Economic network theory is an emerging field of research that applies models from network science to the analysis and interpretation of economic phenomena. Economic network analysis can be seen as a response to the rise of networked organizations over the past few decades, given the proliferation of information technology and globalization. The work of Manuel Castells has identified networks as the emerging fundamental organizational structure within post-industrial economies.
Within the space of just a few decades, the technological infrastructure and institutional superstructure of advanced economies have become increasingly integrated into dense, multimodal networks, from the micro level of individual organizations all the way up to the global level through global cities and the supply chains that they enable. We have increasingly interconnected our economies, bringing in frontier zones like the copper mines in Mongolia or small rural villages in Norway as they connect into the ever-denser set of connections between the global cities. In parallel to this has been the development of financial networks that represent cross linkages between assets and liabilities all over the planet, risk, and returns that are sliced and diced and distributed out into the future through all sorts of derivatives, contracts, and exotic instruments. Making all this happen are of course telecommunication networks on all levels. Previously well-bounded organizations like national economies and corporations are being required to adapt to a massive wave of hyper-connectivity, as we see new IT-enabled network organizations emerge and thrive in this connected economy of the 21st century.
Linear system theory is a component-based theory, meaning it is primarily concerned with the properties of the components in the system, and this will be the same for any science that is using this framework such as standard economics. When we look at a typical map of the world economy, it will show us lots of countries and their GDP. When analyzing a corporation, we describe it by itemizing its gross revenue, number of employees, who the CEO is, and so on. All of these are descriptions of the system through reference to the properties of its components. This is a very valid and important approach to economic analysis, but it is only really relevant when we are dealing with a linear system that has a low level of connectivity. When we turn up the connectivity, the relations between the components start to become the primary factor in determining the system’s overall dynamics. In such a case we need to use an alternative modeling framework that is better able to capture these relations and their structure; this is exactly what network theory does.
Network theory, also called graph theory, is one of the very few major modeling frameworks within complexity theory. It is an abstract formal language which deals with the big idea of connectivity, which is a whole paradigm shift because we are naturally adapted to see things and not so much connections. Thus, this world of connectivity is very different to the one we are used to. It is all about access, where you are in the network, what is the overall structure of that network and what is flowing through it. As is always the case, it is important to understand what the modeling framework is designed to do before you start to use it. Network theory is designed to let us focus on the relations between components and the structure of those relations in both a qualitative and quantitative fashion. As such, we will very rarely be talking about the components themselves. So it offers us just one perspective on the whole system. It is an abstract modeling framework and things can get complex very quickly.
A network in mathematical terms is called a graph. A graph is made of nodes and edges. A node is a thing, like a bank, business or country. An edge is a connection between two things, like the trade of oil between two countries, an investment between a bank and business, or purchase between a retailer and a customer. Both nodes and edges can have values associated with them that denote the size of the node or the volume of exchange within an edge. Edges can be directed or undirected, indicating the net flow of resources along the edge. In a multi-graph, we can have many different edges between any two nodes. For example, each edge might represent the trade of a different good between two nations. This is the very basics of the language of graph theory.
Because network science is all about connectivity, when we are analyzing a particular node in a network, such as a corporation within a supply chain network, our first question will often be how connected is that node, that is to say, how many links does it have with other nodes in the network? This is called its degree of connectivity. The node’s degree of connectivity will define how likely it is to receive whatever is flowing within that network. If the corporation was part of a supply chain network for the production of tractors, its number of links might define how many of the parts for that tractor flow in or out of the organization. Degree distribution will not be the only factor determining its significance within the network but it will be a primary one and the most straightforward one for us to measure.
How important a node is within a network is a function of both how much of the network’s resources are flowing through it and how critical that node is to the network. So the Pearl Delta Economic Zone in Southern China, although only about 3 percent of the nation’s population, represents about 35 percent of the country’s total international trade. Thus, this node plays a very significant role within the economic network due to the sheer volume of resources that flow through it. Panama also plays a critical role in the global supply chain but this time it is because it is the only viable sea route between the Pacific and Atlantic. Thus, it is what is called a bridging link. It performs a differentiated function that the rest of the network requires, giving it significance within the entire network.
A node’s real significance within a network, what is called its centrality, is quite a complex feature to analyze. Added to these two factors previously mentioned, we need to also take account of its location within the overall network, asking how close it is to all the other nodes, thus how quickly it could affect them and also how connected the nodes that this node connects to are. As an example of centrality analysis, we might think about government bailouts during a financial crisis. As the government is interested in maintaining the functionality of the entire network, it needs to ask these questions that we have mentioned: How many links does this bank node have and what volume are those links? Does the node play some critical role within the financial network that no other institution could perform? How closely connected is it to all the other nodes and how important are the other nodes that it is directly connected to? By answering all these questions, they would be able to get some understanding to its importance in maintaining the entire network.
When we are looking at the whole of a network, probably the single most important parameter is the system’s overall density. The density of a network is defined as a ratio of the number of edges to the number of possible edges. As we turn up the probability of there being a link between any two nodes, we will get a more dense network, starting from zero density where no nodes are connected to complete density where all nodes are interconnected. We can then define a parameter for adding links, which is really capturing how easy it is for a node in the network to make a connection with another. We might call this transaction cost. When the expense of making a transaction is high, there will be few connections. As we turn it down, we will get the emergence of a denser network. For example, these transaction costs might represent trade barriers placing a greater transaction cost on international exchange. As we have reduced these trade barriers through liberalization, we have seen the emergence of globally integrated supply chains.
Network density is an important parameter in that it will define the difference between a component-based system at a low level of density, where the value of an element in the system is really in that node; such as a business or some technology, while it is relatively isolated, the value of it is bound within the organization. When we turn up the density and number of connections, this is no longer so. The component’s value is increasingly outside of it in the network of connections it has with other nodes, in the way that a smartphone has value because you can connect to many different services via the Internet, or a business has value derived from its place within a supply chain network. Networks do not always grow in an even linear fashion. Due to the positive feedback of the network effect, we can get nonlinear exponential growth, as we have witnessed with the rise of the Internet, which stayed relatively dormant for a number of decades before reaching a critical mass and then growing rapidly. Now there is a huge amount of value not in any one organization but in the network of connections that gives one access to them. Thus, by reducing transaction costs low enough this can lead to a powerful network effect taking hold.
A second key factor we will be interested in when analyzing an entire network is in asking how close are any two nodes in the network on average – what is called the average path length. This will be a function of both the number of nodes in the network, how interconnected they are and the overall structure of the network. Average path length is important because it defines how close agents are to each other. Agents operating in a system where they are very far from others will create different behaviors to when they have the appearance of being very close to everyone else. Here we might think about globalization, through increased connectivity that has reduced the path length, we suddenly start to feel much closer to everyone else, creating a much greater sense of interdependence. Because of these shorter path lengths, externalities become much more important and immediately perceptive.
So far we have been talking about monoplex networks, meaning that with these relatively simple graphs we are looking at just one type of network; each node and edge in it only serve one type of function.9 But the real world is typically a lot more complex than this. When analyzing a large system like a metropolitan economy, a corporation or the global commodities market, what we are dealing with is a network that is embedded within many other networks. For example, a metropolitan area would be a complex system where economic interactions are embedded within socio-political networks, transportation, and geospatial networks, financial networks; all of which will strongly influence the flow and distribution of economic resources. This is clearly going to add a whole new level of complexity to our representation. But in order to do this, researchers have developed what are called multiplex networks. In a multiplex network, each type of interaction between the nodes is described by a single layer network and the different layers of networks describe the different modes of interaction. Multiplex networks are clearly much more advanced representations of how these systems really operate through the interaction of many different functional domains. They lay at the forefront of contemporary research in network science with a vast amount of untapped potential.
Lastly, as we have tried to illustrate here, connectivity fundamentally changes the dynamics of an economic system. As such, it also alters how we should go about designing and managing them. At a low level of connectivity in a component-based regime, we traditionally try to intervene by directly altering the components in the system. For example, a government tries to improve its economy by starting a big infrastructure project; or we try to get people to buy things by bombarding them with advertisements. But with networks, it is all about designing and managing the connections. You get a person to buy a product by influencing their social network. Your country’s economy grows by connecting it to the global supply chain network. This is economic and financial systems design and management by connecting or disconnecting things. The wealth is in the network. If you want more of something, you restructure its connections and position within the network to make it more receptive to that flow of resources; if you want to diminish it, you disconnect it.
Social and Economic Networks. (2017). Princeton University Press. Retrieved 30 May 2017, from http://press.princeton.edu/titles/8767.html
The Network Society. (2017). Google Books. Retrieved 30 May 2017, from https://books.google.ie/books/about/The_Network_Society.html?id=5SDEQgAACAAJ&redir_esc=y
James B. Glattfelder: Who controls the world?. (2017). YouTube. Retrieved 30 May 2017, from https://www.youtube.com/watch?v=NgbqXsA62Qs
(2017). From Graph Theory to Models of Economic Networks Springer.com. Retrieved 30 May 2017, from https://goo.gl/Sh5epZ
(2017). Maths.ed.ac.uk. Retrieved 30 May 2017, from http://www.maths.ed.ac.uk/~aar/papers/wilsongraph.pdf
(2017). Math.ucr.edu. Retrieved 30 May 2017, from http://math.ucr.edu/home/baez/econ.pdf
(2017). Stat.washington.edu. Retrieved 30 May 2017, from http://www.stat.washington.edu/people/pdhoff/courses/567/Notes/l6_centrality.pdf
(2017). Benkler.org. Retrieved 10 June 2017, from http://www.benkler.org/Benkler_Wealth_Of_Networks.pdf
Multiplex Networks | Cosnet. (2017). Cosnet.bifi.es. Retrieved 10 June 2017, from http://cosnet.bifi.es/network-theory/multiplex-networks/
(2017). Oecd.org. Retrieved 10 June 2017, from https://www.oecd.org/sti/ind/interconnected-economies-GVCs-synthesis.pdf