A Study and Social Network Analysis on the World Business Council for Sustainable Development Theresa Schaffner, Bucknell ‘15 Advisor: Peter Wilshusen Environmental Studies Department Table of Contents Introduction ................................................................................................................................................... 2 History and Importance of the WBCSD ....................................................................................................... 4 Social Network Analysis............................................................................................................................... 7 History ...................................................................................................................................................... 7 Terminology.............................................................................................................................................. 8 Social Network Analysis Software ......................................................................................................... 12 Network Behavior Theories .................................................................................................................... 12 Modes of Analysis .................................................................................................................................. 15 Methods ...................................................................................................................................................... 19 Defining the network .............................................................................................................................. 19 Preliminary networks .............................................................................................................................. 19 Making Ties ............................................................................................................................................ 22 Data Collection ....................................................................................................................................... 23 Future Analysis ........................................................................................................................................... 25 Conclusion and Future Plans ...................................................................................................................... 27 Appendices.................................................................................................................................................. 28 Appendix A: Member Organizations divided by Region ....................................................................... 28 Appendix B: Membership Diagrams ...................................................................................................... 33 Appendix C: Extended Terminology List ............................................................................................... 34 Appendix D: Data Sheet ......................................................................................................................... 35 1 Introduction In the past few decades, there has been an increase in transnational capitalism, businesses, and organizations. Many economists and policy theorists have specifically discussed the rise of a transnational capitalist class; a class that is changing the face and structure of the global economy. The transnational capitalist class (TCC) is commonly thought of as the social network of individuals who leverage tools of the world economy such as transnational organizations and political bodies. The TCC is interested in the world economy as a system of international private commerce to allow free movement of capital across countries (Carroll 2003). However, trade and investment are not sufficient structural conditions for long term sustainable development, nor is neoliberal capitalism a suitable long term vision for a Green Economy. A level of collaboration beyond businesses, region, and sector is necessary so that business leaders may discuss issues of “shared concern, to find common ground, and devise strategies for action” in shaping a global economy (Carroll 2003). The significance of collaborative work areas is made greater by globalization and the search for new types of global economic policies that can support an increasing population and decreasing natural resources. With transnational organizations (TNOs) gaining more influence over global economic, social, and environmental policy there is an increasing interest in the social networks formed between and amongst transnational organizations. This paper will focus specifically on a social network analysis of the World Business Council for Sustainability Development. The World Business Council for Sustainable Development (WBCSD), formed by the United Nations in 1995, is a transnational organization whose goal is to encourage the “global business community to create a sustainable future for business, society and the environment” (WBCSD 2104). The WBCSD as well as other international policy groups such as the World 2 Economic Forum and the Trilateral Commission have been integral to forming the transnational economy by synthesizing public and private elements of capitalism. However, the WEF is focused around a “highly elite core of transnational capitalists” and champions paradigmatic neoliberal structuralism (Carroll 2003). The TC favors regulatory practices to promote capitalism amongst northern Europe, North America, and Japan. The World Business Council for Sustainable Development is the only leading transnational organization committed to creating a sustainable global economy by promoting corporate social responsibility and environmentally sound changes across industries and regions. Conducting a social network analysis on the WBCSD aims to determine whether the structure of the organization is allowing collaboration amongst business leaders as intended. Social network analysis (SNA) can also reveal which actors in the organization are the most influential and how that affects what the organization accomplishes. This study will focus on a social network analysis of the WBCSD to determine which actors, based on centrality measures outlined in diffusion of innovation theory, have the greatest ability to influence the decisions of others and therefore affect the direction and success of WBCSD initiatives. 3 History and Importance of the WBCSD The World Business Council for Sustainable Development started its life as the Business Council for Sustainable Development in 1990. It consisted of forty-eight business leaders who banded together to represent the corporate world at the 1992 Rio Earth Summit. The first BCSD meeting took place in 1991 with Stephan Schmidheiny as president. By 1995 the BCSD merged with the World Industry Council for the Environment to create the WBCSD; organization headquarters were setup in Geneva and Björn Stigson was named the first president. Nearly twenty years later, the WBCSD is made up of 176 member organizations across five continents (WBCSD 2014). See Appendix A for member list and sector details. Companies that are a part of WBCSD are represented in the organization by their CEO or an appointed substitute. The organization is governed by an Executive Committee made up of elected member CEOs who serve on the Committee for two years. The Executive Committee meets three times a year to maintain daily management of the WBCSD, determine priorities of the organization, and discuss issues related to sustainable development (WBCSD 2014). The WBSCD meetings provide a medium for leading business minds to analyze and deliberate aspects of sustainable business management. WBSCD makes advancements in sustainable development by dividing its attention into focus areas separated by business sector. The varying challenges presented by each sector are tackled by projects, systems solutions, and capacity building initiatives. Sector projects are practical initiatives in independent research that investigate how critical industries can meet sustainability challenges. Systems solutions provided an approach that deals with the overlapping challenges presented by sustainable development. Capacity building activities support increased integration of sustainable development into business practices. The overarching goal is to find 4 comprehensive solutions to creating sustainable business practices that can be applied across entire industries. The WBCSD has already contributed to many successes in increasing sustainable practices among member businesses and respective sectors. Since its founding after the 1992 Rio Earth Summit, WBCSD has been promoting the importance business expertise in achieving environmental protection as well as economic growth. Several milestones in sustainable development have since been reached: in 1997 the Rio+5 Convention was held and the WBCSD published Signals of Change, a report describing the progress of business since the 1992 Earth Summit. In 1999 the United Nations established the Global Compact which “asks companies to embrace universal principles and to partner with the United Nations. The Global Compact has grown to become a critical platform for the UN to engage effectively with enlightened global business”(Ki-moon 1999). The most recent achievements include the greenhouse-gas-reducing Kyoto Protocol and the 2015 Millennium Development Goals. Finally, the collective efforts of these initiatives have been defined by the United Nations as working towards a Green Economy. The Green Economy is defined as an economy “that results in improved human well-being and social equity, while significantly reducing environmental risks and ecological scarcities. In its simplest expression, a green economy can be thought of as one which is low carbon, resource efficient and socially inclusive” (United Nations 2014). The WBCSD is continuing to make advancements toward a Green Economy with an increasing number of members and initiatives working around the world. However, an increasing number of companies involved in the council also leads to an increase in personal interests and agendas. As with any organization, there will be some members who are in greater positions of influence than other members. The people in positions of influence will naturally affect the type 5 of initiatives and programs being implemented and completed. Furthermore, the members elected to the Executive Committee may not necessarily have the most influence in the organization. Social network analysis provides an insight to how the structure of an organization, such as WBCSD, influences which members gain the ability to influence others. 6 Social Network Analysis Social network analysis is a means to analyze connections amongst objects. There are two basic concepts that validate social network analysis: homophily and influence. Homophily is the idea that people or things with like characteristics are usually connected. Influence is the idea that those connected people or objects have an effect on one another and the effect is greater with stronger connections (Kadushin 2012). Social network analysis (SNA) can be applied anywhere from small groups to entire global social systems. A classic example uses data pulled from Amazon marketing techniques. Valdis Krebs, a consultant who specializes in SNA, made a network consisting of books related to the 2008 presidential election. He connected books bought by the same person. The results were a connected group of democratic leaning books, a connected group of books pertaining to republican ideas, and a third group related entirely to then presidential candidate Barack Obama (Kadushin 2012). This network showed two things: 1) people who read politically focused books rarely or never read books that are in opposition to their own political leanings and 2) Amazon’s marketing strategy to convince people to buy multiple books on the same subject is highly successful . The point of this example is to show that social network analysis can reveal larger concepts from data that may seem mundane or simple. History Social network analysis has seen a recent resurgence in academic literature but has been developing over several decades. SNA was first promoted by sociologist Georg Simmel in his late 19th century publications about the importance of studying patterns of relationships. In 1954, sociologist J.A. Barnes started using the term “social network analysis” to exclusively and specifically relate to patterns of ties amongst people or objects. One of the most well-known 7 results from a SNA experiment is the concept of six degrees of separation. Stanley Millgram, yet another sociologist, conducted an experiment in the 1960s known as the small-world problem. He randomly selected people from small towns in Kansas and Nebraska to attempt to get a letter to a specific person in upper-class Boston. The participants were instructed to send the letter to someone they knew personally who would have a better chance of getting the letter to the intended person. The results showed that on average it took six people to get the letter from the Midwest to Boston. Millgram never used the phrase “six degrees of separation” – his experiment was simply to show that American people are more connected to each other than they realize- but his experiment promoted the concept nonetheless (Famiglietti 2014). Ten years later in 1971, the International Network for Social Network Analysis (INSNA) was founded as the professional association for researchers interested in social network analysis. SNA has gained enough interest amongst sociologists and other researchers to warrant a professional organization because technological developments have made it possible to use advanced software to analysis larger networks at a faster, more detailed pace (Famiglietti 2014). Terminology In order to fully understand the potential of social network analysis, we have to understand the terminology. A network is defined as a group of objects and a description of the relationships between those objects. The objects, which can range from people to books to anything in between, are called nodes. The relationships between the nodes are called ties or edges. There are four types of relationships between nodes that constitute a tie: 1. Similarities are attributes shared by nodes; for example, holding membership in the same group or committee 8 2. Social relationships are based on feelings or emotions towards another node; for example, feelings of friendship or trust 3. Interactions are ties based on behaviors; for example, voting for a particular candidate 4. Flows are ties that show a flow of information or objects through a network; for example, the way money flows through a retail organization The ties can be either directed or undirected. Directed ties show and imbalance relationship between the nodes, while undirected show a balanced relationship. For example, a directed tie would show that Person A sent a letter to Person B. The direction of the tie would be from person A to person B. An undirected tie would merely show that the two persons communicated with each other (Famiglietti 2014). Directed Tie A B Person A sent a letter to person B Undirected Tie B A Person A and Person B communicated with each other Image 1: directed vs. undirected ties Ties can be binary or weighted. Binary ties merely show the existence of a relationship. Weighted ties show the strength of a relationship based on a quantitative variable. Finally, the number of ties attached to the node is the degree. Nodes in directed networks have in-degrees 9 and out-degrees, which are the number of ties directed towards them or away from them respectively (Prell 2012). Binary Tie A B Person A and Person B are connected Weighted Tie B A C Person A and Person C are more strongly connected than Person A and Person B Image 2: binary vs. weighted ties Directed Degrees of Node B In-degree = 2 Out-degree = 5 Undirected Degree of Node A n=4 A B Image 3: undirected and directed degrees 10 The shape of a network created by the relationships amongst nodes can range from dyads to large encompassing networks. A dyad is two nodes and the ties between them: the simplest form of a network. A triad consists of three nodes with ties only to each other or no ties at all. Larger networks can consist of hundreds of thousands of nodes as long as they are in the range defined by the researcher (Famiglietti 2014). Simple Dyad Triad Variation Image 4: dyad and triad Networks with only one type of node are called single mode; however, it is possible to have multiple types of nodes within a network. A network with two types of nodes is called bimodal, and a network with three or more types of nodes is called k-modal (k=number of different node types). An example of a single mode network is the Amazon example used before: the only type of node in the network is books. An example of a bi-modal network is connecting students (the first type of node) to each other based on if they lived in the same dormitory, as well as connecting them to their respective major departments (the second type of mode). A k-modal network would entail adding another category, such as connecting students to their minor department. Having multiple types of nodes allows for a more advanced analysis but also a more 11 difficult design phase (Famiglietti 2014). Appendix C has a list of more advanced terminology used in social network analysis, but the ones discussed above are sufficient for this study. Social Network Analysis Software After defining the range and type of nodes as well as what ties them together, it is time to enter the data into a database. The database can be an Excel spreadsheet, matrices, or other tables. However, most SNA software programs are easiest to use with Excel spreadsheets. The database needs to be carefully designed before any data is entered. Depending on the analysis, there can be upwards of thousands of nodes and ties. The database must be simple and organized, with clear labels and consistent formatting, in order for the software to process the information. Popular software programs include Cytoscape, UCINET, and Gephi, the latter being used for this study. The software is able to perform a wide range of analyses using advanced algorithms that most people, myself included, would not be able use on their own. These algorithms have helped sociologists develop several theories in explaining network behavior. Network Behavior Theories The purpose of social network analysis is to reveal information about the relationships between nodes that would not be seen by just looking at the nodes as groups or at their intrinsic characteristics. Therefore, all modes of analysis related to SNA focus on the relationships between nodes. The analyses can find patterns, irregularities, strengths and weaknesses, as well as help create theories and “rules” governing networks. General consensus among SNA experts is that some rules apply to all networks (Famiglietti 2014). The first of these is the Power Law. In social network analysis, Power Law states that most nodes have few ties and few nodes have many ties. When put in a two-dimensional graph, the power law curve looks as follows: 12 Undirected Degree of Nodes in Network A F(x)=x^n, n=2 Degree Node Image 5: Power Law Curve The power law tells us that the frequency of nodes with a high degree is very low compared to the frequency of nodes with a low degree. Because of this, if a network were to lose a few nodes chances are that the lost nodes will only have a few ties and the connectivity of the network will not be greatly affected. However, losing a highly connected node can change the entire network structure. If the loss of nodes is a random occurrence, then there is a low probability of losing a node with a high degree and the network is strong. If the loss of nodes is not random, and high degree nodes leave the network more frequently, then the network is more susceptible to change. Depending on the real-world situation, those maintaining a network may wish to flatten the Power Curve (Famiglietti 2014). The second theory I will discuss is known as the small world theory. Small world theory discusses how nodes can be connected to many other nodes in a network despite their number of direct ties. There is a path between nodes via ties across the network. In a small world network, most nodes are not connected to each other, yet can easily reach other nodes in the network via a few hops along ties between other nodes. But what if a network is not a small world? 13 Researchers have defined a “regular” world and a “random” world as well. A regular world network has a high number of ties, like the small world, but the path between nodes is longer. Read: it takes more hops for one node to reach another, but there is a high probability all nodes connect eventually. In a random world network, there are a low number of connections, yet the nodes that are connected have a short path between them (Kadushin 2012). Regular world Small world Random world Increasing Randomness Image 6: (Watts and Strogatz 1998) One of the most important concepts to understand in SNA is the theory of preferential attachment. Preferential attachment is the idea that nodes “prefer” to connect to nodes that already have many connections; that is, nodes are more likely to form a tie with nodes that have a high degree than nodes that have a low degree. This concept supports the Power Law in that, rather than increasing ties on low-degree nodes, new nodes increase the number of ties on a high degree node. One example of this theory in actuality is the fact that highly cited academic papers are more likely to be cited again than a low cited paper (Famiglietti 2014). A more contemporary example is the growing divide between upper and lower economic class and the decrease of a middle class in the United States (Lynch 2011). The “rich” gain more connections and access to 14 capital, while the poor remain the same and become relatively “poorer”. Preferential attachment theory is a focal point in explaining the behavior of networks as they change over time. Modes of Analysis There is a plethora of analyses that can be used to investigate a network, but the analysis you choose is dependent on the type of network and what information you seek. The most common types of analyses are outlined below: Degree: as outlined before, degree is the number of ties connected to a node. Most software will run an average degree as well as the range. Weighted degree: weighted degree is the total weight of all ties connected to a node. Average weighted degree is the average weight per node in the network. Density: the number of ties in a network divided by the total number of possible ties. Neither extremely dense nor sparse networks are useful for analysis. Average path length: the number of ties, on average, it takes to connect each node in the network. A path is defined as a group of connecting nodes where no node is visited more than once. Network Diameter: the longest shortest path through the network, also known as the geodesic distance. This metric is important for optimization analyses. Clustering: how connected a node is to all other nodes in the network. If every node is connected, then there is 100% clustering for that network. The more ties a node has, the higher its chance of gaining a connection to a new node. Average clustering coefficient: The average clustering coefficient (Ci) for a node is the proportion of links between the nodes within the network divided by the number of links that could possibly exist between them (Kadushin 2012). 15 Centrality: the measure of analysis that identifies the most important nodes in the network. Sociologists have defined importance to mean many different things, so there are multiple centrality measures to portray the different meanings of the word. The most basic centrality measure – degree centrality- identifies the most influential nodes in a network by number of ties. Centrality can also measure importance as cohesiveness in the network (Scott 2011). The following three measures are types of centrality measures, as shown in the image below. Image 7: The image shows colors ranging from dark blue to red, with red being most central, illustrating the different types of centrality measures applied to the same network. (Rocchini 2012) A) degree centrality B) closeness centrality C) betweeness centrality D) Eigenvector centrality 16 1) Closeness centrality: measures how long it will take one node to reach all other nodes. In this study, closeness centrality will measure how long it will take information to spread from one WBCSD member to all other members. The “closer” a node is, the faster it is capable of spreading information (Kadushin 2012). Image 8: closeness centrality (Activate Networks, Inc. 2014) 2) Betweeness centrality: measures the “frequency with which a node falls between pairs of other nodes on their shortest paths.” Points central in this respect have potential for control because they control flow of information to many other nodes. In social theory, this would allow nodes with a high betweeness centrality to control what information does and does not pass through the network and manipulate the information to their advantage (Kadushin 2012). 17 Image 9: betweeness centrality (Activate Networks, Inc. 2014) 3) Eigenvector Centrality: shows which nodes are well-connected to the most influential nodes in the network. For example, a person may not have many connections to the political world but he is long-time friends with the vice-president (Famiglietti 2014). Conversely, a CEO may not directly work with many people in her company, but she is in direct contact with the head of each department. When a node has connections to other nodes with high degrees of other centrality measures, it becomes an eigenvector node (Kadushin 2012). Image 10: Eigenvector centrality (Activate Networks, Inc. 2014) 18 Methods The World Business Council for Sustainable Development is an organization made up of global business leaders with excellent communication skills, persuasive abilities, and charismatic personalities that have enabled them to reach the position of influence they currently hold. It only makes sense to use social network analysis to determine which people in this network of corporate magnates are the most influential within WBCSD. This social network analysis will give a sense of how WBCSD, as a policy group, integrates businesses across sector and region allowing the opportunity for actors to be more influential than others based on their positions within the group. The analysis will use the four previously mentioned centrality measures to identify the central actors in the network. Defining the network Let us first review the structure of WBCSD. Every member company is represented in the Council by its CEO or an elected replacement. The company representatives are the set of nodes in the network. Each company is represented once by one person. The nodes will be labeled by company name because the company name has more meaning to a person looking at the network. Preliminary networks Before creating a network of all the nodes and their ties to each other, I created a set of preliminary networks to achieve a more basic view of the organization. The preliminary networks are k-modal with three types of nodes: the company representatives, the region or sector they belong to, and the executive council. The ties between nodes are undirected and binary, and were justified by group membership in either the region, sector, or council. All of the 19 companies based in Europe are connected to the Europe region node. All the companies with seats on the executive council are connected to the council node, etc. Europe Latin America Executive Committee North America Image 11: Network A-company, region,Asia council Africa Image 11: WBCSD companies connected to region where headquarters are located and executive committee 20 Image 12: WBCSD companies connected to industry 21 Making Ties After making the preliminary networks, I moved onto the setup for the network that would encompass all of the WBCSD members and the ties between them. Before I could start data collection, I had to decide what constituted a tie between members. Since the analysis is based on who in the network has influence over others by the ability to spread or manipulate information, a tie should connect people if there is information being exchanged between them. Diffusion of Innovation Theory states that information and new ideas are best spread through interpersonal communication (Valente and Davis 1999). This theory has been applied to many SNA studies in the past, specifically on the concept of board interlock. Board interlock is based on the fact that the board members interact and influence each other to make decisions to benefit the companies they champion (Carroll 2004). Sitting on a board is similar to serving on the same committee or council; it creates a space for communication and discussion. Ideally, I would be able to use personal communication between WBCSD members as justification for a tie between them. However, I am not able to observe every interaction between them or access the archives folder in their email accounts. A “proxy” for observing communication must be used. By looking at the structure of WBCSD as a collection of boards, committees, and work groups we can assume some level of personal communication happened between members. The WBCSD divides their initiatives into several categories and subcategories in order to accomplish the wide range of goals the organization is working towards. The members of WBCSD work specifically on one or multiple projects at a time with other members; approximately five to twelve people make up a working group. The categories and subcategories of initiatives are as follows (WBCSD 2014): Natural Capital o Ecosystems solutions 22 o Forest solutions o Water solutions o Energy and climate o Electric utilities o GHG management Business Applications o Cement sustainability initiative o Chemicals o Tire industry project o Energy efficiency in Buildings 2.0 o Sustainable Mobility 2.0 o Urban infrastructure initiative Social Capital o Inclusive Growths o Performance & Valuation Capacity Building o Consumption & value chain o Education initiatives o People matter o Future leaders team Financial Capital o Reporting and investing In addition to being a part of the committees on the above categories, members are also involved with creating publications, reports, and case studies that are made available to member companies as well as whole industries and the general public. If WBCSD members are in the same committee, have worked on the same publication, or there is evidence of any other sort of collaboration, it is assumed that they would have to directly communicate with each other at multiple points throughout the project. This justifies a tie between two people. The ties will be undirected because the communication between the two people is assumed to be mutual. Ties will be weighted based on the number of times two people have collaborated on an initiative. Data Collection I have begun to collect data by systematically going through the WBCSD website, category by category, recording which company representatives are currently serving on the same work group or have published a document currently available on the WBCSD website. I 23 am using Microsoft Excel to record data because it easily converts into a file that the software will accept. The information in the data collection sheet includes: node identification number, company name, representative name, and type of node. The members are categorized by the committee in which they participate. For an example spreadsheet, see Appendix D. The spreadsheets will be converted to CSV files and imported to Gephi, the open-source software I am using to conduct the analysis. 24 Future Analysis After completing data collection and transferring the data to Gephi, I will use the centrality measures previously discussed to analyze the data. Although I cannot discuss the quantitative results just yet, it is important to understand why the centrality measures work as analysis tool and what they are going to show about the network. The most crucial concept in the analysis is that power and influence are a function of position within the network. More central nodes are more influential, more peripheral nodes are less influential (Cook 1983). Centrality identifies the most “important” nodes that allow information to flow through the network. Degree centrality, the first measure, is the most basic way of measuring the number of nodes one person in WBCSD reaches. However, it gives only one possible perspective on the network. There are more ways of having influence than being in communication with a large number of people. Most people are likely have other sources of information that may conflict with information from the central node. Closeness centrality shows how capable a node is of avoiding control. Having many short connections allows the node to gain information from multiple sources, and therefore be less likely to receive limited or manipulated information (Cook 1983). Nodes with high closeness centrality are less likely to be influenced by other central nodes. Betweeness centrality, a concept introduced by sociologist Linton Freeman, is the opposite of closeness. Introduced as a measure for quantifying the ability of a person to control the communication between other people in a social network, it measures a nodes ability to control the flow of information between different areas of a network (Famiglietti 2014). For example, if one company member is part of both the executive committee and a water solutions publication he or she in theory would be able to control some information that flows between the 25 two groups. The water solutions publication may receive more attention or funding because of the connection to the executive committee. It is easy to see how a node with a high level of betweeness may impact decisions on what and how the WBCSD works on. Eigenvector centrality is the most difficult measure to clearly explain; it is not based on number or length of connections. An eigenvector central node is a node that has a connection to other nodes that are high in other measures of centrality. Eigenvector nodes do not necessarily have influence over a large number of other nodes and usually only have very limited connection to the periphery of the network. However, they are also usually in formal positions of power or authority within the network. Any eigenvector node in the WBCSD network may have a disproportionate amount of influence over the function of the organization (Kadushin 2012). 26 Conclusion and Future Plans All of the analyses discussed tell us about one thing: the ability to control information. Since mass diffusion of innovation leads to change in a network (Valente and Davis 1999), the way information spreads throughout WBCSD and who controls it has a massive impact on the policy tools and mission of the organization. WBCSD is a transnational powerhouse when it comes to shaping the future of the global economy, making it vital that its actions are in the best interest of a sustainable future. The intentions, whether good or bad, of a few individuals should not dictate the changes WBCSD is striving to make. Understanding how the organization works as network via social network analysis will allow global leaders to evaluate the effectiveness of the organization. They can then discuss whether the structure needs to be adapted to better suit a more effective and communicative space in effort to lead the business world into a green economy. I will discuss the specific implications for the WBCSD network once the analysis is done and the central nodes have been identified. I will then begin to look at the specific initiatives WBCSD is producing to see if there is any correlation between the central nodes and the projects that seem most prominent and successful. Additionally, I will discuss any commonalities among the central nodes such as sector or region and then look at the projects specific to any category that stands out in the analysis. I do not have predictions for what I will discover, but I hope to find that the World Business Council for Sustainable Development is a network that allows communication and collaboration amongst world leaders to create a sustainable future for business, society and the environment. 27 Appendices Appendix A: Member Organizations divided by Region Africa South Africa ESKOM Holdings Ltd. Mondi Sasol Europe Austria Andritz AG Borealis Belgium Solvay KBC Denmark Novozymes Finland KONE Oyj Metsä Group Metso Corp Pöyry Oyj UPM-Kymmene Corp France EDF Group GDFSuez Havas Group L'Oréal Lafarge Michelin Kering Schneider Electric Suez Environnement Veolia Environnement Germany BASF Bayer BMW Group Continental AG Daimler AG Deutsche Bahn AG Deutsche Bank Deutsche Post DHL E.ON Evonik Industries HeidelbergCement Henkel RWE AG 28 Siemens Volkswagen Greece Public Power Corp. (PPC) Titan Cement Ireland CRH Italy Eni S.p.A. Italcementi Group Pirelli & Co., S.p.A. Kazakhstan ENRC Group Luxemburg ArcelorMittal The Netherlands Akzo Nobel Arcadis DSM N.V. KPMG Royal Dutch Shell plc. Royal Philips Electronics TNT N.V. Unilever Norway Det Norske Veritas Norsk Hydro Statkraft AS Statoil Portugal ALTRI Group Brisa EDP - Electricidade de Portugal, S.A. Portucel Soporcel Group Sonae SGPS Spain ACCIONA Sweden Skanska Stora Enso Svenska Cellulosa AB (SCA) Sweco Vattenfall AB Switzerland ABB F. Hoffmann-La Roche Firmenich Holcim Nestlé S.A. Novartis Sika 29 SGS Société Générale de Surveillance Holding Syngenta International Turkey Borusan Holdings Eczacibasi Holding United Kingdom Anglo American BG Group BP BT plc Environmental Resources Management Group-ERM SABMiller Vodafone Group Asia China Baosteel Group Corporation China Ocean Shipping (Group) Company, (COSCO) China Petrochemical Corporation (SINOPEC) CLP Holdings Ltd. State Grid Corporation of China (SGCC) India Aditya Birla Group Infosys Technologies Reliance Industries Vedanta Resources ITC Limited Indonesia Asia Pacific Resources International Holdings Ltd (APRIL) Japan AGC Group Bridgestone Corporation Canon Denso Corporation Fujitsu Hitachi Ltd. Honda Motor Co. Komatsu Mitsubishi Chemical Holding Corporation Mitsubishi Corporation Nissan Motor Osaka Gas Sompo Japan Insurance Sumitomo Chemical Company Ltd. Sumitomo Rubber Industries, Ltd. Taiheiyo Cement Corp. Toyo Tire & Rubber Co., Ltd Toshiba Corp. Toyota Motor Corp. The Yokohama Rubber Company 30 Korea GS Caltex Corporation Hankook Tire Co., Ltd KUMHO Tire Co., Inc. Samsung Electronics Malaysia Sime Darby Berhad Saudi Arabia SABIC - Saudi Basic Industries Corp. Taiwan Acer Group CPC Corporation, Taiwan Chunghwa Telecom Thailand PTT Public Company Limited Siam Cement Group (SCG) Latin America Brazil Abril Group Cimentos Liz Fibria InterCement Natura Cosméticos S.A. Petrobras Suzano Papel e Celulose Vale Votorantim Chile Masisa Empresas CMPC Colombia Empresas Públicas de Medellín (EPM Group) Grupo Argos Mexico CEMEX North America Canada Suncor Energy U.S.A 3M Accenture Alcoa Bank of America CH2M HILL Chevron Corporation Cooper Tire & Rubber Company Deloitte Touche Tohmatsu Dow Chemical DuPont 31 Eastman Chemical Eaton Corp. Ernst & Young LLP Fluor Corporation Ford Motor Company General Electric Greif Harley Davidson International Paper Kimberly-Clark Corporation MWV Monsanto Company PepsiCo PricewaterhouseCoopers Procter & Gamble S.C. Johnson & Son The Coca-Cola Company The Goodyear Tire & Rubber Company The New York Times United Technologies Corporation-UTC UPS Weyerhaeuser 32 Appendix B: Membership Diagrams 33 Appendix C: Extended Terminology List State relation: has a degree of permanency or durability that makes it easier for the researcher to see ex. Family, religion Event relation: more temporary and may or may not imply a more durable relation, more like individual occurrences ex. Attending same conference, sending an email Digraph: graph with directed relations, called arcs Boundary: defining which actors are included in the network Isolate: no ties to other nodes in the network Path: walk where no node occurs more than once Semipath: ignores direction of arcs Symmetric matrix: contains data for undirected network Asymmetric matrix: contains data for directed network, records direction of ties Valued matrix: shows weighted values (Prell 2012) 34 Appendix D: Data Sheet Example 35