UNI. Boch) receiving two; and the Institute of Hygiene, Gelsenkirchen, sending a collaboration tie to one actor.

 English translation: Report on the Ruhr water’s quality.  Interviews N° 9 and 10. 60  Interview N° 10. The Ruhrverband also took part in the project “Sichere Ruhr,” an RiSKWa project that assessed the effects of micro-pollutants and possible measures to tackle them (see BMBF (2016, p. 7); BMBF et al. (n.d.)) and in the DSADS project investigating how to prevent pharmaceutical residues from entering the sewage system, see Lippeverband (n.d.). 61  Original wording: Ohne die Kooperation der großen Wasserverbände und dem Ministerium passiert nicht viel (Interview N° 12). 62  Interview N° 12. 58 59

4.3  Actors’ Perceptions of Cooperation

169

Similar to the circle of trust in the Basel region, a so-called Meldekette63 exists for the Ruhr and the Rhine in NRW. The State Office for Nature, the Environment and Consumer Protection NRW (LANUV) monitors both rivers and raises alarm when a substance is detected to exceed the limit value. In an alarm, a fixed communication channel among authorities, service providers, WWTP operators, and polluters is activated, starting upstream, going all the way downstream.64 Although collaborative relations between actors within the management process of micro-pollutants in the Ruhr region as well as outside the Ruhr region’s management process are established and provide ground for common work on the topic, specific lines of conflict regarding the measures to be taken remain. According to two interviewed actors and the Ruhrverband itself, the Ruhrverband does not favor the fourth treatment stage, as its costs are not yet covered.65 At the moment, expenses for new measures are split between the members of the Ruhrverband—something not all members like.66 The Ruhrverband favors preventive and source-directed measures. The NRW Competence Center micro-pollutants, a scientific actor researching on filtering techniques, by contrast favors the fourth treatment stage for wastewater treatment plants. In its view, drinking water providers have no need to take additional measures since concentrations do not exceed limit values for drinking water.67 As opposed to this, a drinking water provider stated that the water from the Ruhrverband still requires specific treatment to comply with the drinking water standards.68 This actor also stressed that state authorities should more decidedly call on polluters, such as agriculture and hospitals, to avoid the entry of substances in the first place. Even the Luxembourgian association of water services (Aluseau) from the Moselle case study, familiar with the situation in NRW, commented that the upgrade of wastewater treatment plants is favored in NRW. However, the success of substances’ reduction in upgraded WWTP varies greatly with the specific substance.69 The representative of the AWWR thought that the focus should be on the sources of micro-pollutants rather than on immediately upgrading WWTPs and drinking

 Literal translation: “report chain.” The main initiator was department IV-5 “Pivotal Issues of Water Management, Surface Water and Groundwater Quality, Water Supply” of the MKULNV (Interview N° 9). An emergency service is available 24/7 for such a contingency (Interview N° 11). 64  Interviews N° 9, 11 and 12. Since the actor LANUV.NRW, the actor that raises the alarm, did not answer the questionnaire entirely and thus is not part of the actor sample, the collaborational connections among the actors of the “Meldekette” cannot be retraced within the collaboration network. 65  Interviews N° 10, 12, and 13. 66  Telephone call N° 2 and Interview N° 13. 67  Interview N° 12. 68  Interview N° 13. 69  Interview N° 17. 63

170

4  Empirical Analysis I: On Cooperation

water plants.70 Another issue of conflict is financing for the water examinations that the drinking water operators are required to conduct. The water fee (Wasserpreis) is fixed by antitrust law and is too low to cover the costs for the examinations, which the drinking water operators thus have to pay themselves.71 These lines of conflict, as already presented in Sect. 4.1.1, are mirrored in the case study actors’ attitudes towards end-of-pipe measures (cf. Fig. 4.2). While state and scientific actors favor it, service providers and consumer organizations do not agree with this type of instrument. The cleavage is also visible in the collaboration network’s factions, which group into one faction made up of mainly scientific and regional state actors (faction 1) and another one consisting mainly of service providers and water associations (faction 2) (cf. Table  4.12, Sect. 4.2.2 and Table  20, Annex X).

4.3.3  … in the Moselle Case Study Actors regard cooperation in the management process of micro-pollutants in the Moselle region as rather poorly established. A scientific actor who did not answer the survey, but did answer several questions via e-mail, gave a clear evaluation of the case’s cross-border cooperation: Mein Eindruck ist, dass es eines großen ‘runden Tischs’ bedarf, um alle [potentiellen] Akteure zusammenzubringen und um Maßnahmen grenzüberschreitend zu koordinieren. Zumindest die fachliche Zusammenarbeit entspricht nach meiner Einschätzung bei weitem nicht dem, was möglich und wünschenswert wäre.72 (e-mail N° 1)

The collaboration network’s faction analysis reflects this cooperational divide into Luxembourgian and German actors: actors in two of the four factions are entirely Luxembourgian and German, respectively (see Table 4.13, Sect. 4.2.2). A European inter-regional project connecting German and Luxembourgian actors of the Moselle case study holds potential for a solution of this situation. The European Union promotes cooperation in Rhine water management by supporting interregional projects between EU and non-EU member states (interreg ABH 2018; interreg Oberrhein 2018b). The Ministry for the Environment, Energy, Food and Forest in RLP (MUEEF.RLP) applied for the support of two so-called interreg proj-

 For the improvement of drinking water operators’ filter techniques, the federal state NRW will invest 300 million Euros (Interview N° 9). 71  Interview N° 13. 72  English translation: “My impression is that there is a need for a ‘round table’ to bring all potential actors together and to coordinate cross-border measures. At least the specialist cooperation does not comply with what could be possible and would be desirable.” 70

4.3  Actors’ Perceptions of Cooperation

171

ects that focus on micro-pollutants.73 The project entitled Schutz der Wasserressourcen von Mosel und Saar vor Belastungen durch Pflanzenschutzmittel74 is based on an action plan for water protection measures at the Moselle and the Saar that was worked out by France, Luxembourg, Wallonia in Belgium, Saarland, and RhinelandPalatinate. The overarching project goal is to establish a transnational network of actors from the agriculture and water sectors aiming at the water protection of the Moselle and the Saar (DLR RLP 2016, p. 32). The application for the interreg project was submitted in 2017. The second project, entitled EmiSûre: Entwicklung von Strategien zur Reduzierung des Mikroschadstoffeintrags in Gewässer im deutsch-­ luxemburgischen Grenzgebiet, was granted the status of an interreg project and is financed for 7 years, starting in 2014.75 The Luxembourgian water authority, Agence de la Gestion de l’Eau, the University of Luxembourg, and the German federal ministries for the environment in RLP and Saarland, MUEFF.RLP and MUV.SAAR, as well as French actors work together in this project and conduct feasibility studies on water sewage treatment techniques reducing micro-pollutants in the cross-border region of the Sauer basin (L’essentiel 2017; MUEEF 2017).76 The only connection between actors that participate in this project and are at the same time part of the case’s actor sample exists between the University of Luxembourg and the MUEEF. RLP (cf. Fig. 4.6, Sect. 4.2.1). In yet another Luxembourgian research project, the scientific actors LIST and the University of Luxembourg worked together assessing the elimination of micro-pollutants in sewage in Luxembourg.77 A collaborational tie between these two actors exists, although it is not certain that it was established because of this very project. Luxembourgian authorities have a specific deficit that contributes to the fact that there are fewer connections across the border than within each territory. Although the country is willing to cooperate in cross-border water management and has done so in the past because of its position as up- and downstream riparian,78 Luxembourg lacks sufficient resources to contribute to cooperation with cross-border partners at all times.79

 Interview N° 19.  English translation: “Protection of water resources from the Moselle and Sauer rivers against pesticides.” 75  English translation: “Development of strategies for micropollutants in German-Luxembourgian waterbodies,” see University of Luxembourg (2017). 76  Interview N° 19 and Telephone call N° 1. 77  Interview N° 19. The project “EmiPoll—Emission profiles of wastewater treatment plants and evaluation of their removal efficiencies in Luxembourg” lasted from 2015 to 2017 and was financed by the Administration de la Gestion de l’Eau, Luxembourg; see University of Luxembourg (2017). 78  The three rivers Alzette, Sauer, and Moselle pass Luxembourgian territory, making Luxembourg to an up- as well as downstream actor among its riparian neighbors. 79  Telepone call N° 1. 73 74

172

4  Empirical Analysis I: On Cooperation

In terms of cooperation between state authorities and polluters on the German side, interviewees rated the cooperation with the agricultural sector as good and stated that point source dischargers, e.g., the chemical enterprise BASF, are directly contacted by the respective authority and held responsible for polluting water resources.80 This statement is visible in faction 2, which is, i.a., made up of the Chamber of Agriculture of RLP and four regional political actors (cf. Table  21, Annex X). To sum up these observations, cooperation is evaluated differently in the three regions. In the management process in the Basel region, actors perceive the ­cooperation between different state levels and between state actors and polluters as positive, whereas they see certain deficits in the cooperation of local service providers and between local service providers and state authorities. A closer look at the actors’ collaboration network affirms these statements. Collaboration between local service providers is rather scattered and connections between them run via the drinking water operator WWB.  Collaboration between state authorities and local service providers are also mainly channeled through single actors that hold a position between service providers and regional state authorities, such as AUEBL and AIBBL. Interactions between cantonal and national state authorities are well established, reflected in the many contacts of the Federal Office for the Environment (FOEN_W) to actors on the cantonal level. In the Ruhr region and beyond, official cooperations between the water, the agricultural, and the industrial sector exist, which aim at the reduction of chemical substances in water bodies. These cooperational structures could only partially be traced in the data, mainly because the actors participating in these cooperatives are beyond the case’s actor sample. The well-established cooperation between scientific actors that was mentioned by one interviewee is also reflected in the case’s collaboration network. The case study actors’ different opinions on measures to be taken were mentioned by the interviewed experts and are mirrored in the data that presents actors’ attitudes towards end-of-pipe measures and reveals the grouping of actors in factions along these attitudes. Actors in the management process in the Moselle region see flaws in cross-­ border cooperation, which is visible in the data through actors’ assignment into factions along their nationalities. The well-functioning cooperation between state authorities and agricultural polluters on the German side is also reflected in the faction analysis. The connections between the case study’s actors that work in EU interreg projects are not evident in the case’s collaboration network. Having outlined the elements of actors’ cooperation in each case, having focused on the collaboration networks from an all-encompassing descriptive network perspective, and having scrutinized the actors’ look at cooperation, the next chapter summarizes these insights and compares them across the cases.

80

 Telephone call N° 3 and Interview N° 19.

4.4  Qualitative Comparison I: Cooperation at Different Stages

173

4.4  Q  ualitative Comparison I: Cooperation at Different Stages I regard the proxy for cooperation, actors’ collaboration, in conjunction with the other constituting elements of cooperation: actors’ shared goal, actors’ information exchange, and actors’ reciprocity in collaboration. By this, I evaluate the intensity of cooperation in each case. The intensity of cooperation varies between the case studies. In the Ruhr case study, there is a line of conflict between actors on the measures to be taken in order to manage the CPR problem micro-pollutants. This is an instance of nonagreement on the same goal. Actors in this case do not fully agree on the goals the micro-­ pollutant management process should have. Their disagreement is also visible in their factions, as will be discussed hereafter. In the management processes of the Basel and the Moselle regions, actors largely agree on the goals of micro-pollutant management in the same way. The Welch’s two-sample t-test further showed for all three case studies that two actors who collaborate have a higher share of agreement on the management process’ goals than two actors who do not collaborate. This relation, however, is not significant in the Ruhr case study, where actors do not fully agree on the management process’ goals. The constituting element (a) aiming at the same goal is thus not entirely given in the Ruhr case study and cooperation less intense than in the other two cases. The other constituting elements of cooperation are considerably similar across the cases. The coordination of actions among actors, i.e., actors’ collaboration, is similarly intense in all case studies. The cases’ collaboration networks are densely linked, their densities amounting to 20.7% (Basel), 23.2% (Moselle), and 24.6% (Ruhr case study). On average, actors entertain ties to between six (Ruhr), seven (Moselle), and seven and a half (Basel) actors in these networks. The pairs of actors that are connected—be it by one tie or via paths—come to a rate of 92% and 94% in the Basel and Moselle cases and to 85% in the Ruhr case; and in all three management processes, about 55% of the actors can reach each other by only one step. Moreover, all three collaboration networks possess one strong component consisting of between 88.5% (Ruhr) and 94% (Moselle) of the case studies’ actors81 and a few single components. The actors comprising these strong components are all connected to each other through directed ties forming interlinking paths. In all three case study regions, a very high proportion of actors thus is entirely connected through directed collaboration ties. This pattern of tight collaborative connection between the case study actors is reflected in the networks’ connectedness that reaches 92% in the Basel, 85% in the Ruhr, and 94% in the Moselle case study. The intense collaboration is also mirrored in the collaboration networks’ core-­ periphery structure. Between 31% (Ruhr) and 39% (Moselle) of the actors form the networks’ cores, where they are densely connected with one another—the collaboration networks’ cores have densities between 50% (Basel) and 84% (Ruhr). The core 81

 The share of actors belonging to the large strong component in the Basel case study is 92%.

174

4  Empirical Analysis I: On Cooperation

actors are highly engaged in the collaboration networks’ activities, i.e., they all have a high share of incoming and outgoing relations with the network actors. Across all three case studies, actors who are in the core have also been judged important for the respective management process. The networks’ peripheral actors are less important, and among the least important ones are single strong components—single actors who do not connect to anyone of the network through directed ties. The meso-level analysis on the actors’ collaboration networks revealed that across the cases, actors group differently in cohesive subgroups. The cases’ faction analysis showed that in the management process in the Basel region, actors partner up along their fields of expertise. The scientific actors predominantly collaborate with service providers. Regional state actors tightly connect with regional wastewater treatment operators and actors from the civil society form their own faction. In the Ruhr region, factions in the management process build along lines of conflict. Service providers who are against end-of-pipe measures make up their own faction. Regional state and scientific actors who are in favor of these measures group in another faction. Actors in the management process in the Moselle region join according to their territorialities, building one dense, entirely Luxembourgian faction; another dense, exclusively German faction; and a third dense faction with actors from both countries. Furthermore, reciprocity of actors’ collaboration ties is equally strong across the cases: it is 29% in the Basel, 29.4% in the Ruhr, and 29.6% in the Moselle case study. This means that in all three cases, about 30% of the ties that actors send out to other actors are reciprocated. The pattern of actors’ exchange of information is also similar across the cases. The exchange of technical and political information is similarly intense among actors in all three management processes, while its intensity differs regarding the type of information. Actors entertain about twice as many connections for technical information exchange than they do for political information: on average, actors receive political information from about three actors in all three case studies and technical information from between six (Ruhr & Moselle) and seven actors; they send political information to two (Ruhr) or three actors (Basel & Moselle) and technical information to about five actors on average. This finding suggests that the environmental problem “micro-pollutants in surface water” requires a higher transfer of technical than of political information. This transfer is highest in the Ruhr region where actors’ technical information exchange network has a density of 22% and lowest in the Basel region, where this network’s share of existing ties in proportion to the potentially possible ones is at 15%. The density of the Moselle case’s technical information flow lies in between, at 17%. In all three cases, about half the actors that are among the most central in sending political information to their peers are also among the most central actors to send technical information to the actors of the respective micro-pollutant management process. Moreover, actors’ information exchange patterns are highly correlated with their collaboration patterns in all three cases; actors who exchange information—no matter which of the two types—are more likely to also collaborate with one another. The examination of actors’ viewpoints on cooperation revealed that from within, actors have a different perspective on cooperation than the perspective revealed by the research’s analysis. Actors in the Basel region have an overall positive impres-

4.4  Qualitative Comparison I: Cooperation at Different Stages

175

sion about cooperation in their management process. However, they see potential for more cooperation between local service providers and between state authorities and local service providers. Although actors in the Ruhr region have different opinions on how to manage the CPR problem, cooperation between certain actor groups works well, as, for instance, between scientific actors, between polluters and state authorities within the context of the alarming system, and between the water and the agricultural sector. Cooperation between agriculture and the state is also evaluated as positive in the German region of the Moselle case. Actors of this case see a clear deficit in cross-border cooperation between German and Luxembourgian actors. The analysis of cooperation within the management process of micro-pollutants in three different regions of the Rhine catchment area produced the following findings. Cooperation, consisting of the aiming towards a joint goal, resource exchange, coordinated action taking, and reciprocity of these coordinated actions exists in all three case study regions: 1. Aim towards a joint goal. 1.1) Actors in the Basel and the Moselle case studies are on the same page regarding the goal the management process should head to. 1.2) Actors in the Ruhr region differ in their convictions about the joint goal and group together along these different opinions when collaborating. The trench runs along the question of whether the management process should focus on the elimination of micro-pollutants at the source or end-of-pipe. Actors from civil society favor the former, state authorities, and scientific stakeholders, the latter. 1.3) Actors’ collaboration correlates significantly with actors’ joint goal in the Basel and Moselle case studies. 2. Information exchange - the type of resource exchange. 2.1) Actors engage more in the exchange of technical than of political information in all three cases. 2.2) Actors’ collaboration positively correlates with actors’ information exchange in all three cases. 3. Coordinating action taking: collaboration. 3 .1) Collaboration is highly developed among actors in all three cases. 3.2) Actors form subgroups when collaborating. These differ across the cases regarding actors’ goals, their competencies, and their territorialities. 4. Reciprocation of coordinated actions lies at roughly 29% in all three case studies. Given these results, actors’ cooperation in micro-pollutant management processes is of equal intensity in two regions of the international Rhine basin, the Rhine basin at Basel, and the Moselle basin on Luxembourgian and German territory. Actors’ cooperation in the micro-pollutant management process in the Ruhr region is less intense since actors approach different goals. However, the other two elements of cooperation and the criterion of reciprocity are equally evolved in this case study as they are in the Basel and Moselle case study.

176

4  Empirical Analysis I: On Cooperation

The intensity of cooperation is one characteristic, its lifespan another. And the lifespans of cooperation are not the same in all cases, as the discussion about the state of the management processes shows. Laws regarding the management of micro-pollutants are in place in all case study regions, with the legislation in the Moselle and Ruhr case studies being orientated towards the EU Water Framework Directive. The elaboration and implementation of measures is, however, at different stages in the three case studies. In the Basel case, rules for tackling micro-pollutants are precise and well elaborated. The measures are being implemented. In the Ruhr case, measures have been agreed upon in laws and management plans and are also being realized. In the Moselle case, the range of implemented instruments is much smaller, the instruments comprising preparatory tools that serve the identification of the extent of the problem and potential prospective measures for micro-pollutant management: monitoring of water quality, feasibility studies, and research on filter techniques. State authorities in the German federal states and Luxembourg are waiting for the EU and in the case of RLP also for the FRG, to propose measures. The implementation process of policies in the Moselle region is thus in an early phase with a few measures aiming at gathering information on the topic, but no managerial instruments decided upon yet. The environmental problem is not yet being managed. As a consequence, cooperation in the management process in the Moselle region is also at an early stage. Actors in this case study have been working on the issue of micro-­ pollutants for a shorter amount of time than their counterparts in the Ruhr and Basel regions; their cooperation is “younger.” In the Basel region, public and private actors have been working together in the implementation process of measures for more than a decade. The measures are strongly legally binding for the addressees. Both actor groups ensure the provision of surface water in a good condition. In the Ruhr case study, private and public actors develop measures for micro-­ pollutant management in the two cooperative systems between the water sector and the agricultural and industrial sector. Instruments are implemented and surface water is provided in cooperation between public and private actors; also, instruments are legally binding and required by law. The analyses in this chapter have shown that actor cooperation in the management process of a CPR problem in the ecological system of the River Rhine exists, although in differing intensities in the three selected case study regions. Moreover, the assessed CPR management processes in the three different Rhine sub-­catchments are in different stages of development. As a consequence, actors’ cooperation in the management processes has different lifespans too. In the Basel case study, actors’ cooperation has existed for more than a decade, being in a mature stage. In the Ruhr case study, actors’ cooperation took off 10 years prior to the moment of data collection, in 2006. Actors’ cooperation in the Moselle case study is, just like the regional management process, at a much earlier stage. Actors’ interactions in the micro-pollutant management process in the Moselle region can be considered a young cooperation, while actors’ cooperation in the management process of the Ruhr case study can be regarded as established and the one in the Basel region considered mature.

References

177

Table 4.22  Cooperation across the case studies

Basel Ruhr Moselle

Actors aim for same goal Yes Lines of conflict Yes

Collaboration Intense Intense

Given information average degree Political Technical 2.8 5.2 1.9 5.4

Reciprocity 29% 29.4%

Intense

2.9

29.6%

5.2

Cooperation Mature Established and less intense Beginning

Table 4.22 summarizes the analyses of the constituting elements of cooperation and the subsequent evaluation of cooperation intensity in each case. The next chapter tests the factors that stimulate these actor cooperations in a typical CPR problem situation.

References Sources e-mail N° 1. Department Raum- und Umweltwissenschaften, field of Analytische und Ökologische Chemie, University of Trier, 10 November 2016 Telephone call N° 2. Department “instrumental analytical chemistry”, Faculty of Chemistry, University Duisburg-Essen, 2 December 2016 Telephone call N° 3. Division “Innovation, Umwelt & Energie”, Industrie- und Handelskammer Rheinland-Pfalz, 12 December 2016

Primary Literature ARW (2018) Über ARW.  ARW  - Teil eines Netzwerks [online]. Arbeitsgemeinschaft Rhein-­ Wasserwerke e.V. (ARW). Available from: http://www.arww.org/ARW4/. Accessed 27 Sept 2019 AWWR, Ruhrverband (2015) Ruhrgütebericht 2015. Arbeitsgemeinschaft der Wasserwerke an der Ruhr (AWWR); Ruhrverband, Essen AWWR, Ruhrverband (2016) Ruhrgütebericht 2016. Arbeitsgemeinschaft der Wasserwerke an der Ruhr (AWWR); Ruhrverband, Essen BAFU (2016) Strategie des BAFU 2030. Bundesamt für Umwelt, Bern. 1 June BMBF (2016) Risikomanagement von neuen Schadstoffen und Krankheitserregern im Wasserkreislauf (RiSKWa) [online]. Available from: http://www.bmbf.riskwa.de/_media/ RISKWA_Praxishandbuch.pdf. Accessed 25 Sept 2019 BMBF, RiSKWa, FONA (n.d.) Sichere Ruhr. Das Projekt Sichere Ruhr  – Was machen wir? [online]. Available from: https://sichere-ruhr.de/. Accessed 25 Sept 2019 Der Schweizerische Bundesrat (1998) Gewässerschutzverordnung. GSchV DLR RLP (2016) Tätigkeitsbericht 2014–2016. Wasserschutzberatung der Dienstleistungszentren Ländlicher Raum in Rheinland-Pfalz. Dienstleistungszentren Ländlicher Raum (DLR) Rheinland-Pfalz

178

4  Empirical Analysis I: On Cooperation

interreg ABH (2018) interreg Alpenrhein, Bodensee, Hochrhein. Wir fördern Europa [online]. Available from: http://www.interreg.org/. Accessed 25 Sept 2019 interreg Oberrhein (2018a) ERMES-Rhein: Monitoring des Eintrags von Spurenstoffen in das Grundwasser. Programm INTERREG V Oberrhein (2014–2020) [online]. interreg Oberrhein. Available from: http://www.interreg-oberrhein.eu/projet/ermes-rhein-entwicklung-derressource-monitoring-des-eintrags-von-spurenstoffen-in-das-grundwasser/?cat=288-279. Accessed 25 Sept 2019 interreg Oberrhein (2018b) interreg Oberrhein/Rhin Supérieur [online]. interreg Oberrhein. Available from: http://www.interreg-oberrhein.eu/page-daccueil. Accessed 25 Sept 2019 L’essentiel (2017) 1,6 million d’euros pour la recherche à l’Uni. L’essentiel, March 9, 2017. Accessed 26 Jan 2020. http://www.lessentiel.lu/fr/luxembourg/story/12911860 Lippeverband (n.d.) Den Spurenstoffen auf der Spur (DSADS). Ein Projekt des Landes Nordrhein-­ Westfalen, der Stadt Dülmen und des LIPPEVERBANDS [online]. Available from: http:// www.dsads.de/worum-geht-es/. Accessed 25 Sept 2019 MUEEF Ministerium für Umwelt, Energie, Ernährung und Forsten, Rheinland-Pfalz, Gewässerschutz/Wasserwirtschaft (2017) Start für grenzüberschreitendes Kooperationsprojekt EmiSûre: Gemeinsam Mikroschadstoffe in Gewässern reduzieren. News release. January 30, 2017. Accessed 26 Jan 2020. https://mueef.rlp.de/de/pressemeldungen/detail/news/News/ detail/start-fuer-grenzueberschreitendes-kooperationsprojekt-emisure-gemeinsam-mikroschadstoffe-in-gewaessern/?no_cache=1 UBA (2014) Forschungsprogramm des Umweltbundesamtes 2015–2017. Umweltbundesamt, Dessau-Roβlau, October UBA (2014, 2015) The UBA. About us [online]. Umweltbundesamt. Available from: https://www. umweltbundesamt.de/en/the-uba/about-us. Accessed 23 Sept 2019 University of Luxembourg. (2017) Projekte mit externer Förderung. Urban Water Management [online]. University of Luxembourg. Available from: https://wwwfr.uni.lu/content/download/105284/1252825/file/Aktuelle%20Projekte.pdf. Accessed 25 Sept 2019

Secondary Literature Analytictech (n.d.) UCINET 6 for windows help contents. Network Centrality Fragmentation [online] Available from: http://www.analytictech.com/ucinet/help/hs4209.htm. Accessed 27 Sept 2019 Borgatti SP, Everett MG, Freeman LC (2002) Ucinet for windows: software for social network analysis. Analytic Technologies, Harvard, MA Borgatti SP, Everett MG, Johnson JC (2013) Analyzing social networks. SAGE, London Braun C, Gälli R, Leu C, Munz N, Schindler Wildhaber Y, Strahm I, Wittmer I (2015) Mikroverunreinigungen in Fliessgewässern aus diffusen Einträgen. Situationsanalyse. Bundesamt für Umwelt, Bern Fischer M, Leifeld P (2015) Policy forums: why do they exist and what are they used for? Policy Sci 48(3):363–382 Hanneman RA, Riddle M (2005) Introduction to social network methods. University of California, Riverside, Riverside Knoke D (1998) Who steals my purse steals trash. The structure of organizational influence reputation. J Theor Polit 10(4):507–530 Laumann EO, Pappi FU (1976) Networks of collective action. A perspective on community influence systems. Academic Press, Inc., New York Plackett RL (1983) Karl Pearson and the chi-squared test. Int Stat Rev 51(1):59–72 Ruxton GD (2006) The unequal variance t-test is an underused alternative to student’s t-test and the Mann-Whitney U test. Behav Ecol 17(4):688–690 Wasserman S, Faust K (1994) Social network analysis: methods and applications. Cambridge University Press, Cambridge

Chapter 5

Empirical Analysis II: On the Emergence of Cooperation

Abstract  In this chapter, I follow Mill’s method of difference, which states that the dependent variables in the studied cases show a difference, different stages of cooperation, and are explained by the differing explaining variable across the otherwise similar cases—the reason why the approach is also called “the most similar research design” (Bennet, Case study methods: design, use, and comparative advantages. In: Sprinz DF, Wolinsky-Nahmias Y (eds) Models, numbers, and cases: methods for studying international relations. University of Michigan Press, Michigan, p.  31, 2004). The chapter starts with a thorough description of the explaining factors’ power to start off cooperation in a CPR problem setting within the context of each case study: a strong perception of the environmental problem’s severeness, actors’ participation in forums, and actors’ shared beliefs. Through the case comparison, I draw the conclusion that actors’ high problem perception contributes to the formation of cooperation, while their participation in forums fosters their cooperation in the management of a surface water pollution in the Rhine basin. The first subchapter presents the ERGM results for each case. The following subchapters discuss the explaining factors in light of the case studies’ contexts to illuminate the explaining factors’ influence on cooperation. Keywords  Exponential Random Graph Model (ERGM) · Problem perception · Ecology of games framework (EGF) · Advocacy Coalition Framework (ACF) Policy beliefs The explaining factors for cooperation in the CPR management processes in the Rhine catchment area have to be viewed in respect of the intensity and stage of cooperation within each management process. Only then, the influence of the explaining factors on the explained phenomenon cooperation becomes meaningful. Firstly, the intensity of cooperation differs across the cases. Actors of the management processes in the Basel region and in the Moselle basin agree on different objectives regarding the management in similar ways. They aim at the same goal. Actors in the Ruhr region disagree on how to tackle the environmental problem at stake. The criterion of cooperation aiming at a joint goal is not fully given and actors’ cooperation in this case thus less intense.

© Springer Nature Switzerland AG 2020 L. M. J. Herzog, Micro-Pollutant Regulation in the River Rhine, https://doi.org/10.1007/978-3-030-36770-1_5

179

180

5  Empirical Analysis II: On the Emergence of Cooperation

Second, actors’ cooperation in the three cases is at different “development” stages. In the Basel case study, actors have been working on the management of micro-pollutants for more than a decade; cooperation between them has evolved throughout this time and is now at a mature stage. The CPR management process in the Moselle region is much younger: instruments to tackle micro-pollutants still need to be elaborated. Cooperation between actors in this case, even if similarly intense as cooperation in the Basel case study, is in its beginning. In the Ruhr case, actors started doing something about micro-pollutants in surface water 10  years prior to this study. Their cooperation, though less intense than in the other two cases, is well established. The dependent variable cooperation thus shows a variance across the cases. The explaining factors do so as well. In this and the next chapter, I follow Mill’s method of difference, which states that the dependent variables in the studied cases show a difference—different stages of cooperation—and are explained by the differing explaining variable across the otherwise similar cases, the reason why the approach is also called “the most similar research design” (Bennet 2004, p. 31). This chapter starts with a thorough description of the explaining factors’ power to start off cooperation in a CPR problem setting within the context of each case study. The next chapter discusses the factors that consolidate cooperation. Through the case comparison—see comparison II in the case study method design (Fig. 5.1)—I can draw general conclusions about why actors cooperate in the management of a CPR problem of over-appropriation in the Rhine basin. The first sub-chapter presents the ERGM results for each case. The following sub-chapters discuss the explaining factors in light of the case studies’ contexts to illuminate the explaining variables’ influence on the dependent variable cooperation.

5.1  The Exponential Random Graph Model The model discerns the relation between an actor’s attribute and the actor’s tie with another actor in the collaboration network of each case study (Model 1). The model’s alpha term was adjusted for each case after checking BIC scores and GOF (Harris 2014, p. 70). Table 5.1 shows the ERGM results of each case study. The ERGM ran on actors’ collaboration networks which serve as a proxy for actors’ cooperation since collaboration is conceptually close to cooperation (cf. Sect. 1.2). I further ran robustness checks of the model on the information exchange networks which showed similar results.1 The robustness checks and the model’s goodness-­of-fit are described in Annex XII. A good goodness of fit is crucial for the

1  For the robustness check models, see Tables 23–26, Annex XI; for their discussion see Annex XII.1.

5.1  The Exponential Random Graph Model

181

Context: CPR problem of micro-pollutants in surface water in the Rhine catchment area

SINGLE-UNIT ANALYSIS —IN EACH CASE STUDY Case study 1: Mgmt. process Case study 2: Mgmt. process Case study 3: Mgmt. process in the Ruhr basin in the Rhine basin at Basel in the Moselle basin Units of analysis:

Units of analysis:

Units of analysis:

UoA 1 Network level— descriptive SNA: cooperation’s constituting elements

UoA 1 Network level— descriptive SNA: cooperation’s constituting elements

UoA 1 Network level— descriptive SNA: cooperation’s constituting elements

UoA 2 Dyadic level— ERGM: collaboration tie between two actors

UoA 2 Dyadic level— ERGM: collaboration tie between two actors

UoA 2 Dyadic level— ERGM: collaboration tie between two actors

ACROSS-UNIT ANALYSIS —CASE COMPARISON Comparison I: Cooperation across the case studies Comparison II: Factors enhancing cooperation across the case studies

Fig. 5.1  Illustration of the book’s case study method design; analytical part 2

model’s results to be valid and interpretable (Cranmer et al. 2017). I further summarize the findings of the different robustness checks I ran to ensure the validity of the models (Annex XII). Both tests indicate that the model is valid. The ERGM results show that different factors relate to collaboration in the different case studies. Actors’ high problem perception (IV 1a) relates to collaboration in the Moselle case study; actors’ participation in forums (IV 2a) does so in the Ruhr case study; actors’ co-participation in forums (IV 2b) correlates with actors’ collaboration in the Basel case study, whereas actors’ similar problem perception (IV 1b) and actors’ similar beliefs (IV 3) do not relate to actors’ collaboration. In the following, I discuss the hypotheses based on the ERGM results and interpret the findings in light of the case studies’ context and the qualitative analysis thereof.

182

5  Empirical Analysis II: On the Emergence of Cooperation

Table 5.1  ERGM results Model 1 IV 1a High problem perception Activity Popularity Heterophily IV 1b Similar problem perception IV 2a Forum participation Activity Heterophily IV 2b Co-participation in forums IV 3 Similar belief CV 1 Regulatory actor Activity CV 2 Implementer Popularity Homophily CV 3 Reputation Activity Popularity CV 4 Pollution-sensitive water use Homophily CV 5 Territoriality Swiss/German Activity Homophily CVs 6 Network effects CV 6.1 Edges

Basel

Ruhr

Moselle

−0.47 ∗ (0.20) −0.05 (0.21) −0.48 . (0.28) −0.24 (1.34)

0.13 (0.26) −0.23 (0.31) 0.00 (0.40) 2.10 (1.87)

1.10 ∗∗∗ (0.28) −0.70 ∗∗ (0.26)

−0.04 (0.04) 0.06 . (0.03) 0.78 ∗∗∗ (0.13) 0.09 . (0.05)

0.24 ∗∗ (0.08) −0.08 (0.08) −0.31 (0.29) −0.02 (0.10)

0.12 . (0.07) −0.02 (0.07) 0.08 (0.17) 0.02 (0.06)

0.29 (0.36)

−0.97 . (0.57)

0.03 (0.37)

−0.32 . (0.18) 0.40 ∗∗∗ (0.12)

−0.42 (0.27) 0.67 ∗∗∗ (0.16)

0.08 (0.22) −0.11 (0.19)

−0.23 (0.19) 0.94 ∗∗∗ (0.20)

−0.37 (0.26) 1.47 ∗∗∗ (0.32)

0.85 ∗∗ (0.30) 0.50 . (0.28)

0.11 (0.16)

0.65 ∗∗ (0.21)

0.31 . (0.16)

−0.51 ∗ (0.21) 0.71 ∗∗∗ (0.16) −4.20 ∗∗ (1.47)

−0.10 (1.77)

0.08 (0.26) 1.09 ∗∗∗ (0.21) −7.98 ∗∗∗ (2.10)

−7.09 ∗∗∗ (1.84) (continued)

5.2  Problem Perception and Cooperation

183

Table 5.1 (continued) Model 1 CV 6.2 Reciprocity CV 6.3 Gwidegree CV 6.4 Gwodegree CV 6.5 Gwesp fixed 0.8

Basel 1.09 ∗∗∗ (0.28) 2.40 ∗∗ (0.82) 1.05 (0.74) 1.22 ∗∗∗ (0.18)

Ruhr 1.12 ∗∗ (0.40) 2.96 ∗∗ (1.05) 0.97 (0.91) 1.46 ∗∗∗ (0.29)

CV 6.6 Gwesp fixed 0.5 BIC Number of nodes Number of edges

1093.30 37 276

629.85 26 160

Moselle 1.54 ∗∗∗ (0.32) 0.10 (0.97) 2.07 . (1.06)

1.44∗∗∗ (0.31) 862.28 31 216

∗∗∗ p < 0.001, ∗∗ p < 0.01, ∗ p < 0.05; coefficients are reported as log-odds; for all results, see Table 22 and Figures 7–9 Annex XI

5.2  Problem Perception and Cooperation One factor that has been stressed by several studies on cooperative action is the recognition of a resource problem (Giest and Howlett 2014; Heikkila and Gerlak 2005; Lubell et al. 2002). A serious threat to an environmental resource might serve as motivator for collaboration, since actors want to avert the high costs arising from the environmental problem by acting united against it (Gerber et al. 2009, p. 807; Giest and Howlett 2014, p. 39; Lubell et al. 2002, p. 150). If actors depend on the resource that is being threatened, the incentive to cooperate might even increase (Ostrom 2000, p. 40; Schlager 2004, p. 152). The costs of not cooperating to act against the resource problem may then exceed the costs of cooperating and averting the negative impact from the environmental problem. To grasp a threat to an environmental resource conceptually—in this study the threat of micro-pollutants to river surface water—I consider actors’ perception of such a threat, i.e., of a CPR problem. Based on these theoretical and empirically assessed assumptions, the first hypothesis states that the higher actors’ problem perception is the more likely the actors are to engage in cooperation (Hypothesis 1a). I further consider actors’ incentive to cooperate based on their similarity in a characteristic (Leifeld and Malang July 2014, p. 8; McPherson et al. 2001, p. 416f.), in this case their similar problem perception. The second hypothesis regarding actors’ problem perception thus puts forward that actors with a similar problem perception are more likely to engage in cooperation with each other (Hypothesis 1b). The ERGM results for the proxy of cooperation, actors’ collaboration, partially confirm Hypothesis 1a in the Moselle case study, but disprove it for the Basel case study. The results across all case studies refute Hypothesis 1b.

184

5  Empirical Analysis II: On the Emergence of Cooperation

The model’s results for the Moselle case study proclaim that actors’ high problem perception (IV 1a) is positive and significantly (p