Competitive and Cooperative Positioning in Supply Chain Logistics Relationships*
By Klein, Richard Rai, Arun; Straub, Detmar W
ABSTRACT Cooperative logistics relationships require the sharing of information, which must be enabled by the integration of disparate information systems across partners. In this article, we theorize business-to-business logistics relationships should be managed using cooperative and competitive postures. Based on data from 91 dyadic relationships using interorganizational information technology (IT), we find that performance gains accrue when parties share strategic information and customize IT; mutual trust enables IT customization and strategic-information flows and equitable relationship-specific investments positively impact IT customization, mutual trust, and performance. Among other scholarly and practical implications discussed, partners should compete on resources for IT customization and cooperate to share strategic information. Managers tend to think of relationships with firms as polar opposites and view them as entirely cooperative or entirely competitive. Our results support active balancing and understanding of both competitive and cooperative stances. Such an approach enables conditions for participation symmetry that yields greater performance gains.
Subject Areas: Collaborative Partnerships, Dyadic Data, Exchange Relationships, Field Studies, Information Sharing, Logistics Service Performance, Structural Equation Modeling, and System Integration.
INTRODUCTION: THE MANAGERIAL PROBLEM OF RELATING TO PARTNERS
Recently, collaborative customer relationships and integrated partnerships have been gaining in popularity in business-to- business (B2B) markets (Day, 2000). Traditional intermediaries for logistics and distribution are changing their roles and value propositions through digital networks (Brousseau, 2002). These service providers are drawing on vast informational resources to exploit market knowledge. While early Internet-based initiatives fostered sales and new transaction opportunities, recent innovations redefine relationships, share processes, and increase collaboration (Wladawsky-Berger, 2000; Patnayakuni, Rai, & Seth, 2006; Rai, Patnayakuni, & Patnayakuni, 2006). Accordingly, the ability to easily, efficiently, and economically access information outside firm boundaries (Hitt, 2000) can generate further efficiencies for participants in exchange relationships (Lee & Whang, 2000; Kotabe, Martin, & Domoto, 2003; Vickery, Jayaram, Droge, & Calantone, 2003).
These generalized informational advantages aside, how should managers sculpt their day-to-day relationships with vendors, particularly logistics providers? In which cases should they be concerned about opportunism and in which cases should they not? Every day, managers make decisions about the posture they need to assume regarding business partners. If the posture is competitive, they bargain for the best deal for their own organization. If the posture is cooperative, then they must take into account the consequences of decisions and actions on their business partner.
Earlier management thinking viewed the vertical supplier-firm and customer-firm relationships as quintessentially competitive. In Porter’s work on industry structure (Porter, 1985; Porter & Millar, 1985), managers cope with the five forces by improving bargaining power with respect to customers or suppliers. In fact, the force of internal rivalry between competitors is viewed in the same light. The environment in which a manager makes decisions about interacting with other firms is seen as being entirely adversarial.
Management thinking has evolved to perceive customers and suppliers as partners and to focus more attention on the relationship (Brandenburger & NaIebuff, 1996). This evolution has occurred across the administrative disciplines. In marketing, for example, the earlier view of customers was an arm’s-length, transactional view (Coviello, Brodie, Danaher, & Johnston, 2002). Later, this was supplemented by a view that not all customers should be held at arm’s length. The term relationship marketing captured this perspective, likely introduced by Berry (1983).
In the management literature, the concept of the relational view has emerged to describe situations in which B2B trading partners are, in effect, competing for resources with each other, but in which the advantages conferred by the relationship, in some cases, outweigh the need to be opportunistic (Dyer & Singh, 1998). What this change in paradigmatic focus suggests is that decision makers can no longer assume that their partners are adversarial and that they should be entirely wary in their dealings with them. The new view posits that such cooperative activities as the extensions of trust and information sharing work well in some situations.
Consider the traditional logistics intermediary relationship that is being transformed. Expanded electronic intermediary functions include sharing private information related to inventory movement or financial flows based on me characteristics of the customer’s supply chain or product. Such information sharing requires the customization of information technology (IT) systems used in an exchange relationship. By integrating the vendor’s logistics IT solutions with their own organizational systems, processes in the client and vendor firms can be aligned (Stein, 1998; Walker, Bovet, & Martha, 2000). In fact, clients can tap into information held by vendors to streamline processes, develop value-added products and services, and strengthen customer ties (Gulati & Kletter, 2005). On the other hand, logistics vendors can pool information about client requirements across time, channels, and services, to globally optimize plans and adapt process execution (Lewis, Rai, Forquer, & Quinter, 2007).
However, will these asset-specific investments lead to mutual benefits and stronger cooperation in relationships? Before exploring this topic, though, we must consider the underlying competitive dynamics for the IT investment required to support the partnership. If firms in logistics exchange relationships intend to cooperate by sharing private information, they must create such customized IT systems. While logistics vendors have developed technology solutions that offer clients value-added, information-based services, employing even the most generic solution requires some integration effort on the part of at least one party to the exchange. The issue that emerges is how partners should contribute to the IT customization, in that the client and vendor firms vary in their specializations in hardware platforms, telecommunications protocols, data formats, enterprise systems, process standards, and employee skill sets (Gnyawali & Madhavan, 2001). These variations generate competitive forces that shape the negotiation between partners with respect to who bears the burden for IT customization or whose resources and capabilities dominate.
Past research indicates that the vendor firm’s positioning, specifically centrality within the industry as a whole, affects the firm’s ability to negotiate and influence how assets, information, and status flow, giving rise to resource asymmetries (Gulati, Nohria, & Zaheer, 2000). Centrality provides the vendor with the power to establish standards for services and to achieve economies of scale. In this scenario, clients are likely to assume a disproportionate amount of the effort to integrate sourced services with their internal systems and processes.
Moreover, vendor firms can achieve growth through the replication of technology and solutions subject to lack of imitation (Kogut & Zander, 1992). Not surprisingly, logistics vendors who are dominant in the industry strive to develop repeatable solutions for supply chain services such as merge-in-transit, cross-docking, inventory management, and order fulfillment (Sonnenfeld & Lazo, 1992; Quelch & Conley, 1997; Rivkin, 1999; Lewis et al., 2007).
Alternatively, vendors not occupying a central position within an industry may serve a niche market segment or specific dominant clients. Accordingly, the vendor may bear the burden of customization. For instance, consider Fritz Company, a freight forwarder and now a part of UPS Supply Chain Solutions. This vendor assumed the IT customization onus to establish the necessary visibility for the just-in-time delivery solution that it rolled out for Apple Computers (Healy, Laschinger, & Shroff, 2002; Morton, 2002).
Finally, a client firm that holds a central position within its own industry and represents a significant potential revenue stream can motivate the vendor to bear a significant portion of the customization onus (Subramani & Venkatraman, 2003). In fact, the vendor can improve its own network position, enabling it to tap into additional resources and opportunities by establishing ties with just such a high-centrality client (Gulati, 1998). In addition, a dominant client firm may be so central within its industry that it holds far greater negotiating power than the vendor, as in the case of Wal-Mart (Rigby & Haas, 2004; Yoffie & Mack, 2005).
Common to all of the noted scenarios, one party within the exchange relationship bears a disproportionate burden for IT customization required for information flows. This situation gives rise to tension between parties as each takes a posture with respect to its investments for the customization effort, even when they are to cooperate for information sharing. It becomes important then to understand the conditions under which safeguards exist in a relationship to enable firms to make relationship-specific investments in IT customization, which should facilitate both firms sharing strategic information. In sum, best practice demands that managers make conscious decisions about safeguarding their own positions while also opening themselves to the possibility of gaining Ricardian rents through benefits generated via interorganizational relationships. Given the previous discussion, our study examines two related managerial questions: (i) How can managers safeguard disproportionate investments for IT customization and cooperation for the flows of strategic information in logistics exchange relationships? (ii) What are the performance implications for an exchange relationship that includes a conscious strategy for maintaining competitive and cooperative postures for IT customization and information sharing, respectively? In addressing these questions, we draw upon the relational exchange perspective in marketing.
To test our model, we gathered dyadic data in a field study of a Global 500 logistics vendor’s client-vendor exchange relationships employing interfirm technology solutions. These vendor offerings are specifically aimed at facilitating interfirm information flows between client and vendor. We demonstrate how logistics firms and their clients make relationship-specific investments to customize IT for a relationship and how they share information. Mutual trust and the symmetry in contributions to relationship-specific investments create an environment of cooperative governance, enabling cooperative information sharing while safeguarding disproportionate IT customization investments. Moreover, we find that this duality of competition for disproportionate investments in IT customization and cooperation for information sharing yields mutually beneficial performance gains.
This article proceeds by detailing the theoretical rationale of our research model that specifies the competitive and cooperative initiatives in exchange relationships for logistics services. Subsequently, we outline specific research hypotheses and then describe the research methodology, data analysis, and results. In the final sections, we interpret the findings and their implications for theory, practice, and future research.
Marketing exchange theories tend to focus on relationship exchanges, specifically on (i) behaviors of buyers and (ii) sellers directed at consummating exchanges, (iii) institutional frameworks directed at facilitating exchanges, and (iv) societal consequences of buyers, sellers, and institutions (Hunt, 1983). These B2B exchange relationships range from transactional to collaborative (Day, 2000), with the transactional end of the spectrum characterized by anonymous transactions and automated purchasing and the collaborative end exhibiting high integration between supplier and customer/channel partners.
To engage in exchange relationships, firms must develop capabilities that are multifunctional and cover numerous organizational levels, including, but not limited to, information sharing and process integration (Day, 2000). Business strategies aimed at customer offerings that meet basic transactional requirements (i.e., physical delivery within the logistics function) are required for purely transactional exchanges. By contrast, collaborative exchanges with customers and channel partners require technology and process integration. The current evolution in technological capabilities and depth of information resources available to logistics providers has positioned these vendors to formulate offerings that increase their level of collaboration with partners (Brousseau, 2002).
Discussions about the content of exchange have also evolved in die B2Brelationship theoretical literature. Past research points to an increasing shift in these relationships from the exchange of tangible goods to the exchange of intangibles, namely specialized skills, knowledge, and processes (Vargo & Lusch, 2004). Consistent with the service-centric view, customers in a logistics exchange relationship are coproducers, with vendors engaging client firms in customizing offerings that maximize value for them while continuing to meet their immediate transactional needs. There are, however, unique challenges to managing servicecentered exchange relationships. Namely, “firms must learn to be simultaneously competitive and collaborative” (Vargo & Lusch, 2004, p. 13).
Strategic-information sharing is a key capability required under the servicecentric perspective for collaborative exchange relationships. However, achieving some degree of equity in strategic- information sharing requires a certain level of IT customization. Importantly, it is not just the magnitude of customization that is relevant here. Competitive positioning in achieving the requisite IT customization should result in partners efficiently using their technological resources and capabilities, ideally avoiding duplication of effort for customization. Thus, the degree of exchange partners’ collective IT customization determines whether competitive forces have been applied to the configuration of this asset for the exchange relationship.
Based on this line of reasoning, we suggest that earning Ricardian rents requires firms to adopt cooperative positions with their partners for some initiatives and to assume competitive positions for others. Performance benefits accrue to relationships that efficiently manage this delicate dynamic between cooperating and negotiating hard with partners. This is in contrast to past research that focuses on either competitive or cooperative relationships, and capabilities for each of these types of relationships, but not on managing these domains simultaneously (see, however, Brandenburger & Nalebuff, 1996). For example, Rai, Borah, and Ramaprasad (1996) suggest strategic alliances that cooperate to share information can result in a win-win game. Similarly, Joskow (1988) and Nooteboom (1996) show that competitive positioning shapes the unequal nature of contributions made by parties with respect to time, organizational resources, and effort for IT customization. Thus, there is a theoretical gap in evaluating how exchange relationships can be structured to simultaneously cooperate in certain domains and to compete in others.
To begin to fill the gap in how managers should structure logistics exchange relationships, we draw on relational models that suggest cooperative governance architectures safeguard asymmetric investments by one partner and promote cooperation in resource sharing. Specifically, mutual trust (Das & Teng, 1998) and relationship-specific investments are two key elements of a cooperative governance orientation. The stock of investments in relationships includes financial costs and/or investments in time and effort expended to develop the exchange relationship (Joskow, 1988; Dasgupta & Sengupta, 1993; Bensaou & Anderson, 1999).
In our focusing on the relational exchange perspective, we should note why transaction cost economic (TCE) theory is an inadequate theoretical basis from which to understand how logistics exchange relationships should be structured to be mutually beneficial. First, TCE posits that firms must safeguard against opportunism within interorganizational exchange relationships. Thus, TCE is not designed to explain value cocreation through collaboration in exchange relationships. Second, TCE identifies mechanisms, such as monitoring and contractual governance, that can be applied to safeguard against opportunism, especially under conditions of high- asset specificity (Williamson, 1985). Logistics exchange relationships involve collaboration patterns, which are hard to control through formal monitoring and contracting. In these relationships, the client and the vendor share private information, the vendor has visibility to physical and financial flows between its client and their customers, and the vendor can learn about the growth or decline in its client’s business with customers and across channels and products based on order and shipment patterns. Prior marketing research efforts found similar limitations in applying TCE to the manufacturer-principal agent channel relationship (Heide & John, 1988).
Accordingly, we draw on the relational-exchange perspective to posit that the degree and symmetry of trust and relationship- specific investments should promote cooperation between partners for information sharing while safeguarding IT customization partners undertake. In turn, high IT customization levels and mutual information sharing should promote mutual performance gains for the client and vendor.
ASSESSING DEGREE AND SYMMETRY IN RELATIONAL DYADS
Traditional monadic models conceptualize constructs on either one side of the business relationship or the other. That is, the researcher tests supplier/vendor-only or buyer/client-only models. Our research considers pairs, or dyads, of firms as the unit of analysis, which is the most natural approach for studying interfirm exchange relationships (demons & Row, 1993; Anderson, Hakansson, & Johanson, 1994; Dyer & Singh, 1998; Chen & Paulraj, 2004). Practical difficulties associated with dyadic research designs often lead researchers to collect, and subsequently examine, only one side of the relationship. Kotabe et al. (2003), for example, explored how relationship duration in customer and buyer dyads improved performance in U.S. and Japanese auto firms. Yet, likely because of the complications of acquiring customer data matched with supplier data, monadic supplier performance measures served as the basis for their data analysis. Examining both sides of exchange relationships at once allows for measures of the relational dyadic symmetry. However, measuring only symmetry in a relationship fails to capture the degree or extent of the construct values. The empirically modeled and tested technique (Straub, Rai, & Klein, 2004) employed here conceptualizes the degree symmetry of constructs by assessing both symmetry and value (i.e., degree or magnitude). Table 1 details the derivation of degree-symmetric constructs. In brief, (i) summing all measures for a given construct and standardizing to a value between O and 1 yields the magnitude for the client, C^sub C^, and vendor, C^sub V^. Next, (ii) the mean value of the client and vendor magnitudes, C^sub C^ and C^sub V^, yields the degree value, C^sub DV^. Conversely, (iii) dividing the lesser magnitude by the greater yields a standardized value between 0 and 1, reflecting the symmetric value of the construct, C^sub SV^. Ultimately, (iv) the mean value of C^sub DV^ and C^sub SV^ yields the degree-symmetric value for the construct, C^sub DS^.
We capture dyadic cooperative behaviors through measures that simultaneously assess both degree and symmetry within the logistics provider-client dyad through degree-symmetric (C^sub DS^) constructs. We apply the degree symmetry formulation to other constructs for which we hypothesize the existence of higher levels of a variable for both parties for the sake of parity across the exchange relationship. Within the current work, we examine performance outcomes, specifically focusing on the extent to which these outcomes achieve higher levels and equivalence for both vendor and client. Hence, we apply this previously validated approach (Straub et al., 2004) to assessing the total magnitude mathematically combined with symmetry as a surrogate for the dyadic relationship-specific performance-dependent variable. Moreover, in considering relative levels of trust that have developed in the other party and each party’s investment in developing the relationship, we similarly consider the effects for higher levels of the constructs that are also equivalent across the dyad.
Finally, the degree value (C^sub DV^) measures the level of IT customization in the exchange relationship, irrespective of each firm’s contributions. Consistent with the service-centric view, vendors treat each relationship independently of others potentially absorbing the customization for some relationships, while in other instances deferring to clients or some combination in between. This construct formulation represents the result of competitive positioning for IT customization within dyadic exchange relationships.
MODEL OF INTERFIRM LOGISTICS EXCHANGE RELATIONSHIPS
Figure 1 depicts our proposed model of competitive and cooperative positioning in logistics exchange relationships. The current research greatly extends prior work proposing degree- symmetric constructs (Straub et al., 2004) by exploring both competitive and cooperative positioning within logistics exchange relationships. Specifically, we investigate the relationships between IT customization resulting from competitive positioning and strategic-information flows (H^sub 1^) resulting from interfirm cooperation. In addition, we focus on performance outcomes and examine antecedents to equitable higher levels of performance, namely, strategicinformation flows (H^sub 2^) and relationship- specific investments (H^sub 7^). Our model further considers the relationship between mutual trust (H^sub 3^) and IT customization. In addition, we hypothesize that mutual trust (H^sub 4^) and relationship-specific investment (H^sub 6^) are associated with strategic-information flows. Finally, we posit a relationship between relationship-specific investments and mutual trust (H^sub 5^).
Table 1: Degree and degreee-symmetric construct derivations.
Figure 1: Model of competitive and cooperative positioning in logistics exchange relationships.
Competitive Positioning Enabling Cooperative Behaviors
Within the context of logistics exchange relationships, vendors focusing on developing service-centered, value-added initiatives with clients tend to tap into information-sharing opportunities in an effort to advance strategic partnerships. Traditionally, information sharing within exchange relationships ranges from open lines of communication to complete digital integration of partners, often accomplished through electronic data interchange (EDI) (Day, 2000). Internet-based electronic business applications represent a form of interorganizational systems that enable buyers to interact digitally with suppliers. EDI and electronic business systems share common features, although EDI is far more expensive and falls out of the reach of smaller enterprises (Zhu & Kraemer, 2005). The proliferation of Internetbased business applications overcomes many financial and technical constraints inherent with EDI.
Researchers identify operational, strategic, and strategic/ competitive classes of private information shared between partners in supply chain relationships (Seidmann & Sundararajan, 1997). Shared operational information leverages the capabilities of the other partner. Consider vendor-managed inventory systems, in which a buyer shares inventory position data. These data leverage the supplier’s capabilities witii respect to replenishment and inventory management. In turn, buyers develop and refine the functionality, service levels, and cost structures of their operations and offerings. Strategic information yields specific value to the owning firm and has the potential to accrue operational benefits to business partners through sharing. For example, point-of-sale history offers limited and specific value to a firm, but enhances the business partner’s ability to successfully forecast demand and subsequently improves service efficiency to client firms. Finally, strategic/competitive information enables the owning party to allow partners to derive additional benefits through sharing across strategic areas such as product/service development as well as sales and marketing.
This work additionally notes the existence of order information, which more closely resembles data than information (Seidmann & Sundararajan, 1997). Such information most often describes the transactions necessary for the execution of business exchanges. The operational, strategic, and strategic/competitive information classes by contrast go beyond order information, with such order- extrinsic information possessing the ability to create value-added gains for firms engaged in the relationship. For the sake of simplicity, we use the term strategic in referring to private, order- extrinsic information.
Information sharing among partners calls for some level of IT customization due to the varying specializations or differences in capabilities across firm boundaries (Carlile & Rebentisch, 2003). IT- related resource differences in hardware platforms, telecommunications protocols, data formats and process standards, enterprise systems, and employee skill sets arise from the differentials in the flow of resources and capabilities among partners (Gnyawali & Madhavan, 2001). In attempting to achieve the necessary levels of IT integration, partners invariably vie to have their unique resources, such as standards and legacy systems, drive these relationship-specific asset investments and/or otherwise mitigate their expended effort and investment.
The core theoretical argument that we make is that buyers and sellers within exchange relationships pursue investments in relationship-specific technology assets that enable increased collaboration between them and possess a joint profitmaximizing benefit (Ring & Ven, 1992; Gulati, 1995). Based on this logic, the customized IT infrastructure that results from the relationship- specific IT investments should facilitate the exchange of information between parties in the relationship. We make no claim with respect to which party assumes the greater burden for the IT customization. Circumstances might dictate that this role falls to either party, a stance that makes the hypothesis more generalizable. We posit that a greater collective level of IT customization effort within the logistics exchange relationship is associated with greater levels of strategic-information sharing coupled with equity in flows across the relational dyad, as stated in our first hypothesis:
H^sub 1^: The greater the IT customization dv, the greater the strategicinformation flows ^sub DS^.
The Information Systems Success Model (DeLone & McLean, 1992,2003) formulates the presence of both individual and organizational performance with potential intermediate levels at different points in between (e.g., the business unit). Using a similar logic, relationship-specific performance seeks to examine outcomes realized by two interacting but independent organizations that have recurring business interactions. The performance results derived through such an exchange relationship are, dierefore, a viable and relevant construct (Straub et al., 2004). Relying on this prior formulation of degree-symmetric constructs, we measure performance by aggregating the magnitude of specific tangible and intangible outcomes for each party to the relationship and the subsequent equity, or symmetry, in these outcomes.
We contend that rent generation requires firms to adopt cooperative positions with their partners for interfirm strategic- information flows. Such an argument is consistent with the notion that managers in strategic alliances promoting cooperative information sharing see performance benefits for both participants (Rai et al., 1996). Similarly, increased operational improvements are noted outcomes of supply chain coordination (Robinson, Sahin, & Gao, 2005). Moreover, information exchanges that support joint planning and forecasting at multiple levels can yield operational performance gains (Saeed, Malhotra, & Grover, 2005). Accordingly, the cooperative-strategies perspective lends further insight into the relationship between information exchanges and performance. Contemporary theoretical work in cooperative strategies evolved from older theories of games and economic behavior (Von Neumann & Morgenstern, 1953). This stream of research posits how businesses cooperate for mutual benefit or, conversely, the concept of noncooperative games (Nash, 1951). Cooperative strategies provide a basis for theorizing how both significant levels and symmetry in information sharing within strategic supply chain relationships can result in greater performance across the dyad. In addition, some work contends that effectively leveraging complimentary resource endowments (i.e., information) generate economic rents for both parties (Hamel, 1991; Hill & Hellriegel, 1994; Walker, Kogut, & Shan, 1997; Dyer & Singh, 1998). Hence, we posit that, when equitable and at sufficient levels, such strategic, order-extrinsic information flows between partners in a logistics relationship equate to similar mutual performance gains, as stated in the following hypothesis:
H^sub 2^: The greater the strategic-information flows ^sub DS^, the greater the relationship-specific performance ^sub DS^.
By investing in customized IT for business partners, firms commit to time, money, and effort with the expectation of enabling gainful future interactions (Ahmad & Schroeder, 2001). All tilings being equal, trust is a necessary but not sufficient condition for repeated market transactions (Ring & Ven, 1992). We contend that mutual trust, although essentially a cooperative mindset, also affects competitive decisions and actions. Research supports this viewpoint, with some work concluding that interorganizational trust serves to mitigate conflict between parties (Zaheer, McEvily, & Perrone, 1998). Work studying network dyads finds trust present in complex interorganizational integration efforts (Larson, 1992). In fact, firms are more likely to make disproportionate IT investments in a relationship if they have established prior trust in their partner’s capabilities and motivations, and vice versa. Such mutual trust potentially safeguards against opportunistic behavior and creates conditions for dedicated investments in a relationship. Hence, our third hypothesis states that both higher levels and parity in trust are correlated with higher levels of IT customization.
H^sub 3^: The greater the mutual trust ^sub DS^, the greater the IT customization dv
Researchers hypothesize that effective governance such as mutual trust between partners can generate economic rents within exchange relationships by lowering transaction costs and/or providing for incentives to engage in value-added initiatives (Dyer & Singh, 1998). Moreover, work emphasizing transaction value highlights the influence of trust in advancing partnering relationship initiatives (Zajac & Olsen, 1993). Other findings suggest that interorganizational trust reduces die negotiation costs for enhancing performance within interfirm relationships (Zaheer et al., 1998). Mutual trust is an effective governance mechanism for lowering transaction costs and reinforcing cooperative behavior between participating firms (Dyer & Singh, 1998). Prior studies also recognize how trust influences cooperation and teamwork within organizations (Jones & George, 1998). Other work also recognizes mutual trust as important in strategic supply chain cooperative relationships when partners work to establish an adequate level of confidence in each other (Das & Teng, 1998). Empirical research has determined that imbalances in dyadic manufacturer and distributor relationships, when one party browbeats the other, negatively influence perceived levels of mutual trust (Anderson & Weitz, 1989). Moreover, recent work suggests that cooperative activity is a natural effect of earned trust (Johnston, McCutcheon, Stuart, & Kerwood, 2004).
Hence, we view the existence of mutual trust as a key ingredient for cooperative behavior. With regard to interfirm exchange relationships, information sharing constitutes one form of cooperative behavior that is potentially promoted through mutual trust. Consistent with these arguments, we assert that higher levels of mutual trust in logistics relationships are associated with higher levels of cooperative behavior by both parties, namely, increased degree and symmetry of strategic-information flows between them.
H^sub 4^: The greater the mutual trust ^sub DS^, the greater the strategic-information flows ^sub DS^.
To sustain and grow relationships, firms make relationship- specific investments in time, money, and effort (Joskow, 1988). For example, to develop relationships, firms dedicate resources to respond to requests for proposals, to market solutions to clients, for customer support, and toward similar initiatives (Dwyer, Schurr, & Oh, 1987). When these costs cannot be readily transferred to otiier vendor relationships, such costs can only yield benefits through continuous interaction.
What pattern of relationship-specific investment should promote mutual trust? Ring and Van de Ven (1992) note that trust emerges as a consequence of recurring market transactions between buyers and sellers, as repeated prior contact enables them to learn about each other and develop norms (Shapiro, Sheppard, & Cheraskin, 1992). Moreover, through these repeated interactions, organizations see the prospect of future interactions, which discourages attempts to seek short-term opportunism and promotes mutual trust (Maitiand, Bryson, & van de Ven, 1985). Thus, firms making investments in a relationship should be far more inclined to trust their partner’s intentions when the partner also makes commensurate investments. Hence, we suggest that a greater degree and symmetry in relationship- specific investments are associated with higher mutual trust between parties.
H^sub 5^: The greater the relationship-specific investment ^sub DS^ the greater the mutual trust ^sub DS^.
Relationship-specific investments can yield productivity gains as specialization of assets constitutes “a necessary condition for rent” (Amit & Schoemaker, 1993, p. 39). Some researchers observe that firms often elect to develop their own specialized assets in conjunction with partners’ assets in an effort to gain market advantages (Klein, Crawford, & Alchian, 1978). Moreover, nonrecoverable investments within strategic interorganizational alliances (i.e., relationship-specific asset investments) have been shown to be positively related to firm performance (Parkhe, 1993). Empirical work also demonstrates that commitment of nonrecoverable resources and effort within strategic alliances is positively related to performance (Parkhe, 1993; Dyer, 1996). Thus, greater degree and symmetry in relationshipspecific investments in logistics exchange relationships should relate to higher relationship- specific performance for both parties, as stated in the following hypothesis.
H^sub 6^: The greater the relationship-specific investment ^sub DS^, the greater the relationship-specific performance ^sub DS^.
As noted earlier, relationship-specific investments in time, money, and effort can serve to sustain and grow cooperation in exchange relationships (Joskow, 1988). Such mutual investments enhance each firm’s capabilities to leverage the partner’s resources. As a result, we suggest that mutual relationship- specific investment should be associated with mutual information sharing (i.e., degree and symmetry in strategic-information flows) between partners in the relationship.
H^sub 7^: The greater the relationship-specific investment ^sub DS^, the greater the strategic-information flows ^sub DS^.
Our study involved both exploratory and subsequent confirmatory phases. During the exploratory phase, both clients and vendors provided qualitative data that led to a validated survey instrument for the primary data collection for theory confirmation (Stone, 1978; Kaplan & Duchon, 1988; Creswell, 1994). The initial phase employed a case technique (Yin, 1994), while the second, confirmatory, phase used a field study methodology (Creswell, 1994). The confirmatory field study examined customizations of integrated logistics supply chain functions between an outsourced service provider and its clients and the other constructs of interest to our study.
Information obtained through the exploratory phase served as the basis for the development of a three-item scale for the IT customization construct as well as a three-item scale for relationship-specific investment not directed at a specific class of assets, such as IT. Table 2 details the items used for all constructs as well as the descriptive statistics from the confirmatory stage for their derived degreesymmetric values. For measures of trust, we adopted a validated 11-item scale of omnibus measures of trust from prior works (McKnight, Cummings, & Chervany, 1998; McKnight, Choudhury, & Kacmar, 2002). This multidimensional measure captures trusting beliefs of interest in the study, namely, ability, benevolence, and integrity, and is based on the integrated model of organizational trust (Mayer, Davis, & Schoorman, 1995).
For strategic-information flows, the qualitative data served as the basis for development of a five-item scale of order-extrinsic (strategic) information shared, specifically, cost structures, inventory/capacity planning, margin structures, marketing strategies, and production schedules. To assess each partner’s performance emerging from the relationship, we created an eight- formative-item scale capturing performance outcomes, namely, improved asset management, improved capacity planning, improved resource control, increased flexibility, increased productivity, lower operating costs, and reduced workflow. Confirmatory Phase
Prior research emphasizes the importance of dyadic research designs to investigate interfirm relationships (Clemons & Row, 1993; Anderson et al., 1994; Dyer, 1996; Chen & Paulraj, 2004). However, practical difficulties associated with such research designs have often led to monadic data collection. Consistent with Dyer (1996), we employed a dyadic research design for our field study. We focused on strategic relationships that employed the vendor’s Internet- based technology solutions for interfirm information flows. As we are interested in strategic interfirm relationships, we considered relationships that were managed through dedicated account managers, as opposed to those served through call center operations.
A secure Internet-based survey tool facilitated final data collection. As with Dyer’s (1996) approach for gathering dyadic data, we worked through a senior executive within the vendor’s e- commerce marketing organization. This executive contacted all account managers assigned to overseeing accounts employing Internet- based IT interfirm solutions. The executive encouraged each of them to complete the survey with regard to a single relationship. As a single account manager oversaw multiple customers, selection of the specific relationship was random with respect to client size, tenure, profitability, and/or nature of recent interactions. Subsequently, primary account representatives at 183 different buyer organizations received the matching survey instrument. In total, 132 of the 183 different vendorfirm account managers responded to the survey for a response rate of 72% for the vendor side of the survey. With respect to clients, 91 of the 183 client contacts responded for a response rate of 49% for the client side of the survey. Therefore, the final matching of client and vendor responses resulted in 182 completed client and vendor surveys or 91 usable matched-pair dyadic responses. Averaging the response rate for the vendor side with the client side yields a 61% overall rate.
Table 2: Summary of descriptive statistics of measures.
Matching responses from firms on both sides of an exchange relationship to form dyads is a difficult undertaking. Given that the average for other research employing this same strategy is approximately 58% (Dyer, 1996; Johnson, Cullen, Sakano, & Takenouchi, 1996; Fein & Anderson, 1997), our study achieved a reasonably high number of usable dyadic data. Consider that in Dyer’s 1996 study of suppliers within the automobile manufacturing industry, he captured 83 usable pairs, reporting a 61% response rate for the supplier side and 77% for manufacturer side. Employing this same survey strategy in studying international cooperative alliances, researchers report achieving a 44% overall response rate with 98 matched dyads (Johnson et al., 1996). Finally, in examining territory and brand choices witiiin manufacturer-distributor relationships, Fein and Anderson (1997) realized 362 usable pairs, with a reported overall response rate of 72%. Thus, our response rate is reasonable for a dyadic interfirm research design.
Sample demographics and descriptive statistics
The sample demographics reflect a diverse representation of both client and vendor respondents with respect to overall work, IT, and business relationship-management experience. Moreover, client firm respondents were primarily from mid to senior management levels, with firms representing a cross-section of major industry sectors. Table 3 provides a profile of the relationships examined within this study.
Methodological literature frequently cites a 60% response rate as a reasonable assurance of an absence of systematic bias from respondents (Bailey, 1978). Again, averaging the vendor and client sides yields a 61% overall rate, meeting the prescribed threshold. Moreover, comparing construct means between the early wave of respondents and those who responded during the fourth and final week of data collection provided an additional insight into nonresponse bias. This wave technique treats late respondents as a proxy for nonrespondents (Bailey, 1978). Exactly 43 of the 182, or 23%, of the total respondents forming the 91 dyads completed the survey during the latter period. Analyses of variance for differences across waves on key characteristics (primary industry, primary location by region, number of employees, tenure, individual respondent’s gender, work experience, IT, and relationship-management experience) revealed no significant differences.
Table 3: Firm demographics.
Common method bias
In specifying measures, our analysis sought to safeguard against common metiiod bias by employing different types of measures for some key constructs and different scale types for certain measures. In addition, we evaluated common method variance by applying the Harmon one-factor test (Podsakoff & Organ, 1986). Five factors were extracted from the data on both sides of the dyad. No single factor accounted for the bulk of the covariance, suggesting that common method bias is not a significant issue for our data.
To complement the Harmon test, we conducted an additional analysis outlined by Podsakoff et al. (2003) and Williams, Edwards, and Vandenberg (2003). This procedure specifies, besides substantive factors, a common method factor whose indicators include all the principal construct items in the partial least squares (PLS) model. The result is the proportion of the variance explained by the common method. As reported within Appendix A, our results show that the average explained variance of substantive indicators is .68, while the average method-based variance is .02. The subsequent ratio of substantive variance to method variance is 34:1, with no significant method factor loadings detected for all but three items at p
Analysis and Results
Quantitative analysis included measurement validation and hypothesis testing. The proposed research model involved multiple interdependent relationships and some formative constructs. This combination of factors was conducive to PLS (Gefen, Straub, & Boudreau, 2000), which allows for the examination of both measurement and structural models (Barclay, Higgins, & Thomson, 1995; Chin, 1998b). The presence of both formative and reflective constructs calls for assessing the convergent and discriminant validity of measures through variants of the standard validation techniques. Formative measures included strategic-information flows and relationship-specific performance. Reflective measures included IT customization, mutual trust, and relationship-specific investments. The measurement model provides for the primary assessment of instrument validity within PLS, through the specification of the relationship between the observed variables, or indicator variables, and the latent variables (Igbaria, Guimaraes, & Davis, 1995).
As noted, this study adopts previously validated measures of strategic-information flows and relationship-specific performance from Straub et al. (2004) and of mutual trust from McKnight et al. (2002). However, as suggested in methodological research, the current study fully retested instrumentation (Straub & Carlson, 1989). Cronbach’s alphas for reflective measures, namely, IT customization, mutual trust, and relationship-specific investment, all exceed the prescribed 0.7 threshold (Nunnally & Bernstein, 1994). For formative measures, specifically, strategicinformation flows and relationship-specific performance, Cronbach’s alphas are not an appropriate test (Chin, 1998a, b; Gefen et al., 2000). Table 4 reports all findings of the reliability and validity analysis.
Average variance extracted (AVE) measures the percentage of overall variance in indicators captured by a latent construct through the ratio of the sum of captured variance and measurement error (Hair, Anderson, Tathem, & Black, 1998). When using PLS as an analytical tool (Gefen et al., 2000), AVE serves as a means for assessing discriminant validity, or the extent to which indicators differentiate among constructs. When the square root of the AVE of a measure exceeds the correlations between the measure and all other measures, adequate discriminant validity exists (Gefen et al., 2000). In Table 4, the pattern of intercorrelations and square roots of AVEs reflect no discriminant validity issues among reflective constructs.
AVE analysis assumes reflective measures; therefore, in assessing formative constructs, we examined the pattern of correlation between items and constructs. We adopted a technique based on work proposing that formative items should correlate with a “global item that summarizes me essence of the construct” (Diamantopoulos & Winklhofer, 2001, p. 272). PLS provides item weights that reflect the influence of individual formative construct items (Bollen & Lennox, 1991). We normalized item weights and men computed and summed the products of item values and weights for each construct, a technique previously proposed (Bagozzi & Fornell, 1982). Next, we examined interitem and item-to-construct correlations. A second version of this technique takes the product of normalized item scores and their PLS weights to derive weighted item scores (Ravichandran & Rai, 2000). Employing either technique, items for each construct correlate higher with one another than with other item measures or constructs, finding no significant discriminant validity issues.
In further assessing discriminant validity, we examined the correlations among the three control variables, specifically client’s industry and organization size as well as relationship longevity. These should not correlate remarkably with study variable measures. Model variables show no substantial correlation with the controls. Of the 11 trust items, only 3 correlated at a low level with the number of employees (i.e., client organization size) at a significant level. The analysis finds no other significant correlations between constructs and control variables. Campbell and Fiske (1959) note that normal statistical distributions in a large matrix often result in a few meaningless exceptions. The overall patterns of results allow us to conclude that reasonable discriminant validity exists among constructs. Convergent validity
In assessing convergent validity, measures believed to be related to the same construct should correlate at a significant level with one another (Campbell & Fiske, 1959). As suggested by Fornell and Larcker (1981), each reflective construct should explain 50% or more of the variance of each of their measurement items with observed item loadings in excess of .7; our data pass all of these thresholds.
Table 4: Reliability, intercorrelations, and average variance extracted (AVEs).
Figure 2: Path coefficients and explained variance in the structural model.
Figure 2 summarizes the results of the PLS analysis. Competitive positioning through IT customization ^sub DV^ has a significant positive direct effect on cooperative strategic-information flows ^sub DS^, supporting H^sub 1^ Moreover, with respect to H4, mutual trust ^sub DS^ also positively influences strategic-information flows ^sub DS^, with both antecedents explaining 54.9% of the construct’s variance. In addition, mutual trust ^sub DS^ has a positive direct effect on IT customization D^sub V^, with 31.2% of the variance explained, supporting H^sub 3^. Strategic-information flows ^sub DS^ and relationship-specific investment ^sub DS^ both have a positive direct effect on relationship-specific performance ^sub DS^. finding for H^sub 2^ and H^sub 7^ explaining 36.7% of the variance of performance. Finally, in testing H^sub 5^, relationship- specific investment ^sub DS^ positively affects mutual trust ^sub DS^. with an explained variance of 20.9%. Our analysis found no support for H^sub 6^, a relationship between relationship-specific investment ^sub DS^ and mutual strategicinformation flows ^sub DS^.
Our proposed research model includes potential mediation effects. Specifically, strategic-information flows may serve to mediate the impact of both (i) mutual trust ^sub DS^ and (ii) IT customization ^sub DV^ on relationship-specific performance Ds- We tested for mediation effects within our work through two complementary procedures (Subramani, 2004). The first, which assesses the exploratory power of competing models, compares the research model that proposes full mediation against a competing, partially mediated model and proposes both direct and mediated effects. The second procedure employs mediation analysis techniques (Hoyle & Kenny, 1999; Subramani, 2004) and provides information on the significance of mediation effects.
In comparing the research model with full mediation to partially mediated nested models, PLS can be employed to statistically compare results (Chin, Marcolin, & Newsted, 2003; Subramani, 2004). We compute two alternative, partially mediated models by adding one path for each to the fully mediated model. One path is added from (i) mutual trust ^sub DS^ to relationship-specific performance ^sub DS^ and, in the other, a path is added from (ii) IT customization ^sub DS^ to relationshipspecific performance ^sub DS^. Appendix B depicts the results for both models. The R^sup 2^ for relationship- specific performance ^sub DS^ for the full model is 36.7%. The R^sup 2^s for relationship-specific performance ^sub DS^ for the partially mediated models are 37.2% for mutual trust ^sub DS^ [arrow right] relationship-specific performance ^sub DS^ and 37.5% for IT customization ^sub DV^ [arrow right] relationship-specific performance ^sub DS^.
To understand the additional contribution of paths, we examine the incremental changes in R^sup 2^. A procedure for measuring the effect size and significance of the change in R^sup 2^ between models is an f^sup 2^ statistic calculated by dividing (R^sup 2^ ^sub partially mediated^ – R^sup 2^ ^sub full mediation^) by (1 – R^sup 2^ ^sub partially mediated^). Subsequently, a pseudo F test for the change in R^sup 2^ with 1 and (n – k) degrees of freedom was calculated by multiplying the f^sup 2^ statistic by (n – k – 1). In applying this formula, the f^sup 2^ statistic for mutual trust ds – > relationship-specific performance ^sub DS^ is 008, while IT customization ^sub DV^ [arrow right] relationship-specific performance ^sub DS^ is 012, with insignificant pseudo F(1,86) statistics of 0.68 and 1.02, respectively. These results are summarized in Table 5. Accordingly, the additional variance explained by the inclusion of either of the direct paths does not significantly add to the exploratory power of the overall model.
The second procedure for mediation analysis is based on the path coefficients and standard errors of the direct paths between (i) independent and mediating variables (i.e., iv [arrow right] m) and (ii) mediating and dependent variables (i.e., m [arrow right] dv). The magnitude of mediation is computed as the product of path coefficients between iv ^sub DS^ m and m ^sub DS^ dv. Hence, the magnitude of mediation for mutual trust ^sub DS^ ^sub DS^ relationship-specific performance ^sub DS^ is -201, and for IT customization ^sub DV^ ^sub DS^ relationship-specific performance ^sub DS^ is .123. An approximation of the standard error of the mediated path is computed based on standardized path coefficients and standard deviations for the direct paths, (i) iv ^sub DS^ m and (ii) m ^sub DS^ dv (as summarized in Table 6 and discussed in detail in Hoyle and Kenny (1999)). The resulting z statistics are 3.76 and 2.32 for mutual trust ^sub DS^ and IT customization ^sub DV^ to relationship-specific performance ^sub DS^, respectively; both are significant at p
Table 5: Test of mediation, nested model analysis.
Table 6: Test of mediation, mediated path analysis.
Managers sculpt their relationships with other firms by determining when to engage in hard bargaining and when to compromise for mutual benefit. In this article, we argue that strategizing for these positions can and should take place simultaneously. In some aspects of each relationship, cooperative disclosure is useful, which, in this case, is information sharing. In other aspects of the interorganizational connection, the parties must competitively negotiate a settlement as to who will make the most sizable IT investments for customizing their transactional exchanges. Positioning oneself properly calls for a recognition of, as Kenny Rogers sings, “When to hold ‘em and when to fold ‘em.”
Our study found tiiat, indeed, effective logistics exchange- relationship partners cooperate with respect to the sharing of strategic information while competing with respect to investments for IT customization. Offering one of the first empirical studies capturing both domains within a single nomology, our work suggests that these relationships result in symmetric Ricardian performance gains for both partners. Hypothesized elements of symmetry and/or degree played out cleanly in all of our findings. From a managerial perspective, evolving exchange relationships have seen the emergence of Internet-based technology solutions that enhance strategic- information flows between partners. We find that such information sharing results in subsequent positive performance results for both parties. In adopting tiiese solutions, parties face differences with respect to IT resources and capabilities, giving rise to tension in the relationship while determining who will bear the burden of integration.
Implications for Managerial Decision Making
From a practical perspective, this research provides a number of insights into how managers should make decisions about how they approach evolving servicecentered exchange relationships. Importantly, decision makers need to manage those less tangible aspects of their relationships from both competitive and cooperative postures. In particular, results indicate that firms must rethink the strategy of acting purely in their own self-interest and not cooperating with partners. In managing strategic information, natural inclinations lean toward carefully guarding external disclosures. Our results suggest achieving some level of magnitude and parity in these behaviors gives rise to outcomes that might not otherwise have been attainable, strengthening the basic tenet set forth in the prisoner’s dilemma (Poundstone, 1992).
Managers seem to feel their way toward achieving parity in what critical information each party is willing to share. When the parties trust one another, this process is smoother, but, in any case, the data suggest that, in the higherperforming relationships, the parties are relatively balanced in the extent to which they cooperate with information requests from their partners. Decision makers faced with choices about which alliances to pursue and how to structure these alliances should keep this finding in mind when they negotiate contracts. The loss of parity in the sharing of strategic information will lead to decremental mutual performance. It is in the interest of both parties to ensure that the other party believes it is receiving the information necessary for the relationship to work.
Besides pursuing a cooperative approach for sharing of strategic information, managers must realize that they compete with one another with respect to IT customization investments. While vendors often develop a variety of technology solutions aimed at furthering partner relationships, they find a vast majority of clients possess varying specializations or differences in capabilities (Carlile & Rebentisch, 2003). As noted, IT-related resource asymmetries emerge with respect to hardware platforms, telecommunications protocols, data formats and process standards, enterprise systems, and, most significantly, knowledge-based skill sets (Gnyawali & Madhavan, 2001). Despite their best efforts to deploy universally applicable innovations, the adoption process within the context of individual relationships necessitates some investment in developing customizations that make use of vendor solutions. From a practical perspective, managers should embrace tension in the course of interactions and recognize this as the natural course of business, actively negotiating to ensure efficient use of resources across the interfirm dyad, and agreeing to bear a disproportional share of the burden where justified and within their capabilities. The best managerial choice in this situation is to analyze the relative capability of the partner to provide the required IT customization. If the customization is much less burdensome for one party than the other, or if one party needs the relationship much more than the other, the choice can be clearer and mutual agreement less complicated. But if the investment will be difficult and costly for either, and if bargaining power between the parties is relatively equal, the resolution of the situation will call for a healthy competitiveness in bargaining. In the final analysis, our results suggest that performance metrics that represent the partnership will not suffer with an imbalance in who provides the IT infrastructure. But the requisite infrastructure must be supplied by one party and/ or the other in order to experience Ricardian rents.
From an academic perspective, our research examined recent propositions set forth in theories of exchange relationships to better understand the cooperative versus competitive phenomena and subsequent consequences within interorganizational supply chain relationships. The service-centric view highlights the limitations of examining discrete transactions (Vargo & Lusch, 2004). Hence, our work focuses on the overall characteristics of the exchange process and relationship within which these transactions occur. Moreover, our findings also demonstrate applicability of derived degree- symmetric constructs (Straub et al., 2004). Benefits from cooperation and competition in strategic partnerships inevitably lead to studying constructs core to such relationships, both in terms of their symmetry as well as degree or extent. Traditional monadic research models limit our ability to capture data for constructs to only one side of the business relationship or the other. The current study employs a dyadic approach to observing the constructs on both sides and focusing the unit of analysis on the relationship itself. Within business disciplines, marketing studies have been among the first to adopt dyadic approaches as their research questions have focused on interorganizational constructs (Anderson & Narus, 1990; Anderson et al., 1994; Fein & Anderson, 1997). The current study joins a limited number of interdisciplinary focused research efforts (demons & Row, 1993; Dyer, 1996; Kirsch, Sambamurthy, Ko, & Purvis, 2002) and employs a dyadic research design to investigate key interorganization phenomenon.
Besides these major theoretical and methodological contributions, our work offers numerous specific findings that should inform and influence future research. From the standpoint of understanding the role of IT in this context, the present research investigates IT integration as suggested by past work. Prior research hypothesizes that the investment climate in supply chain relationships shapes such integration (Klein et al., 1978; Maddigan, 1981; Dewan, Michael, & Min, 1998; Frohlich, 2002; Vickery et al., 2003). Our findings support the view that specializations of IT assets play an important role in the creation of economic rents (Amit & Schoemaker, 1993), in addition to supporting suggestions that asset-specific investments enable richer forms of interfirm collaboration (Joskow, 1988).
Limitations and Future Research
The focus on a logistics vendor in this work limits the potential generalizability to similarly structured business relationships in other supply chain processes. Indeed, the relationship between a logistics vendor and its clients might differ from the relationship that exists between an organization’s internal IT department and internal clients. Moreover, the findings might restrict generalizability beyond the logistics function, given that different externally sourced functional relationships might produce different results.
As noted, the current study specifically examines the combining of IT assets and resources. Other specific resources and/or capabilities within different relational settings might support a cooperative, as opposed to competitive, domain interpretation. Future efforts should further explore the IT customization construct in an effort to gain a greater understanding of partners’ contribution to interfirm integration. Prior work on network externalities, sponsorship of technology and standards by firms, and technology compatibility (Katz & Shapiro, 1986) might provide a useful basis to inform such investigations.
In addition, the current study uses a dyadic research design and surveys 91 logistics exchange relationships from the perspective of the vendor and its customers. While our response rate is comparable to past research employing a similar dyadic data collection strategy, the sample size constitutes a limitation of this work. Given the inherent difficulties in dyadic data collection, researchers should investigate research designs