2 Literature Review
2.2 Non-standard Supply Chains
2.2.3 Complex Adaptive Systems
As demonstrated in the previous discussion, interactions in modern SCs exhibit a much higher complexity than merely linear causality where an effect can be directly related to one cause, but rather a network of interacting variables, thus making them systems according to the literature of systems thinking (von Bertalanffy 1968, Emery 1969, Checkland 1981, Checkland 1983). Indeed, the added complexity that arises from spatially and temporally separating cause and effect along a SC with a multitude of actors has been noted as necessitating a good understanding of the interdependencies and causal relationships in SCs (Holmberg 2000). Systems thinking offers a way to describe and analyse such complex problems, an enables communication of them and has thus been utilised in operations research and SC and logistics management (Checkland 1983, Holmberg 2000, Rigby et al. 2000, Sterman 2000).
Systems thinking has been applied to SC research since Jay Forrester’s work in 1961 (Mason-Jones and Towill 1999b). The central concept of a system embodies the idea of a set of elements connected together to form a whole that exhibits properties that are different than those of its component parts (Checkland 1981). SC management encompasses this systems thinking approach based on the fundamental understanding that the whole is greater than the sum of its parts (Mentzer et al. 2001, Christopher 2005, Grant et al. 2006, Patel et al. 2013, Ellram and Cooper 2014). Systems thinking focuses on the interactions of the parts of a system (Wiener 1948, Ashby 1963, von Bertalanffy 1968). This thesis employs systems theory and the conceptual framework developed here is based on it, particularly on CAS.
CAS are systems that function without any central control and lack a permanent, fixed structure, but are nonetheless distinguishable from their surroundings, such as ecosystems, immune systems, or cities (Holland 1995). Originally, CAS or complexity theory emerged in the physical sciences (Prigogine and Stengers 1984, Lewin 1993, Kauffman 1995), but were applied to the social sciences almost simultaneously (Kiel
1991, Wheatley 1999). The underlying idea is that CAS mimic living organisms in their ability to organise, to learn, and to evolve (Capra 1996, Lansing 2003). A CAS consists of “populations of individual adaptive agents whose interactions result in complex non-
linear dynamics, the results of which are emergent system phenomena” (Brownlee
2007). The concept of CAS as self-organising systems is at a very high level of abstraction, transcending academic disciplines and being applied across both natural and social sciences (Levin 1998, Lansing 2003). It has seen use in such diverse fields as mathematics, psychology, anthropology, evolution, ecology, information systems, and business management, although there is no unified theory of CAS (Brownlee 2007).
It is challenging to study CAS by conventional means, as it is impossible to study parts of the system in isolation and still retain a sense of the overall functionality of the CAS. CAS are also highly dependent on their history, thus making it difficult to compare and forecast (Holland 1995). This is in sharp contrast to the mechanistic approach of Taylorism and related schools of thought that dominated much of early management philosophies, which aim to subdivide and manage individual parts to approach the whole of an organisation (Dooley 1997). Instead, the CAS approach sits within the domain of systems thinking, which has been developing since the 1940s and is centred around the idea of interactions between parts of a system (Wiener 1948, Ashby 1963, von Bertalanffy 1968), leading scientists in the field of management to see organisations as quasi-organisms (Morgan 1986). With the focus on a collective of interacting adaptive agents in CAS, a reductionist approach is not feasible, as it would over-simplify the inherent complexity of such systems (Gell-Mann 1994). In CAS, as well as in systems thinking as a whole, the whole is always more than the sum of its parts, as interactions play a key role in determining overall functionality and ability to adapt to changes.
In the preceding discussion of SC types and in particular of non-standard SCs, interconnectedness of individual elements within a SC and the ability to react flexibly to changes played a key role in modern SCs. The concept of CAS thus appeals as an underlying idea for the study of such SCs. Individual organisations within a SC can all be seen as interacting adaptive agents that are united in a common purpose if following the alignment demanded in literature (Lee 2004, Christopher 2005). While a fixed and somewhat permanent structure and centralised control might be desirable in some SCs, others, such as the aforementioned non-standard SCs cannot achieve this and, due to the high uncertainty they face, would probably be ill served by such a structure as it is likely to be established at the expense of their flexibility.
Indeed, CAS have been applied in the study of SCs, regarding them as “a
collection of firms that seek to maximize their individual profit and livelihood by exchanging information, products, and services with one another” (Choi et al. 2001, p.
365) and stressing that interactions between adjacent firms determine the level of control of the system exhibited by any one firm and the behaviour of the SC as a whole, while remaining emergent, dynamic and unpredictable. Employing the established theory of CAS is regarded as a major step towards understanding how highly-complex modern SCs can be governed and how adaptive, flexible and coherent collective behaviour can be coordinated (Surana et al. 2005). The key elements of a SC in terms of CAS are (Day, 2014):
• Entities: Decision-making organisations within the CAS that exhibit dynamic learning based on information obtained through their relationships. • Topology: The network created by connections between organisations,
including the flow of resources, finance, and information, as well as decisions made.
• Environment: The broader context in which entities, topology, and the system as a whole exist, and which affects them.
• System: Emergent from the interactions between entities within the CAS, but also interactions with the environment, displays system behaviours.
Research indicates that interactions with the environment, as well as path dependencies within the system are crucial in determining how CAS SCs evolve over time and respond to changes (Pathak et al. 2009). A CAS view of SCs conforms with the earlier discussion of modern SCs and the challenges they face, as the role of the overall business environment and the alignment of SC partners is highlighted.
It is primarily the connectivity amongst entities that determines the topology of the overall SC (Day 2014). However, there is an inherent dichotomy of cooperation and competition as each entity seeks to maximise their individual profit (Surana et al. 2005). While it may seem counterintuitive, such competition is actually regarded as positive in the context of CAS as it aids the creation of emergent and highly dynamic new solutions within the system (Choi et al. 2001). Such dynamic change is to be embraced, as it can increase the resilience of the CAS SC (Pathak et al. 2007, Pathak et al. 2009, Day 2014) through the development of new, shorter paths within the system (Strogatz 2001), as well as higher levels of path redundancy (Albert 2005). Viewing SCs as a CAS therefore requires some level of trust in the properties of the system rather than immediate interference through control measures. While managers should control the
course of action towards overall goals, their key role in a SC CAS is to observe emergent issues and to make flexible and dynamic changes accordingly (Choi et al. 2001).
From a research perspective, CAS provide both guidance for managerial approaches towards a self-organising system without central control, and the opportunity to model SCs using insights from a wide range of fields that study complexity, thus bundling expertise and learning to enhance understanding of SCs that display such a high level of complexity that they are difficult to grasp with standard problem solving methodologies (Pathak et al. 2007, Pathak et al. 2009). Particular attention should be paid to environmental factors that are regarded as crucial in determining how a particular CAS reacts to inevitable changes (Choi et al. 2001, Pathak et al. 2009).