OW DO UPPLIERS ENEFIT FROM NFORMATION · PDF fileSubramani/Benefits from IT Use in Supply Chain Relationships MIS Quarterly Vol. 28 No. 1/March 2004 47 and theory testing in this paper

Subramani/Benefits from IT Use in Supply Chain Relationships

MIS Quarterly Vol. 28 No. 1, pp. 45-73/March 2004 45

RESEARCH ARTICLE

HOW DO SUPPLIERS BENEFIT FROMINFORMATION TECHNOLOGY USE INSUPPLY CHAIN RELATIONSHIPS?1

By: Mani SubramaniInformation and Decision SciencesCarlson School of ManagementUniversity of MinnesotaMinneapolis, MN [email protected]

Abstract

Supply chain management systems (SCMS)championed by network leaders in their suppliernetworks are now ubiquitous. While prior studieshave examined the benefits to network leadersfrom these systems, little attention has been paidto the benefits to supplier firms. This study drawsfrom organizational theories of learning and actionand transaction cost theory to propose a modelrelating suppliers� use of SCMS to benefits. It pro-poses that two patterns of SCMS use by sup-pliers�exploitation and exploration�create con-texts for suppliers to make relationship-specific

1Ron Weber was the accepting senior editor for thispaper.

investments in business processes and domainknowledge. These, in turn, enable suppliers toboth create value and retain a portion of the valuecreated by the use of these systems in interfirmrelationships.

Data from 131 suppliers using an SCMS imple-mented by one large retailer support hypothesesthat relationship-specific intangible investmentsplay a mediating role linking SCMS use tobenefits. Evidence that patterns of informationtechnology use are significant determinants ofrelationship-specific investments in businessprocesses and domain expertise provides a finer-grained explanation of the logic of IT-enabledelectronic integration. The results support thevendors-to-partners thesis that IT deployments insupply chains lead to closer buyer-supplierrelationships (Bakos and Brynjyolfsson 1993). Theresults also suggest the complementarity of thetransaction-cost and resource-based views,elaborating the logic by which specialized assetscan also be strategic assets.

Keywords: Buyer-supplier relationships, inter-organizational systems (IOS), EDI, supply chainmanagement systems (SCMS), transaction costeconomics, intangible asset specificity, IT use,exploration, exploitation


46 MIS Quarterly Vol. 28 No. 1/March 2004

Introduction

Early forms of interorganizational systems (IOS)primarily supported the automation of manualprocesses such as ordering and settling accounts.A range of new features for information sharing,communication, and collaboration has subse-quently enhanced these systems (Icasati-Johanson and Fleck 2003). Estimates suggestthat over 30,000 IOS are currently in use, sup-porting a large proportion of business-to-businesstransactions (Harris 2001).

In recent years, network leaders such as Chrysler,Dell, Ford, and Wal-Mart have made significantefforts to derive the benefits of coordination andcollaboration with their suppliers by using aparticular form of IOS: supply chain managementsystems (SCMS2). A greater understanding of thebenefits that these systems provide to supplierfirms is therefore an issue of interest to bothresearchers and practitioners. Nonetheless,despite their significance, research on SCMS hasbeen scant and fragmented. Prior work hasfocused largely on the benefits derived fromSCMS by network leaders. Little attention hasbeen paid to the benefits derived from SCMS bysuppliers and mechanisms that enable suppliersto realize benefits.

Supplier networks are characterized by a largenumber of supplier firms working with a dominantnetwork leader. Relationships between suppliersand network leaders are largely asymmetric.Network leaders play a central role in orches-trating suppliers. They often champion theintroduction of SCMS in their supplier networks.Benefits from these information technologies aredistributed unevenly and skewed in favor of thenetwork leader (Riggins and Mukhopadhyay1994). Although supplier participation is neces-sary for network leaders to derive benefits,supplier firms appear unlikely to benefit fromnetworks. Instead, network leaders seem tobenefit at the expense of supplier firms (Carter

1990; Clemons and Row 1993), often by shiftingactivities and costs to them. For instance, themove to vendor-managed inventories shifts tasksrelated to monitoring and managing retailinventories to suppliers, creating benefits fornetwork leaders while adding to the tasksperformed by suppliers. Similarly, quick-responseprograms create benefits for auto manufacturersand retailers while burdening suppliers withmaking more frequent deliveries and incurringhigher inventory holding costs (Mukhopadhyay etal. 1995). Network leaders can also use theirsuperior bargaining power to appropriate supplierbenefits from streamlining interfirm processes.For instance, they may specify annual costreduction targets in supply contracts (Ghosh andJohn 1999). SCMS can thus be an unfortunatestrategic necessity for suppliers (Barua and Lee1997).

The few studies that have examined supplierbenefits from interorganizational systems (e.g.,Lee et al. 1999; Mukhopadhyay and Kekre 2002)focus on systems developed by large supplierfirms. We know little of benefits in the moreprevalent instance of suppliers adopting SCMSchampioned by a dominant network leader. Thissituation is troubling because the rollout of SCMSby network leaders has gained momentum in awide range of industries. The question con-fronting suppliers is often not whether they shoulduse SCMS but how they can take advantage ofthese systems and benefit from their use.

This paper draws on organizational theories andtransaction cost economics to examine supplierbenefits from SCMS use. This work contributes tothe literature in three ways. First, it focuses on thesupplier perspective in IT-mediated supplier-retailer interactions and highlights the benefits tosuppliers from IT use. Second, it theorizes twopatterns of use of supply chain technologies�forexploitation and for exploration�and highlightsthe implications of these appropriation choices forrelationship-specific supplier investments andsupplier benefits. Third, it theorizes the role ofrelationship-specific intangible investments inenabling suppliers to both generate value andobtain benefits from SCMS use. Theory building

2In this paper, SCMS are viewed as instances of infor-mation technologies employed in interorganizationalcontexts to mediate buyer-supplier transactions.


MIS Quarterly Vol. 28 No. 1/March 2004 47

and theory testing in this paper are grounded inthe context of supplier-retailer relationships in aretail distribution channel. Specific features of thecontext, such as asymmetric relationships ofsuppliers with the network leader and charac-teristics of supplier-retailer interactions, are usedto articulate the relationships between SCMS use,specialized investments, and benefits.

Theoretical Background

Figure 1 presents the model examined in thisresearch. For suppliers in an SCMS, it showsrelationships that are hypothesized to exist amongpatterns of IT use, relationship-specific invest-ments, and benefits.

Patterns of SCMS Use

Prior examinations of IOS reveal multiple goalsmotivating their use: providing management sup-port, reducing operational costs, improvingcustomer service, and gaining competitive advan-tages (El Sawy et al. 1999; Premkumar et al.1997). However, prior conceptualizations fail toarticulate patterns of IOS use that help explaindiverse outcomes. Prior research has highlightedvolume, diversity, breadth, depth, scope, andintensity as dimensions of IOS use (Bensaou andVenkatraman 1995; Massetti and Zmud 1996).Such descriptive features are of limited value inrelating system use to variations in outcomesbecause the use of IOS in different contexts�even if similar in breadth, depth, or intensity�clearly can be motivated by different goals. Aconceptualization of IOS use that reflects inten-tionality of use is missing.

In this regard, the concept of appropriation pro-posed by DeSanctis and Poole (1994) providesuseful insights. The term appropriation refers topatterns of IT use. Differing appropriations canlead to diverse outcomes, even when the contextof use and underlying technologies are similar(DeSanctis and Poole 1994). While this constructhas been useful as a means of understanding how

group decision support systems (GDSS) are used,it has not been applied in the context of SCMSuse. Appropriations of SCMS can reveal inten-tionality and help relate SCMS use to outcomes.

To this end, the paper draws from a theory oflearning and action that suggests actions inorganizations can be categorized as eitherexploitation or exploration (March 1991).Exploitation is the extension or elaboration of oldcertainties. It is the class of actions whose goal isto improve operational efficiencies (e.g., throughincreased standardization, tighter processcontrols, and reduced manual intervention). Incontrast, exploration is the pursuit of newpossibilities. It is the class of activities whose goalis to learn about the environment and discovernovel ways of creating value or solving oldproblems. These two classes of action alsoincorporate differing task orientations.Exploitation involves the application of variance-reducing strategies to streamline activities,perform them efficiently with a high level ofconsistency, and achieve greater control overprocess execution. In contrast, explorationinvolves the application of variance-seekingstrategies to reassess current approaches to prob-lems and to develop novel solutions. Explorationreflects risk taking, experimentation, and inno-vation. Inherently, it involves activities where theimmediate payoffs are less certain but which, overmultiple periods, can help build a firm�s ability todevise superior solutions to problems.

Consistent with this view, SCMS use forexploitation and SCMS use for exploration areconceptualized as two complementary patterns ofappropriation of supply chain technologies. Thedistinction between exploitation and exploration inthe use of supply chain technologies has con-siderable appeal. It parallels the fundamentaldifferences between automating and informating,which are two broad motives for using informationsystems. These patterns of SCMS use areexpected to explain variations in benefits to sup-pliers using identical supply chain technologies.Table 1 lists activities comprising SCMS use forexploitation and for exploration.



IT Use for Exploitation

IT Use for Exploration

Business-processSpecificity

Domain-knowledgeSpecificity

Operational Benefits

Strategic Benefits

CompetitivePerformance

Pattern of ITAppropriations

Relationship-SpecificInvestments

First-OrderBenefits

Second-OrderBenefits

Control Variables:Uncertainty, Dependence, Size,Years of Association

IT Use for Exploitation

IT Use for Exploration

Business-processSpecificity

Domain-knowledgeSpecificity


Strategic Benefits

CompetitivePerformance

Pattern of ITAppropriations

Relationship-SpecificInvestments

First-OrderBenefits

Second-OrderBenefits

Control Variables:Uncertainty, Dependence, Size,Years of Association

Figure 1. IT Use, Relationship-Specific Investments, and Supplier Benefits

Relationship-Specific IntangibleAssets

Transaction cost economics suggests thatrelationship-specific investments are importantsources of value creation in interfirm exchanges(Williamson 1995). Such supplier investmentsinclude customized business processes cateringto the requirements of a particular buyer and thedevelopment of expertise unique to an exchange(such as a detailed understanding of an auto-maker�s engineering practices). Within the con-text of an exchange, specialized, relationship-specific assets create more value than non-specialized, generic assets.3 They are an impor-tant source of interorganizational competitiveadvantage (Dyer and Singh 1998). Subramaniand Venkatraman (2003) highlight that enhancedvalue creation enabled by intangible relationship-specific investments confers supplier firms with

advantages over competitors operating withoutsuch assets.4 They represent a powerful meansfor suppliers to create value and to positionthemselves strategically to claim an equitableproportion of the value created. The use ofrelationship-specific investments by suppliers cancreate exit barriers for the retailer. They enhancethe bargaining power of suppliers, thus becomingpart of the strategizing calculus in relationships(Ghosh and John 1999).

The focus in this paper is on supplier investmentsthat create two types of intangible asset speci-ficity: business-process specificity and domain-knowledge specificity. Business-process speci-ficity arises from the development of relationship-specific routines or standard operating proceduresfor efficient task execution. Domain-knowledgespecificity arises

3Relationship-specific assets are commonly viewed asassets that are of lesser value when redeployed inalternative exchanges. An alternative perspectiveemphasizing their constructive role is that specializedassets deliver greater value than non-specialized,generic assets in the context of a particular relationship(Ghosh and John 1999).

4The paper focuses on the role of intangible assets thatare characterized by greater causal ambiguity thanphysical assets and are thus less amenable to imitationby competing suppliers. The arguments indicate thatintangible, relationship-specific investments by supplierscan create lock-in effects, even without reciprocal assetcommitments by retailers.



Table 1. Patterns of IT Use by Suppliers IT Use for Exploitation: Execution of

Structured Interfirm ProcessesIT Use for Exploration: Execution of

Unstructured Interfirm Processes

Goals: Improving, applying, and incremen-tally refining firm capabilities.

Outcomes: Clearly definable benefits (e.g.,cost reduction, process consistency, processefficiency).

Goals: Creating new capabilities, devising novelsolutions to current problems.

Outcomes: Soft benefits that are difficult toevaluate in advance (e.g., shared understanding,clearer picture of cause-effect relationships,greater understanding of operating environment).

Examples of IT Use for Exploitation Examples of IT Use for Exploration

Request for Quotation (RFQ) receivedelectronically by suppliers. Supportdocuments such as detailed part drawingsand quality specifications accessed online.

Analysis of point-of-sale data to understandpatterns in customer preferences, patterns in thesale of complementary products.

Analysis of product-return data to detect issues tobe addressed at retail store level (e.g., problems inhandling, displaying products).

Electronic Transmission of Purchase Orders,Electronic Transmission of Advanced Ship-ment Notification (ASN).

Scheduling delivery windows at warehouseloading docks.

Communicating with repair and service personnelof retailer to diagnose problems encountered inthe field.

Interactions between supplier�s engineering groupand retailer�s service division on diagnosingcomponent failures to inform product redesign,new product design.

Electronic Payment Settlement Advice, issuedin predefined intervals after material receipt.

Communicating with retail documentation group toimprove service manuals and customer manuals.

Transmission of order status reports (e.g., toalert retailer about unanticipated delays, abilityto advance-ship hot items).

Communication between product designers andretail buyers related to new features to be incor-porated to improve product�s appeal

Electronic notification of changes (e.g., pro-duct specifications, changes to packaging andshipping procedures, specification of pur-chase order quantities).

Interaction between retail planners and supplier�smanagers to decide stocking levels for differentproducts, size allocations to different stores.

Inventory alerts based on preset triggers tocommunicate stocking levels of products inwarehouses.

Ad hoc communications between supplier andretailers� buying and merchandising groups (e.g.,to signal competitive activity such as specialpromotions or to respond to stock-outs fromunanticipated product demand).



from the development of a context-sensitiveunderstanding of cause-effect relationships thatfacilitate effective action and resolution ofambiguities in task planning and execution.

Business-Process Specificity

Business-process specificity is the degree towhich a supplier�s key business processes suchas operating processes, administrative processes,and quality-control processes are particular to therequirements of the focal firm5 in the relationship(Subramani and Venkatraman 2003). Business-process specificity is an important factor linked toperformance in interfirm relationships (Dyer1996b; Mukhopadhyay and Kekre 2002; Zaheerand Venkatraman 1994). The fieldwork con-ducted for this study provides an example. Amattress supplier made significant changes tomanufacturing processes in multiple plants todevelop a customized make-to-order process thatenabled a retailer to discontinue the relativelyinefficient practice of maintaining mattressinventories. The supplier assigned one primarymanufacturing plant and one backup plant in eachregion to local clusters of retail stores. Ordersplaced by customers in the retailer�s stores weretransmitted to the designated plants. The plantsmanufactured the mattresses within 12 hours anddelivered them directly to customers by the nextday. The move to a make-to-order systeminvolved many changes to the supplier�s planning,manufacturing, and shipping processes. Theseintangible investments were useful only in therelationship with the particular retailer.6 Theycreated exit barriers for the retailer becauserelationship termination involved foregoing bene-fits created by them as well as incurring the costsof reestablishing the discontinued warehouse

stocking and customer delivery processes.Business-process specificity thus enhances theretailer�s dependence on the supplier andincreases the supplier�s bargaining power in therelationship (Subramani and Venkatraman 2003).

Domain-Knowledge Specificity

Domain-knowledge specificity is the degree towhich a supplier�s critical expertise such ascompetitive analysis and strategy formulation andnew-product development are particular to therequirement of the focal firm in the relationship(Subramani and Venkatraman 2003). Domain-knowledge specificity is reflected in instances offirms relying on suppliers for innovations and forinputs in critical decisions (Dyer and Singh 1998).Interviews with suppliers in the distributionchannel revealed several manifestations ofdomain-knowledge specificity. For example,managers at a supplier of designer women�sclothing indicated that their analysis of a retailer�ssales helped them better understand regionalvariations in size and color preferences. Thisunderstanding of the retailer�s customers enabledthem to work with the retailer to create region-levelsize profiles and even store-level merchandiseforecasts for their products. As a result, theretailer reduced the level of markdowns, improvedtheir margins on the product line, and reinforcedthe high-end image of the supplier�s products.Managers at the supplier firm also used theirknowledge to plan new design collections cus-tomized for different geographic areas such asFrench-speaking Canada, the Pacific Northwest,and the far North. This example reflects thevalue-creating potential of relationship-specificdomain knowledge such as an intimate under-standing of the preferences and tastes ofcustomer segments addressed by the retailer.7

Because these benefits are unavailable to theretailer if they switch to an alternative supplier,domain-knowledge specificity enhances theretailer�s reliance on the supplier.

5The network leader, viewed from the perspective ofsupplier firms, is termed the focal firm.

6The physical assets�the machinery used to make themattress�were general purpose and not particular tothe retailer. In this paper, the focus is primarily onintangible investments by suppliers that Bakos andBrynjolfsson (1993) highlight as being non-contractibleand thus contributing to contact incompleteness in IT-mediated supply relationships.

7Takeishi (2002) describes similar instances ofrelationship-specific expertise investments by suppliersin the auto industry.



Linking SCMS Use to Relationship-Specific Intangible Assets

The following section discusses the patterns ofassociation of SCMS use by suppliers and theirrelationship-specific intangible asset investments.

Business-Process Specificity and SCMSUse for Exploitation

Greater levels of SCMS use for exploitation(ITExploit)�using the system to perform struc-tured, repetitive tasks�are accompanied bychanges to interorganizational business pro-cesses. These changes (e.g., to manufacturing,quality control, and shipment processes) are bothprompted and facilitated by SCMS use. Forinstance, making direct-to-store deliveries (ratherthan warehouse deliveries) involves use of theSCMS to facilitate the complex but structured taskof disaggregating the retailer�s order into multiplestorewise orders and creation of storewise ship-ment and billing documents. As a result, suppliershave an opportunity to enhance their benefits bymaking complementary changes to their produc-tion and inventory management processes. Theycan thus more efficiently execute high-variety, low-volume store-level orders. The value of suchretailer-specific process reconfigurations oftenbecomes evident to suppliers in the course ofSCMS use. These changes occur subsequent tothe basic set of process changes involved inadopting and using the SCMS.

Similarly, use of the SCMS in the structured taskof tracking retail warehouse inventories createsthe opportunity for suppliers to redesign theirplanning and manufacturing processes to deriveefficiency benefits (e.g., by linking materialprocurement processes to retailer orders or byincorporating information on retailer promotions intheir production planning processes). Eventhough none of these changes are mandated orrequired to work with the retailer, supplierschoosing to make them derive greater benefitsthan those operating without specialized pro-cesses. These arguments suggest that ITExploitis an enabler of change in a manner that custo-

mizes suppliers� process to a specific exchange.Higher levels of SCMS use for exploitation(ITExploit) are therefore likely to be associatedwith higher levels of business-process specificity:

H1: The higher the level of SCMS usefor exploitation, the greater the level ofbusiness-process specificity in theexchange.

Domain-Knowledge Specificity andSCMS Use for Exploration

SCMS use for exploration (ITExplore) occurs viathe reporting and messaging features of theSCMS. Greater levels of ITExplore reflectincreasing reliance on an SCMS for unstructuredtasks by suppliers. For instance, ITexploreenables suppliers to understand patterns incustomer preferences, develop new perspectivesthrough IT-mediated interactions with retailerpersonnel, and develop novel approaches to field-service problems. Greater levels of ITExplore arelikely to develop and refine a supplier�sunderstanding of the retailer�s market and theretailier�s customers and their preferences,leading to suppliers developing greater levels ofrelationship-specific domain knowledge.

Buyer-supplier relationships in the distributionchannel and the auto industry comprise ongoingsupply relationships between the network leaderand a set of supplier firms selected afterscreening and verification by the network leader.SCMS provide features to allow network leadersto provide suppliers with an array of reports on aregular basis. The level of ITExplore reflects theuse of such information by suppliers inunstructured tasks. For instance, suppliers canuse retail sales and return reports to infer regionalpatterns in customer tastes (e.g., stylepreferences and color preferences of women�sswimsuits) as well as product and marketcharacteristics (e.g., sensitivities of products topromotions).

ITExplore thus enhances a supplier firm�s domainknowledge in a manner that is particular to the



relationship. It also reflects use of the messagingand collaboration features in SCMS. Byenhancing the level of supplier-retailer interactionbeyond that possible in non-IT-mediated settings,it helps supplier firms become sensitive toidiosyncratic features of the retailer�s require-ments. For instance, in the auto industry, supplierfirms use SCMS features such as e-mail anddiscussion forums to understand details of theautomaker�s requirements not completely con-veyed in formal documentation (Takeishi 2002).In the distribution channel, engineers in a supplierfirm can use the SCMS to interact with a retailer�sfield service group providing after-sales services.These interactions help them learn how customersuse their products and how the service grouphandles field failures. These arguments suggestthat ITExplore influences the development of asupplier�s domain knowledge and makes it moreattuned to the nuances of a particular exchange.Higher levels of SCMS use for exploration(ITExplore) are therefore likely to be associatedwith higher levels of domain-knowledge specificity:

H2: The higher the level of SCMS usefor exploration, the greater the level ofdomain-knowledge specificity in theexchange.

Relative Influence of SCMS Use onRelationship-Specific Investments

H1 and H2 highlight the association of ITExploitwith business-process specificity and ITExplorewith domain-knowledge specificity. The followingsubsections discuss the relative influence of thetwo patterns of SCMS use on business-processspecificity and domain-knowledge specificity.

Relative influence on business-process speci-ficity: ITExplore enables suppliers to learn abouta variety of issues in the relationship that caninfluence the level of business-process specificity.For instance, higher levels of ITExplore (e.g.,greater communication with the merchandisinggroup or the quality-control group) allow suppliersto become aware of shortcomings in their busi-ness processes and learn about the advan-

tageous business practices adopted by othersuppliers (Dyer and Nobeoka 2000; Dyer andSingh 1998). Suppliers thereby recognize oppor-tunities for improving their current processes or forcreating new processes. Higher levels ofITExplore can thus lead to greater levels ofbusiness-process specificity. Because the linkbetween ITExplore and business-process specifi-city is indirect and often serendipitous, however,the association between ITExplore and business-process specificity is likely to be weaker than theassociation between ITExploit and business-process specificity. ITExploit is likely to be morestrongly associated with suppliers� business-process specificity:

H3a: The association of ITExploit withbusiness-process specificity is strongerthan the association of ITExplore withbusiness-process specificity.

Relative influence on domain-knowledgespecificity: Higher levels of ITExploit can haveinformational benefits and positively influence thelevel of domain-knowledge specificity. Forinstance, ITExploit can reveal information aboutretailer processes that are articulated andformalized in implementing the SCMS. Even inroutine use of SCMS for structured interorgani-zational processes (e.g., to transmit quality-controlreports), suppliers may become aware over timeof the retailer�s internal processes for formulatingquality standards, subjective judgments involvedin quality-control procedures, preferences ofindividuals charged with making these judgments,and the discretion to permit exceptions possessedby individuals in the quality-management group.Higher levels of ITExploit thus enable suppliers todevelop higher levels of relationship-specificdomain knowledge in the exchange. However,this association is likely to be constrained by gapsin communication and knowledge sharing amongfunctions within supplier firms (e.g., personnel inthe manufacturing function and designers involvedin creating products) and differences in theirinterpretation of information. Consequently, of thetwo patterns of SCMS use, ITExplore is likely to bemore strongly associated with domain-knowledgespecificity than ITExploit:



H3b: The association of ITExplore withdomain-knowledge specificity is strongerthan the association of ITExploit withdomain-knowledge specificity.

Overall, these hypotheses highlight the differentconsequences of suppliers� SCMS appropriationchoices for the profile of their relationship-specific,intangible investments. These hypotheses repre-sent a context-specific elaboration of the linkbetween SCMS use by suppliers and theirdevelopment of intangible, IT-enabled resourcesand capabilities (Bharadwaj 2000).

First-Order and Second-Order Benefits

This work draws upon a two-stage model ofbenefits (Barua et al. 1995; Mukhopadhyay andKekre 2002) in which information technologies areviewed as creating direct, first-order benefits,which in turn generate indirect, second-orderbenefits. First-order benefits are related to firmactions and can be influenced directly by firms. Incontrast, second-order benefits are competitiveoutcomes and incorporate the influence ofexternal factors such as competitors� moves andenvironmental changes that are beyond thecontrol of an individual firm.

Operational benefits and strategic benefits are twofirst-order benefits of SCMS use (Mukhopadhyayand Kekre 2002). Operational benefits arise fromlowered transaction and production costs throughSCMS use. Examples include faster invoicing andpayment settlement, more-efficient inventorymanagement, and automating and rationalizingbusiness processes. In contrast, strategic bene-fits arise through firms positioning themselves totake advantage of opportunities arising in therelationship. These include the development ofnew products and services, a richer understandingof the partner and nuances of the exchange(Mukhopadhyay and Kekre 2002), and the abilityto recognize and respond to changes in therelationship. This parallels the distinction betweenoutcomes linked to cost reduction and to end-product enhancement in interfirm relationships(Ghosh and John 1999).

A supplier�s competitive performance�reflectingthe supplier�s success relative to competitors�isviewed here as a second-order benefit influencedby operational and strategic benefits created byinformation technologies. This view is consistentwith the suggestion that competitive performanceis a long-term strategic benefit of SCMS use(Mukhopadhyay and Kekre 2002).

Linking SCMS Use to Benefits: Mediation by Relationship-Specific IntangibleInvestments

The logic of benefits from IT use proposed here isthat patterns of SCMS use by suppliers influencethe nature of relationship-specific assets, whichlead to supplier benefits. This mediating role forrelationship-specific investments is suggestedfrom the perspective of both value creation andvalue retention. The value-creation perspec-tive�that value creation by firms through infor-mation technologies is linked to complementarychanges in organizational processes and busi-ness strategies�has been examined in prior work(Duliba et al. 2001; Sabherwal and Chan 2001).

In contrast, the value-retention perspective hasreceived less attention in the IS literature. Theability of suppliers to claim benefits from IT use expost and prevent benefits from being appropriatedby the other party is an important consideration inasymmetric interfirm relationships (Ghosh andJohn 1999).8 In this respect, relationship-specific,intangible investments represent an importantsource of advantage (Subramani and Venka-traman 2003). For instance, a supplier�s IT-enabled business processes that enable them tomanage offshore garment manufacturers and theirIT-enabled domain knowledge that builds onindustry knowledge possessed by firm employeesare manifested as complex capabilities that

8The strategizing calculus through which firms contendfor a share of benefits is a key distinction between ITuse in interorganizational contexts and IT deploymentswithin firms.



present barriers to imitation, emulation, andsubstitution by competitors. As a result, a supplierfirm can both constrain the focal firm�s ability toswitch to alternative suppliers and benefit from thevalue created by relationship-specific assets (Dyer1996a, 1996b; Subramani and Venkatraman2003). Williamson (1995, p. 230) terms this out-come a fundamental transformation: �the transfor-mation of what had been a large numbers biddingcompetition at the outset into one of bilateralexchange during contract execution and atcontract renewal intervals.� Relationship-specificinvestments, therefore, enhance a supplier�sability to retain an equitable proportion of thevalue generated by IT use�value that the focalfirm could otherwise appropriate by the crediblethreat of switching to alternative suppliers. Thislogic is consistent with observations that interfirmrelationships involve a complex interplay of exante cooperation to jointly create value and expost self-interested bargaining to claim value(Ghosh and John 1999). It is also consistent withresearch suggesting that the prospect of ad-versely affecting the value created by idiosyncraticand intangible supplier assets in future periodsattenuates ex post opportunistic behavior by focalretailers (Subramani and Venkatraman 2003).

Overall, these arguments, from the value-creationand value-retention perspectives suggest thatsupplier benefits are positively related to the levelof relationship-specific intangible investments(which, in turn, are related to patterns of SCMSuse). The roles of business-process specificityand domain-knowledge specificity in creating andretaining benefits are not differentiated becauseprior theory on this issue is limited. In conjunctionwith the arguments for H1, H2, and H3, thisanalysis leads to

H4: The higher the level of SCMS usefor exploitation, the higher the levels ofoperational and strategic benefitsachieved through the leverage ofrelationship-specific business processes.

H5: The higher the level of SCMS usefor exploration, the higher the levels ofoperational and strategic benefits

achieved through the leverage ofrelationship-specific domain knowledge.

Consistent with the two-stage model of benefits,higher levels of operational and strategic benefitsare likely to lead to higher levels of competitiveperformance:

H6: Higher levels of operational benefitsin the exchange are associated withhigher levels of competitive performance.

H7: Higher levels of strategic benefits inthe exchange are associated with higherlevels of competitive performance.

Control Variables

To discount rival hypotheses, the model incor-porates four variables influencing supplier bene-fits: product uncertainty, retailer replaceability,supplier size, and length of association betweenfirms. Hypotheses related to these variables arenot proposed because the paper does not attemptto develop theory related to their effects.However, they are included in the model to assessthe effects of the model�s independent variableson dependent variables, beyond those attributableto these control variables.

Uncertainty arising from shorter product life cyclesincreases information processing demands inrelationships (Bensaou and Venkatraman 1995).It also increases the likelihood of a supplierencountering unanticipated contingencies. Forexample, unexpected product improvements bycompetitors can catch a supplier by surprise andadversely affect its sales. Therefore, productuncertainty is expected to be negatively related tosupplier benefits.

The level of retailer replaceability reflects the easewith which a supplier can make the transition toworking with other customers for their goods in theevent they are not able to sell to the focal retailer.The level of retailer replaceability is inverselyrelated to the level of dependence of the supplieron the focal retailer. Low levels of retailer replace-



ability can reflect a cooperative climate in therelationship and therefore be positively related tosupplier benefits (Dyer and Singh 1998). Yet lowlevels of retailer replaceability can also make thesupplier more vulnerable to exercises of power bythe retailer (Hart and Saunders 1997) andadversely affect its performance. Including retailerreplaceability in the model helps control for theseeffects of dependence on supplier benefits.

Including firm size in the model controls for factorssuch as relative bargaining power and size of theresource base that can affect supplier benefits(Zaheer and Venkatraman 1994). Larger sup-pliers may be more willing to make investments intraining personnel. They may also have moreexperience in information systems use. Benefitsto larger suppliers may therefore be systematicallyhigher than those to smaller suppliers (Lee et al.1999). However, larger suppliers may haveexisting IT and manufacturing operations thatneed considerable redesign to integrate with theretailer�s processes (an issue likely to be lessproblematic for smaller suppliers). Incorporatingsize in the model controls for these extraneouseffects.

Including the length of association as a controlvariable has two advantages. First, it helpscontrol for the potential effects of relationshipduration on supplier benefits. Second, it controlsfor recursive relationships, if any, between depen-dent and independent variables (Subramani andVenkatraman 2003). For instance, greater supplierbenefits in one period might lead to higher levelsof IT use and investments in specialized intangibleassets in subsequent periods. Including length ofassociation in the model helps control for suchtemporal patterns that may otherwise confoundthe results.

Methods

This section presents details about the context ofthe study, the procedures used to develop thesurvey items, and the procedures used for datacollection.

Data

This study was conducted in Canada with thecooperation of Alpha, a leading Canadian retailerthat pioneered the use of information technologiesin their supply chain. Alpha has a reputation forcarrying high-quality products and using rigoroussupplier qualification and audit processes. Beinga supplier to Alpha is recognized as a mark ofdistinction and provides reputational benefits.Alpha�s relationships with suppliers were asym-metric with Alpha being more powerful, which is acharacteristic typical of buyer-supplier relation-ships in the retail distribution channel. The datacollected for this study were part of a larger datacollection effort aimed at examining the manage-ment of buyer-supplier relationships.

In the first phase of fieldwork, eight day-longstrategy sessions in which senior Alpha managersand selected suppliers shared information aboutmarket developments and discussed their short-and long-term plans were observed. Then 27semi-structured interviews were conducted withAlpha and supplier managers on the nature ofinteractions in their relationship. Some of theseinterviews were conducted at supplier premises,which allowed observation of suppliers� use of theSCMS.

The SCMS was a proprietary system developedby Alpha. The application was based on elec-tronic data interchange (EDI) technologies. Itinterfaced with Alpha�s internal merchandising,quality control, retail operations, logistics, andaccounting systems. In addition to supportingstandard functional processes, the SCMS pro-vided a range of status reporting and messagingfacilities. The supplier module of the SCMS ranon PCs. Suppliers signed up with value-addednetwork service providers for communicationsfacilities. The system let Alpha provide supplierswith a range of periodic reports including districtand region-wide sales, product forecasts, storereturns, markdowns, promotion calendars, andinventory alerts. The SCMS interfaced with thedatabases of Alpha�s large post-sale service



organization and allowed suppliers access to field-service records for their products. The SCMS wasused to ensure coordinated supplier interactionsby Alpha�s merchandising, quality-control,accounting, and after-sales service groups. Sup-pliers were assigned e-mail IDs and providedmessaging facilities through the SCMS. Themessaging module interfaced with Alpha�s internale-mail systems. Suppliers could send e-mailmessages to individuals in Alpha listed in theaddress book. The features of this SCMS aretypical of systems currently in use. Although sys-tems based on Internet technologies are currentlyavailable, recent evidence suggests that 90percent of firms still use EDI-based systems(Shah et al. 2002). These latter systems are ex-pected to continue to play a key role in mediatingsupply relationships (Chwelos et al. 2001).

A structured questionnaire was developed basedon the fieldwork and a review of prior studies. Theinstrument was refined in pretests with five seniorbuyers in Alpha and five senior managers insupplier firms. The sampling frame was the set of640 supplier firms that had provided more than 0.5percent of the annual requirement of any ofAlpha�s departments during the prior calendaryear. Over 90 percent of the retailer�s purchasesin the prior year were made from this set.

The strategy of collecting data from one focalretailer�s population of suppliers, all of which usethe same SCMS technology, reduces extraneousvariations that might otherwise confound results.Features of the technology and implementation,like the variety of processes supported and thedegree of integration with the retailer�s businessprocesses, are similar across suppliers sampled.This also minimizes variations in retailer attributes,such as supplier involvement in decision making,the level of retailer assistance to supplier firms,and differences in incentives for SCMS adoption.Furthermore, survey data could be supplementedwith information from Alpha�s supplier databases.Although sampling a specific supply networkintroduces limitations (discussed later), the advan-tages in this case outweigh the disadvantages.

Response Rate andNonresponse Bias

The final survey instrument was mailed tomanagers in the 640 supplier firms. The responserate was 33 percent (211 valid responses). Acomparison of early and late respondents using at test (p < 0.10) revealed no significant differencesbetween the two groups. They also did not differin their average annual sales to Alpha, years ofassociation with the retailer, or the number ofstock-keeping units (SKUs) supplied. The groupsusing Alpha�s supplier database were alsocompared. There were no significant differencesin dollar volume of purchases in the prior year byAlpha or the number of purchase-order infractions.In addition, 5 percent of the nonrespondents,picked at random, were called. Their responsesalleviated the concern that systematic factorsmight underlie nonresponse. Of the 211 respon-dents, 131 (62 percent) had been using the SCMSfor interaction with Alpha for at least 24 months.The analyses reported here are based on theresponses of this set.9

To avoid common-methods bias, independentassessments of suppliers� competitive perfor-mance were obtained through a survey of Alpha�sbuying group.

Measures

For most constructs, measures validated inprevious studies were adapted. For constructsunique to the model, multiple operational mea-sures based on field interviews were developed.Details of the measures and their sources are inAppendix A.

9Supplier interviews suggest that it took about sixmonths for SCMS implementation and usage tostabilize. This set of 131 firms thus comprises firmsusing the system for retailer interactions routinely for atleast 18 months, thus enabling valid assessments ofpatterns of SCMS use, relationship-specific investments,and benefits.



SCMS Use

The format of items for SCMS use was adaptedfrom Boynton and Zmud (1994). The content ofitems was based on details of SCMS use col-lected in supplier interviews.

SCMS use for exploitation (ITExploit): TheSCMS supported transaction processing activitiesby suppliers such as electronic order receipt andinvoicing, management of packaging, andshipping. The most prominent features in the sys-tem were order management involving electronicorder receipt, order acknowledgement, andinvoice generation. Suppliers needed to usethese functions for basic EDI compliance. Thesystem also provided facilities for suppliers tomaintain quality-inspection data, create advance-shipment notices (ASN), and reserve deliverywindows for supplier trucks at warehouses. It alsosupported manufacturing planning and procure-ment activities and allowed suppliers to managetheir warehouse stocks, in-process inventories,and in-transit inventories. Suppliers� use of thesefeatures reflects increasing degrees of theincorporation of the SCMS in structured tasks.The items for ITExploit tapped the extent of SCMSuse for order processing, invoicing, settlingaccounts, managing inventories, and exchangingshipment and delivery information.

SCMS use for exploration (ITExplore): Theitems for ITExplore assessed the level of SCMSuse to support nonroutine, unstructured tasks.The broad array of information regularly availablethrough the system made it possible for suppliersto understand market trends and customer pre-ferences and to use that knowledge in variousways. Field-service reports made it possible forsuppliers to evaluate the performance of theirproducts after they were sold and gain insights forproduct modifications and new product design.Retail sales and merchandise-return reportsprovided information that suppliers could use tounderstand the distribution of customer pre-ferences and reasons for product returns. Thesummary information in retail sales reports allowssuppliers learn about sales of related products.For example, a manufacturer of stationary exer-

cise bicycles had access to sales summaries fortreadmills. Similarly, a supplier of men�s garmentshad access to data on sales of men�s accessories.The SCMS messaging facilities allowed suppliersto contact individuals and groups within Alphadirectly. For example, a supplier could e-mail aregional field-service group to have defective partsreturned for examination. The items for ITExploreassessed the extent to which suppliers used theSCMS for the unstructured tasks highlighted infield interviews as being important for supplierperformance. One item assessed the extent of itsuse for understanding product and market trendsand customer preferences. A second itemassessed SCMS use for field service informationand the extent to which suppliers used inputs fromAlpha�s post-sale service organization to improvedesign and manufacturing functions.10 A third itempertained to the extent of SCMS use to supportthe creation of new business opportunities toextend the scope of the relationship with Alpha.

Relationship-Specific Investments

The format of items for business-process speci-ficity and domain-knowledge specificity wasadapted from Zaheer and Venkatraman (1994)and Bensaou and Venkatraman (1995).Subramani and Venkatraman (2003) also usedthese items.

Business-process specificity: Alpha repre-sented the SCMS as a system that suppliers coulduse to interact not only with it but also with otherretailers. Consequently, the relationship speci-ficity of supplier investments depends on theextent to which the SCMS and related applications(e.g., converters to interface existing stock-keeping systems to the SCMS) were used exclu-sively to interact with Alpha as opposed to beingused uniformly across other retailers with whichthe supplier worked. Such overlap between Alpha

10Technicians in Alpha�s service group had a reputationfor devising innovative solutions to field problems. Thisdivision also maintained detailed data on part failure andpart replacement gathered from field service calls andcarry-in repairs.



and other retailers occurred largely in billing andinventory management, and one item focused onthe relationship specificity of these processes. Asecond item captured the relationship-specificnature of supplier investments in vendor selection,cost accounting procedures, and other adminis-trative procedures. Relationship-specific changesto operating procedures in manufacturing andshipping functions were highlighted as anothersource of business-process specificity. Forinstance, the mattress manufacturer using theSCMS to support the move to a make-to-ordersystem introduced a variety of changes in itsmanufacturing, packaging, and shipping pro-cesses. These supplier investments were usefulonly in working with Alpha. The third item cap-tured the extent of such relationship-specificinvestments in operating procedures.

Domain-knowledge specificity: The items cap-tured the specialized intangible investments bysuppliers in understanding Alpha�s requirementsand the unique context of the interaction. Theyfocus on the specificity of both component andarchitectural knowledge and cover expertisedeveloped for new-product planning, productconception and design, and pricing. These threeareas reflect a supplier�s understanding andknowledge of Alpha�s market positioning andcustomer expectations.

Supplier Benefits

The items focused on benefits highlighted insupplier interviews and in prior literature.Because Alpha was expected to favor vendorsadopting the SCMS, suppliers were likely to derivecost efficiencies from higher sales volumes. Thisargument is consistent with the observation byMukhopadhyay and Kekre (2002, p. 1312) thatsuppliers adopting business-to-business systemschampioned by customer firms are �rewarded withhigher sales volumes.� Suppliers were unlikely toderive benefits in the form of price increasesbecause Alpha negotiated price reductions insupply contracts, which is a pattern also observedin other studies (Dyer and Nobeoka 2000).Suppliers could also benefit from process

improvements and the creation of new processes.For example, online ordering could help eliminateorder-entry errors, electronic booking of deliveryslots could reduce trucks� idle times at Alpha�swarehouses, and submitting invoices electroni-cally could enable timely payments by customers.Suppliers could also increase their overallprofitability through efficiencies created by SCMSuse and use of information made available by thesystem. The three items for operational benefitstapped the extent to which suppliers derivedbenefits from these sources.

Measures of strategic benefits assessed out-comes that positioned suppliers more advan-tageously in their relationship with Alpha. Theseincluded learning about Alpha�s customer seg-ments and their preferences, the extent to whichsuppliers were able to create new products orenhancements for Alpha, and the extent to whichsuppliers developed new business opportunities inthe exchange. Alpha�s merchandising groupsused these three criteria to nominate firms for�Supplier of the Year� awards.

Suppliers� competitive performance was mea-sured using two items focusing on trends in salesto Alpha and trends in the suppliers� market sharewith Alpha. Managers in Alpha�s merchandisinggroup provided assessments of suppliers� per-formance11 with respect to competing suppliers inthe product category (e.g., small appliances,footwear, and luggage). Because none of thesuppliers were sole providers, there was always acomparison set available.

Control Variables

Measures of uncertainty, retailer replaceability,length of association, and size were adapted fromprior studies. Appendix A lists the items used andtheir sources.

11Obtaining independent assessments from informantsin Alpha guards against common-methods bias that canarise when an informant provides assessments of bothdependent and independent variables.



Data Analysis

Structural equation modeling procedures imple-mented in PLS Graph were used to perform asimultaneous evaluation of both the quality ofmeasurement (the measurement model) andconstruct interrelationships (the structural model).PLS Graph provides the ability to model latentconstructs even under conditions of non-normalityand small- to medium-size samples (Chin et al.1996). The sample of 132 cases is adequate forPLS analysis. It satisfies the heuristic that thesample size be at least 10 times the largestnumber of structural paths directed at any oneconstruct.12

Testing Direct Effects

Hypotheses postulating direct effects betweenconstructs (H1, H2, H6, and H7) were testedbased on the magnitude and significance of pathscomputed by PLS Graph. Hypotheses regardingdifference in the strengths of multiple paths (H3a,H3b) were tested by comparing the path magni-tudes calculated by PLS.

Testing Mediated Effects

The mediation hypotheses (H4, H5) were tested intwo ways. The first approach compares theresearch model (with mediated paths from SCMSuse to benefits) to a competing model (incor-porating a direct link between the constructs).Because the two models are nested, model-comparison procedures using PLS results enablestatistical conclusions to be reached regardingmodel fit. The second approach uses mediation-analysis techniques (Hoyle and Kenny 1999) tocalculate the magnitude and significance ofindividual mediated paths based on values ofstandardized direct paths computed in PLS.

Appendix B describes the two complementaryapproaches to test mediated effects.

Results

This section presents the measurementproperties, sample demographics, and results ofhypothesis testing.

Measurement Properties

Internal consistency was assessed by examiningρc, a measure of composite reliability. The ρc

values for constructs are all above the suggestedthreshold of 0.7, indicating reliable measurement13

(Appendix A). The values of average varianceextracted (AVE)�the ratio of the constructvariance to the total variance among indica-tors�are all above the recommended threshold of0.50. The measures exhibit satisfactory conver-gent and discriminant validity. The values of thesquare root of the AVE (reported on the diagonalin Table 2) are all greater than the inter-constructcorrelations (the off-diagonal entries in Table 2).

Informant and Sample Demographics

Of the informants, 78 percent were senior mana-gers (titles such as general manager, vicepresident, or CEO) with an average tenure of 14years. The sample consists of well-establishedfirms with an average of 32 years of operation and260 employees in a range of industries, such asgarments, fashion accessories, luggage, furniture,lighting, sports equipment, and hardware. Thefirms had longstanding associations with Alpha(mean of 19 years), consistent with prior suppliersamples in the distribution channel (Heide 1994).The suppliers were predominantly small firms.

12The largest number of paths to any construct in theresearch model is six. This count includes the pathsfrom the four control variables that are not shown inFigure 2.

13All the loadings of items on constructs used tocalculate ρc were significant at p < .01 and wereuniformly high. Twenty-one of the 27 loadings wereabove 0.8, with the lowest loading being 0.71.



SCMS Use Relationship-Specific Intangible Investments

First-Order Benefits

Second-Order Benefits

Control Variables

Notes:* p < .05, ** p < .01� All R2 values are significant at p < .01.� Non-significant paths in the model indicated with dashed lines.� For clarity, control variables are not individually indicated in the

model. Each of these was modeled with a direct path to bothOperational and Strategic Benefits.

R2 = 0.30

SCMS Use for Exploration

SCMS Use for Exploitation

Domain-knowledge specificity

Strategic Benefits


Competitive Performance

R2 = 0.21

R2 = 0.29

R2 = 0.24

R2 = 0.19

0.22*

0.05

0.20*

0.44*

0.05

0.27**

0.01

0.37**

0.23*

0.36**

Business-processspecificity

Size, Years ofAssociation, Retailer

Replaceability, Uncertainty

SCMS Use Relationship-Specific Intangible Investments

First-Order Benefits

Second-Order Benefits

Control Variables

Notes:* p < .05, ** p < .01� All R2 values are significant at p < .01.� Non-significant paths in the model indicated with dashed lines.� For clarity, control variables are not individually indicated in the

model. Each of these was modeled with a direct path to bothOperational and Strategic Benefits.

R2 = 0.30

SCMS Use for Exploration

SCMS Use for Exploitation

Domain-knowledge specificity

Strategic Benefits


Competitive Performance

R2 = 0.21

R2 = 0.29

R2 = 0.24

R2 = 0.19

0.22*

0.05

0.20*

0.44*

0.05

0.27**

0.01

0.37**

0.23*

0.36**

Business-processspecificity

Size, Years ofAssociation, Retailer

Replaceability, Uncertainty

Figure 2. Research Model Linking SCMS Use, Relationship-Specific Investmentsand Supplier Benefits

Over 55 percent had annual sales between $11and $50 million. On average, Alpha accounted for18 percent of the annual sales of firms in thesample. Only 6 percent indicated that Alphaaccounted for over 50 percent of their sales. Over80 percent of the annual sales to Alpha weremade under Alpha�s home brands and privatelabels. This statistic is interesting, consideringthat most firms (55 percent) indicated that theirproducts were among the top three brands in theircategories.14 On average, the firms supplied 235different SKUs.15 The retail prices of the items

ranged from $20 to $75. The products involvedlittle customization for Alpha (median of 2 on aseven-point scale), were relatively complex(median of 4 on a seven-point scale), and weremore fashion and style-driven than utilitarian(median of 3 on a seven-point scale).

The means, standard deviations, and correlationof constructs are in Table 2. The significancelevels of paths in the research model weredetermined using PLS�s bootstrap resamplingprocedures. Overall, the results suggest asatisfactory fit of the model to the data. The R2

values for the dependent constructs range from0.19 to 0.30. They are all significant at p < .01.The results of the PLS analysis are in Table 3 andFigure 2. The results of the mediation analysesare in Tables 4 and 5.

Table 3 and Figure 2 provide the magnitude andsignificance of inter-construct relationships.Retailer replaceability is inversely related to both

14This high figure suggests that Alpha was successful inleveraging the expertise of leading suppliers in supplyingAlpha�s house brands. House brands in general areassociated with higher margins than supplier brands.

15In many instances, this number does not necessarilyreflect a diversity of products. Supplying dress shirts ofone style in four different sizes and seven differentpattern/color combinations amounts to 28 SKUs. Withtwo styles, the number rises to 56.

Table 2. Means, SD, Inter-Construct Correlations ,and Average Variance Extracted (n = 131)Mean(SD)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)

IT Use fo Exploitation (1)

5.12(1.61)

0.93

IT Use for Exploration(2)

2.13(1.53)

0.188 0.82

Business Process Specificity (3)

5.68(1.15)

0.321* 0.039 0.82

Domain-knowledgespecificity (4)

2.91(1.35)

0.329* 0.468* 0.200 0.88

Operational Benefits(5)

2.80(1.57)

0.179* 0.343* 0.163 0.550* 0.81

Strategic Benefits (6) 2.85(1.80)

0.258* 0.352* 0.257* 0.410* 0.489* 0.84

Competitiveperformance (7)

4.33(1.48)

0.086 0.005 0.013 0.158 0.173 0.274* 0.93

Uncertainty (8) 3.38(1.37)

0.049 0.198 0.077 0.197 -0.028 -0.044 0.131 0.79

Retailer Replaceability(9)

3.82(1.59)

0.028 0.07 -0.10 -0.111 -0.23** -0.31** -0.67** 0.13 0.89

Size (10) 3.44(1.30)

0.132 0.165 -0.152 0.205* -0.019 -0.015 -0.01 0.328** 0.19* 1.00

Yrs of Association (11) 18.98(12.75)

0.137 0.117 -0.187* 0.097 -0.002 0.055 0.007 0.133 -0.069 0.32** 1.00

Notes: a) Figures in shaded diagonal are values of square root of the AVE b) * = p < .05, ** = p < .01



Table 3. Path Coefficients

Path/HypothesisPath

Coefficient t Value

ITExploit ! Business-process specificity (H1) 0.22 2.42**

ITExploit ! Domain-knowledge specificity 0.20 2.13*

ITExplore ! Business-process specificity 0.05 0.071

ITExplore ! Domain-knowledge specificity (H2) 0.44 5.77**

Business-process specificity ! Operational Benefits 0.05 0.71

Business-process specificity ! Strategic Benefits 0.23 2.54**

Domain-knowledge specificity ! Operational Benefits 0.36 4.77**

Domain-knowledge specificity ! Strategic Benefits 0.37 4.63**

Operational Benefits ! Competitive performance (H6) 0.01 0.18

Strategic Benefits ! Competitive performance (H7) 0.27 3.29**

Uncertainty ! Operational Benefits -0.15 0.60

Uncertainty ! Strategic Benefits -0.11 1.64

Retailer Replaceability ! Operational Benefits 0.17 1.77*

Retailer Replaceability ! Strategic Benefits 0.21 2.79**

Size ! Operational Benefits -0.10 0.98

Size ! Strategic Benefits -0.08 0.20

Years of Association ! Operational Benefits -0.09 0.54

Years of Association ! Strategic Benefits 0.08 0.58

Note: 1. * = p < .05, ** = p < .01 in one-tailed tests2. Only the hypotheses tested based on individual path magnitudes (H1, H2, H6, H7) are listed

here.

Table 4. Nested Model Comparison

Direct Path

R2 in MediatedModel (no

direct path)

R2 withDirectPath

f 2

ValuePseudo F F(1,130) Conclusion

ITExploit ! OperationalBenefits

0.238 0.244 0.008 1.000 Not significant

ITExploit ! StrategicBenefits

0.300 0.305 0.007 0.913 Not significant

ITExplore !Operational Benefits

0.238 0.246 0.011 1.323 Not significant

ITExplore ! StrategicBenefits

0.300 0.317 0.025 3.109 Not significant



Table 5. Significance of Mediated Paths from IT Use to Benefits (n = 131)

Indirect Effect Row Mediated PathsGraphical

Representation Patha z stat

ITExploit !OpBen

a ITExploit ! BPS ! OpBen 0.011 0.680

b ITExploit ! DKSpec ! OpBen 0.073 2.435*

ITExploit !StrBen

c ITExploit ! BPS ! StrBen 0.050 1.962*

d ITExploit ! DKSpec ! StrBen 0.075 2.417*

ITExplore !OpBen

e ITExplore ! BPS ! OpBen 0.003 0.443

f ITExplore ! DKSpec !OpBen

0.161 3.674**

ITExplore !StrBen

g ITExplore ! BPS ! StrBen 0.012 0.560

h ITExplore ! DKSpec !StrBen

0.165 3.613**

Note: * = p < .05, ** = p < .01 in one-tailed tests a: Standardized path magnitude

operational benefits (path = -0.17) and strategicbenefits (path = -0.21). This negative relationshipbetween retailer replaceability and supplier bene-fits (or expressed differently, the positive rela-tionship between supplier dependence and sup-plier benefits) is consistent with the sampling of aset of longstanding, ongoing supplier-retailerrelationships. Uncertainty, size, and years ofassociation are not significantly related to either ofthe dependent variables. The results of hypo-thesis testing are summarized below.

Hypothesis 1, Supported: The path fromITExploit to business-process specificity (path =0.22, t = 2.42, p < .01) is positive and significant(Table 3).

Hypothesis 2, Supported: The path fromITExplore to domain-knowledge specificity (path =0.44, t = 5.77, p < .01) is positive and significant(Table 3).

Hypothesis 3a, Supported: The path fromITExploit to business-process specificity (path =



0.22, t = 2.42, p < .01) is positive and significant,but that from ITExplore to business-processspecificity (path = 0.05, t = 0.71) is positive but notsignificant. A one-sided test of this directionalhypothesis indicates that the magnitude of thepath from ITExploit to business-process specificityis greater than that from ITExplore to business-process specificity (difference = 0.173, t = 1.86, p< .05).

Hypothesis 3b, Supported: The path fromITExplore to domain-knowledge specificity (path =0.44, t = 5.77, p < .01) is positive and significant,as is the path from ITExploit to domain-knowledgespecificity (path = 0.20, t = 2.13, p < .05). A one-sided test for the directional hypothesis indicatesthat the magnitude of the path between ITExploreand domain-knowledge specificity is greater thanthat between ITExploit and domain-knowledgespecificity (difference = 0.243, t = 3.30, p < .01).

Overall, all hypotheses relating patterns ofsuppliers� SCMS use to relationship-specificintangible investments are supported by the data.

Hypothesis 4, Supported: The results of nested-model comparison (Table 4) indicate that theeffect of ITExploit on benefits is completelymediated by relationship-specific assets. Theresearch model incorporating mediated paths ispreferred to competing models that include directpaths from ITExploit to operational benefits (Table4, row 1) and strategic benefits (Table 4, row 2).The analysis of individual mediated paths (Table5) indicates that three of the four mediated pathsfrom ITExploit to benefits are positive andsignificant (Table 5, rows b, c, and d).

Domain-knowledge specificity mediates the linkbetween ITExploit and both operational benefitsand strategic benefits. Both the indirect paths arepositive and significant (Table 5, rows b and d).Business-process specificity mediates the linkbetween ITExploit and strategic benefits. Theindirect path is positive and significant (Table 5,row c). However, the indirect path from ITExploitto operational benefits through business-processspecificity is positive but not significant (Table 5,row a).

Hypothesis 5, Supported: The nested modelcomparison (Table 4) indicates that the effect ofITExplore on benefits is completely mediated byrelationship-specific assets. The research modelincorporating mediated paths is preferred tocompeting models including direct paths tooperational benefits (Table 4, row 3) and tostrategic benefits (Table 4, row 4). The analysis ofindividual mediated paths indicates that two of thefour mediated paths from ITExplore to benefits arepositive and significant (Table 5, rows f and h).

Domain-knowledge specificity mediates the effectof ITExplore on both operational and strategicbenefits. The indirect paths are positive andsignificant (Table 5, rows f and h). The indirectpaths to operational and strategic benefits throughbusiness-process specificity are both positive butnot significant (Table 5, rows e and g).

Overall, the hypothesized mediated effects ofITExploit and ITExplore on benefits are supported.

Hypothesis 6, Not Supported: The pathbetween operational benefits and competitiveperformance (Table 3) is not significant (path =0.010, t = 0.1, ns).

Hypothesis 7, Supported: The path betweenstrategic benefits and competitive performance(Table 3) is positive and significant (path = 0.265,t = 3.291, p < .01).

Discussion and Conclusions

This paper develops and tests theory relatingsupplier firms� use of information technologies insupply chains to the profile of their relationship-specific intangible investments and outcomes forthese firms. The study contributes to the literatureby providing evidence that (1) suppliers canbenefit by participating in supply chain manage-ment initiatives of network leaders, (2) benefitsfrom SCMS use are mediated by suppliers�deployment of relationship-specific business pro-cesses and domain knowledge in the exchange,



and (3) the profile of relationship-specific invest-ments by suppliers is linked to their patterns ofSCMS use. These results highlight the impor-tance of recognizing variations in patterns ofappropriation of supply chain technologies�forexploitation and exploration�to articulate benefitsto suppliers from these systems.

The study contributes to theory by highlighting therole of relationship-specific assets in the dynamicsof value creation and value retention in contexts ofIT-mediated buyer-supplier interactions. Whileprior research has largely emphasized the partplayed by relationship-specific assets in enablingvalue creation by suppliers (Dyer 1996a, 1996b),this study extends current theory by highlightingtheir complementary role in value retention. Thisrole is particularly salient in the context of SCMSuse where both suppliers and a dominant buyercontend for the benefits created by the use ofinformation systems. As patterns of SCMS usethat create value are recognized and adopted bycompeting suppliers and become standard prac-tice in supplier networks, the benefits created bysuch uses of SCMS are bargained away bysuppliers in negotiations or appropriated by focalfirms (Bakos and Brynjyolfsson 1993). This logicmay explain why prior studies found few benefitsto suppliers from systems championed by focalfirms and cast the adoption of these systems asan unfortunate strategic necessity for suppliers(Barua and Lee 1997). When suppliers combineSCMS use with investments in relationship-specific intangible assets, however, the causalambiguity of the combination raises barriers toimitation and enables system use to become alever for differentiation. In effect, relationship-specific intangible investments that cloak SCMSuse in idiosyncratic features prevent competingsuppliers and the retailer from unpacking thecomplex causal structure contributing to supplierperformance in the exchange. This causal ambi-guity enables suppliers to offset the asymmetry inbargaining power in the exchange, which in turnallows them to retain some of the benefits createdby SCMS use.

The strategic combination of SCMS use andrelationship-specific investments thus enhances

suppliers� ability to benefit from SCMS use. It alsoillustrates how suppliers can create negativeexternalities for competitors in IT-mediated rela-tionships to improve their own relative advantage.This logic of IT-enabled competitive advantage ininterfirm contexts thus reflects an interestinginstance of the complementarity of the prescrip-tions of transactions-cost and resource-basedviews, providing a context-based explanation ofhow relationship-specific assets can also bestrategic assets. In this respect, the data suggestthat domain-knowledge specificity is more potentthan business-process specificity as a basis forderiving strategic advantage.

The results provide empirical support for Bakosand Brynjolfsson�s (1993) vendors-to-partnersthesis whereby IT use in buyer-supplier ex-changes leads to closer cooperative relationships.The results further refine this insight, indicatingthat the nature of intangible, relationship-specificinvestments is influenced by patterns ofappropriations of SCMS by suppliers. SCMS usefor exploitation, which focuses on achievingexchange efficiency, is dominant in enabling thecreation and deployment of co-specialized busi-ness processes. SCMS use for exploration,involving the exchange of unstructured informa-tion, collaborative interactions, and analysis ofdata to infer cause-effect relationships, is domi-nant in enabling the development and use ofspecialized domain knowledge.

This view is consistent with suggestions thatinformation technologies are complex artifactsproviding multiple affordances to users. Thesetechnologies can be interpreted and used indifferent ways by different actors (Pentland 1992).The results of the current research indicate thatthe affordances introduced by SCMS can pertainboth to structures of interfirm interaction incor-porated in routines (business processes) and tocompetence structures underlying action (domainknowledge). In the context of suppliers� SCMSuse, the results demonstrate that different appro-priations of SCMS direct attention to differing setsof issues in the interaction, thereby leading todifferent choices of intangible investments andultimately to different outcomes.



This study has several limitations. The data werecollected from the supplier network of one largeretailer. While this sampling choice minimizesconfounds by holding focal-firm characteristicsconstant, it limits the generalizability of the results.Reliance on the report of a single informant withineach supplier firm is also a limitation. Further-more, while the theory linking SCMS use,relationship-specific assets, and benefits iscausal, and path analysis represents relationshipsbetween constructs as causal links, the paperuses cross-sectional data to empirically assessthese relationships. The results thus reflect asso-ciations rather than causal links between con-structs. Moreover, while the paper draws on priortheory and evidence to model SCMS use as aprecursor to relationship-specific investments,rival arguments that suppliers may makerelationship-specific investments in advance ofSCMS use cannot be discounted. Biases fromomitted variables and the possibility of mutualinfluence among constructs in the model are alsolimitations.

The constructs in the model and their inter-relationships are independent of the technologyunderlying the SCMS examined in this study. Thefindings are therefore likely to be generalizable toother contexts with different underlying tech-nologies. For instance, the emergence of theInternet as a medium for business-to-businesstransactions has led to the development of privateexchanges�SCMS implementations based onopen standards and Web technologies�wherenetwork leaders control supplier participation(Harris 2001). Recent work suggests that systemfeatures and business dynamics in Internet-basedprivate exchanges are similar to those in theSCMS studied here (Icasati-Johanson and Fleck2003). The results are generalizable to suchcontexts. The results need to be applied withcaution, however, as developments in informationtechnologies continue to expand the variety andsophistication of SCMS features. Exchanges pro-viding richer transactional and collaborativefeatures may exhibit characteristics different fromthe system studied here. Furthermore, the resultsapply only to settings in which the fundamental

assumption of this study holds: that relationship-specific intangible assets developed in ongoinginteractions create more value than generic,nonspecific assets available in spot markets. Thissituation may not be the case in public exchangesor electronic markets (Choudhury et al. 1998) inwhich firms can aggressively seek suppliersoffering the lowest price each time they need tobuy goods or services (Harris 2001). When buyersand sellers engage in plug-and-play interactionsand history effects are unimportant, modularityand flexibility are more valuable than relationshipspecificity (Lee 2002). Nonetheless, the con-structs developed in this paper can serve as auseful point of departure to examine valuecreation in these contexts.

This study highlights a variety of issues for futureinvestigation. The finding that the path fromITExploit to operational benefits mediated bybusiness-process specificity is not significantwarrants further scrutiny. Moves toward stan-dardization and modularization of businessprocesses in contexts of IT mediation possiblymake them less viable as a source of benefits.Organizational process reconfigurations andexpertise development are time- and resource-intensive. The cost-effectiveness of such invest-ments needs further study. The finding in thisstudy that operational benefits are not linked tocompetitive benefits in the form of enhancedrevenues and market share also deserves furtherexamination. While this result may reflect thevalue Alpha placed on cultivating suppliers thatemphasized new product development andinnovation, it may also indicate that havingsuppliers focus exclusively on operational benefitsfrom SCMS use may be counterproductive.Future research on the role of SCMS in enablingexpertise coordination (Faraj and Sproull 2000),strategic learning (Thomas et al. 2001), anddeveloping functional capabilities (Dutta et al.2003) at the team and organizational levels withinsupplier firms can also yield useful insights.Evolving a richer conceptualization of domainknowledge specificity incorporating the influenceof constructs such as social capital and functionalexpertise (Subramaniam and Youndt 2003) is yetanother promising direction for future work.



The results have several implications for practice.Suppliers can anticipate the relationship-specificinvestments in adopting SCMS and their links topatterns of SCMS use. They can recognize theadvantages of deploying co-specialized intangibleassets in IT-mediated exchanges and thepossibility of employing such assets strategicallyto lock larger customers into bilateral relation-ships. In this way, they can enhance their abilityto appropriate an equitable share of the valuecreated in the exchange. For focal firms pursuingthe goal of leveraging the resource endowmentsof their supplier networks, the results suggest thatinfluencing patterns of SCMS use by suppliers isan important means to achieve their objective.The results highlight the opportunity for networkleaders to leverage supplier capabilities throughpolicies that encourage non-contractible, rela-tionship-specific supplier investments. They alsosuggest that advantages accrue from managingtheir dependence upon suppliers through rela-tional rather than contractual processes.

The contribution of information technologies to thecreation and deployment of capabilities inorganizational networks is an important theme inIS research. This study highlights that patterns ofIT use in interfirm relationships are antecedents ofco-specialized supplier investments in businessprocesses and domain knowledge (resources thatcreate competitive advantages for supply chains).For suppliers in asymmetric relationships with anetwork leader, these investments, which com-plement IT use, play an important part. They arecentral to value creation. They are also keyelements of the strategic calculus of claimingvalue post hoc in the relationship. As firmsincreasingly rely on IT-mediated interfirm rela-tionships to develop and deploy capabilities,recognizing the complex interplay of contextualfeatures, appropriations of technology, and firmstrategies is important in explicating the roleplayed by information technologies.

Acknowledgements

This research was supported by a grant-in-aidfrom the Office of the Vice President and the Dean

of the Graduate School of the University of Minne-sota. The project was also supported by theSystems Research Center at Boston University.I am thankful to John Henderson and N. Venkatra-man of Boston University, Gurbux Singh, andmanagers at the retailer organization for theirsupport and assistance at different stages of thisproject.

The comments and suggestions of participants inworkshops at the University of Minnesota�sInformation & Decision Sciences Department andMarketing Department, the University of Kansas atLawrence, the University of Maryland at CollegePark, and Michigan State University were veryuseful in developing the ideas in the paper.

I am grateful for detailed comments on earlierdrafts of the paper from Cynthia Beath, MarkBergen, Gordon Davis, Samer Faraj, RobKauffman, Tridas Mukhopadhyay, Fred Riggins, V.Sambamurthy, Harry Sapienza, HuseyinTanriverdi, Stephanie Watts, and Aks Zaheer.Wynne Chin and David Gefen provided valuablehelp with the PLS analyses.

References

Bakos, J. Y., and Brynjyolfsson, E. �FromVendors to Partners: Information Technologiesand Incomplete Contracts in Buyer SellerRelationships,� Journal of Organizational Com-puting (3:3), 1993, pp. 301-329.

Barua, A., Kriebel, C. H., and Mukhopadhyay, T.�Information Technologies and Business Value:An Analytic and Empirical Investigation,�Information Systems Research (6:1), 1995, p.1995.

Barua, A., and Lee, B. �An Economic Analysis ofthe Introduction of an Electronic Data Inter-change System,� Information SystemsResearch (8:4), 1997, pp. 398-422.

Bensaou, M., and Venkatraman, N. �Con-figurations of Interorganizational Relationships:A Comparison Between U.S. and JapaneseAutomakers,� Management Science (41:9),1995, pp. 1471-1492.



Bharadwaj, A. �A Resource Based Perspective onInformation Technology Capability and FirmPerformance: An Empirical Investigation,� MISQuarterly (24:1), 2000, pp. 169-196.

Boynton, A. C., and Zmud, R. W. �The Influenceof IT Management Practice on IT Use in LargeOrganizations,� MIS Quarterly (18:3), 1994, pp.299-319.

Carter, J. R. �The Dollars and Sense of ElectronicData Interchange,� Production and InventoryManagement (31:2), 1990, pp. 22-26.

Chin, W. W., Marcolin, B. L., and Newsted, P. R.�A Partial Least Squares Latent VariableModeling Approach for Measuring InteractionEffects: Results from a Monte Carlo SimulationStudy and Voice Mail Emotion/Adoption Study,�in Proceedings of the Seventeenth InternationalConference on Information Systems, J. I.DeGross, A. Srinivasan, and S. L. Jarvenpaa(eds.), Cleveland, OH, 1996, pp. 21-41.

Choudhury, V., Hartzel, K. S., and Konsynski, B.R. �Uses and Consequences of ElectronicMarkets: An Investigation in the Aircraft PartsIndustry,� MIS Quarterly (22:4), December1998, pp. 471-507.

Chwelos, P., Benbasat, I., and Dexter, A. S.�Research Report: Empirical Test of an EDIAdoption Model,� Information SystemsResearch (12:3), 2001, pp. 304-321.

Clemons, E. K., and Row, M. �Limits to InterfirmCoordination through Information Technology:Results of a Field Study in Consumer GoodsDistribution,� Journal of Management Infor-mation Systems (10:1), 1993, pp. 73-95.

DeSanctis, G., and Poole, M. S. �Capturing theComplexity in Advanced Technology Use:Adaptive Structuration Theory,� OrganizationScience (5:2), 1994, pp. 121-147.

Duliba, K. A., Kauffman, R. J., and Lucas, H. C.�Appropriating Value from ComputerizedReservation System Ownership in the AirlineIndustry,� Organization Science (12:6), 2001,pp. 702-728.

Dutta, S., Zbaracki, M. J., and Bergen, M. �PricingProcess as a Capability: A Resource-BasedPerspective,� Strategic Management Journal(24:7), 2003, pp. 615-631.

Dyer, J. H. �Does Governance Matter? KeiretsuAlliances and Asset Specificity as Sources of

Japanese Competitive Advantage,� Organi-zation Science (7:6), 1996a, pp. 649-666.

Dyer, J. H. �Specialized Supplier Networks as aSource of Competitive Advantage: Evidencefrom the Auto Industry,� Strategic ManagementJournal (17:4), 1996b, pp. 271-292.

Dyer, J. H., and Nobeoka, K. �Creating and Man-aging a High Performance Knowledge-SharingNetwork: The Toyota Case,� Strategic Manage-ment Journal (21:3), 2000, pp. 345-367.

Dyer, J. H., and Singh, H. �The Relational View:Cooperative Strategy and Sources of Inter-organizational Competitive Advantage,�Academy of Management Review (23:4), 1998,pp. 660-679.

El Sawy, O. A., Malhotra, A., Gosain, S., andYoung, K. M. �IT-Intensive Value Innovation inthe Electronic Economy: Insights from MarshallIndustries,� MIS Quarterly (23:3), 1999, pp.305-335.

Faraj, S., and Sproull, L. �Coordinating Expertisein Software Development Teams,� Manage-ment Science (46:12), 2000, pp. 1554-1568.

Ghosh, M., and John, G. �Governance ValueAnalysis and Marketing Strategy,� Journal ofMarketing (63), 1999, pp. 131-145.

Harris, N. ��Private Exchanges� May Allow B-to-BCommerce to Thrive After All,� Wall StreetJournal, March 16, 2001, p. B1.

Hart, P., and Saunders, C. �Power and Trust:Critical Factors in the Adoption and Use ofElectronic Data Interchange,� OrganizationScience (8:1), 1997, pp. 23-42.

Heide, J. B. �Interorganizational Governance inMarketing Channels,� Journal of Marketing(58:1), 1994, pp. 71-85.

Hoyle, R. H., and Kenny, D. A. �Statistical Powerand Tests of Mediation,� in Statistical Strategiesfor Small Sample Research, R. H. Hoyle (ed.),Sage Publications, Newbury Park, CA, 1999.

Icasati-Johanson, B., and Fleck, S. J. �Impact ofE-Business Supply Chain Technology on Inter-organizational Relationships: Stories From theFront Line,� presentation at the Sixteenth BledElectronic Commerce Conference, Bled, Slo-venia, 2003.

Lee, H. G., Clark, T., and Tam, K. Y. �Can EDIBenefit Adopters?,� Information SystemsResearch (10:2), 1999, pp. 186-195.



Lee, H. L. �Aligning Supply Chain Strategies withProduct Uncertainties,� California ManagementReview (44:3), 2002, pp. 105-119.

March, J. G. �Exploration and Exploitation inOrganizational Learning,� Organization Science(2:1), 1991, pp. 71-87.

Massetti, B., and Zmud, R. W. �Measuring theExtent of EDI Usage in Complex Organizations:Strategies and Illustrative Examples,� MISQuarterly (20:3), 1996, pp. 331-345.

Mukhopadhyay, T., and Kekre, S. �Strategic andOperational Benefits of Electronic Integration inB2B Procurement Processes,� ManagementScience (48:10), 2002, pp. 1301-1313.

Mukhopadhyay, T., Kekre, S., and Kalathur, S.�Business Value of Information Technology: AStudy of Electronic Data Interchange,� MISQuarterly (19:2), June 1995, pp. 137-157.

Pentland, B. T. �Organizing Moves in SoftwareSupport Hot Lines,� Administrative ScienceQuarterly (37:4), 1992, pp. 527-548.

Premkumar, G., Ramamurthy, K., and Crum, M.R. �Determinants of EDI Adoption in theTransportation Industry,� European Journal ofInformation Systems (6:2), 1997, pp. 107-121.

Riggins, F. J., and Mukhopadhyay, T. �Inter-dependent Benefits from InterorganizationalSystems: Opportunities for Business PartnerReengineering,� Journal of MIS (11:2), 1994,pp. 37-57.

Sabherwal, R., and Chan, Y. E. �AlignmentBetween Business and IS Strategies: A Studyof Prospectors, Analyzers, and Defenders,�Information Systems Research (12:1), 2001,pp. 11-33.

Shah, R., Goldstein, S. M., and Ward, P. T.�Aligning Supply Chain Management Charac-teristics and Interorganizational InformationSystem Types: An Exploratory Study,� IEEETransactions on Engineering Management(49:3), August 2002, pp. 282-292.

Subramani, M., and Venkatraman, N. �Safe-guarding Investments in Asymmetric Interorga-

nizational Relationships: Theory and Evi-dence,� Academy of Management Journal(46:1), February 2003, pp. 46-62.

Subramanian, M., and Youndt, M. A. �TheInfluence of Intellectual Capital on the Types ofInnovative Capabilities,� Working Paper,Boston College, 2003.

Takeishi, A. �Knowledge Partitioning in theInterfirm Division of Labor: The Case of Auto-motive Product Development,� OrganizationScience (13:3), 2002, pp. 321-338.

Thomas, J. B., Sussman, S. W., and Henderson,J. C. �Organizational Learning, KnowledgeManagement, and Sensemaking,� OrganizationScience (12:3), 2001, pp. 331-345.

Williamson, O. E. �Transaction Cost Economicsand Organization Theory,� in: OrganizationTheory: From Chester Barnard to the Presentand Beyond, O.E. Williamson (ed.), OxfordUniversity Press, New York, 1995.

Zaheer, A., and Venkatraman, N. �Determinantsof Electronic Integration in the InsuranceIndustry: An Empirical Test,� ManagementScience (40:5), 1994, pp. 549-566.

About the Author

Mani Subramani is an assistant professor in theInformation and Decision Sciences Department atthe Carlson School of Management, University ofMinnesota. His research focuses on the strategicrole of information technology within the organi-zation and in interorganizational relationships. Hiscurrent areas of research are knowledge manage-ment and the leveraging of organizational capa-bilities using information technologies. His workhas been published in Academy of ManagementJournal, Communications of the ACM, Journal ofManagement Information Systems, InformationSystems Research, and Sloan ManagementReview.



Appendix A

Details of Measures

Strategic Benefits (ρρρρc = 0.87)

Please indicate the extent to which you are receiving the following benefits as a result of your relationshipwith Alpha:

� Learning about customers and markets for our products� Creation of new products, product enhancements� Development of new business opportunities

Scale: Little or none of this benefit � Some Level of this benefit � High level of this benefit (1-7 Scale)

Operational Benefits (ρρρρc = 0.85)

Please indicate the extent to which you are receiving the following benefits as a result of your relationshipwith Alpha:

� Cost Efficiencies from higher sales volumes� Improvements to current processes or creation of new processes� Increased profitability

Scale: Little or none of this benefit � Some Level of this benefit � High level of this benefit (1-7 Scale)

Business-Process Specificity (ρρρρc = 0.86)

The extent to which the software and applications used (e.g., billing, inventory management, EDI, etc.) insupplying Alpha are relatively similar or are significantly different from what you use with other retailers.

The extent to which the administrative procedures used (e.g., vendor selection, cost accounting proceduresetc.) in supplying Alpha are relatively similar or are significantly different from what you use with otherretailers.

The extent to which the operating procedures used (e.g., manufacturing, bar-coding, packaging, shippingprocedures etc.) in supplying Alpha are relatively similar or are significantly different from what you use withother retailers.

Scale: Relatively Similar as with other Retailers � Moderately Customized � Significantly Customizedfor Alpha (1-7 Scale)

(Zaheer and Venkatraman 1994; Bensaou and Venkatraman 1995)



Domain-Knowledge Specificity (ρρρρc = 0.91)

The extent to which the knowledge and understanding used in planning for new products, programs forAlpha is significantly specific to the relationship (i.e., customized for Alpha) or is relatively similar to whatyou use with other retailers.

The extent to which the knowledge and understanding used in product conception and design for Alphais significantly specific to the relationship (i.e., customized for Alpha) or is relatively similar to what you usewith other retailers.

The extent to which the knowledge and understanding used in determining product pricing for Alpha issignificantly specific to the relationship (i.e., customized for Alpha) or is relatively similar to what you usewith other retailers.

Scale: Relatively Similar as with other Retailers � Moderately Customized � Significantly Customizedfor Alpha (1-7 Scale)

(Zaheer and Venkatraman 1994; Bensaou and Venkatraman 1995)

IT Use for Exploitation (ρρρρc = 0.94)

Please indicate the extent to which you use specific Information Technology based support for the followingin your relationship with Alpha:

� The extent to which you use specific IT based support for order Processing, invoicing and settlingaccounts

� The extent to which you use specific IT based support for exchange of shipment and deliveryinformation

� The extent to which you use specific IT based support for managing warehouse stock andinventories

Scale: Minimal Use � Some Use � Significant Use (1-7 Scale)

(Boynton and Zmud 1994)

IT Use for Exploration (ρρρρc = 0.86)

Please indicate the extent to which you use specific Information Technology based support for the followingin your relationship with Alpha:

� Understanding trends in sales and customer preferences with Alpha� Integrating your functions (e.g. design and manufacturing) with Alpha�s service organization� Leveraging your firm�s expertise to create new business opportunities

Scale: Minimal Use � Some Use � Significant Use (1-7 Scale)

(Boynton and Zmud 1994)



Uncertainty (ρρρρc = 0.83)

What is the likelihood of major changes occurring in this product category over the next 12 months?

� Extensive Style Changes� Major Product Innovations� Key Manufacturing/Quality Innovations

Scale: Very Unlikely � Likely � Very Likely (1-7 Scale)

(Subramani and Venkatraman 2003)

Retailer Replaceability (ρρρρc = 0.93)

� We could easily find other customers who would offer as much supplier assistance as providedby Alpha.

� We could easily find other customers to replace the margin levels with Alpha.� We could easily substitute for the loss of reputational effects of being a Alpha supplier.

Scale: Strongly Disagree � Neither Agree nor Disagree � Strongly Agree (1-7 Scale)

(Subramani and Venkatraman 2003)

Size

What is the annual sales revenue of your firm?

Years of Association

For how many years has your firm been associated with Alpha in Canada? ______ years

Product Characteristics (7-point semantic differential scale anchored at bipolar endpoints)

Please indicate the position on the following scale that best describes this product category:

� Standard (low customization for Alpha) � Specialized (highly customized for Alpha)� Simple � Complex� Fashion oriented and style driven � Utility oriented, not style driven

(Bensaou and Venkatraman 1995)

Competitive Performance (αααα = 0.79)

� The trends in your purchases from SuppCo (retail and catalog) over the last 18 months� The trends in SuppCo�s market share in the category over the last 18 months

Scale: Significant reduction � No Change � Significantly Increase (1-7 Scale)

Note: Assessments of suppliers� competitive performance were obtained in an independent survey ofAlpha�s buying group.


16f 2 is calculated as (R2full - R2excluded)/(1 - R2full). The pseudo F statistic is calculated as f 2 * (n � k � 1), with 1,(n � k) degrees of freedom where n is the sample size and k is the number of constructs in the model.

17The standard error of the mediated path is approximated as sqrt(b2sa2 + a2sb

2 + sa2sb

2), where a and b and are themagnitudes of the paths between x, M, and y, and sa and sb are the standard deviations of a and b.


Appendix 2

Tests of Mediation

Comparing Nested Models

The PLS analysis was run for the research model (proposing that the relationship between IT use andoutcomes is completely mediated) and for a set of competing models (each of which incorporated a directlink between one type of IT use and one of the benefits). The two models are nested, and competingmodels have one path more than the research model (e.g., a direct link between ITExploit use andoperational benefits). The magnitude and significance of the difference in the R2 statistics of the twomodels (e.g., the R2 values for operational benefits are 0.238 for the research model and 0.244 for thecompeting model with a direct link between ITExploit and operational benefits) reflect the increasedexplanation of the dependent variable (operational benefits) by the inclusion of the direct link. Thesignificance of the extra path is assessed using a procedure similar to that employed to test nested modelsin stepwise linear regression. The f2 statistic is computed based on the R2 difference; the significance ofthe f2 is assessed based on a pseudo F test16 (Chin et al. 1996). The results of these tests are presentedin Table 4.

Analyses of Individual Mediated Paths

The results of the PLS analysis�the magnitudes and the variance of the direct paths among independentvariable, mediator, and dependent variable�are used to calculate the extent to which a construct mediatesthe relationship between the independent variable and the dependent variable (Hoyle and Kenny 1999).For instance, the magnitude of the mediation effect between ITExplore (variable x) and Strategic Benefits(variable y) mediated by Domain-knowledge specificity (variable M) is the product of the standardized pathsbetween x and M and between M and y. The standard deviation of the mediated path can be computedbased on the magnitudes and the variance of the paths among x, M, and y.17 The results of the analysesof paths in the model are in Table 5.

The two tests for mediation are complementary. The comparison of nested models highlights the utility ofincorporating a direct path in addition to the indirect paths between constructs. It provides no information,however, about individual mediated paths through which the indirect effect may be occurring. The analysisof individual mediated paths provides detailed information on the magnitude and significance of individualindirect paths underlying the overall indirect effect. However, it does not indicate if incorporating a directpath in addition to the indirect paths would improve explanation of the dependent variable.

Documents

OW DO UPPLIERS ENEFIT FROM NFORMATION · PDF fileSubramani/Benefits from IT Use in Supply Chain Relationships MIS Quarterly Vol. 28 No. 1/March 2004 47 and theory testing in this paper