• No results found

Chapter 2 Theory Development

2.3 Theory Base

2.3.1 Network Theory

Food production and distribution is a multi-firm process. It is also not a linear process as food supply chains can be thought of as complex organizations. The expertise and performance of each actor involved is important and these actors are dependent on each other’s performance to run their businesses. A such, these actors and their relationships form a large web termed as a network (Håkansson & Ford, 2002). Fundamentally, network theory examines the methods and processes that interact with network structures to produce certain results for individual actors or groups and also to understand the behavior of the network as a whole (Barabási, 2009). Network theory considers these structures as purposefully interconnected. These interrelated entities can be a single firm, a dyad or a triad (G. Li et al., 2010). A triad is the smallest unit of a network (Choi & Wu, 2009). Most of the early research on networks is from an individual perspective (Merton, 1957). Recently, however, scholars have begun to

study triads in networks (Hearnshaw & Wilson, 2013; Pathak et al., 2007; Pathak, Wu, & Johnston, 2014).

2.3.1.1 Basic Elements of Network Theory

Scholars use this network theory to understand the structure and connections among individual actors, dyads and triads. In the case of supply chains, at a very basic level, two types of network can be recognized depending on the characteristics of the nodes and the nature of the links; an actors’ network and a physical network. The actor level consists of the firms that operate and work together in a given environment. These firms exchange goods, information, knowledge and money (the links). The physical network comprises of warehouses and other storage places which can be accessed by different forms of transport. Together these networks make a supply network.

A network consists of two main elements: nodes and arcs/links. Different nodes are connected together through the arc. Nodes are also called points or vertices and arcs are also called ties and edges. Nodes represent the actors, whereas the arc is the relationship between them. How the nodes are interlinked through arcs defines the structure of any network. Network will be called a complete network if all the nodes are connected to each other. Nodes can be directly connected to another node or they can be indirectly connected through other nodes and arcs. Any limited sequence of nodes and arcs is called a walk and if each node is unique on this walk, it will be called a path. In one network, multiple paths can be present and on each path, the flow of goods, money and information may differ (Wasserman & Faust, 1994).

The interpersonal links and ties among different firms are called social networks. Social capital is a vital intangible asset which can increase the overall efficiency of the supply network (Krause, Handfield, & Tyler, 2007). Social capital covers issues of trust, rules, norms and beliefs that are part of any society. These networks influence the way firms perform in a given context. Firms are embedded in these networks and the degree of embedding decides the relationship with other firms in the same network. This aspect of network theory can be analyzed using social capital theory. Social capital affects firm performance and relationships (Johnson, Elliott, & Drake, 2013). It is different from physical resources, as it is not located at a certain place, but rather, is embedded in relationships (Wills-Johnson, 2008). Social capital can be defined as all the available resources within a network and that are derived from the relationships between different firms or individuals (Nahapiet & Ghoshal, 1998).

Network perspective is important for the study of supply chains as it provides a way to understand the structural characteristics of ties, collaboration and also the power distribution in various organizations. Choi and Wu (2009) view a supply network as a network of organizations engaged together in producing and distributing products. Specifically, they underscore the importance of dynamism in these networks. These networks are not static in nature, but rather, are ever evolving as they are affected by numerous events (disasters included). These supply networks adapt, as organizations try to adjust their position in order to survive and fulfill the demand of customers in these events (Pathak et al., 2007). As mentioned before, triads are the smallest unit within networks (Dubois & Fredriksson, 2008). To understand supply networks, it is necessary to understand the relational actions between different triads. Simmel (1950), a pioneer in the field, discussed triads in a sociological context. Later on, Burt and Celotto (1992) studied the behaviors of nodes in various social contexts. In any network, two nodes may have no direct link, but through the third node which is common to these two. The third nodes play the role of a broker and this disconnection is called a structural hole (Burt & Celotto, 1992). The two disconnected nodes may be aware of each other but do not form any relationship with each other. This triadic block can have two separate relational strategies. One is ‘tertius gaudens’ as introduced by Simmel (1950), that means a third party benefits from conflict between two others. In this strategy, the third node controls the two nodes by actively separating them, thus deliberately creating a structural hole. However, there is another relational strategy which may be present between two nodes which is called ‘tertius iungens’. This term means the third party unites the other two. In this strategy, a buyer forms a bond a between two of his/her suppliers thus increasing collaboration within the network (Wilhelm, 2011).

Based on these two main relational strategies, Obstfeld (2005) shows the evolution of social networks (Figure 2.3 ). Where A is a primary buyer, B and C are first-tier suppliers, D, E, F, G are second-tier suppliers. Similarly, this can be seen in the case of natural disasters where, in the response and recovery phases, new ties evolve (or break) within the network.

Figure 2.3 An Example of an Evolving Triadic Network. A B F C G D E A B F C G D E A B F C G D E

Initial State (The third who benefits): A is broker

between B & C and similarly B brokers D & E, C

brokers F & G

Tertius iungens introduction stage: A forces

a collaboration between B & C

Collaboration Between B & C generates new structural

holes

A new tertius iungens develops between A and E

A B F C G E

Source: Adapted From (Obstfeld, 2005)

Key actors in these structures control and coordinate to ensure that customer demand is fulfilled (Choi & Krause, 2006). Based on this perspective, Pathak et al. (2014) introduced four elements in supply networks that are interrelated and are common to each network: firm level activities, ties among firms, network level goals and network governance. Firm level activities include, the routine tasks performed by a firm such as procurement, product packaging, product development and distribution. Ties refers to the coordination among different firms to achieve the daily operational objectives of individual firms, that ultimately fulfils the network level objective to collectively deliver the products to end customers (Batt & Purchase, 2004). Governance refers to controlling and managing the behavior of individuals or group of actors, present in the network. (Provan & Kenis, 2008) argue that there are three ways by which this can be managed: shared (where individual firms present in the network govern accumulatively), lead (where one single lead firm controls and manages the network) and network administered (some separate entity or governing body manages the network). These factors refer to the governance or the administration of the network, however networks have inherent properties themselves. Indeed, supply networks will be efficient if overall coordination and governance is satisfactory throughout the network. There are three main properties that are evident in efficient supply networks, these are: short characteristic path length, existence of power law distribution and high clustering coefficient. This being true for scale free networks (Barabási, 2009; Ramasco, Dorogovtsev, & Pastor-Satorras, 2004).

Discussing these three properties in turn, the characteristics path length points to the average detachment between two random nodes selected from a network. In terms of the supply chain, it depicts an average number of firms from all of the tiers that must be traversed

between two randomly selected nodes. If a large number of intermediaries are involved, then the average length from the initial supplier to the end customers will be obviously high. An efficient supply chain will have a short path length: in short, there will be a smaller number of traverses between the two nodes (Hearnshaw & Wilson, 2013). This short characteristics path length demonstrates the small world property originally coined by Milgram (1967).

Next, the clustering coefficient refers to the probability of two suppliers of a given buyer as being attached to each other (see triadic structural holes definition above). It is necessary to examine these different triadic relationships in any network. A high clustering coefficient means there are more (triadic) links between contiguous nodes, potentially more collaboration, and hence more efficiency at a network level (Pathak et al., 2014).

Finally, the Power law distribution means that there are only a few nodes in the network with the highest number of connections present. These are generally referred to as ‘hub’ firms (Barabási, 2009). Human and Provan (2000) state that the presence of a hub firm in a supply chain network is significant. Like the channel leader, these firms tend to control and manage the network wide coordination of supply chains. It should be noted that only ‘scale free’ networks possess a Power law distribution. Other distributions may reflect other network structures, yet it is not necessary for a Power law to be present to establish the presence of hub nodes.

Traditionally, networks are modeled as either regular or random. Today, there are typically a large numbers of actors involved in supply chains and thus their complexity has subsequently increased. This is why, Choi and Krause (2006); Hearnshaw and Wilson (2013); Pathak et al. (2007) assert that the supply chain should not be considered as a simple system, but rather, as a complex adaptive system. Hearnshaw and Wilson (2013) believe that complex network models capture the properties of efficient supply chains in a more holistic-systems way.

Two common and important complex network models are the Watts-Strogatz’s (WS) model and the Barabasi- Albert (BA) model. Based on the random network typology, the WS model suggested that high clustering coefficients and the ‘small world’ properties better represents the efficient network (Watts & Strogatz, 1998). But random connections is not an accurate depiction of the formation of various supply chain relationships developed in real supply chains. It means that supply chain are systems with a fixed number of firms and relationships (Hearnshaw & Wilson, 2013). Indeed, the BA model provides an alternative, a scale free

network, that deals with supply chain complexity. In comparison to the random attachment model, where pairs of nodes are randomly connected to each other, scale free networks evolve through the mechanism of preferential attachment. That is, new nodes entering the network will ‘choose’ those nodes that already have a high number of connections. In reality, this mechanism produces a ‘rich-gets-richer’ phenomenon (Besanko, Dranove, Shanley, & Schaefer, 2009). This formation mechanism helps explain the existence of hub firms in the network and is a unique feature as compared to other models.

Scale free networks demonstrate an amazing robustness and resilience against any random disruption (Hearnshaw & Wilson, 2013). This property is embedded in their inhomogeneous topology, in that, the removal of small nodes (that are plentiful) in the event of some disaster would not affect greatly the overall network. Hence, the network will keep on working or will return to the same state as it was before the disaster. This is because the majority of the links are associated with the hub firms (Albert, Jeong, & Barabási, 2000). However, if there is a planned attack on the hub, then the whole network would become vulnerable to this threat. This is an inherent drawback of these types of networks based on hub firms.

Another important element of resilience is adaptability. Researchers have revealed the relationship between adaptability, size of the network and persistence (Palla, Barabási, & Vicsek, 2007). Smaller networks are more adaptable and persistent when they have fixed and stable hub firms. In contrast, larger network tend to be adaptable, particularly if new nodes are constantly entering and exiting the hub node (Cowan & Jonard, 2007). For achieving resilience and adaptability, firms need to contrive and position themselves in these networks to access information and resources. These resources, and any competitive advantage through these, can be explained with the help of resource-based theory.