Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
INTERDEPENDENCE AND NETWORK-LEVEL TRUST IN SUPPLY CHAIN NETWORKS: A COMPUTATIONAL STUDY USING THE NK FRAMEWORK Antonio Capaldo S.E.GEST.A. Department of Management Catholic University of the Sacred Heart 1, Largo Gemelli – Milan 20123, Italy e-mail: [email protected] Ilaria Giannoccaro Department of Mechanical and Management Engineering Polytechnic University of Bari 182, Viale Japigia – Bari 70126, Italy e-mail: [email protected] Paper submitted to Decision Sciences Acknowledgements Earlier versions of this paper were accepted for presentation at the 2010 Strategic Management Society Annual International Conference and at the 2011 Academy of Management Annual Meeting. We thank two anonymous AOM reviewers for useful comments. 1 INTERDEPENDENCE AND NETWORK-LEVEL TRUST IN SUPPLY CHAIN NETWORKS: A COMPUTATIONAL STUDY USING THE NK FRAMEWORK Antonio Capaldo and Ilaria Giannoccaro ABSTRACT We investigate the impact of the supply chain interdependence structure on network-level trust in the supply chain (i.e., supply chain trust). We employ the NK framework to model the supply chain network as a set of interdependent decisions interacting among each other according to a specific pattern reflecting the overall supply chain interdependence structure. We argue that supply chain networks can reveal in practice the 10 patterns identified by Rivkin and Siggelkow (2007) in a recent study on patterned interactions in complex systems. Thus, we perform computational analysis to evaluate, for each considered interdependence pattern, the risk of opportunism by the participating firms. This allows us to compare the patterns on the level of supply chain trust. We show that supply chain trust is a positive (negative) function of the number of uninfluenced (uninfluential) partners, that are, partner firms whose decisions are not influenced by (do not influence) the decisions made by the remaining partners. We also find that, for each examined pattern, the higher the degree of interdependence in the supply chain, the lower supply chain trust. INTRODUCTION Supply chains are networks of interfirm alliances among interdependent firms operating along an industry value chain. As competition is increasingly being fought supply chain vs. supply chain, managing the overall supply chain network becomes vital to competitive success (Greis & Kasarda, 1997; Ketchen & Giunipero, 2004). The supply chain is indeed more than its composing dyads, and looking at it as a sum of dyadic relationships does not allow managers and researchers to take into account the complex web of interdependencies which characterizes the supply chain and influences behaviors and performance outcomes in real-world supply chains (Nair, Narasimhan, & Choi, 2009). Thus, doing impactful supply chain management research may benefit from shifting the level of analysis from the dyad to the overall network. A critical element for achieving effective supply chain management resides in establishing and nurturing trust among the participating organizations (Handfield & Bechtel, 2002; Panayides & Lun, 2009). A significant literature has indeed pointed out the beneficial impact of trust on 2 supply chain management and showed that trust is a powerful antecedent of effective cooperation and a significant predictor of positive performance outcomes and competitive advantage in supply chain alliances (Kumar, 1996; Monczka, Petersen, Handfield, & Ragatz, 1998; Dyer & Chu, 2003; Johnston, McCutcheon, Stuart, & Kerwood, 2004). However, most studies on trust in supply chain contexts have focused on dyadic relationships (Zaheer, McEvily, & Perrone, 1998; Dyer & Chu, 2003; Johnston et al., 2004; Laaksonen, Jarimo, & Kulmala, 2009). Conversely, there has been a paucity of research on trust at the level of the overall supply chain network. For the purposes of this study, network-level trust (i.e., supply chain-level trust, or simply supply chain trust) expresses the diffusion of trust within the network, that is, the extent to which trust is pervasive across the supply chain. In line with the need of adopting a network (vs. dyadic) perspective to supply chain research, the present study focuses on supply chain-level trust. This resonates with previous literature suggesting that, in order to reach a better understanding of how trust affects supply chain outcomes, we need to concentrate on network-level trust (Ireland & Webb, 2007). Indeed, when trust is pervasive across the supply chain, ideas, knowledge, products, and services can freely flow to help design, manage, and perform processes and activities aimed at creating value, with positive effects on several performance outcomes (Uzzi, 1997; Dyer & Singh, 1998; McCarter & Northcraft, 2007). Commenting upon supply chain management challenges, Galaskiewicz (2011: 6) has put it this way: “the management problem is how to create a social network that would enable the parties to trust one another up and down the supply chain”. To contribute to solve this puzzle, we focus on one major factor influencing trust in supply chains– that is, interdependence. Specifically, we investigate the differential impact of different possible supply chain interdependence patterns on supply chain-level trust. By doing so, we aim at providing useful insights for supply chain architects and, more generally, for those interested in interorganizational network building and management. Interdependence refers to whether, and the extent to which, two or more economic actors depend upon one another for product and process accomplishments and/or for strategically relevant resources and capabilities owned by their partners (Pfeffer & Salancik, 1978). Both types of interdependence occur recurrently in supply chains. While previous research has 3 typically looked at the impact exerted on trust by the degree (i.e., intensity) of interdependence, we focus on the overall pattern of interdependencies that characterizes the supply chain. We adopt an opportunism-based view of trust, according to which trust and opportunism are the opposite of one another (Barney & Hansen, 1994; Gambetta, 1988). Specifically, we assume that the higher the average risk of opportunism across the supply chain, the lower trust at the supply chain level. Based on this, we advance that the supply chain interdependence pattern affects supply chain-level trust by influencing whether, and the extent to which, the local interests of the supply chain members are aligned with the global interests of the supply chain, which in turn impact the risk that partners behave opportunistically. A notable feature of this study is its drawing on the literature on complex adaptive systems (CASs) (Holland, 1995). Previous scholars suggested that supply chains can be conceptualized as CASs (Choi, Dooley, & Rungtsusanatham, 2001; Surana, Kumara, Greaves, & Raghavan, 2005) and that doing so has the potential to move the field of supply chain management studies beyond a dyadic buyer–supplier framework (Pathak, Day, Nair, Sawaya, & Kristal, 2007), thus generating valuable insights into how to manage the overall supply chain network (Nair et al., 2009). However, Pathak, Day, Nair, Sawaya, and Kristal (2007) noted that this potential has been exploited by only a small number of scholars in the Operations Management and Supply Chain Management fields and urged the supply chain management research community to leverage the CAS perspective in order to deepen our understanding of the supply chain and its inherent complexity. They also argued that “recent advancements made by Rivkin and Siggelkow (2007) toward extending CAS research of organizations (…) using the NK model of fitness from theoretical biology (…) could have important lessons for the study of supply chains” (Pathak et al., 2007: 557) (see also Choi & Krause, 2006). Responding to Pathak, Day, Nair, Sawaya, and Kristal’s (2007) call, we draw on previous research, that has employed Kauffman’s (1993) NK framework to analyzing management problems (e.g., Ganco & Hoetker, 2009; Levinthal, 1997), to model the supply chain in terms of the 10 interaction patterns identified by Rivkin and Siggelkow (2007) in their study on patterned interactions in complex systems. We then perform computational analysis to measure, for each examined pattern, the risk of opportunism by the participating firms. This allows us to gauge the 4 level of supply chain trust associated with each pattern and compare the patterns on such a level. Based on our findings, we conclude that supply chain trust is positively (negatively) affected by the number of uninfluenced (uninfluential) partners, i.e., supply chain partners whose decisions are not influenced by (do not influence) the decisions made by the remaining partners. We also find that supply chain trust is negatively affected by the degree of interdependence. The primary purpose of this study is to contribute to our knowledge of trust in supply chain networks by focusing on the role of the supply chain interdependence pattern. Indeed, although we have a good theoretical understanding of the constructs of trust and interdependence, the relationships between them, and in particular the impact that interdependence patterns exert on trust at the supply chain level, need in-depth investigation. We thus look at the supply chain as a CAS and employ the NK framework to construct a computational representation of the supply chain as a network of interdependent decisions and related performance outcomes, from which we infer the probability that supply chain partners behave opportunistically, that in turn affects the level of supply chain trust. In so doing, in line with Davis, Eisenhardt, and Bingham (2007), we aim to develop fresh theoretical insights by leveraging the ability of computational analysis to enrich extant undeveloped theory. TRUST, OPPORTUNISM, AND INTERDEPENDENCE IN SUPPLY CHAIN NETWORKS Trust and Opportunism Trust has largely been examined in management research (e.g., McEvily, Perrone, & Zaheer 2003; Seppänen, Blomqvist, & Sundqvist, 2007; Wolf & Muhanna, 2011) and dozens of definitions have been offered in previous studies (Hosmer, 1995; Mayer, Davis, & Schoorman, 1995; Rousseau et al., 1998; Dirks & Ferrin, 2001). Trust occurs in economic exchange when economic actors are willingly vulnerable to the behavior of other parties because of expected cooperation or benevolence from them (Rousseau et al., 1998; McCarter and Northcraft, 2007). In the case of interorganizational relationships, trusting organizations are willing to depend on partner organizations whom they trust because they expect those trusted partners will not behave opportunistically (Bradach & Eccles, 1989; Barney & Hansen, 1994). This is consistent with the 5 idea of trust as the probability that economic actors will make decisions and take actions that will not be detrimental to others (Gambetta, 1988: 217). Following this line of reasoning, trust is the opposite of opportunism (Barney & Hansen, 1994: 176). In fact, empirical studies have found that trust and opportunism are substitutes for one another (Zaheer & Venkatraman, 1995; Dyer & Chu, 2003; Cavusgil, Deligonul, & Zhang, 2004). Williamson defined opportunism as “a lack of candor or honesty in transactions, to include self-interest seeking with guile” (Williamson, 1975: 9). In interorganizational relationships, opportunism occurs when one or more parties exploit the vulnerabilities of other parties to seek their own unilateral gains at the substantial expense of the other parties and/or of the overall relationship (Das, 2006; Luo, 2007; Das & Rahman, 2010). Opportunistic behaviors include, among others, violation of written contracts, failing to honor informal agreements, falsification of information, quality shirking, and breach of distribution contracts (see Wathne & Heide, 2000 for a comprehensive review of the original and emergent conceptualizations of opportunism). Network-level Trust in Supply Chain Networks The present paper focuses on trust among organizations involved in supply chain networks. Previous research that has investigated trust in networks has typically conceptualized trust as a dyadic construct. Thus, attention has been drawn to trust within the individual dyadic relationships the examined networks are composed of (e.g., Zaheer et al., 1998; Capaldo, 2007). This perspective has led to major insights concerning, among others, the influence of the overall network on trust in its component dyads (Granovetter, 1985; Coleman, 1990) and the impact exerted by trust-based dyadic relationships on network-level outcomes such as innovativeness and competitive advantage (Dyer & Nobeoka 2000). Conversely, little attention has been paid to network-level trust and a clear conceptualization of the construct is still lacking. Indeed, scholars have argued for the need of focusing on trust at the network level (Provan, Fish, & Sydow 2007) and for the importance of network-level trust in supply chain contexts (Ireland & Webb, 2007), without, however, providing a definition of network-level trust, let alone using it for research purposes. To fill these gaps, we concentrate on trust as a network-level construct and define network-level trust as the extent to which trust is pervasive across the network.. We contend that 6 trust is pervasive across a supply chain network when, and to the extent that, the interests of the supply chain partners are aligned with the interests of the overall network, and therefore partners will not make decisions detrimental to the supply chain. Thus, in accordance with the opportunism-based definition of trust outlined above, we infer the existence of trust from the lack of opportunism. Specifically, we assume that the lower opportunism by supply chain partners, the higher trust across the overall supply chain network (i.e., supply chain trust) (McCarter & Northcraft, 2007). This is consistent with extant studies arguing that, under high trust conditions, the proclivity of economic actors to be opportunistic decreases, whereas under low trust conditions economic actors are more likely to behave opportunistically (Williamson, 1975; Zaheer et al., 1998; Nooteboom, 2002). Interdependence and Trust in Supply Chain Networks “Management scholars use the term interdependence to suggest [that two or more parties] are dependent on each other to achieve their desired outcomes” (Wicks, Berman, & Jones, 1999: 104). Two conceptualizations of interdependence have been identified in previous studies (Victor & Blackburn, 1987; Astley & Zajac 1991). Workflow interdependencies occur in systems of interdependent activities wherein each system unit performs specialized activities that are instrumental to the achievements of the overall system. Resource interdependencies occur when two or more system units are dependent on the resources they receive from each other. In a seminal contribution, Thompson (1967) distinguished between interdependence occurring within the organization’s boundaries (i.e., internal interdependence) and interdependence of the organization with the task environment (i.e., external interdependence). We focus on external interdependence (simply interdependence, hereinafter) exclusively, and in particular on interdependencies occurring among organizations. Both workflow and resource interdependencies can occur among firms involved in interorganizational relationships (Blankenburg Holm, Eriksson, & Johanson, 1999). This is particularly so in supply chain networks, where firms are “operationally, strategically, and technologically integrated” (Hult, Ketchen, & Slater, 2004: 241) and therefore make large numbers of interacting decisions (Choi et al., 2001). Workflow interdependencies are primarily due to the division of labor along the supply chain. Each supply chain partner performs indeed 7 one or more phases of the entire value creation process from the supply of raw materials to the delivery of the final products and/or services to the customer. Resource interdependencies occur when supply chain firms share their tangible and/or intangible resources for pursuing mutual strategic goals such as exploiting scale economies, reaching product/process improvements, or developing new products/processes (Miles & Snow, 2007). Interdependence exerts a significant influence on firm behavior and performance in supply chain contexts and may affect competitive success (Nair et al., 2009; Mahapatra, Narasimhan and Barbieri 2010). One way by which this occurs is through the influence that interdependence exerts on trust. Indeed, although the linkages between interdependence and trust have been variously addressed in the literature on interorganizational relationships (Wicks et al., 1999; Yilmaz, Sezen, & Ozdemir, 2005; Krishnan, Martin, & Noorderhaven, 2006; Laaksonen et al., 2009), previous supply chain studies suggest that interdependence is a major determinant of trust. Kumar, Scheer, and Steenkamp (1995) found that the more interdependence in channel relationships, the more trust. The authors contend that growing interdependence makes it increasingly dangerous for channel partners to engage in opportunistic behavior, which in turn creates the conditions for the development of trust. In a similar fashion, Crook and Combs (2007) suggest that growing interdependence, and the ensuing increasing coordination needs, reduce opportunism in supply chain relationships by inducing stronger supply chain members to forbear from using their bargaining power to appropriate disproportionate shares of the overall supply chain profits in order to avoid conflict and resentment that might undermine the effectiveness of their and their partners’ coordination efforts. However, using survey data from a sample of buyer-supplier relationships, Ryu, So, and Koo (2009) rejected the hypothesis that interorganizational interdependence is positively related to trust. The above studies are illustrative of a literature that has looked at the impact of the degree of interdependence on trust, reporting mixed evidence. Much less attention has been placed on the impact exerted on trust by the overall structure of interdependencies that characterizes the supply chain. A notable exception is a recent contribution by McCarter and Northcraft (2007). The authors identify three possible types of interdependence structure (i.e., asymmetrical, extended, and constellational) among supply chain partners and speculate that the three types 8 differentially affect uncertainty about exchange hazards within the supply chain, and so the fear that participants in the supply chain behave opportunistically, which in turn influences trust. The present study attempts to contribute to the extant literature on the relationships between interdependence and trust in supply chain networks by looking primarily at the supply chain interdependence structure. This is in line with previous conceptual and empirical research showing that the structural characteristics of interorganizational networks, such as the overall pattern of relationships that characterizes the network, do influence trust (Coleman, 1990; Capaldo, 2007). Thus, we build on previous empirical studies showing that interactions among organizational decisions, and the underlying interdependencies, are typically highly patterned (Choi et al., 2001; Rivkin & Siggelkow 2007; Rosenkopf & Schilling 2007). Specifically, we focus on a number of possible supply chain interdependence patterns to evaluate whether and to what extent they differentially influence trust at the supply chain level. Interdependence Patterns, Opportunism, and Supply Chain Trust Firms participating in a supply chain network commit themselves to cooperate in order to sustain the supply chain overall performance (Christopher, 1992; Bowersox, Closs, & Stank, 1999). However, what is good for the supply chain as a whole may not be best for each single partner (Lee, Padmanabhan, & Whang, 1997; Crook & Combs, 2007). Indeed, globally beneficial decisions may be locally inefficient, that is, the local interests of the supply chain members may not be aligned with the global interests of the supply chain. In such cases, being primarily interested in maximizing their individual performance (Choi et al., 2001), partner firms tend to depart from the globally beneficial behavior in order to pursue their local optimum (Lee, 2004; Narayanan & Raman, 2004). In other words, they behave opportunistically. Thus, opportunism occurs in a supply chain when partners decide and act in their own best interest (i.e., pursue their local optimum) exclusively, even though this is detrimental to the overall supply chain interest (i.e., to pursuing the global optimum). A major factor influencing the risk of opportunism among supply chain members is the overall pattern of interdependence relationships that characterizes the supply chain. Determining whose decisions influence each member firm’s decisions, the supply chain interdependence pattern affects the performance of both each single partner and the overall supply chain–and, by 9 doing so, whether the local interests of the supply chain members are aligned with the global interest of the supply chain. Building on this thread, we propose to gauge the influence that a number of possible different supply chain interdependence patterns exert on supply chain-level trust by quantitatively analyzing the impact of the selected patterns on the alignment between local and global interests. To do so, we draw on the literature on complex adaptive systems (CASs). THE SUPPLY CHAIN AS A COMPLEX ADAPTIVE SYSTEM CASs are networks of adaptive agents that emerge over time into coherent forms through continuous interaction among the participating agents, without any singular entity or central control mechanism deliberately managing or controlling the overall system (Holland, 1995). Recently, scholars interested in CASs theory have started to address the concerns of management academics and practicing managers (Watts, 2003; McKelvey & Andriani, 2005; Amaral & Uzzi, 2007). Along this way, scholars have argued that CASs theory can be a useful tool to study supply chains (Choi et al., 2001; Surana et al., 2005; Pathak et al., 2007). Indeed, supply chains are complex networks characterized by large numbers of interdependence relationships and repeated interactions among multiple suppliers, manufacturers, assemblers, distributors, and retailers (Surana et al., 2005; Pathak et al., 2007), that coevolve over time with the rugged and dynamic environment in which they exist (Choi et al., 2001). Therefore, supply chains can be conceptualized as CASs. Choi, Dooley and Rungtusanatham (2001: 364) have observed that framing the supply chain as a CAS can significantly improve our understanding of the supply chain and has the potential to generate valuable insights into how to manage the supply chain as a whole. We add that looking at the overall network of interdependence relationships among all participating firms in the supply chain, rather than at the supply chain as a sum of dyads, can be especially useful for the purposes of this paper, which adheres to a network perspective to supply chain research and therefore focuses on the impact of the overall supply chain interdependence pattern on supply chain-level trust. Previous research that has applied CAS theory in supply chain studies has typically not adopted computational tools, however (for a notable exception see Nair 10 et al., 2009). To fill this gap, in the present study we use Kauffman’s NK model (Kauffman, 1993). MODELING SUPPLY CHAIN NETWORKS The NK Model in Organization Studies The NK model advanced by Kauffmann (1993) in the context of evolutionary biology defines a family of fitness landscapes which can be tuned by two parameters, N and K. In particular, the stochastic procedure proposed by Kauffmann to design fitness landscapes has subsequently become popular to model organizational decision problems (McKelvey, 1999; Gavetti & Levinthal, 2000; Rivkin, 2001; Nickerson & Zenger, 2004; Ghemawat & Levinthal, 2008; Ganco & Agarwal, 2009). In these studies, the firm’s policy choices are represented by a vector composed of N interacting decisions. A particular N-digit string represents a specific set of choices (i.e., a choice configuration) d=(d1, d2, …, dN), in which it is typically assumed that di=0 or 1 (i=1,…N). K (<N) is the average number of interactions among the decisions di. The pattern of interactions among the decisions is contained in an NxN influence matrix where each x in the (i,j) position means that the column decision j influences the row decision i. The different ways in which the firm’s choices (0-1) about the decisions are combined generate 2N possible choice configurations, each of which is associated a fitness value for the overall system, i.e., a firm overall payoff P(d). For each choice configuration, each single decision di offers a contribution Ci to the overall payoff, which in turn is calculated by averaging the N contributions. Therefore, P(d)= ⎡⎢ N ⎤ ∑ C (d )⎥⎥⎦ / N . The contributions Ci are drawn randomly from a ⎢⎣ i =1 i uniform distribution over [0,1]. Note that each Ci depends not only on the corresponding decision but also on how the decisions interacting with it are resolved. The map showing, for each possible choice configuration, the values of the corresponding Cis and P(d) represents the fitness landscape. The goal of organizational search is to identify the choice configuration that yields the highest firm overall payoff, or in other words, to reach the highest peak of the landscape (i.e., the global peak). The firm is thus engaged in an adaptive walking on the landscape in search of the global peak. In addition to its global peak, however, the landscape contains local peaks. A local 11 peak corresponds to a particular configuration d such that no other configuration exists that differs from d in only one decision and has a higher payoff than d. Note that, while corresponding to suboptimal configurations, the local peaks of a landscape still represent favorable, yet second-best, positions on it. NK Modeling of the Supply Chain Whereas previous research has typically employed the NK framework at the firm level, we use the framework to model the supply chain network as a set of decisions that interact among each other according to a specific pattern reflecting the overall interdependence pattern of the supply chain. This is consistent with recent research that has applied the NK framework to interorganizational settings such as interfirm alliances (Aggarwal, Siggelkow, & Singh, 2011). In fact, due to workflow and/or resource interdependencies, the decisions made by each member firm in a supply chain are influenced by those made by the other supply chain partners interacting with it and generate differing outcomes depending on how those other decisions are resolved. We model the supply chain through an NK model characterized by N decisions (to keep the analysis manageable, each partner is assumed to take one single decision) and K interactions among the decisions. A representative set of the interdependence patterns the supply chain can assume in practice is considered. Each interdependence pattern corresponds to an influence matrix, from which, using the stochastic procedure referred to above, different types of fitness landscapes are generated, one for each selected value of K. Each landscape contains, for each of the 2N possible choice configurations associated to the underlying influence matrix, the contributions Ci and the overall payoff P(d). In our model, the Cis are the performances of the individual supply chain firms, while the P(d)s are the overall supply chain performances. Thus, for each considered interdependence pattern and each selected value of K, the corresponding landscape conveys information about the performance of each single firm (Pfi) and the total supply chain performance (Psc) in all the possible combinations of decisions made by the supply chain firms fi. An illustrative supply chain fitness landscape and the associated influence matrix are reported in Figure 1. Note that, in Figure 1a, c is the configuration that yields the global peak, while c and d correspond to the local peaks. 12 <Insert Figure 1 about here> THE SUPPLY CHAIN INTERDEPENDENCE PATTERNS We argue that supply chain networks can reveal in practice the 10 patterns identified by Rivkin and Siggelkow (2007) in a recent study on patterned interactions in complex systems. Examples of influence matrices corresponding to the 10 patterns and the associated network graphs are shown in Figure 21. For each pattern, Figure 2 also reports one or more real-world supply chain examples. <Insert Figure 2 about here> Random In the random pattern, interactions among decisions occur on a random basis. This pattern reflects one of the two original specifications of the NK model. It is also deeply rooted in the history of network theory and specifically in the work of Erdos and Renyi (1959) on random neworks. An important prediction of random-network theory is that most nodes will have approximately the same number of links, while very few nodes will show a considerably lower or higher number of links than the average. Thus, despite the random placement of links, the resulting network will be deeply ‘democratic’ (Barabasi & Bonabeau, 2003). For our purposes here, in supply chain networks where interdependencies are arranged according to the random pattern, each member firm is affected by the decisions made by approximately the same number of other firms. Accordingly, this pattern describes well supply chains within ‘classic’ Italian industrial districts, i.e., localized concentrations of interlinked firms that are characterized by an extended division of labor among numerous SMEs collaborating intensively among each other vertically and/or horizontally and having approximately the same influence within the network (e.g., Piore & Sabel, 1984). This is the case for the assembling supply chains within the metalengineering district of Lumezzane in Northern Italy. Here, in order to fully fulfill customer orders that typically exceed the internal capacity of each single assembler, orders are split among 1 In the interaction matrices, each x in the (i, j) position means that the decisions made by the row supply chain firm i are affected by the decisions made by the column firm j. In the corresponding network graphs, where each node stands for a supply chain firm, each x in the (i, j) position of the interaction matrix is represented as an influence tie sent from firm j to firm i. 13 several assemblers, who therefore affect each other’s decisions and exert approximately the same influence within the supply chain. Local The other original specification of the NK model is reflected in the local pattern. Here, each decision is influenced by its K/2 neighboring decisions on either side of it. Therefore, if K=2, decision i is affected by decisions i-1 and i+1. Decisions are thus assumed to lie on a “ring”, where each decision is affected by both the decision that precedes it and the decision that follows it, In supply chain contexts, the local interdependence pattern may result from the adoption of just-in-time strategies. In just-in-time supply chains, partner firms typically adopt a single sourcing policy. Thus, when seen from a materials flow perspective, the supply chain network appears as a chain of repeated supplier-customer relationships wherein each partner buys from the upstream node and supplies to the downstream one. In addition, since production planning decisions are triggered by customer orders, information flows along the supply chain in the opposite direction to the materials flow. Therefore, overall, the decisions of each partner firm are directly affected by those made by the adjacent ones, both upstream and downstream. Further cases in point are the “ring networks” found by Liu and Brookfield (2000) in Taiwan’s machine tools industry. For example, consider how a top class Taiwanese machine tool manufacturer has organized the production of electrical boxes. The manufacturer orders metal forms from a supplier that, after readying the sheet metal, delivers the boxes directly to a painting firm which in turn, after processing, delivers them to an electrical layout supplier who install all the necessary components before delivering the finished part back to the manufacturer. Small-world In the small-world pattern, most interactions among decisions are ‘local’ (i.e., adjacent), yet a few interactions exist between decisions that are ‘distant’ (i.e., not adjacent) from each other. This pattern is based on the work by Milgram (1967) and by Watts and associates (Watts & Strogatz, 1998; Watts, 1999), who found that networks tend to organize in clusters of densely connected nodes with low average path length. Small-world properties have been found in the Broadway creative artist network (Uzzi & Spiro, 2005), in the alliance networks of several hightechnology manufacturing industries (Schilling & Phelps, 2007), and in the Indian railway 14 network (Sen, Dasgupta, Chatterjee, Sreeram, Mukherjee, & Manna, 2003) among others. Examples of the small-world interdependence pattern can be found in just-in-time supply chain networks adopting a ring structure yet the phases by which the overall process is composed are shared among small clusters of interconnected partners, rather than being performed by individual firms. Indeed, since the clusters are directly connected between them and/or share common contacts, the supply chain resembles a small-world network. Block-diagonal In the block-diagonal pattern, decisions are grouped so that, while decisions belonging to each block (i.e., group) all affect each other, no interactions occur among decisions belonging to different blocks. This pattern characterizes supply chains producing and distributing modular products or services such as personal computers, stereo systems, bicycles, and financial services. For example, the typical desktop PC supply chain is composed of a final assembler and a number of module suppliers, each of which is fully responsible for producing the assigned module and, to do so, coordinates and controls its own network (i.e., block) of sub-assemblers and parts suppliers. Sub-assemblers and parts suppliers belonging to the same block depend on each other’s decisions and interact repeatedly in order to build the modules required for final production. Conversely, being the modules nearly independent subsystems of the overall product, there are no interdependencies among sub-assemblers and parts suppliers belonging to different blocks (Baldwin & Clark, 1997). Preferential Attachment Whereas random network theory assumes that the probability that two nodes in a network are connected is random and uniform, in many real networks a preferential attachment mechanism is at work, or in other words, network nodes preferentially attach to other nodes having large numbers of connections (Barabasi & Albert, 1999). The preferential attachment pattern considered here is based on this observation. It resembles a “core-periphery” network structure (Borgatti & Everett, 1999), with the network core composed of a handful of highly influential decisions, i.e., decisions influencing large numbers of other decisions, and the network periphery composed of a larger number of poorly influencing decisions, i.e., decisions influencing a few other decisions or only themselves. The preferential attachment interdependence pattern can be 15 recognized in vertical keiretsu supply chain networks characterizing the Japanese economy, where the decisions made by a small number of ‘core’ manufacturers and/or assemblers influence large numbers of ‘peripheral’ affiliated suppliers, which in turn have scant influence on the core companies and little or no influence on each other (Gerlach, 1992). Scale-free A basic tenet of modern network theory is that, due to the preferential attachment mechanism previously discussed, the development of networks resolves into a scale-free state where the distribution of the nodes connectivity follows a power law (Barabasi & Albert, 1999). Analogously, in the scale-free pattern considered here, the degree distribution of the decisions (the degree of each decision being equal to the number of other decisions it affects) follows a power law in that, while a few decisions are very influential, most decisions affect only a restricted number of other decisions. The scale-free interdependence pattern resembles the huband-spoke networks that are very popular in operations and supply chain literature. Examples of hub-and-spoke supply chain networks include, among others, distribution/transportation systems (O’Kelly, 1998), airline networks (Bryan & O’Kelly, 1999), and the Indian auto component industry (Parhi, 2005). In addition, although the ‘classic’ Italian district model described earlier, which resembles the random pattern, has long been considered dominant, several industrial districts, and the supply chains therein, are in fact organized according to the hub-and-spoke model, and therefore are dominated by one or a few large, vertically integrated firms, surrounded by smaller and less powerful suppliers. Examples include, among others, the Seattle region and Toyota city (Gray, Golob, & Markusen, 1996; Markusen, 1996). However, note that the scalefree pattern also resembles the underlying network structure of some Italian industrial districts and the supply chains within them. Accordingly, studies have shown that, far from being social communities wherein influence and power are symmetrically distributed, some Italian districts are similar to hub-and-spoke networks where larger firms play the leading role and organize production among groups of smaller subcontractors (Albino, Garavelli, & Schiuma, 1999; Lazerson & Lorenzoni, 1999). Centralized 16 In the centralized pattern, a few decisions (a single decision, in extreme cases) influence(s) all the remaining decisions, while other decisions only affect themselves. Thus, the notion of highly influential decisions, on which both the preferential attachment and scale-free patterns are based, is taken here to the extreme. The centralized interdependence pattern occurs in supply chain networks where inventories are managed according to Vendor-Managed Inventory (VMI) or Continuous Replenishment (CR) agreements. Under these agreements, inventory management is indeed centralized by manufacturers who de facto manage the inventories of partner retailers by making decisions concerning replenishment time and order quantity (Simchi-Levi, Kaminsky, & Simchi-Levi 2000). The centralized pattern can also be observed in Prato textile industrial district, where the operational decisions of the “impannatori”–who purchase raw materials, allocate orders and give instructions to specialist subcontractors, and market the final products– influence all the remaining supply chain firms, whose decisions are independent among each other and do not influence the impannatori (Paniccia, 1998). Other examples of the centralized interaction pattern are convergent supply chains, used for assembling several types of products (e.g., household electrical appliances), where the assembler makes production planning decisions that affect all the suppliers, whose decisions in turn have little influence within the network. Hierarchical The hierarchical pattern assumes that decisions are hierarchically ordered so that high-ranked decisions influence the decisions below them but not the decisions above them. Thus, the decision ranked 1 is the most influential (i.e., the one influencing the highest number of other decisions). This pattern can be recognized in assembling supply chains where the first assembler makes production planning decisions for the entire network and relies on a back-up assembler when needed. In such a case, the first assembler’s decisions influence those of the remaining supply chain members but are not affected by the decisions of any of them, while the decisions of the back-up assembler, that are affected by the first assembler’s production planning decisions, influence the decisions of all or a major part of the suppliers but not those of the first assembler. Diagonal Analogously to the hierarchical pattern, in the diagonal pattern decisions are ordered so that lowranked decisions do not affect high-ranked decisions. However, differently from the hierarchical 17 pattern, in the diagonal pattern decision 1 is not necessarily the most influential. This may happen in project supply chains, for example those characterizing the construction and the oil and gas industries, where a main contractor (whose decisions are assigned the highest rank) subcontracts the work to a first layer of subcontractors (whose decisions are assigned median ranks), who in turn subcontract their work to second-layer subcontractors (whose decisions are assigned low ranks). The diagonal interdependence pattern also occurs where the leading firm in a supply chain network selects a small number of suppliers to whom full responsibility for coordination of subsuppliers is given. This is the case for several OEMs in the automotive industry, who directly select and manage their first-tier suppliers while delegating to their first-tier and second-tier suppliers the tasks of selecting and managing larger numbers of second-tier and third-tier suppliers, respectively (e.g., Choi & Hong, 2002). In these lead-firm multi-tiered supply chain networks, the leading firm’s decisions directly affect the decisions made by the first-tier suppliers, which in turn directly affect those of the second-tier suppliers but not those of the leading firm. Analogously, the decisions made by the second-tier suppliers directly influence those made by the third-tier suppliers without, however, directly affecting the first-tier suppliers’, let alone the leading firm’s, decisions. Dependent In the dependent pattern, a handful of decisions are affected by all the remaining decisions, while exerting little influence themselves. This pattern occurs in supply chain networks characterized by several partner firms who influence, with their decisions, a small number of other firms, whose decisions in turn exert little influence. This happens, for example, in supply chains using postponement strategies, where product differentiation is delayed at the distribution stage so that standard not-customized products are produced in several manufacturing sites and then shipped to a few regional warehouses for customization according to the users’ needs and requirements. While production decisions made at the manufacturing sites do influence those made at the warehouses, decisions made at the warehouses exert only scant influence on those made at the manufacturing sites. In a similar fashion, the dependent pattern characterizes distribution supply chains composed of few large warehouses serving numerous independent retailers. In such 18 conditions, the retailers’ decisions (i.e., orders) influence the warehouses’ requirement planning decisions, which in turn have slight or no influence on the retailers’ decisions. Also supply chains aimed at designing new products characterized by an integral architecture can exhibit a dependent pattern. This happens when one or a few assemblers involve a number of suppliers in the design and development process of a new product according to the black-box approach (Clark & Fujimoto, 1991). In such cases, the suppliers assume the sole responsibility of designing, engineering, and developing components, or even entire subassemblies, on behalf of their customers (Koufteros, Cheng, & Lai, 2007). Thus, while adhering to the cost/performance requirements set by the assemblers, the suppliers make their design decisions independently from them. Conversely, the integral nature of the product architecture requires the assemblers to make their design decisions at the architectural level by taking simultaneously into account the decisions taken at the component level by all the suppliers. COMPUTATIONAL ANALYSIS We performed computational analysis to measure the impact of the interdependence patterns described above on the risk of opportunism by supply chain partners, which in turn affects supply chain-level trust. Note that, whereas scholars adopting the NK framework in management studies typically analyze the shape (i.e., single-peaked vs. multi-peaked) of the fitness landscapes and simulate the adaptive walk undertaken by the examined systems (typically firms) on the landscapes in search of better positions, a simulation analysis was not needed for the purposes of our study. Indeed, rather than being interested in understanding whether and/or under what conditions the supply chain would reach the global peak or be trapped in local peaks, we were interested in gauging network-level trust by the alignment between local and global interests in supply chain contexts. In order to do so, we considered two different scenarios in which the supply chain satisfies its global interest, albeit at different degrees. This increased the robustness of our findings. In the first-best scenario, the supply chain performs at its best, or in other words, reaches the global peak of the landscape. Thus, it takes the choice configuration that yields the highest possible supply chain performance. In the second-best scenario, the supply chain remains 19 trapped in local peaks. Thus, it takes choice configurations corresponding to suboptimal positions on the landscape. We designed, for each of the 10 interdependence patterns described in the preceding sections, three types of landscapes. Each type was generated by selecting one of the three different values of K (2; 3; 5) that we considered in order to test the robustness of our results as K varies, through a sensitivity analysis. In total, 30 different types of landscapes were designed. In all the landscapes, N=12. To guarantee the statistical significance of the results, the relevant measures were computed over 600 landscapes of each type. As discussed earlier, the risk of opportunism in a supply chain network depends on the extent to which the local interests of the supply chain members are aligned with the global interest of the supply chain. The higher the alignment, the lower the risk of opportunism. Thus, in the first-best scenario, for each type of landscape, the risk of opportunism by firm i (i=1,…12) was gauged by the percentage of the 600 landscapes in which the firm reached its local optimum while the overall system was in the global peak. Averaging this probability over the N supply chain partners we obtained the Supply Chain Trust index for the first-best scenario (SCT_fb). The higher this index, the more the local and global interests were aligned, and therefore the lower the risk of opportunism in the overall network, and hence the higher the level of supply chain trust. In the second-best scenario, for each type of landscape, the local peaks were first identified. Next, for each firm i, the percentage of times in which the firm reached its local optimum while the overall system was in one of the identified local peaks was computed over the 600 landscapes and then averaged over the N supply chain partners. This restituted the Supply Chain Trust index for the second-best scenario (SCT_sb). Analogously to the first-best scenario, the higher this average value, the lower the risk of opportunism in the overall network, and hence the higher trust at the supply chain level. RESULTS AND DISCUSSION Focusing on the first-best and the second-best scenario respectively, Tables 1 and 2 report, for each type of landscape, the probability that each supply chain partner reaches its local optimum while the overall supply chain reaches a favorable position on the landscape (i.e., the global peak, in the first-best scenario; the local peaks, in the second-best scenario), and therefore will 20 tend not to behave opportunistically. The values of the Supply Chain Trust index are shown in the last column of each Table. <Insert Table 1 and Table 2 about here> Influence of the Interdependence Patterns on Supply Chain Trust For all the examined patterns, in the first-best scenario supply chain trust is systematically higher when compared to the second-best scenario. Moreover, in both scenarios and for all the patterns, the higher K the lower the level of supply chain trust. However, results of the first-best and second-best scenario analysis, and those obtained for each of the three selected values of K, display similar (although not identical) trends. We thus focus on the results of the first-best scenario analysis (K=2) exclusively. We note that the level of supply chain trust, expressed by the SCT_fb index, is rather different across the examined interdependence patterns. We thus compare the 10 patterns on the level of supply chain trust. The dependent pattern is associated with the highest level of trust. Most firms participating in supply chains based on this pattern (e.g., firms 1 to 10 in Figure 2) display a considerable degree of alignment (83.52% on average) between their local interests and the supply chain’s global interest, while only a few of them (e.g., firms 11 and 12) have strong incentives to behave opportunistically, since they almost never reach their local optima. Thus, overall, the supply chain is characterized by a substantial level of trust. This result suggests that trust tends to be widespread across supply chain networks using postponement strategies. Indeed, although customizing firms are very likely to behave opportunistically, this exerts a negligible impact on supply chain trust. In addition, the above result is consistent with recent findings suggesting that supply chains producing integral products are likely to be associated with longterm trust-based relationships between OEMs and their suppliers (Ulku & Schmidt, 2011). We add that this is especially true when suppliers are involved in the design and development of new products according to black-box approaches. Interacting scantily with other supply chain members, black-box suppliers have their decision space hardly constrained by others. This increases the probability for them to make decisions yielding the local optimum, which renders the risk that they behave opportunistically considerably lower than that associated with grey-box approaches. 21 The diagonal pattern is ranked second on supply chain trust. However, when compared to the dependent pattern, the diagonal pattern has a substantially lower value of the SCT_fb index, that reaches well below 50%. Firms making high-ranked decisions (e.g., firms 1 to 3 in Figure 2) display a considerable degree of alignment between their local interests and the global interest (73.83% on average). Conversely, for firms making medium- to low-ranked decisions (e.g., all the remaining firms), such alignment is lower. This implies that firms lying on the lowest tiers of multi-tiered supply chain networks have considerable incentive to be opportunistic. This is consistent with previous research arguing that, being clearly focused on benefiting the buyer, the tiered supplier partnership model is substantially prone to the risk of opportunism by lower-tier firms (Stuart, Deckert, McCutcheon, & Kunst, 1998. See also Dyer, 1996). The preferential attachment pattern is ranked third on supply chain trust, while the random pattern is ranked fourth. This resonates with previous empirical studies showing that both Japanese automotive supply chains based on the Keiretsu model, that are reminiscent of the preferential attachment pattern, and supply chains within ‘classic’ Italian districts, which resemble the random pattern, are characterized by moderately high levels of trust (Piore & Sabel, 1984; Dyer, 1996). We also note that, in core-periphery supply chain networks, core firms (e.g., firms 3, 6, 9, and 11 in Figure 2) have a considerably higher tendency to be opportunistic, when compared to peripheral firms (e.g., firms 2, 4, 7, and 12). Conversely, in random supply chain networks, all partners are associated moderate risks of opportunism. The small-world, local, scale-free, and block-diagonal patterns display similar but slightly decreasing and rather low values of the SCT_fb index. Our results for the small-world pattern are surprising in that small-world networks have been credited of considerable levels of trust in previous literature (Uzzi & Spiro, 2005). Conversely, we found that, having a degree of alignment between their local interests and the supply chain’s global interest ranging between 31,33% and 42,17%, firms involved in small-world supply chains have a rather high tendency to be opportunistic. Analogously, in the local pattern, all member firms appear to be inclined to opportunism. We thus caution that just-in-time supply chains, such as those resembling both the small-world and the local pattern, are vulnerable not only to accidental events (e.g., fires and machine failures), due to the low inventory levels, but also to opportunism. This is especially 22 relevant in the light of previous literature pointing out the need of considerable levels of trust for just-in-time strategies to fully deliver their promises (e.g., Frazier, Spekman, & O’Neal, 1988). The scale-free pattern is ranked seventh on supply chain trust. Indeed, previous literature has not credited hub-and-spoke networks with high levels of trust (Gray et al., 1996). This is also true for hub-and-spoke supply chains in Italian industrial districts, where trust is often carefully interspersed with control. For example, Lazerson and Lorenzoni (1999) reported that district manufacturers in Italy employ a number of mechanisms (e.g., quality-control inspectors) to exercise control even over trusted subcontractors. The block-diagonal pattern shows a low level of supply chain trust as well. Focusing on the relationships between final assemblers and module suppliers, studies have argued that product modularity reduces opportunism by module suppliers (Hoetker, 2006). However, our results show that, where interdependencies are arranged according to the block-diagonal pattern, the probability for partner firms to satisfy their local interests while the supply chain reaches the global optimum is quite low (34.35% on average), which renders opportunism a frequent accompaniment of this pattern. To reconcile this apparent contradiction, we caution that opportunism may be nested within the blocks. Indeed, reciprocal interdependence among the decisions taken by partner firms in each block increases the risk of opportunism within the blocks by rendering each partner’s decisions dependent on those made by all the remaining partners belonging to the same block. Overall, this suggests that the advantages of modular product architectures, which have been pointed out by a significant literature (Langlois & Robertson, 1992; Garud & Kumaraswamy 1995; Baldwin & Clark, 1997), should be weighted against the considerable risks of opportunism by sub-assemblers and parts suppliers participating in supply chains producing and/or distributing modular products. The hierarchical pattern is ranked ninth on supply chain trust. While the local interest of the most influential firm in the network (e.g., firm 1 in Figure 2) is aligned with the global interest of the supply chain in 63.67% of cases, such alignment is considerably lower for the remaining partners (31,39% on average). We thus caution first assemblers in assembling supply chains resembling the hierarchical pattern to set up mechanisms such as contracts and 23 performance schemes to protect themselves and the supply chain from the risk that their partners behave opportunistically. The centralized pattern is associated with the lowest level of supply chain trust. In only 32.76% of cases on average pursuing the supply chain interest allows the participating firms to satisfy their private interest. Thus, the risk of opportunism is high for all partners. Accordingly, studies have shown that, under Vendor-Managed Inventory and Continuous Replenishment contracts, the risk of opportunism by vendors is considerably high, so that retailers are advised to adopt information and communication technologies and other mechanisms to protect themselves from their opportunism (Narayanan & Raman, 2004). Other studies further support our finding that supply chains wherein interdependencies are arranged according to the centralized pattern are particularly prone to opportunism. Paniccia (1998) has questioned that the relationships between impannatori, traders or wool-mills, and subcontractors in Prato textile industrial district are governed by trust. In fact, impannatori maintain strong bargaining power over weaving and spinning firms, so increasing the threat of opportunism in the district’s supply chains. Explanation Drawing on Rivkin and Siggelkow (2007), we submit that a major explanation for the different levels of trust we found across the 10 examined supply chain interdependence patterns resides in the different numbers of ‘uninfluenced’ and ‘uninfluential’ partners the patterns contain, or in other words, in the different numbers of supply chain members whose decisions are not influenced by / do not influence the decisions made by the remaining partners. Having their decision space not constrained by others, uninfluenced partners are highly likely to make decisions that allow them to reach their local optimum while the supply chain takes the choice configuration that yield the global peak (in the first-best scenario), or one of the configurations yielding local peaks (in the second-best scenario), of the fitness landscape. Therefore, the higher the number of uninfluenced partners, the lower the average risk of opportunism across the supply chain, and hence the higher the level of supply chain trust. Conversely, as the number of uninfluential partners increases, influence concentrates in a decreasing number of firms. Thus, the decisions made by a small number of partners constrain the decision space of the remaining partners, thereby significantly reducing the probability for the latters to reach their local optimum 24 while the supply chain reaches the global optimum. At the supply chain level, this increases the risk of opportunism and decreases trust. To illustrate, Figure 3 contrasts the two interdependence patterns which showed the highest and lowest levels of supply chain trust, i.e., the dependent vs. the centralized pattern. Compared with the remaining patterns, the former contains the highest number of uninfluenced partners, while the latter is characterized by the highest number of uninfluential partners. <Insert Figure 3 about here> Based on the above reasoning, we advance that trust in supply chain networks is a positive function of the number of uninfluenced partners and a negative function of the number of uninfluential partners. To test this hypothesis, we regressed the SCT_fb index on the number of uninfluenced partners and the number of uninfluential partners characterizing the examined patterns. To do this, we fixed K=32 and divided the 10 patterns into two groups. The first group included the local, centralized, block-diagonal, hierarchical, and dependent patterns. The second group included the random, small-world, preferential attachment, scale-free, and diagonal patterns. For the patterns belonging to the first group, the xs have fixed positions in the corresponding matrices. Conversely, for the patterns belonging to the second group, chance influences where precisely the xs lie in the matrices, and therefore the number of uninfluenced and uninfluential partners may vary. We thus generated one influence matrix for each pattern belonging to the first group and 50 matrices for each pattern belonging to the second group. Next, for each influence matrix, we generated 600 landscapes and calculated the SCT_fb index corresponding to each pattern as described in a previous section. Finally, we performed the multiregression analysis. Results reported in Table 3 confirm the above hypothesis and also show that the positive impact exerted on supply chain trust by the number of uninfluenced partners is substantially higher than the negative impact of the number of uninfluential partners. <Insert Table 3 about here> Influence of the Degree of Interdependence on Supply Chain Trust 2 We chose K=3 for regression analysis purposes because we had already employed K=2 to generate our hypothesis while choosing K=5 would have required disproportionate computational effort. 25 While our research concentrates primarily on the influence of the supply chain interdependence structure on supply chain trust, the sensitivity analysis that we performed by considering different values of K has implications for the influence exerted on supply chain trust by the degree of interdependence. Indeed, an aside yet notable result emerging from Tables 1 and 2 is that, for each considered pattern, the higher K the higher the risk of opportunism across the supply chain, and hence the lower the level of supply chain trust. This result is at odds with previous research suggesting that the degree of interdependence positively affects trust (Kumar et al., 1995). However, it is consistent with recent claims that the more interconnected the firms participating in a supply chain, the more the risk that they behave opportunistically (McCarter & Northcraft, 2000). In adding to this literature, we argue that the negative influence of the degree of interdependence on network-level trust in supply chain networks is a consequence of the negative impact that interdependence exerts on the alignment of the local interests of the supply chain members with the global interest of the supply chain. The more interdependent the partners, the more they constrain each others’ decisions, and therefore the less likely it is that they are able to optimize their individual performance while the overall supply chain performance is at its maximum. This, in turn, increases the probability that supply chain partners behave opportunistically. Nevertheless, increasing interdependence does not reduce supply chain-level trust to the same extent in all the patterns. Table 4 shows that, as K grows from 2 to 3 and from 3 to 5, the corresponding decrease in the SCT_fb index is considerably lower for the hierarchical, diagonal, and dependent patterns. <Insert Table 4 about here> CONCLUSION Since management scholars have discovered that trust is a key lubricant of interorganizational relationships and a major driver of superior performance (Uzzi, 1997; Krishnan et al., 2006), an overriding concern in supply chain management has become one of trust and the problem of opportunism (Galaskiewicz, 2011: 5). In particular, as the supply chain has become a key competitive unit, the level of trust across the overall supply chain network, rather than within its composing dyads, has come under the spotlights. Thus, in order to shed some light on trust as a 26 network-level construct and on its determinants, we have shown that the overall pattern of interdependencies that characterizes the supply chain influences trust at the supply chain level. While previous literature has neglected the impact exerted on trust by the supply chain interdependence structure, we have focused on a representative set of the interdependence patterns supply chains may assume in practice and found that the 10 considered patterns differentially affect trust across the supply chain. Our analysis has showed that a major explanation for this resides in the different numbers of uninfluenced and uninfluential partners characterizing the patterns. Specifically, we have argued that supply chain trust is positively related to the number of uninfluenced partners and negatively related to the number of uninfluential partners. In addition, while previous studies that have focused on the degree of interdependence have found a positive relationship between interdependence and trust (Kumar et al., 1995) or no direct impact of interdependence on trust (Ryu et al., 2009), we have found that the degree of interdependence negatively influences trust at the supply chain level. Thus, our overarching conclusion is that three aspects of a supply chain network influence network-level trust in supply chains i.e., the number of uninfluenced partners, the number of uninfluential partners, and the degree of interdependence. Our research also contributes to the assessment of a number of supply chain strategies and architectures, and their value creation potential. Indeed, resulting in different supply chain interdependence patterns, different supply chain strategies and architectures differentially affect the risk of opportunism across the supply chain, and hence supply chain trust. We have shown that both hub-and-spoke and just-in-time supply chains are rather vulnerable to opportunism. This is more so in supply chains producing and distributing modular products or services, wherein opportunism tends to be nested within the blocks of sub-assemblers and parts suppliers. Moreover, where inventories are managed according to the VMI and CR agreements, and in convergent supply chains, supply chain-level trust is at its minimum. Conversely, vertical Keiretsu supply chains and supply chains within ‘classic’ Italian industrial district are characterized by moderately high levels of supply chain trust. Analogously, multi-tiered supply chain networks are associated high levels of supply chain trust, yet firms lying on the lowest tiers have considerable incentives to behave opportunistically. Finally, supply chains using 27 postponement strategies and integral product supply chains with black box suppliers are associated the highest level of supply chain trust. Building on extant literature, and in line with the overall purposes of our study, the research strategy and methodology used here have the following distinctive features. First, our study responds to a previous call by Pathak, Day, Nair, Sawaya, and Kristal (2007) for leveraging the CAS perspective in supply chain management research. In particular, looking at the supply chain through the lenses of CAS theory has allowed us to adopt a network perspective in our study, and specifically to focus on the overall network of interdependence relationships among all participating firms in the supply chain, rather than on the supply chain as a sum of dyads. Second, using the NK framework to model the supply chain as a network of interdependent partners and their decisions has allowed us to investigate the impact of the overall supply chain interdependence pattern on supply chain trust. In so doing, our study capitalizes on the interpretive potential of the NK framework in the field of management studies, that has been identified yet hardly exploited so far in supply chain management research (Choi & Krause, 2006; Pathak et al., 2007). Finally, in line with Davis, Eisenhardt and Bingham (2007), relying on a computational approach has led us to add fresh insights to our understanding of the relationships between interdependence and trust in supply chain networks by leveraging the strengths of computational analysis for theory development. Implications for Practice In an attempt to help closing the research-practice gap in supply chain studies (Carter, 2008), our research contributes to a strand of literature aimed at addressing managerially relevant issues at the intersection between strategy, marketing, and supply chain management (Hult, Ketchen and Arrfelt 2007; Stock, Boyer, & Harmon, 2010). In order to do so, we have adhered to the need of conceiving of the supply chain as a complex web of interdependent relationships, and thus to adopt a network perspective to supply chain management research (Nair et al., 2009). The study has at least two major managerial implications. First, we have shown that, in order to benefit from high levels of supply chain trust, managers should pay particular attention to the interdependence pattern of the overall supply chain networks in which their firms operate and try to implement supply chain patterns that are 28 less conducive to opportunism. To help them to do so, we have focused on 10 patterns and compared them on the level of supply chain trust. Second, our work has implications for strategic management that go beyond the supply chain field. In particular, our findings speak to those interested in strategic networks (Gulati, Nohria, & Zaheer 2000), and specifically in the rising fields of network management and relational capabilities (Capaldo, 2007; Allred, Fawcett, Wallin, & Magnan, 2011). We argue that network architects should design interorganizational alliance networks characterized by: (1) small numbers of partners whose decisions do not influence the decisions of the remaining ones (i.e., uninfluential partners) and, more importantly, large numbers of partners whose decisions are not influenced by those made by the remaining ones (i.e., uninfluenced partners); and by (2) low interdependence among the participating organizations. Facilitating the alignment of the partners’ private interests with the interest of the overall network, these two aspects decrease the risk that partner firms behave opportunistically, which in turn has the potential to increase network-level trust. Limitations and Future Research Our study has several limitations that might be addressed in future research. First, we have cautioned that interdependence constrains the decision space of supply chain firms and ultimately increases the probability that they behave opportunistically. This, in turn, may negatively influence performance in supply chain networks. However, previous research on interorganizational relationships has insisted that interdependence represents a major source of value for allied organizations (e.g., Zajac & Olsen, 1993; Dyer & Singh, 1998). The two perspectives do not exclude each other, however. Future research should consider simultaneously–and evaluate the net effect of–the constraining role of interdependence and its value-creating potential in supply chain networks. Second, while focusing on the structural features of the supply chain, our laboratory test-bed study has not allowed us to take into account other variables that have been showed to influence trust in interfirm networks. Future research might consider how social embeddedness, shared identities, and contracts among others interact with the supply chain interdependence pattern to influence supply chain trust. Third, although we have concluded that supply chain architects should maneuver interdependence in order to reduce opportunism and generate trust across the supply chain, the 29 present study does not consider the impact of interdependence, and of the ensuing levels of supply chain trust, on performance. Simulation studies are needed to investigate this point. Finally, while we have argued that interdependence affects trust across the supply chain network, the opposite might be true as well. In other words, network-level trust might influence network structure. For example, the more trust is widespread across the supply chain, the less the supply chain leader will exert tight control over the remaining firms, which in turn may increase the number of uninfluenced partners. In addition, a climate of trust may encourage partners to collaborate, so increasing the degree of interdependence within the network. Future longitudinal studies are needed to shed light on the direction of causality, as well as on the interplay, between network-level trust and interdependence in supply chain contexts. REFERENCES Aggarwal, V., Siggelkow, N., & Singh, H. (2011). Governing collaborative activity: Interdependence and the impact of coordination and exploration. Strategic Management Journal, 32, 705-730. Albino, V., Garavelli, A.C., & Schiuma, G. (1999). Knowledge transfer and inter-firm relationship in industrial districts: The role of the leader firm. Technovation, 19, 53-63. Allred, C. R., Fawcett, S. E., Wallin, C., & Magnan, G. M. (2011). A dynamic collaboration capability as a source of competitive advantage. Decision Sciences, 42, 129–161. Amaral, L.A.N., & Uzzi, B. (2007). Complex systems – A new paradigm for the integrative study of management, physical, and technological systems. Management Science,. 53, 1033-1035. Astley, W.G., & Zajac, E.J. (1991). Intraorganizational power and organizational design: Reconciling rational and coalitional models of organization. Organization Science, 2(4), 399-411. Baldwin, C.Y., & Clark, K.B. (1997). Managing in an age of modularity. Harvard Business Review, 75(5), 84-93. Barabasi A.L., & Bonabeau, E. (2003). Scale-free networks. Scientific American, 288, 60-69. Barabasi, A.L., & Albert, R. (1999). Emergence of scaling in random networks. Science, 286, 509-512. Barney, J.B., Hansen, M.H. 1994. Trustworthiness as a source of competitive advantage. Strategic Management Journal, 15, 175-190. Blankenburg Holm, D., Eriksson, K., & Johanson, J. (1999). Creating value through mutual commitment to business network relationships. Strategic Management Journal, 20, 467-486 Borgatti, S.P., & Everett, M.G. (1999). Models of core/periphery structures. Social Networks, 21, 375395. Bowersox, D.J., Closs, D.J., & Stank, T.P. (1999). 21st Century Logistics: Making Supply Chain Integration a Reality. Oak Brook, IL: Council of Logistics Management. Bradach, J.L., & Eccles, R.G. (1989). Price, authority, and trust. From ideal types to plural forms. Annual Review of Sociology, 15, 97-118. Bryan, D.L., & O’Kelly, M.E. (1999). Hub-and-spoke networks in air transportation: An analytical review. Journal of Regional Science, 39, 275–295. 30 Capaldo, A. (2007). Network structure and innovation: The leveraging of a dual network as a distinctive relational capability. Strategic Management Journal, 28, 585-608. Carter, C.R. (2008). Knowledge production and knowledge transfer: Closing the research-practice gap, Journal of Supply Chain Management, 44(2), 78-82. Cavusgil, S.T., S.Z. Deligonul & C. Zhang (2004). Curbing foreign distributor opportunism: An examination of trust, contracts, and the legal environment in international channel relationships. Journal of International Marketing, 12, 7–27. Choi, T.Y., & Krause, D.R. (2006). The supply base and its complexity: Implications for transaction costs, risks, responsiveness, and innovation. Journal of Operations Management 24, 637-652. Choi, T.Y., & Hong, Y. (2002). Unveiling the structure of supply networks: Case studies in Honda, Acura, and DaimlerChrysler. Journal of Operations Management, 20, 469–493. Choi, T.Y., Dooley, K., & Rungtusanatham, M. (2001). Supply networks and complex adaptive systems: Control versus emergence. Journal of Operations Management, 19, 351–366. Christopher, M. (1992). Logistics & Supply Chain Management. London, UK: Pitmans. Clark, K.B., & Fujimoto T. (1991). Product Development Performance. Strategy, Organization, and Management in World Auto Industry. Harvard Business School Press, Boston. Coleman, J.S. (1990). Foundations of Social Theory. Harvard University Press, Cambridge. Crook, T.R., & Combs, J.G. (2007). Sources and consequences of bargaining power in supply chains. Journal of Operations Management, 25, 546-555. Das, T.K. 2006. Strategic alliance temporalities and partner opportunism. British Journal of Management, 17, 1-21. Das, T.K., & Rahman, N. (2010). Determinants of partner opportunism in strategic alliances: A conceptual framework. Journal of Business Psychology, 25, 55-74. Davis, J.P., Eisenhardt, K., & Bingham, C.B. (2007). Developing theory through simulation methods. Academy of Management Review, 32, 480-499. Dirks, K.T., & Ferrin, D.L. (2001). The role of trust in organizational settings. Organization Science, 12, 450-467. Dyer J.H., & Chu W. (2003). The role of trustworthiness in reducing transaction costs and improving performance: Empirical evidence from the United States, Japan, and Korea. Organization Science 14, 57-68. Dyer, J.H. (1996). How Chrysler created an American keiretsu. Harvard Business Review, 74, 42-56. Dyer, J.H., & Nobeoka, K. (2000). Creating and managing a high-performance knowledge-sharing network: The Toyota case. Strategic Management Journal, 21, 345-367. Dyer, J.H., & Singh, H. (1998). The relational view: cooperative strategy and sources of interorganizational competitive advantage. Academy of Management Review, 23, 660-679. Erdos, P., & Renyi, A. (1959). On random graphs. Publicationes Mathematicae, 6(1), 290–297. Frazier, R., Spekman, E., & O’Neal, C.R. (1988). Just-in-time exchange relationships in industrial markets. Journal of Marketing, 52, 52-67. Galaskiewicz J. (2011). Studying supply chains from a social network perspective. Journal of Supply Chain Management, 47, 4-8. Gambetta, D. (1988). Can we trust trust? In D. Gambetta (Ed.), Trust: Making and Breaking Co-operative Relations. Oxford, UK: Basil Blackwell, 213–237. Ganco, M., & Agarwal, R. (2009). Performance differentials between diversifying entrants and entrepreneurial start-ups: A complexity approach. Academy of Management Review, 34, 228-252. 31 Ganco, M., & Hoetker, G. (2009). NK modeling methodology in the strategy literature: Bounded search on a rugged landscape. In D.D. Bergh & D.J. Ketchen (Eds.), Research Methodology in Strategy and Management, 5, 237-268. Garud, R., & Kumaraswamy, A. (1995). Technological and organizational designs for realizing economies of substitution. Strategic Management Journal, 16, 93-109. Gavetti, G., & Levinthal, D. (2000). Looking forward and looking backward: Cognitive and experiential search Administrative Science Quarterly, 45, 113-137. Gerlach, J.M. (1992). The Japanese corporate network: A block model analysis. Administrative Science Quarterly, 37, 15-139. Ghemawat, P., & Levinthal, D. (2008). Choice interactions and business strategy. Management Science, 54, 1638-1651. Granovetter, M. (1985). Economic action and social structure: The problem of embeddedness. American Journal of Sociology., 91(3), 481-510. Gray, M., Golob, E., & Markusen, A. (1996). Big firms, long arms, wide shoulders: The `hub-and-spoke' industrial district in the Seattle region. Regional Studies, 30, 651-666. Greis, N.P., & Kasarda, J.D. (1997). Enterprise logistics in the information age. California Management Review, 39(3), 55–78. Gulati, R., Nohria, N., & Zaheer, A. (2000). Strategic networks. Strategic Management Journal. 21, 203215. Handfield, R.B., & Bechtel, C. (2002). The role of trust and relationship structure in improving supply chain responsiveness. Industrial Marketing Management, 31, 367-382. Hoetker, G. (2006). Do modular products lead to modular organizations? Strategic Management Journal, 27, 501-518. Holland, J.H. (1995). Hidden Order. Reading, UK: Addison-Wesley. Hosmer, L.T. (1995). Trust - the connecting link between organizational theory and philosophical ethics. Academy of Management Review, 20(2), 379-403. Hult, G.T.M., Ketchen, D.J., & Slater, S.F. (2004). Information processing, knowledge development, and strategic supply chain performance. Academy of Management Journal, 47, 241–253. Hult, G.T.M., Ketchen, D.J., & Arrfelt, M. (2007). Strategic supply chain management: Improving performance through a culture of competitiveness and knowledge development. Strategic Management Journal, 28, 1035-1052. Ireland R.D., & Webb J. W. (2007). A multi-theoretic perspective on trust and power in strategic supply chains. Journal of Operations Management, 25, 482-497. Johnston, D.A., McCutcheon, D.M., Stuart, F.I., & Kerwood, H. (2004). Effects of supplier trust on performance of cooperative supplier relationships. Journal of Operations Management, 22, 23-38. Kauffman, S. (1993). The Origins of Order: Self-Organization and Selection in Evolution. New York, US: Oxford University Press. Ketchen, D.J., & Giunipero, L. (2004). The intersection of strategic management and supply chain management. Industrial Marketing Management, 33, 51-56. Koufteros, X.A., Cheng, T.C.E., & Lai, K. (2007). “Black-box’’ and ‘‘gray-box’’ supplier integration in product development: Antecedents, consequences and the moderating role of firm size. Journal of Operations Management, 25, 847-870. Krishnan, R., Martin, X., & Noorderhaven, N. (2006), When does trust matter to alliance performance? Academy of Management Journal, 49, 894-917. 32 Kumar, N. (1996). The power of trust in manufacturer-retailer relationships. Harvard Business Review, November-December, 92-106. Kumar, N., Scheer, L.K., & Steenkamp, J.E.M. (1995). The effects of perceived interdependence on dealer attitudes. Journal of Marketing Research, 32, 348-356. Laaksonen T., Jarimo T., & Kulmala H. (2009). Cooperative strategies in customer–supplier relationships: The role of inter-firm trust. International Journal of Production Economics, 120, 79-87. Langlois, R.N., & Robertson, P.L. (1992). Networks and innovation in a modular system: Lessons from the microcomputer and stereo component industries. Research Policy, 21(4), 297-313. Lazerson, M.H., & Lorenzoni, G. (1999). The Firms that feed industrial districts: A return to the Italian source. Industrial and Corporate Change, 8, 235-266. Lee, H.L. 2004. The triple-a supply chain. Harvard Business Review, 82, 102-111. Lee, H.L., Padmanabhan, V., Whang, S. 1997. Information distortion in a supply chain: The bullwhip effect. Management Science, 43, 546–558. Levinthal, D.A. (1997). Adaptation on rugged landscapes. Management Science, 43, 934–950. Liu, R., & Brookfield, J. (2000). Stars, rings and tiers: Organisational networks and their dynamics in Taiwan’s machine tool industry. Long Range Planning, 33, 322-348. Luo, Y. (2007). Are joint venture partners more opportunistic in a more volatile environment? Strategic Management Journal, 28, 39-60. Mahapatra, S.K., Narasimhan, R., & Barbieri, P. (2010). Strategic interdependence, governance effectiveness and supplier performance: A dyadic case study investigation and theory development. Journal of Operations Management, 28, 537-552. Markusen, A. (1996). Sticky places in slippery space: A typology of industrial districts. Economic Geography, 72, 293-313. Mayer R.C., Davis J.H., & Schoorman, F.D. (1995). An integrative model or organizational trust. Academy of Management Review, 20, 709-734. McCarter M.W., Northcraft G.B., 2007. Happy together? Insights and implications of viewing managed supply chain as social dilemma. Journal of Operations Management, 25, 498-511. McEvily, B., Perrone, V., & Zaheer, A. (2003). Trust as an organizing principle. Organization Science, 14, 91-103. McKelvey, B. (1999). Avoiding complexity catastrophe in coevolutionary pockets: Strategies for rugged landscapes. Organization Science, 10, 294–321. McKelvey, W., Andriani, P. (2005). Why Gaussian statistics are mostly wrong for strategic organization. Strategic Organization, 3, 219-228. Miles, R.E., & Snow, C.C. (2007). Organization theory and supply chain management: An evolving research perspective. Journal of Operations Management, 25, 459-463. Milgram, S. (1967). The small world problem. Psychology Today, 1, 61–67. Monczka, R.M., Petersen, K.J., Handfield, R.B., & Ragatz, G.L. (1998). Success factors in strategic supplier alliances: The buying company perspective. Decision Science, 29, 553-577. Nair, A., Narasimhan, R., Choi, T.Y. (2009). Supply networks as a complex adaptive system: Toward simulation-based theory building on evolutionary decision making. Decision Sciences 40, 783815. Narayanan, V., & Raman, A. (2004). Aligning supply chain incentives. Harvard Business Review, 82 94103. Nickerson, J.A., Zenger, T.R. (2004). A knowledge-Based theory of the firm: The problem-solving perspective. Organization Science 15, 617-632. 33 Nooteboom, B. (2002). Trust: Forms, Foundations, Functions, Failures and Figures. Cheltenham, UK: Edward Elgar. O’Kelly, M. (1998). A geographer’s analysis of hubs-and-spoke networks. Journal of Transport Geography, 6, 171-186. Panayides, P. M., & Lun, Y.H.V. (2009). The impact of trust on innovativeness and supply chain performance. International Journal of Production Economics, 122, 35-46. Paniccia, I. (1998). One, a hundred, thousands of industrial districts. Organizational variety in local networks of small and medium-sized enterprises. Organization Studies, 19, 667-699. Parhi, M. (2005). Dynamics of inter-firm linkages in indian auto component Industry: A social network analysis. Proceedings of DRUID Winter Conference. Pathak, S.D., Day, J.M., Nair, A., Sawaya, W.J., & Kristal, M.M. (2007). Complexity and adaptivity in supply networks: Building supply network theory using a complex adaptive systems perspective, Decision Science, 38, 547–580. Pfeffer, J., & Salancik, G. (1978). The External Control of Organizations: A Resource Dependence Perspective. New York, US: Harper and Row. Piore, M., & Sabel, C. (1984). The Second Industrial Divide. New York, US: Basic Books. Provan, K.G., Fish, A., & Sydow, J. (2007). Interorganizational networks at the network level: A review of the empirical literature on whole networks. Journal of Management, 33, 479-516. Rivkin, J.W. (2001). Reproducing knowledge: Replication without imitation at moderate complexity. Organization Science, 12, 274–293. Rivkin, J.W., Siggelkow, N. (2007). Patterned interactions in complex systems: Implication for exploration. Management Science, 43, 1068-1085. Rosenkopf, L., Schilling, M.A. (2007). Comparing alliance network structure across industries: observations and explanations. Strategic Entrepreneurship Journal, 1, 191–209. Rousseau, D.M., Sitkin, S.B., Burt, R.S., & Camerer, C. (1988). Not so different after all: A crossdiscipline view of trust. Academy of Management Review, 23, 393-404. Ryu, I., So, S., & Koo, C. (2009). The role of partnership in supply chain performance. Industrial Management & Data Systems, 109, 496-514. Schilling, M. A., & Phelps, C.C. (2007). Inter-firm collaboration networks: The impact of large-scale network structure on firm innovation. Management Science, 53, 1113–1126. Sen, P., S. Dasgupta, A. Chatterjee, P. Sreeram, G. Mukherjee, & Manna, S. (2003). Small world properties of the Indian railway network. Physical Review E, 67, 036106. Seppänen, R., Blomqvist, K., & Sundqvist S. (2007). Measuring inter-organizational trust – a critical review of the empirical research in 1990-2003. Industrial Marketing Management, 36, 249-265. Simchi-Levi, D., Kaminsky, P., & Simchi-Levi, E. (2000). Designing and Managing the Supply Chain: Concepts, Strategies and Case Studies. Singapore: McGraw-Hill International. Stock, J.R., Boyer, S.L., & Harmon, T. (2010). Research opportunities in supply chain management. Journal of the Academy of Marketing Science, 38, 32-41. Stuart, I., Deckert, P., McCutcheon, D., & Kunst, R. (1998). Case study—A leveraged learning network. Sloan Management Review, 39, 81–93. Surana, A., Kumara, S., Greaves, M., & Raghavan, U.N. (2005). Supply-chain networks: A complex adaptive systems perspective. International Journal of Production Research, 20, 4235-4265. Thompson, J.D. (1967). Organizations in Action. New York, US: McGraw Hill. Ülkü, S., & Schmidt, G.M. (2011). Matching Product Architecture and Supply Chain Configuration. Production and Operations Management, 20, 16-31. 34 Uzzi, B. (1997). Social structure and competition in interfirm networks: The paradox of embeddedness. Administrative Science Quarterly, 42, 35-67. Uzzi, B., & Spiro, J. (2005). Collaboration and creativity: The small world problem. American Journal of Sociology, 111, 447–504. Victor, B., & Blackburn, R.S. (1987). Interdependence: An alternative conceptualization. Academy of Management Review, 12, 486-498. Wathne, K.H., & Heide, J.B. (2000). Opportunism in interfirm relationships: Forms, outcomes, and solutions. Journal of Marketing, 64, 36–51. Watts, D.J. (1999). Networks, dynamics, and the small-world phenomenon. American Journal of Sociology, 105, 493-527. Watts, D.J. (2003). Six Degrees. New York, US: W.W. Norton & Co.. Watts, D.J., & Strogatz, S.H. (1998). Collective dynamics of ‘small-world’ networks. Nature, 393, 440442. Wicks, A.C., Berman, S.L., & Jones, T.M. (1999). The structure of optimal trust: Moral and strategic implications. Academy of Management Review 24, 99-116 Williamson, O.E. 1975. Markets and Hierarchies. New York, US: The Free Press. Wolf, J.R., & Muhanna, W.A. (2011). Feedback mechanisms, judgment bias, and trust formation in online auctions. Decision Sciences, 42(1), 43-68. Yilmaz, C., Sezen, B., & Ozdemir, O. (2005). Joint and interactive effects of trust and (inter) dependence on relational behaviors in long-term channel dyads. Industrial Marketing Management, 34, 235248. Zaheer, A., & Venkatraman, N. (1995). Relational governance as an interorganizational strategy: An empirical test of the role of trust in economic exchange. Strategic Management Journal, 16, 373-392. Zaheer, A., McEvily, B., Perrone, V. (1998). Does trust matter? Exploring the effects of interorganizational and interpersonal trust on performance. Organization Science, 9, 141-159. Zajac, E.J., Olsen, C.P. (1993). From transaction cost to transactional value analysis: Implications for the study of interorganizational strategies. Journal of Management Studies, 30, 131-145. 35 Figure 1. Illustrative supply chain fitness landscape (N=3; K=1) and the associated influence matrix. 1.a – Fitness landscape 1.b – Influence matrix Choice configurations f1 Partner firms (fi) f2 f3 Partner firms’ performance (Pfi) Pf1 Pf2 Pf3 Supply chain performance (Psc) a 0 0 0 0.13 0.34 0.96 0.48 b 0 0 1 0.13 0.68 0.34 0.38 c 0 1 0 0.57 0.70 0.96 0.74 d 1 0 0 0.54 0.34 0.91 0.60 e 1 1 0 0.15 0.70 0.91 0.59 f 1 0 1 0.54 0.68 0.08 0.43 g 0 1 1 0.57 0.49 0.34 0.47 h 1 1 1 0.15 0.49 0.08 0.24 36 f1 f1 x f2 f3 x f2 x f3 x x x Figure 2. The interdependence patterns.a Patterns Examples Random x x x x x x xx xx x xx x x x x x x x x x x Supply chains within “classic” Italian industrial districts. x xx x x x x x x xx x x Local Just-in-time supply chains. xx x xxx xxx xxx xxx xxx xxx xxx xxx xxx xxx x xx “Ring” production networks. Small-world Just-in-time supply chains wherein the phases of the overall process are shared among clusters of interconnected partners and the clusters are directly connected between them and/or share common contacts x x x xxx xxx x xx xxx xxx xxx x x x xxx xxx x xx x xx Block-diagonal xxx xxx xxx Modular product supply chains. xxx xxx xxx xxx xxx xxx xxx xxx xxx Preferential attachment x x x x x Vertical keiretsu supply chains. xx x x x x x x x x x xx x xx x xxx x x x x xx x x x x x 37 Scale-free xx xx xx xxx x x x x Hub-and-spoke distribution/transportation systems. x x x x x Hub-and-spoke industrial district supply chains. xx x x xx xx x x xxx x x x xx Centralized x x x x x x x x x x x x Supply chain networks adopting centralized inventory control programs such as VMI and CR. xx xx xx x x x x x x x x x x x x x x x x x x Prato textile district supply chains. Convergent supply chains. Hierarchical x x x x x x x x x x x x x x x x x x x x x x x x xx x x x x Assembling supply chains where the first assembler makes production planning decisions for the entire network and relies on a back-up assembler when needed. x x x x x x Diagonal x Project supply chains where a main contractor subcontracts the work to a first layer of subcontractors, who in turn subcontract their work to second-layer subcontractors. x x xxx x x x x xxxxx x x xx x x x xx x x x x x xx x x x xx Multi-tiered supply chain networks. Dependent x Supply chains using postponement strategies. x x x x x x Distribution supply chains with small numbers of large warehouses serving large numbers of independent retailers. x x xxx xxxxxxxxxxxx xxxxxxxxxxxx Integral product supply chains with black-box suppliers. a The 10 illustrative matrices are those reported by Rivkin and Siggelkow (2007). All matrices have N = 12, K = 2, and the same number of total interactions (N * (K + 1) = 36). 38 Table 1. Results – first-best scenarioa Supply Chain Partners Patterns 1 2 3 Random K=2 35.17% 32.17% 35.67% K=3 19.67% 25.83% 17.67% K=5 5.67% 6.33% 6.33% Local K=2 35.17% 37.17% 33.17% K=3 24.50% 22.00% 22.67% K=5 5.83% 6.17% 5.33% Small-world K=2 37.00% 37.33% 39.33% K=3 22.83% 24.17% 23.00% K=5 5.83% 6.50% 5.33% Block-diagonal K=2 36.67% 33.00% 33.17% K=3 21.00% 17.00% 20.67% K=5 4.17% 6.00% 5.67% Preferential attachment K=2 30.17% 100.00% 18.50% K=3 16.83% 22.00% 10.33% K=5 12.50% 20.33% 9.00% Scale-free K=2 36.17% 27.67% 34.17% K=3 18.83% 13.50% 33.83% K=5 4.17% 3.83% 1.33% Centralized K=2 28.00% 29.83% 23.50% K=3 17.00% 12.33% 15.00% K=5 3.33% 5.50% 4.67% Hierarchical K=2 63.67% 40.00% 27.67% K=3 68.67% 38.33% 25.83% K=5 69.83% 49.67% 28.50% Diagonal K=2 78.17% 70.50% 72.83% K=3 72.17% 65.17% 45.00% K=5 70.83% 47.83% 28.83% Dependent K=2 87.50% 86.33% 85.33% K=3 85.00% 84.17% 85.33% K=5 79.33% 81.67% 79.50% a Results are computed over 600 landscapes. 4 5 6 7 8 9 10 11 12 SCT_fb 35.50% 19.83% 6.00% 36.50% 22.50% 7.83% 37.83% 19.83% 5.67% 35.00% 17.00% 5.67% 36.83% 21.83% 5.17% 36.33% 18.00% 5.17% 38.50% 25.33% 4.67% 38.67% 17.00% 6.83% 36.67% 24.00% 5.67% 36.24% 20.71% 5.92% 37.33% 20.17% 6.67% 34.67% 23.83% 6.00% 33.17% 17.50% 6.67% 36.50% 21.50% 7.33% 34.67% 23.33% 4.67% 38.00% 20.00% 6.17% 36.00% 22.50% 7.17% 38.00% 20.50% 6.50% 35.00% 22.33% 6.67% 35.74% 21.74% 6.26% 35.17% 20.17% 4.67% 33.33% 19.17% 5.67% 31.33% 21.17% 6.83% 33.50% 20.33% 6.17% 35.00% 19.17% 6.00% 39.67% 20.17% 7.67% 42.17% 23.17% 5.00% 36.67% 19.50% 8.17% 33.67% 19.00% 6.33% 36.18% 20.99% 6.18% 34.33% 21.50% 4.83% 34.17% 19.00% 5.67% 35.50% 19.50% 7.50% 34.83% 20.33% 3.50% 36.50% 19.50% 6.50% 32.00% 20.83% 7.00% 33.83% 20.67% 5.67% 34.17% 17.00% 6.17% 34.00% 21.00% 7.00% 34.35% 19.83% 5.81% 65.50% 18.83% 4.17% 32.33% 9.50% 1.33% 21.67% 30.83% 5.50% 41.50% 11.50% 24.33% 31.83% 29.33% 10.50% 19.50% 6.67% 1.33% 25.33% 41.83% 6.00% 21.83% 32.33% 7.00% 76.17% 48.33% 1.00% 40.36% 23.19% 8.58% 24.67% 17.33% 7.33% 23.17% 37.50% 4.00% 62.17% 13.00% 4.33% 42.00% 11.83% 12.33% 29.83% 9.50% 5.17% 44.50% 22.17% 4.00% 22.67% 38.67% 11.33% 37.17% 16.50% 9.33% 34.50% 18.67% 19.83% 34.89% 20.94% 7.25% 31.83% 14.33% 3.83% 33.83% 17.33% 4.00% 37.00% 19.17% 4.33% 38.00% 20.67% 5.83% 34.50% 22.33% 6.67% 33.83% 20.83% 5.50% 32.00% 17.33% 6.50% 36.83% 21.00% 6.67% 34.00% 20.67% 5.67% 32.76% 18.17% 5.21% 21.67% 15.33% 18.33% 20.50% 10.67% 9.33% 24.17% 10.50% 5.00% 34.83% 11.67% 3.83% 33.67% 10.83% 2.00% 34.33% 12.67% 1.50% 35.17% 11.83% 1.00% 39.00% 18.67% 0.83% 34.33% 21.17% 1.00% 34.08% 21.35% 15.90% 32.17% 30.17% 36.83% 49.50% 33.83% 10.50% 53.33% 40.33% 5.50% 7.50% 5.33% 4.83% 24.50% 6.50% 0.33% 27.50% 14.50% 1.17% 36.50% 6.50% 0.17% 25.50% 13.33% 1.50% 21.50% 4.33% 0.00% 41.63% 28.10% 17.36% 87.50% 83.50% 80.50% 86.00% 85.33% 82.67% 88.17% 81.50% 83.83% 86.17% 82.83% 10.50% 87.83% 86.83% 0.00% 84.67% 33.67% 0.17% 55.67% 0.00% 0.00% 0.00% 0.17% 0.00% 0.17% 0.17% 0.17% 69.61% 59.04% 41.53% 39 Table 2. Results – second-best scenario a Supply Chain Partners Patterns 1 2 3 4 Random K=2 25.14% 23.39% 25.33% 25.64% K=3 12.76% 15.02% 10.72% 12.39% K=5 3.32% 3.86% 3.22% 2.96% Local K=2 25.79% 26.58% 26.61% 24.65% K=3 13.60% 14.32% 13.34% 14.00% K=5 3.30% 3.31% 3.17% 3.20% Small-world K=2 26.32% 28.26% 29.12% 23.06% K=3 13.91% 14.15% 13.63% 12.03% K=5 3.21% 3.23% 3.18% 3.01% Block-diagonal K=2 27.90% 26.20% 21.80% 27.80% K=3 14.90% 11.90% 14.60% 12.60% K=5 3.40% 3.10% 3.00% 3.50% Preferential attachment K=2 21.33% 100.00% 14.24% 52.25% K=3 11.78% 11.53% 6.95% 11.79% K=5 5.38% 9.43% 5.00% 2.20% Scale-free K=2 31.72% 18.98% 19.96% 15.90% K=3 13.15% 7.59% 22.90% 9.05% K=5 2.47% 1.68% 1.02% 4.01% Centralized K=2 19.22% 20.49% 19.31% 27.75% K=3 10.22% 9.36% 9.84% 10.71% K=5 2.32% 2.41% 2.31% 2.84% Hierarchical K=2 52.01% 27.51% 16.66% 15.63% K=3 52.21% 27.20% 14.66% 8.22% K=5 53.61% 28.68% 15.15% 8.43% Diagonal K=2 57.35% 57.15% 58.18% 22.45% K=3 55.05% 53.83% 29.96% 19.99% K=5 53.35% 29.24% 16.24% 16.68% Dependent K=2 57.29% 57.14% 57.16% 57.09% K=3 54.96% 55.57% 55.45% 55.28% K=5 53.57% 53.86% 54.09% 54.27% a Results are computed over 600 landscapes. 5 6 7 8 9 10 11 22.74% 14.15% 3.73% 29.35% 13.35% 3.06% 23.89% 11.61% 3.14% 25.93% 15.26% 3.19% 24.36% 12.54% 3.48% 25.87% 15.03% 3.16% 27.91% 11.32% 3.39% 24.96% 13.34% 3.41% 26.66% 12.86% 3.46% 26.51% 14.12% 3.11% 25.72% 13.04% 3.30% 24.93% 12.96% 3.14% 27.59% 26.34% 26.94% 26.11% 13.51% 13.57% 12.62% 13.44% 3.40% 3.16% 3.32% 3.27% 23.41% 12.97% 3.44% 24.53% 13.56% 3.44% 25.12% 13.20% 3.45% 24.14% 12.33% 3.39% 29.15% 14.83% 3.26% 28.53% 14.04% 3.08% 22.89% 11.87% 3.13% 24.61% 25.76% 12.68% 13.27% 3.50% 3.28% 25.70% 13.30% 3.30% 26.10% 12.50% 3.50% 25.10% 14.70% 2.90% 29.90% 12.00% 3.10% 24.60% 12.60% 3.00% 29.00% 12.60% 3.50% 29.00% 13.00% 3.80% 27.80% 14.00% 3.100% 22.12% 8.18% 0.90% 13.86% 18.36% 3.43% 27.91% 7.23% 11.43% 20.18% 17.19% 5.33% 13.10% 5.38% 1.17% 15.73% 25.57% 2.77% 12.53% 18.93% 3.21% 63.63% 31.41% 33.45% 14.70% 1.01% 4.27% 17.99% 27.60% 2.41% 54.30% 9.75% 2.34% 32.91% 7.17% 7.64% 18.85% 8.19% 2.61% 31.39% 13.23% 2.53% 15.86% 28.77% 5.24% 27.79% 9.19% 4.71% 27.75% 26.12% 12.26% 14.07% 10.18% 3.90% 25.98% 14.44% 2.73% 26.67% 13.36% 2.77% 26.67% 13.68% 3.80% 26.76% 13.74% 3.80% 26.37% 13.57% 4.01% 26.37% 13.41% 3.87% 27.45% 13.30% 3.90% 26.76% 24.98% 13.74% 12.45% 3.82% 3.22% 14.61% 7.99% 4.34% 15.90% 7.48% 2.31% 27.44% 7.82% 1.32% 27.65% 7.28% 0.73% 26.62% 7.85% 0.62% 26.69% 7.52% 0.64% 26.01% 14.29% 0.53% 26.89% 25.30% 14.36% 14.74% 0.50% 9.74% 32.78% 16.81% 5.11% 40.55% 19.97% 2.73% 4.76% 3.80% 2.60% 14.42% 3.98% 0.80% 17.26% 7.90% 0.87% 24.39% 3.57% 0.43% 15.48% 7.78% 0.52% 11.08% 29.65% 2.94% 18.80% 0.10% 10.72% 56.85% 55.64% 54.17% 57.08% 55.43% 53.96% 56.91% 55.63% 2.51% 57.09% 55.58% 0.06% 57.02% 9.36% 0.06% 18.00% 0.10% 0.06% 0.16% 0.11% 0.05% 0.14% 44.33% 0.08% 37.77% 0.07% 27.23% 40 12 SCT_sb 26.66% 25.52% 13.47% 13.13% 3.14% 3.31% 26.80% 13.20% 3.30% Figure 2. Uninfluenced and uninfluencial partners in the dependent and centralized patterns. Centralized Dependent Uninfluenced partners Uninfluencial partners Table 3. Regression results for supply chain-level trust (first-best scenario; K=3). Dependent Variable: SCT_fb Index Intercept Uninfluenced partners Uninfluential partners Multiple R R2 Adjusted R2 * p<.01. ** p<.00001. Coefficients .23925 .03520 -.00407 Standard Error .00492 .00257 .00194 T-stat 48.62345** 13.72117** -2.10198* .8284 .6863 .6793 Table 4. Decrease (%) of supply chain-level trust (SCT_fb index) following increases in K. K=2 Î K=3 K=3 Î K=5 Random Local 42.85 71.41 39.17 71.21 Smallworld 41.98 70.56 Blockdiagonal 42.27 70.70 41 Pref. attach. 42.54 63.00 Scalefree 39.98 65.38 Centralized 44.54 71.33 Hierarchical 37.35 25.53 Diagonal 32.50 38.22 Dependent 15.18 29.66