Traceability and normal accident theory: how does supply network complexity influence the traceability of adverse events?
Thinking about complex supply networks rather than simple supply chains has become a preferred approach to the challenges of managing the flow of goods and information. Authors such as Lee (2002), Choi, Dooley and Rungtusanatham (2001) and Madhavan, Gynwali and He (2004) have proposed that in most cases the supply chain metaphor leads managers and researchers to oversimplify the problems of managing flows of goods, services and information. This over-simplification has prevented scholars and managers from fully understanding the causes of defects and adverse events that occur in supply relationships. This is because the causes of some of the defects and adverse events that occur in supply networks are complex because the network itself is complex. For example, defects and adverse events that result from interactions between a changing mix of participants in supply systems that depend on both relational and transactional governance mechanisms can be exceptionally difficult to understand. Shifting the unit of analysis from the simple supply chain to the complex supply network should enable managers and researchers to engage quality and control problems that have thus far proved intractable.
One of these problems has to do with traceability. Traceability, which we define as the ability to identify and verify the components and chronology of events at all stages of a process chain, has long been an important construct in control-oriented theories of quality (Juran 1979; Cheng and Simmons 1994; Ramesh 1998; Linderman, Schroeder, Zaheer and Choo 2003). Complete information about process chains is necessary in order to verify conformance to specifications on one hand, and to trace the causes of failures and adverse events on the other. As processes and relationships become more complex, achieving traceability becomes an elusive goal. For example, efforts to build traceability into design processes, so that features in the end product can be traced back to the requirements they fulfill, have sometimes broken down as requirements and development processes have become more complex (Griffin 1992; Ramesh 1998). As a result traceability is an issue in the management and improvement of processes in both designed and emergent systems. As defects or adverse events occur, traceability is called into play to uncover the causes of variation, which may in turn lead to the development of improved controls or improved processes. What we propose in this study is that as processes (supply processes in particular) become more complex, there are greater barriers to traceability. Overcoming these barriers is critical for quality management in complex supply networks (Roth, Tsay, Pullman and Gray 2008).
The purpose of this study is to make progress toward a more encompassing theory of barriers to traceability, based on recent work advancing the idea that supply network complexity changes the range of possibilities for managing material and information flows. In this paper we develop the idea that traceability depends on the degree of tight coupling in information flows and the transparency of buyer supplier relationships, in addition to supply network complexity. All of these play a role in the ability of users to identify and verify the sources and sequence of events in supply networks (Roth et al. 2008). Our theory development is informed primarily by normal accident theory (Sagan 1993; Perrow 1999; Weick 2005), which is best known for the proposition that accidents are nearly inevitable in complex, tightly coupled production systems. We take the normal accidents approach because, while catastrophic failures may not be inevitable (Roberts and Rousseau 1989), less critical adverse events, defects and near misses (Sagan 1993) are common enough to be the basis of widely applied theories of continuous improvement and methodologies like Six-Sigma (Linderman et al. 2003). Our use of normal accident theory departs from prior work on normal accidents because we are concerned with the problem of how network structure obstructs tracing the causes of adverse outcomes after the fact, rather than with showing how catastrophic system accidents could (or could not) be prevented.
Our interest in understanding how traceability is related to supply network complexity leads us to include the concept of transparency in tins study. Transparency at the network level is the extent to which information about sources, processes and relationships is readily accessible to counterparties in an exchange, and to outside observers (Lamming, Caldwell and Harrison 2004; Roth et al. 2008). We include transparency because, under some conditions, network complexity has different effects on transparency than on traceability. Since transparency is a factor in traceability (Roth et al. 2008) things that influence transparency can indirectly constrain traceability.
To make progress toward a more encompassing theory of barriers to traceability, we review the literature on supply network complexity, traceability and transparency and develop definitions for these constructs. We show how these relate to normal accident theory and explain how the construct of tight coupling fits with them. This leads us to propositions about the relationships between network complexity, tight coupling, transparency and traceability and an application of Perrow's (1999) typology of system types to supply networks. Drawing on events in food supply networks, we provide examples of the different types of complex supply network structure and show how each type influences traceability. We end by discussing managerial and scholarly implications for the design of traceability systems and supply networks.
Traceability and Transparency
Traceability, the ability to identify and verify the components and chronology of events in all steps of a process chain, is an explicit concern in fields such as total quality management, measurement technology, software and product design, logistics and food safety. Complete information about process chains is necessary in order to verify compliance with specifications on one hand, and to trace the causes of failures and adverse events on the other, whether the process is software development, replenishing a blood or food supply or assembling products from high value or critical parts.
Traceability has been proposed as a design principle in requirements traceability in software development (Ramesh 1998), QFD in product development (Griffin 1992), and manufacturing traceability systems (Cheng and Simmons 1994). Scholars have advocated for this principle in order to ensure that the outcomes of design processes comply with specifications.
Traceability becomes an operational issue when an adverse event, defect or any other deviation from specification occurs in a process. Some defects and adverse events are minor while others are catastrophic, but the difference is primarily one of degree. As defects or adverse events emerge in a process traceability is called into play to uncover the causes of variation, which may in turn lead to the development of improved controls or improved processes. It is primarily as an operational issue that traceability is an issue for theories of quality improvement in complex supply networks (Cheng and Simmons 1994; Charlier and Valceschini 2007; Golan, Krissoff, Kuchler, Calvin, Nelson and Price 2008; Roth et al. 2008).
There have been many attempts to define traceability, which can be categorized as logistics or attribute-oriented (Folinas, Manikas and Manos 2006). Where logistics traceability is only concerned with the physical movement of a product, attribute-oriented traceability also tracks information relating to quality and safety. The majority of traceability definitions are logistics-oriented. Our definition is attribute-oriented (Golan et al. 2008) because it involves tracking a product's flow and its attributes throughout the entire supply network. Examples of attribute traceability would be preserving data about whether an aircraft part comes from a particular ingot that received a particular heat treatment, whether a feature in a software package meets multiple requirements, or whether a food product has been in contact with genetically modified material.
Barriers to traceability arise when information about component movement and attributes is ambiguous, lost, distorted or hidden. While participants in supply networks might be presumed to know what they are exchanging and with whom, this is not always the case. Complex exchange relationships create opportunities for unintended errors and accidents to occur. All one has to consider is the frequency with which material goods are separated from the information that should go with them to understand that complex networks create unintended barriers to traceability. Complex exchange relationships also create opportunities for deliberate opportunism. Counterfeiting, misrepresentation, substitution of substandard materials or labor, adulteration and diversion of goods and information are all more likely to go undetected when exchange networks are complex. Although Jones, Hesterly and Borgatti (1997) make valid arguments about the collective sanctions that an industry network can bring to bear on participants who cheat, their arguments presume that participants know who they are dealing with and exactly what is being exchanged. As Lamming and his colleagues have demonstrated, supply relationships can be opaque, with only minimal information exchange; translucent, with limited exchange; or transparent, with full and free exchange (Lamming, Caldwell, Harrison and Phillips 2001; Lamming et al. 2004).
Following Lamming et al. (2001), we define transparency in dyadic supply relationships as "the two-way exchange of information and knowledge between customer and supplier" (p. 4). As more exchange partners open up to each other, it becomes more likely that information will be accessible throughout die network. When we move to the network level of analysis, we define transparency as the extent to which information about sources, processes and relationships is readily accessible to counterparties in an exchange and to outside observers (Lamming et al. 2004; Roth et al. 2008).
The level of transparency determines the amount of effort required to achieve traceability. Some degree of transparency is necessary in order for traceability to be effective but the relationship is not simple and linear. Minimizing information exchange results in an opaque network where high levels of effort are required to trace defects. Institutionalizing information exchange may lead to translucence, expressed as an emphasis on specific kinds of information and knowledge that are easily understood and exchanged (Bowker and Star 1999). In complex systems, this may lead participants to overlook information and knowledge that is ambiguous, hard to understand or that results from complex interactions between participant actions. Translucence can therefore create barriers to traceability that require additional effort to overcome.
Complete transparency may be an elusive goal because ensuring total openness requires participating organizations, agencies and governments to invest in information systems and share sensitive information that is not of immediate use, but that creates immediate costs. Although there can be no doubt that information standards and systems have reduced costs and improved information capture, retention and transmission, we are doubtful that they can be effective in all networks. Because the investments required to develop effective systems are large (Agrawal, Cheung, Kailing and Schonauer 2009), because the costs of adverse events do not reveal themselves on a regular basis, and because free and open information exchange creates occasions for opportunistic behavior, organizations tend to resist investments in full transparency (Lamming et al. 2004; Roth et al. 2008). Thus, while traceability is partly enabled by transparency, in all but the ideal case traceability always faces barriers created by network complexity.
Before moving on to supply network complexity we must recognize limits on the kind of factors we consider as creating barriers to traceability. The ability to trace comes partly from the investigator but we are not concerned with investigative skill in the current study. We assume a competent investigator. We also put barriers to traceability associated with time and distance outside the limits of this study. This is because traceability will be impaired regardless of the complexity of the supply network when more time has passed or information is more remote (Roth et al. 2008).
Supply Network Complexity
Operations management researchers have begun to theorize supply networks as complex systems (Choi et al. 2001; Madhavan et al. 2004; Pathak, Day, Nair, Sawya and Kristal 2007), involving many types of participants exchanging many classes of material items (Ford and Hakansson 2006) and even more types of information in transactions that are driven by a wide variety of considerations (Turnbull, Ford and Cunningham 1996). For the most part the typologies that have resulted from a network approach have concerned themselves with the ego networks of focal organizations, rather than with overall network complexity (Harland, Lamming, Zheng and Johnsen 2001; Choi and Krause 2006). For an excellent discussion of these typologies, the reader is referred to Harland et al. (2001).
We define supply network complexity by starting with Choi and Krause's (2006) definition of supply base complexity as the product of three factors: the number of suppliers, the differentiation of suppliers and the level of interrelationship between suppliers. Supply network complexity is defined as a function of the number of participants in the whole chain of relationships that ultimately connect consumers to the means of production for specific goods and services, the level of differentiation between participants, and the level and types of interrelationships that exist between participants. The complex supply network is the theater for the agency of the organizations participating in it. Making this change in level of analysis allows us to propose that organizations' ability to act independently depends on the structure of the overall network as well as on their local relationships (Turnbull et al. 1996; Ford and Hakansson 2006).
Larger supply networks are often more complex. They are more likely to have a larger number of process stages connected by more frequent exchanges. They are more likely to be populated by a wider variety of participants interacting in a greater variety of ways. Smaller networks are usually less complex. They usually have fewer stages and exchanges, involving fewer types of participants. When small networks have many stages, there are likely to be fewer participants at any stage, reducing the number of possible interactions and the potential for variation within participant classes. The size of the network thus contributes to network complexity in part as a constraint on the other two factors, participant variety and interrelationship variety.
All supply networks are populated by multiple classes of participants in exchange relationships. An assembler is different from its tier one suppliers and from its downstream distributors. Differences between classes make networks more complex because they expand the variety of possible exchange structures. For example, if there are only a few members of an upstream class exchanging with a numerous downstream class, bargaining power will be out of balance (Porter 1980; Havila, Johanson and Thileniusm 2004). In addition to differences between classes, supply networks are made more complex by differences within class: participating organizations are larger or smaller, local or offshore, technologically advanced or backward and well funded or strapped for cash.
Supply network participants interact and interrelate in a variety of ways that influence the structure of information exchange and thus transparency. Some supply relationships are persistent, highly specified or rigidly controlled, while others are poorly defined, open to ambiguity or highly adaptable. Variations are common because participants in supply networks are often related to each other in ways that do not directly involve exchange but nonetheless influence it. Participating entities may be related by ownership, family ties, relative power and status, or national or religious interest. Different types of participants may have similar relationship structures (e.g, tier 1 suppliers are connected by exchange to customers and tier 2 suppliers) but these structural norms may not be universal, as when a customer establishes a direct relationship to a tier 2 supplier. As the relationships that constitute a supply network become more varied, the supply network also becomes more complex.
Evaluating network complexity can be facilitated by expressing the supply network dimensions in more formal terms. One way to do this is through a census of the population of triads (Madhavan et al. 2004; Contractor, Wasserman and Faust 2006; Hanneman and Riddle 2009). A transitive triad has all three participants connected (A is connected to B, who is connected to C, who is connected to A), while an intransitive triad has an open link (A is connected to B, and A is connected to C, but B is not connected to C). Transitivity increases the number and variety of ways participants can interrelate (Havila et al. 2004; Wu and Choi 2005; Choi and Wu 2009), and this complexity increases when there are more types of participants so that there are more types of triads. A network composed mainly of intransitive triads is more likely to be sparsely connected and more likely to be populated by fewer types of participants.
The Choi and Krause (2006) idea of differentiation as a key to complexity can be extended by noting that there can be different motives for forming transitive triads. Madhavan et al. (2004) propose that there are two basic motives for transitive triad formation: countering and clustering. In countering, two of the parties develop a relationship to counter the greater bargaining power of the third party. For example, an end user and a second tier supplier separated by a first tier broker (Burt 2005) might develop a relationship to reduce the power of the broker (Choi and Wu 2009). In clustering, all three parties develop relationships in order to pool resources and create comparative advantage for the triad. A buyer working with two suppliers in product codevelopment might encourage the suppliers to also develop a relationship. Clustering and countering can also take forms involving more than triads. For example, all producers of a certain type may belong to an association that promotes standard practices or pools outputs for market control.
Normal Accident Theory and Supply Network Complexity
Perrow (1999) based normal accident theory on contrasts between systems dominated by complex or linear interactions and between systems characterized by tightly coupled or loosely coupled subsystems. He combines these two dimensions to develop the proposition that system accidents are likely to occur in complexly interactive, tightly coupled systems, regardless of the intentions of systems designers. He is primarily concerned with catastrophic failure, but both he and Sagan (1993) argue that catastrophic failures are only ordinary defects that run out of control in systems where detection and correction are rendered difficult by system structure and where tight coupling permits component failure to spread between subsystems. The underlying principle is that system structure creates barriers to the detection, comprehension and correction of variations in outcomes. In other words, normal accident theory is at the heart a theory of traceability.
Perrow (1999) argues that interactive complexity is important because it multiplies the opportunities for hard to understand accidents to occur. For Perrow, large, differentiated systems with subsystems that interact in a variety of ways are more likely to have unplanned, unexpected, unfamiliar, ambiguous or incomprehensible sequences, and are made obscure by complicated transformation processes, feedback loops and branching. Events in linearly interactive systems, with fewer, less differentiated components interacting in simple sequences are easier to predict and understand. Less complex systems are more easily controlled because interactions are simple, transformations are fewer and easier to understand and the number and variety of participants involved is less.
The other dimension of normal accident theory is the degree of coupling between system components. Weick defines coupling in terms of variables shared between subsystems: "Loose coupling occurs when two separate subsystems have few variables in common, or when the common variables are weak compared with other variables that influence the system" (1979, p. 111). Tight coupling occurs when subsystems share many variables or the common variables are influential. When coupling is tight, events in one subsystem influence processes in other subsystems in unintended and unexpected ways. Perrow (1999) extends this definition by proposing that tightly coupled systems will be characterized by hierarchical authority that creates unambiguous performance standards, low tolerance for delays, invariant sequences, rigid and little slack in supplies, equipment or personnel. Substitutions and buffers are limited. These characteristics, which are also associated with lean production systems (Dyer and Nobeoka 2000), can make tight coupling very efficient. Loosely coupled systems tend to have ambiguous or limited performance standards, allow for delays, do not have fixed sequences or relationships, retain slack resources and exploit fortuitous substitution possibilities.
As Perrow (1999) recognizes, these dimensions of system structure are not wholly independent. Interactive complexity makes tight coupling more difficult to achieve. The conclusion of normal accident theory is that attempting to realize this combination increases the risk of catastrophic system accidents. Scholars from the high reliability organizations school of safety theory have identified a limited number of examples of relatively safe, tightly coupled interactively complex systems. These tend to be systems that are costly to operate, segregated from the rest of the world and that operate under military or military like discipline, such as aircraft carriers, atomic power plants and space shuttle operations (Roberts and Rousseau 1989; Sagan 1993; Weick and Roberts 1993; Perrow 1999). These scholars argue that a centralized authority, isolation and strict discipline are required to make such systems possible. Even when national interest is not involved, tightly coupled complex supply networks require the participation of powerful players who act as the strategic centers of the extended enterprise (Roberts and Rousseau 1989; Lorenzoni and Baden-Fuller 1995; Harland et al. 2001). Dyer and Nobeoka's (2000) study of the Toyota supply network is an illustration of this principle. To be effective, the influence of strong central participants must ensure knowledge sharing while protecting less powerful participants from opportunism. As these authors point out, linear interactions make tight coupling easier to achieve, and because tightly coupled systems are often more efficient, loosely coupled or complex network systems tend to evolve toward tight coupling and linearity.
The Choi and Krause (2006) dimensions of supply network complexity map directly onto Perrow's construct of complexly interactive systems. Supply network complexity is a spectrum, anchored at the complex end by networks with many and varied transitive triads, evolving as participants maneuver by clustering and countering. In complex supply networks, both number and variety of transitive triads would be high, reflecting the many forms of interaction between participants. Complicated transformation processes, feedback loops and branching would be common. Interactions between heterogeneous participants in multiple modes based on heterogeneous material and information flows make defects and adverse events hard to predict, diagnose and trace. At the linear end of the spectrum would be networks with only a few types of participants and simple material and information flows within supply chains that are mostly segregated from each other. A linear interaction structure makes defects and adverse events easier to predict, diagnose and trace.
The difference between loose and tight coupling in the supply network context is essentially the same as in complex production systems. Just as different subsystems in a nuclear power plant, aircraft carrier, dam or traffic control system can be highly integrated and complexly interrelated, so that single events have influence across subsystem boundaries, so too in supply networks (Weick 1979). In contrast, loosely coupled subsystems are connected by sparsely defined interfaces that isolate different subsystems from each other (Sanchez and Mahoney 1996), permit easy substitution, allow for delays and allow for casual or intermittent relationships. We define tight coupling in supply networks in the following way: Supply relationships and supply networks are loosely coupled when participants base exchanges on a minimal specification of product or service attributes, or when the specification is unimportant relative to other goals, and tightly coupled when participants base exchanges on detailed specifications that are more important than other goals. In a tightly coupled supply network, parties are more likely to make strong commitments to enduring exchange relationships, including reciprocal transaction specific investments that increase the efficiency of the relationship (Alchian and Demsetz 1972; Uzzi 1997). Performance standards are likely to be unambiguous, delays not tolerated, methods well defined and slack severely limited. In a loosely coupled supply network, the lack of detail in exchange specifications will be associated with networks characterized by intermittent transactions, lack of commitment, delays, flexible sequences and methods and partner substitution.
While the form of coupling has obvious implications for transparency and traceability, the constructs are not coeval. In loosely coupled relationships, information exchange and retention are haphazard and retrieval is difficult. Both traceability and transparency should be weak in loosely coupled environment. In contrast, while in some tightly coupled relationships information exchange is standardized, unambiguous and enforced, it is plausible to argue that many tightly coupled parties will exchange relatively little information. This is because increasing levels of familiarity and trust, which often characterize enduring relationships based on shared goals, tend to be associated with decreasing levels of monitoring and formal information exchange (Weick 1979; Sagan 1993; Perrow 1999; Langfred 2004, 2007). Information may be retained for later retrieval as a matter of obligation, but not actually exchanged. This means that we should expect transparency to exhibit a nonlinear relationship with tight coupling, while traceability would tend to show a positive relationship.
We continue to refer to tight or loose coupling to maintain consistency with Perrow (1999); however, the formal construct we propose is the degree of tight coupling in a supply network, expressed as the level of detail in exchange specifications and the importance of specifications relative to the other goals of exchange participants.
A THEORETICAL MODEL OF NETWORK COMPLEXITY, TRANSPARENCY AND TRACEABILITY
The relationships between network complexity, the tightness of network coupling, transparency and trace-ability are shown in Figure 1, which depicts our theoretical model.
[FIGURE 1 OMITTED]
Network Complexity and Coupling
Like Perrow (1999) and Sagan (1993) we view network complexity and the degree of tight coupling in supply networks as distinct but not entirely orthogonal constructs. Tight coupling is more difficult to achieve in complex supply networks. The variety of participants and relationships in a complex network makes it more difficult to fully specify the exchange criteria that are the basis for tight coupling. Although we can easily imagine islands of tight coupling in a complex network, we think that very complex networks are likely to be characterized primarily by loose coupling. If managers pursue tight coupling, networks are likely to move toward a more linear structure. We therefore propose that there is a negative relationship between network complexity and the degree of tight coupling. We think that this relationship is reciprocal, as shown in Figure 1.
Proposition 1: The degree of tight coupling in a supply network is negatively related to network complexity.
Traceability and Network Complexity
Network complexity is negatively related to traceability. Multifarious participants and modes of connection interacting at multiple levels, engaged in many transformation processes, combine to make traceability more difficult to establish. In a complex network there are simply more opportunities for information to be lost, omitted or corrupted than there are in a linear network. For example, in networks rich in transitive triads, relationships in which two parties are trying to counter the power of a third party will lead both sides to withhold information from each other. Clustering relationships in which the resources of the triad are combined to create collaborative advantage will tend to produce joint outcomes that make inputs and contributions more difficult to trace. In linear networks the number and variety of participants and interactions is reduced, making intransitive triads more common and traces more likely to succeed.
Proposition 2: Supply network complexity is negatively related to traceability.
Traceability and Coupling
Coupling matters for traceability because the degree of tight coupling alters the likelihood that data integrity will be compromised. In tightly coupled networks product and service attributes are more likely to be fully specified and information exchange protocols are more likely to be detailed. Since most protocols are designed to catch relatively common errors, they facilitate traceability for those errors. Tightening specification to include more attribute information will enable the detection of rare errors, but as errors become rare, the payoff to the new investment, while positive, will be less pronounced payoff. In loosely coupled networks information exchange is likely to be informal, indirect and ad hoc. Traceability will therefore be unambiguously positively related to the degree of tight coupling in a supply network.
Proposition 3: The degree of tight coupling in a supply network is positively related to traceability.
Transparency at the network level depends partly on network complexity and the extent of tight coupling. As with traceability, network complexity makes it less likely that the network as a whole will be characterized by transparency. Participants in complex networks often work with numerous counter-parties in a wide variety of exchanges governed by a variety of modes of relationship. This variety means that while some relationships may be transparent, most will not because the conditions for establishing transparency are not met by short run, ephemeral relationships. For example, transparency in transitive triads would require trust to develop in all three relationships. This will be more difficult because the motives for transitive triad formation balance countering the competitive advantage of one of the parties against clustering or pooling resources to create competitive advantage for the triad (Madhavan et al. 2004). Countering obviously makes trust less likely throughout a triad be* cause two members "gang up" on the third. Since successful clustering creates rents that need to be divided, even a cooperative motive for complex relationships has the potential to reduce transparency.
Proposition 4: Supply network complexity is negatively related to transparency.
Transparency and Coupling
Transparency is positively related to the degree of tight coupling--up to a point. That point of inflection arises because, by specifying in detail the kinds of information to be reported, network participants may conclude that they are reporting everything that needs to be reported. Strong specifications may create a translucent exchange relationship in which managers overlook or misinterpret unanticipated causes, or suppress or overlook information that will be of use in understanding infrequent or unanticipated adverse events (Sagan 1993; Sanchez and Mahoney 1996; Perrow 1999). Bowker and Star (1999) argue that infrastructure classification schemes like the International Classification of Diseases are subject to political influences that lead to the suppression of information about new diseases--examples of this are the infrastructure related tracing problems that occurred during the West Nile virus (Weick 2005) and SARS outbreaks, and the reporting problems that continue to plague HIV/AIDS tracing in many countries. In addition, if tight coupling is associated with higher levels of trust, there may be a decrease in monitoring and formal information exchange (Sagan 1993; Perrow 1999; Langfred 2004, 2007).
We propose that transparency is likely to be positively influenced as coupling begins to become tighter. As relationships move from opaque to translucent, they are likely to expand to include new types of information about additional attributes. As coupling becomes tighter and trust increases, it may begin to restrict the transparency of a relationship. When counterparties rigidly specify the information they exchange, or monitor each other less, we think it becomes less likely that additional sensitive information will be revealed. As a result, coupling can create opportunities for adverse events to go undetected. This would suggest that the relationship between transparency and tight coupling is asymptotic, increasing initially and then leveling off.
Proposition 5: The degree of tight coupling in a supply network is related to transparency such that the relationship has an inverted U shape, increasing up to some point and then leveling off as coupling becomes increasingly tighter.
Transparency and Traceability
Transparency is positively related to traceability, rather than vice versa. We argue that transparency partly determines traceability because making more information more readily available will clearly make tracing adverse events easier. Having high levels of traceability, however, does not necessarily make a network more transparent because an adverse event might be easily traced in a linear, loosely coupled system because the network is simple and short.
Proposition 6: The transparency of a supply network is positively related to traceability.
This model predicts a state space in which the tightly coupled, complex network part of the space is restricted. Traceability will be strongest in tightly coupled linear networks and fairly strong in some loosely coupled linear networks, due to their simplicity, and sometimes transparency. Traceability will be moderately strong in any tightly coupled complex networks that exist, and will be weakest in loosely coupled complex networks.
FOOD SUPPLY NETWORKS AND TRACEABILITY
The remaining challenge for this paper is to demonstrate that this framework provides insights into the problem of traceability as it occurs in the real world. In this section of the paper, we apply Perrow's (1999) two dimensional matrix of system types to food supply networks, illustrate each quadrant of the framework with examples, and discuss the barriers to traceability present in each.
Traceability has become a dominant topic in the literature on the global food supply (Charlier and Valceschini 2007; Roth et al. 2008; Stone and Richtel 2009). Trace-ability matters because when adverse events occur in the food supply, consumers become ill or die, markets are disrupted, risks of lawsuits increase and the value of brands is eroded. When the causes of adverse events can be quickly and convincingly traced, economic consequences can be limited (Golan et al. 2008). Although there are supply networks where traceability is desirable because it verifies prestigious origin or improves supply chain management (Pouliot and Sumner 2008), trace-ability is a concern in the food supply primarily because of its role in limiting the consequences of adverse events (U.S. Food and Drug Administration [USFDA] 2008a).
Figure 2 depicts the two dimensional space defined by the constructs of supply network complexity and the degree of tight coupling, as applied to food supply networks. The matrix is broken into four quadrants, representing archetypes of food supply networks. On the horizontal axis, supply networks are depicted as linear or complex, while on the vertical axis they are represented as loosely or tightly coupled. Because convention argues for a square matrix, while our propositions suggest that network complexity and the degree of tight coupling are not orthogonal, we truncate the state space by excluding the unpopulated region shown in the upper right of Figure 2.
[FIGURE 2 OMITTED]
The archetype of a tightly coupled complex supply network would consist primarily of transitive triads, made up of many types of actors, arranged in networks in which connections are relatively dense and governed by rigorously enforced contracts specifying detailed information flows. Our research has not discovered an example of a food supply network that precisely fits this archetype, which makes sense given the competing pressures of network complexity and tight coupling we describe in Proposition 1. In a tightly coupled, complex network, traceability will be facilitated by tight coupling and impaired by network complexity. Transparency will tend to be reduced relative to the sheer amount of attribute information available for exchange, as participants are likely to rely on formal, standardized data formats. This means, following Proposition 6, that transparency should contribute minimally to traceability for this archetype. The influence of coupling and structure on traceability should be a mixed proposition, with familiar types of adverse events easily detected and traced, and novel or unfamiliar types very hard to trace.
The food supply networks that come closest to this archetype are characterized by the participation of large, powerful players such as global food companies or global food retailers who try to detail complex specifications for many suppliers and multiple processing stages. Since tight coupling in a complex network requires information intensive relationships with a greater variety of counterparties, there is a risk of information overload for the central actor. As illustrated in reports concerning Wal-Mart's recently announced sustainability and food safety initiatives (Rosenbloom 2008) this kind of effort is likely to result in an evolution from a loosely coupled complexly interactive supply network to one with fewer suppliers operating in more tightly coupled linear relationships.
An example of a typical adverse event in a fairly complex, fairly tightly coupled food supply network is the 2007 recall of several hundred thousand units of a pasta and meatball product by ConAgra (USFDA 2008b). The product is complex and its ingredients are sourced from a wide variety of suppliers, each of which has its own complex supply base. Supply bases often overlap. The product and its ingredients go through several combinations and transformations. Exchange specifications appear to be formalized, standardized and strongly enforced. Like many other adverse events in this kind of network, the problem was detected by the producer and confirmed by information exchange with the suppliers involved. This pattern is characteristic of the majority of recalls of processed food products in the United States and European Union, and results from shared investments in testing protocols that detect familiar types of adverse events.
As an archetype, a tightly coupled linear supply network would consist primarily of intransitive triads (such as retailer, processor, or grower) arranged in chains that are only sparsely linked to each other, where relationships are governed by reciprocal investments and enforced contracts that specify detailed information flows. Because information flows and material flows are tightly linked, data are more likely to retain their integrity. Traceability will be easier in these networks because tight coupling enables strong information exchange and the linear network structure reduces the information management problem and improves transparency. Better transparency should make unfamiliar adverse events easier to trace in this archetype.
An example of this type of network is the California supply network for fresh packaged leafy greens (California Food Emergency Response Team 2008). In this network, growers from multiple regions supply a processor/packer without having ties to each other, except through the packer. The processor/packer supplies multiple retailers, who are not related to each other or to the growers. Supply networks for different products (e.g., salad mixes) sometimes intersect at the producer/packer stage. Because this supply chain is dominated by retailers who support upstream investments in information systems, coupling is fairly tight. As a result, most defects are either detected in quality assurance processes or minimized by processing.
The Dole fresh bagged baby spinach E. coli O157:H7 contamination event of 2006 is an example of a catastrophic food safety event in this type of supply chain. A number of people died from E. coli Ol57:H7 infections while many others became severely ill. Product from a single packer was distributed through a wide variety of retailers, who share commitments to transparency with processors, growers and regulators. Traceability was supported by high levels of transparency and through the widespread use of data capture systems using standardized formats. "Using the product codes on the bags, and employing DNA fingerprinting on the bacteria from the bags, the investigators were able to match environmental samples of E. coli O157:H7 from one field" (USFDA 2008c).
In contrast, a loosely coupled linear supply network would consist of mostly intransitive triads, without the constraint of detailed specifications or committed relationships. In a network of this type the simplicity of the network can make traceability easy, while loose coupling means that information and material flows are less likely to be formalized or enforced. These networks are the most likely to be transparent since linearity increases transparency and transparency is positively affected when coupling is relatively loose. In aggregate, traceability will therefore be moderately easy and will depend strongly on transparency.
Examples of this type of network can be found in the supply of locally produced foods, particularly in spot markets and seasonal products. There are typically only a few types of participants in these networks, usually small growers, processors and consumers who interact casually, without necessarily developing relationships or specifications for exchange. We have to be careful to distinguish linear, casual, loosely coupled networks from equally local and linear artisanal networks with strong goal alignment that enhances transparency, such as the "farm to table" local supply movement (Paxson 2008; Stone and Richtel 2009). In an artisanal network, a distributor or retailer of premium shellfish or fine wine benefits by making commitments to his or her suppliers and aligning with their goals, which creates an implicit strong specification. This produces a slightly more tightly coupled, transparent system in which information is accessible as part of the appeal to the consumer, rather than as the result of formal specifications. This type of network is formalized in the European system of Appellation d'Origine Controlee, in which geographic limits on the origin of produce favor linear, transparent supply relationships.
In a casual network, buyer and supplier can be oblivious to each others' goals. Because these networks tend to be local, the adverse events resulting from them also tend to be local. The 2008 listeriosis outbreak in Quebec associated with locally produced raw milk cheeses is a good example. Provincial regulations to permit the general sale of locally produced raw milk cheeses were established in the summer of 2008. The change in regulations was quickly followed by multiple incidents of Listeria poisoning (White 2008). This occurred despite the fact that the new regulations were designed to enforce an artisanal set of goals in what had previously been casual networks of commercial producers who bought homogenized milk in a spot market. The regulations required each cheese maker to know his or her milk suppliers, to align goals around higher standards of cleanliness than the casual network had been accustomed to and to implement rigorous testing protocols that the casual network had not required. Because these regulations were alien to the producers and regulators who entered raw milk cheese production and because the relationships were new, goals and actions were not aligned across the stages of the supply network, setting the stage for the outbreak.
The archetype of a loosely coupled complex supply network would consist primarily of transitive triads, made up of many types of actors, arranged in networks in which connections are relatively dense, and connected by relationships that are intermittent and governed by minimal, flexible specifications. Loose coupling and complexity combine to interfere with traceability. In these networks very moderate increases in the tightness of coupling will enhance transparency, but at the same time, complexity will impair it. To the extent that it can be achieved, transparency will improve traceability. In a complex network transparency is confounded by the problem of knowing where in the network to begin looking.
This is perhaps the most common type of food supply network and is the type in which traceability is weakest. An example of this type of network is the broker-based network for bulk fresh produce. In broker networks growers, carriers, processors and distributors enact supply networks designed to accommodate a large variety of products that come in varying degrees of quality. Products and subsystems interact in a variety of ways, specifications are minimal, delays are common and substitutions are frequent.
A specific example that illustrates the difficulty of tracing contamination through a complex loosely coupled network is the recent Salmonella saintpaul contamination in the United States. This outbreak was first associated by investigators with tomatoes and then with peppers (USFDA 2008d). The cause of this outbreak was ultimately traced to a sample of irrigation water from a Serrano pepper farm in Tamaulipas, Mexico. Although epidemiologists were aware early on that they were dealing with one strain of Salmonella saintpaul, indicating a common source, the complexity of the network led investigators to initially implicate the most commonly exchanged product in the network, tomatoes. This led to the loss of several weeks of production for the tomato industry, as consumers were warned away from the wrong product. Only when a tomato source could not be identified did the investigation begin to turn toward Serrano and Jalapeno peppers and eventually to the identification of the contamination source. Similar recent examples of catastrophic hard to trace events in complex loosely coupled networks include the adulteration of global supplies of milk powder and animal feed with melamine (Kaufman and Weiss 2007; Chao 2008) and the recent peanut butter Salmonella contamination in the United States (Moss 2009).
DISCUSSION AND CONCLUSION
Our model of the relationship between traceability and supply network complexity is an additive one in which transparency partly mediates the main relationships. While it is tempting to propose that network complexity moderates the effect of tight coupling, we think that the relationship between network complexity and coupling is more fundamental. Network complexity and tight coupling constrain each other, making the rightly coupled complex supply network a nearly unattainable ideal unless both isolation and strict discipline can be achieved. Our research suggests that these conditions are not regularly met in the food supply, and may be equally rare in other supply networks.
The range of possible complex supply network types is also under pressure in the lower left quadrant. In the food supply and other networks, casual loosely coupled linear supply networks are the site of many adverse events, leading to additional pressure to move toward a more tightly coupled state. In the food supply, regulators recognize that standards like the Hazard Analysis and Critical Control Point (HACCP) methodology or ISO 22000 are less likely to be implemented in small organizations and local networks (Paxson 2008). In response they have attempted to develop alternative standards and protocols (Food Safety and Inspection Service 2008; Stone and Richtel 2009) that would lead to more goal alignment and tighter coupling and help avert incidents like the Ontario listeriosis event.
One of the most important implications of this article for managers is that traceability systems must be tailored to the supply networks they are deployed in. Formal information systems are uncommon in loosely coupled linear networks, in part because simple linear patterns of relationship mean that such networks are more likely to be characterized by transparency. Formal systems are less necessary because local, linear, casual networks tend to be the simplest of all. In a farmer's market supply network or an AOC one, there are very few steps between the raw material and the final consumer, which make relying on less formal systems like "farm to table" values more effective than implementing formal systems in terms of traceability.
In contrast, tightly coupled linear networks present the fewest challenges to systems developers. It therefore comes as no surprise that effective traceability systems are already common in these networks. Systems for these relatively simple networks do not necessarily scale up to meet the challenges of complex supply networks, including methods for preserving information through transformation processes, comingling of shipments and the other information degrading processes that characterize complex networks. In relatively tightly coupled networks, reducing complexity and assigning an enforcement role to a central player would make trace-ability systems possible, but only at a high cost in terms of system resources, data quality and the opportunity costs of committing to a smaller supplier base for any product. In a loosely coupled complex network, we think information systems-based solutions are limited to very simple logistics approaches. There is a global consortium designing a standard protocol for logistics traceability based on RFID tags (Agrawal et al. 2009), but systems that follow this protocol will do little more than permit tracking between transformation points. While it might be possible to make inferences about transformations and other complex interactions from a global picture of logistics data, doing so would require access to sensitive data about multiple products and suppliers. The more intrusive these systems are, the more traceability needs collide with the costs of transparency.
Some scholars have argued that transparency, rather than traceability is the critical idea in complex supply networks (Lamming et al. 2001, 2004; Roth et al. 2008). If transparent access to information was a universal condition of doing business, so that the costs of being open were borne by the system rather than disproportionately by voluntary participants, this might be a valid position. Achieving this condition would seem to require an agency that is capable of universally effective enforcement. Although regulatory agencies worldwide make the food supply much safer than it would otherwise be, we do not think any of these agencies exhibit this capability. Complex global trade, property rights, competing interests, political influence and naked corruption all interfere. Scholars have argued that if organizations would only recognize that it is in their own interest to be transparent, the situation would improve. This is the basis for the idea that central players should reward counterparties who cooperate. One consequence of rewarding cooperative partners with more business is that this will tend to move complex supply networks away from relatively efficient open markets toward more linear, quasivertically integrated networks. Whether or not this is a good policy choice, it does seem to be an idea that is gaining currency among the very large organizations that have the most to lose from adverse food safety events.
One goal of a theoretical study like this one should be to frame the theory in such a way that it lends itself to being tested against reality. Data on adverse events in the food supply and the processes of tracing their causes are maintained by regulatory agencies as public records, so that operationalizing a dependent variable such as ability to trace would seem fairly straightforward. The major challenge for testing this theory would be in operationalizing supply network complexity and coupling. We have indicated how triad census methods could be used to evaluate network complexity, and transaction cost oriented research has developed methods for measuring the degree of coupling.
In practice, accurate information on the number of participants in a network and the relationships among them is likely to become more difficult to acquire as networks become more complex or more loosely coupled. This would suggest limiting tests of the theory to relatively narrow segments of the population of networks, such as the comparison between tightly coupled and loosely coupled linear networks. Doing so would make research more feasible, but would constitute a very limited test of the theory. An alternative might be to begin with the raw materials end of the network, focusing on products that feed into a variety of supply networks, with the goal of predicting variations in traceability across the range of networks. The challenge here would be in finding data related to a source material that has both a wide range of uses in a variety of networks and is associated with a significant number of adverse events, such as ground beef.
Another avenue for future research would be to investigate the actual variety of supply network structures present in the global economy. Our theory, borrowing from normal accident theory, proposes that complex tightly coupled networks will be rare. A question we have yet to ask, which a supply network census could answer, is whether the complex loosely coupled network is actually common. If these networks are less common than we think, the barriers to improving traceability would be less difficult to overcome. Given the regulatory pressure that is applied to loosely coupled linear networks and the presence of major participants with an interest in trace-ability, it is also possible that the population of food supply networks is evolving toward a much more homogenous, less complex, more tightly coupled, safer range of structures. Detecting such an evolution, or resistance to it, would be an interesting contribution.
In conclusion we have argued that supply network complexity, the degree of tight coupling in supply networks and the transparency of buyer-supplier relationships play a role in the ability of users to identify and verify the components and chronology of events in all of the steps of a process chain. While our theory, along with normal accident theory, might seem to suggest that traceability in complex networks is a lost cause, we do not think so. The fact that catastrophic adverse events in the food supply and elsewhere are so heavily reported is partly a function of their actual rarity, given the enormous number of transactions that occur each day in global supply networks. Managers and scholars should be concerned with these rare events because they have important consequences in themselves, and because they represent the tip of the iceberg of minor defects and near misses that occur. Taken cumulatively minor events and ordinary defects that are hard to trace represent major inefficiencies in supply networks. This study makes its primary contribution to the goal of greater traceability by trying to show how the conditions that enable or impede traceability emerge from supply network complexity, which should help direct attention to solutions that will lead to continuous improvement even in the most complex of supply networks.
Agrawal, R., A. Cheung, K. Railing and S. Schonauer 'Towards Traceability across Sovereign, Distributed RFID Databases," 2009. Available at http://www.almaden.ibm.com/cs/projects/iis/hdb/Publications/papers/ideas06.pdf (Accessed March 20, 2009).
Alchian, A. and H. Demsetz. "Production, Information Costs, and Economic Organization," American Economic Review, (62), 1972, pp. 777-795.
Bowker, G.C. and S.L. Star. Sorting Things Out: Classification and Its Consequences, MIT Press, Cambridge, MA, 1999.
Burt, R.S. Brokerage and Closure: An Introduction to Social Capital, Oxford University Press, Oxford, NY, 2005.
California Food Emergency Response Team. "Investigation of an Escherichia coli O157:H7 Outbreak Associated with Dole Pre-Packaged Spinach," 2008. Available at http://www.dhs.ca.gov/fdb/Iocal/PDF/2006%20Spinach%20Report%20Final%20redacted. PDF (Accessed November 12, 2008).
Chao, L. "China Bolsters Dairy Supply Oversight in Effort to Rebound from Scandal." The Wall Street Journal, October 7, 2008, p. A19.
Charlier, C. and E. Valceschini. "Coordination for Traceability in the Food Chain. A Critical Appraisal of European Regulation," European Journal of Law and Economics, (25), 2007, pp. 1-15.
Cheng, M.J. and J.E.L. Simmons. 'Traceability in Manufacturing Systems," International Journal of Operations and Production Management, (14:10), 1994, pp. 4-16.
Choi, T.Y., K.J. Dooley and M. Rungtusanatham. "Supply Networks and Complex Adaptive Systems: Control Versus Emergence" Journal of Operations Management, (19:3), 2001, pp. 351-366.
Choi, T.Y, and D.R. Krause. 'The Supply Base and Its Complexity: Implications for Transaction Costs, Risks, Responsiveness, and Innovation," journal of Operations Management, (24:5), 2006, pp. 637-652.
Choi, T.Y. and Z. Wu. 'Triads in Supply Networks: Theorizing Buyer-Supplier-Supplier Relationships," Journal of Supply Chain Management, (45), 2009, pp. 8-25.
Contractor, N.S., S. Wasserman and K. Faust. 'Testing Multitheoretical, Multilevel Hypotheses about Organizational Networks: An Analytic Framework and Empirical Example," Academy of Management Review, (31), 2006, pp. 681-703.
Dyer, J.H. and K. Nobeoka. "Creating and Managing a High-Performance Knowledge-Sharing Network: The Toyota Case," Strategic Management Journal, (21), 2000, pp. 345-367.
Folinas, D., I. Manikas and B. Manos. 'Traceability Data Management for Food Chains," British Food Journal, (108), 2006, pp. 622-633.
Ford, D. and H. Hakansson. "IMP--Some Things Achieved: Much More to Do," European Journal of Marketing, (40), 2006, pp. 248-259.
Food Safety and Inspection Service. "Small and Very Small Plant Outreach," 2008. Available at http://www.fsis.usda.gov/Science/Small_Very_Small_Plant_Outreach/index.asp (Accessed November 12, 2008).
Golan, E., B. Krissoff, F. Kuchler, L. Calvin, K. Nelson and G. Price "Traceability in the U.S. Food Supply: Economic Theory and Industry Studies," 2008. Available at http://www.ers.usda.gov/Publications/AER830/ (Accessed November 15, 2008).
Griffin, A. "Evaluating QFDS Use in United-States Firms as a Process for Developing Products," Journal of Product Innovation Management, (9), 1992, pp. 171-187.
Hanneman, R. and M. Riddle "Introduction to Social Network Methods," 2009. Available at http://faculty.ucr.edu/~hanneman/ (Accessed March 20, 2009).
Harland, C.M., R.C. Lamming, J. Zheng and T.E. Johnsen. "A Taxonomy of Supply Networks", Journal of Supply Chain Management, (37:4), 2001, pp. 21-27.
Havila, V., J. Johanson and P. Thileniusm. "International Business-Relationship Triads," International Marketing Review, (21), 2004, pp. 172-186.
Jones, C, W. Hesterly and S. Borgatti. "A General Theory of Network Governance: Exchange Conditions and Social Mechanisms," Academy of Management Review, (22), 1997, pp. 911-945.
Juran, J. Quality Control Handbook, 3rd ed., McGraw-Hill, New York, 1979.
Kaufman, M. and R. Weiss "FDA to Test Imported Additives for Melamine," The Washington Post, April 25, 2007, p. A07.
Lamming, R., N. Caldwell and D. Harrison. "Developing the Concept of Transparency for Use in Supply Relationships," British Journal of Management, (15), 2004, pp. 291-302.
Lamming, R., N. Caldwell, D. Harrison and W. Phillips. 'Transparency in Supply Relationships: Concept and Practice," Journal of Supply Chain Management, (37:4), 2001, pp. 4-11.
Langfred, C.W. 'Too Much of a Good Thing? Negative Effects of High Trust and Individual Autonomy in Self-Managing Teams," Academy of Management Journal, (47), 2004, pp. 385-399.
Langfred, C.W. 'The Downside of Self-Management: A Longitudinal Study of the Effects of Conflict on Trust Autonomy, and Task Interdependence in Self-Managing Teams," Academy of Management Journal, (50), 2007, pp. 885-900.
Lee, H.L. "Aligning Supply Chain Strategies with Product Uncertainties," California Management Review, (44:3), 2002, pp. 105-119.
Linderman, K., R.G. Schroeder, S. Zaheer and A.S. Choo. "Six Sigma: A Goal-Theoretic Perspective," Journal of Operations Management, (21), 2003, pp. 193-203.
Lorenzoni, G. and C. Baden-Fuller. "Creating a Strategic Center to Manage a Web of Partners," California Management Review, (37:3), 1995, p. 146.
Madhavan, R., D.R. Gynwali and J. He.'Two's Company, Three's a Crowd? Triads in Cooperative-Competitive Networks," Academy of Management Journal, (47), 2004, pp. 918-927.
Moss, M. "Peanut Case Shows Holes in Safety Net," The New York Times, February 9, 2009, p. Al.
Pathak, S.D., J.M. Day, A. Nair, W.J. Sawya and M.M. Kristal. "Complexity and Adaptivity in Supply Networks: Building Supply Network Theory Using a Complex Adaptive Systems Perspective," Decision Sciences, (38), 2007, pp. 547-580.
Paxson, H. "Post-Pasteurian Cultures: The Micro-biopolitics of Raw-Milk Cheese in the United States," Cultural Anthropology, (23), 2008, pp. 15-47.
Perrow, C. Normal Accidents: Living with High Risk Technologies, Princeton University Press, Princeton, NJ, 1999.
Porter, M.E. Competitive Strategy, New York Free Press, New York, 1980.
Pouliot, S. and D. Sumner. 'Traceability, Liability, and Incentives for Food Safety and Quality," The American Journal of Agricultural Economics, (90), 2008, pp. 15-27.
Ramesh, B. "Factors Influencing Requirements Traceability Practice," Communications of the ACM, (41:12), 1998, pp. 37-44.
Roberts, K.H. and D.M. Rousseau. "Research in Nearly Failure-Free, High-Reliability Organizations: Having the Bubble," IEEE Transactions on Engineering Management, (36:2), 1989, pp. 132-139.
Rosenbloom, S. "Wal-Mart to Toughen Standards," The New York Times, October 22, 2008, p. Bl.
Roth, A.V., A.A. Tsay, M.E. Pullman and J.V. Gray. "Unraveling the Food Supply Chain: Strategic Insights from China and the 2007 Recalls," Journal of Supply Chain Management, (44), 2008, pp. 22-40.
Sanchez, R. and J. Mahoney. "Modularity, Flexibility and Knowledge Management in Product and Organization Design," Strategic Management Journal, (17:Winter Special Issue), 1996, pp. 63-76.
Sagan, S.D. The Limits of Safety: Organizations, Accidents and Nuclear Weapons, Princeton University Press, Princeton, NJ, 1993.
Stone, B. and M. Richtel "Forging a Hot Link to the Farmer Who Grows the Food," The New York Times, March 28, 2009, p. Bl.
Turnbull, P., D. Ford and M. Cunningham. "Interaction, Relationships and Networks in Business Markets: An Evolving Perspective," The Journal of Business and Industrial Marketing, (11:3/4), 1996, pp. 44-60.
U.S. Food and Drug Administration (USFDA). "Guidance for Industry: Guide to Minimize Microbial Food Safety Hazards of Fresh-cut Fruits and Vegetables," 2008a. Available at http://www.cfsan.fda.gov/~dms/prodgui4.html (Accessed October 15, 2008).
U.S. Food and Drug Administration (USFDA). "Pennsylvania Firm Recalls Pasta and Meatball Meals Due to Possible Underprocessing," 2008b. Available at http://events/Recall_011_2007_Release/index.asp (Accessed November 12, 2008).
U.S. Food and Drug Administration (USFDA). "FDA Finalizes Report on 2006 Spinach Outbreak," 2008c. Available at NEWS/2007/NEW01593.html (Accessed November 12, 2008)
U.S. Food and Drug Administration (USFDA). "Salmonella saintpaul Outbreak," 2008d. Available at http://www.fda.gov/oc/opacom/hottopics/tomatoes.html (Accessed October 16, 2008).
Uzzi, B. "Social Structure and Competition in Inter-firm Networks: The Paradox of Embeddedness," Administrative Science Quarterly, (42:1), 1997, pp. 35-67.
Weick, K.E. The Social Psychology of Organizing, 2nd ed., McGraw-Hill, New York, 1979.
Weick, K.E. "Managing the Unexpected: Complexity as Distributed Sensemaking." In R.R. McDaniel Jr. and D.J. Driebe (Eds.), Uncertainty and Surprise in Complex Systems, Springer, Berlin, NY, 2005, pp. 51-65.
Weick, K.E. and K.H. Roberts. "Collective Mind in Organizations: Heedful Interrelating on," Administrative Science Quarterly, (38:3), 1993, pp. 357-381.
White, M. "More Cheeses Recalled In Quebec Listeriosis Outbreak," 2008. Available at http://www2.canada.com/montrealgazette/news/story.html?id=b019el3c-710e-4f0c-9680-28790ffl59aa (Accessed November 12, 2008).
Wu, Z. and T.Y. Choi."Supplier-Supplier Relationships in the Buyer-Supplier Triad: Building Theories from Eight Case Studies," Journal of Operations Management, (24), 2005, pp. 27-52.
Paul F. Skilton (Ph.D., Arizona State University) is an assistant professor in the Morrison School of Management and Agribusiness at Arizona State University--Polytechnic in Mesa, AZ. Dr. Skilton's principal area of research is the strategic management of project-based enterprises. Recently he has focused on strategic issues in the management of creative projects, especially questions about the nature and structure of supply relationships. This includes research on the effects of network structure on supply relationships, supplier bargaining power, offshore supply relationships for knowledge-based services and the role of social capital in supplier careers.
Jessica L. Robinson (Ph.D., Candidate) is a second year doctoral candidate at Arizona State University--Polytechnic in Mesa, AZ. Her concentration is in agribusiness with a primary focus in supply chain management. Having prior education and work experience relating to manufacturing-specific supply chain, her research interests consist of incorporating manufacturing-based supply chain strategies with food supply chains, particularly in view of food safety concerns.
PAUL F. SKILTON AND JESSICA L. ROBINSON
Arizona State University-Polytechnic
|Printer friendly Cite/link Email Feedback|
|Author:||Skilton, Paul F.; Robinson, Jessica L.|
|Publication:||Journal of Supply Chain Management|
|Date:||Jul 1, 2009|
|Previous Article:||Triads in services outsourcing: bridge, bridge decay and bridge transfer *.|
|Next Article:||Investigating population and topological evolution in a complex adaptive supply network *.|