Message Board

Collab-U

CMC Play

E-Commerce

Symposium

Net Law

InfoSpaces

Usenet

NetStudy

VEs

---

Table of Contents

• Abstract
  
• Introduction
  
• Research Questions
  
• Methods
  
• Results
  
• Discussion
  
• Acknowledgments
  
• Footnotes
  
• References
  
• Appendix A
  
• About the Authors
  

---

Abstract

One view holds that organizations are virtual to the extent that they outsource key components of their production processes, and that electronic networks make it easier to do this. The goal of the present paper is to examine explicitly the effects that use of electronic networks for transactions with suppliers have on firms' degree of virtualization. In so doing, we also highlight factors that influence the use of networks for coordination with suppliers, and the impact such use has on coordination success. Contrary to much recent speculation, the use of electronic networks for transactions was not associated with increased outsourcing, but rather greater dependence on internal production. Moreover, the use of interpersonal relationships for coordination, which many think of as an alternative to electronic network use, was positively associated with greater network use. Surprisingly, use of electronic networks was negatively associated with such outcomes as order quality and efficiency, and satisfaction with suppliers, while more reliance on personal linkages was associated with better outcomes and mitigated the negative consequences of using electronic networks.

Introduction

Much of the literature on virtual organizations rests on two assumptions. The first is that firms adopt virtual forms in order to gain benefits of acquiring goods and services from specialized producers, who are able to make these inputs more efficiently (Davidow & Malone, 1992). The second assumption is that modern computer and telecommunications networks sufficiently reduce the costs of coordination, allowing firms to achieve these production benefits without incurring the higher transaction costs traditionally associated with buying from an external supplier (Malone, Yates & Benjamin, 1987). The goal of this article is to examine the validity of these assumptions, by empirically testing some of their implications.

We had intended to start our research with case studies of virtual companies— firms that were successfully using computer and telecommunications networks to link themselves with other companies, where important design, management, and production for the firms’ products were being conducted. To this end, during the summer and fall of 1995 we conducted interviews in fourteen large companies in four industries--apparel manufacturing, pharmaceuticals, magazine publishing, and advertising--where we thought virtual organizations were common. In each firm, the informants included a senior manager in operations or manufacturing and a senior manager in charge of information systems. Eleven firms were interviewed on-site, and three were interviewed by telephone. In these interviews, informants described the production of their firm's most important product or service, their rationale for outsourcing or keeping in-house key elements of production, and the role that information technology in general and computer networks in particular played in production. Interviews in a single firm typically lasted about three hours and involved two to four informants.

However, we found few virtual companies in these industries. While many apparel firms outsourced key elements of manufacture and even design, they zealously guarded other production elements (e.g., selection of fabric). Magazine publishers used some free lance reporters and outsourced printing, but kept editorial decisions, layout, and a substantial amount of the writing in-house. These observations led us to reconsider the definition of virtual organization we had been using and to revisit the relevant transaction cost literature, which provides a rationale for the existence of the traditional firm and suggests some of the constraints on virtual organizing. We start our introduction by proposing that virtual organizing is a matter of degree.

Even in cases where the responsibility for design and production was spread across multiple firms, we were surprised at how unimportant computer-to-computer networking seemed to be in coordinating production. Telephones, fax machines, and express mail were much more prevalent than computer networks, and where networks were used, electronic mail connecting people was far more common than electronic document interchange connecting machines. When we identified interesting examples of computer networking (e.g., a news weekly sending page layout information to typesetting machines in a printing plant or graphic designers in New York and Los Angeles collaborating by using shared computer files and shared computer screens), the examples were as likely to occur within a single firm as between them. These observations led us to consider the interplay between two mechanisms for coordinating production processes, personal relationships and electronic networks, and to examine the conditions under which firms use them for coordination. We focused on these two mechanisms because of their potential importance for interfirm coordination (e.g., Granovetter, 1985; Keen & Cummins, 1993) and because much of the literature on virtual organizations assumes that the two mechanisms substitute for each other and have different effects on important business outcomes. (e.g., Kekre & Mudhopadhyay, 1992). We review the literature on electronic networks and interpersonal communication for coordination in the next section of the introduction.

Our literature review, below, leads to research questions on the conditions under which firms use electronic networks, on the effects of electronic networks on virtualization, and on the effects of coordination mechanism choices and virtualization on performance outcomes. We examined these questions through an empirical investigation of how firms in four industries use personal connections and electronic networks to support transactions with suppliers of important inputs to their production processes. The data come from a national survey of 250 firms in apparel manufacturing, pharmaceuticals, magazine publishing, and advertising. These vary in their technological sophistication and involve products that range from tangible to information-intensive. We examine both the conditions under which firms use different coordination mechanisms and the relationships between the use of these mechanisms, organization form, and business process outcomes.

Organizational Forms

Defining virtual organizations. The term "virtual organization", as used in the literature, has no consistent meaning. The term has been applied to movie production, in which personnel come together for the duration of a project; just-in-time manufacturing operations, in which subcontractors simultaneously act as a manufacturing firm's supplier and warehouse; "adhocracies", in which specialized task forces and work groups arise and disband on demand; and informal regional consortia, in which material and personnel flow through the companies in a geographic area. Although these examples do not provide a tight definition, they suggest some of the features that underlie the concept a virtual organization. First, production processes transcend the boundaries of a single firm, and as a result, are not controlled by a single organizational hierarchy. Second, and perhaps as a result, production processes are flexible, with different parties involved at different times. Third, the parties involved in the production of a single product are often geographically dispersed. And finally, given geographic dispersion, coordination is heavily dependent on telecommunications and data networks rather than physical travel, at least for the people involved.

For most firms, being virtual is a matter of degree. The production of any complex good or service requires combining various raw materials and modifying them in many stages, with each step adding value as the product wends its way towards the final consumer (Porter, 1980). At one extreme, a firm is virtual to the extent that each of these steps is performed outside the core firm’s boundaries, with the firm acting as coordinator. Some publishing operations approach this extreme, with no writing, editing, printing, and distribution done within the firm itself. But this non-value-adding structure will rarely arise in competitive business environments. Even book publishing houses typically perform manuscript selection and marketing in-house. The other extreme, the ‘traditional’, fully integrated organization--in which a single firm performs all aspects of management, production, sales, and distribution--is also unlikely to arise. In reality, most firms perform some steps internally, and make contractual and logistical arrangements to have other activities performed by one or more external supplier(s). In the case of physical production processes this involves deciding whether to make or to buy each component of the product. For services, it is often the choice between using in-house staff or external consultants, freelancers, or specialty firms. Most organizations are situated between these extremes. Rather than expect to find virtual organizations, "virtualization" of organizations is better viewed as a continuum. Firms become more virtual when a larger proportion of important production processes occur outside of traditional organizational boundaries.

When do firms become more virtual? By defining virtual organization in terms of the number and importance of cross-boundary transactions, one links this phenomenon to the substantial theoretical and empirical literature on transaction cost economics. Beginning with Coase (1937) and Williamson (1975), transaction cost theorists have focused on understanding why certain activities are kept within the boundaries of the firm while others are performed outside. Transaction cost models of organizations have proposed that firms make decisions about the location of business processes to minimize the combined cost of production and governance.

If only production costs were considered, many theorists would argue that these costs can be reduced as more production activities are performed outside the firm (i.e., greater virtualization, e.g., Malone, 1987; Davidow and Malone, 1992). By moving production outside of the core firm to other, specialized companies, the core firm gains access to more experience, makes better use of available production facilities, and capitalizes on economies of scale, all of which can lead to lowered production costs. If enough potential suppliers exist, a firm can shop around for the best combination of price, quality, or other desirable attributes. As a result, the procuring firm can take advantage of the capabilities of the most efficient producers available and also avoid being held hostage to the opportunistic behavior of any single supplier. For instance, a dress manufacturer could buy designs, fabric, buttons, and assembly services in the open market, and may find "bargains" in each area. As a result, unlike firms that follow an in-house production strategy, firms that are more virtual are expected to be more efficient, flexible, and effective due to their ability to re-form as the environment changes.

However, in making decisions about the degree of virtualization to adopt, firms must also take into account the costs of executing various transactions as well as the costs of production. Costs of governance are expected to be higher when firms purchase goods and services in the open market rather than producing them in-house (Williamson, 1975; 1985; 1996). When using external suppliers, firms incur costs as they search for appropriate partners, specify agreements, enforce contracts, and handle financial settlement. For complex components or services, firms may have difficulty specifying what they want, and suppliers incur costs when advertising the availability of their goods and services to potential customers (Malone et al, 1987). After the most appropriate supplier has been identified, governance processes, such as arriving at and enforcing a contractual agreement, monitoring and controlling quality, processing orders, and settling payment, all create transaction costs.

Moreover, firms have bounded rationality, and cannot know all contingencies that might arise during a transaction. Hence when the goods or services that a firm needs change from order to order or when a firm doesn't know what it will need until the last minute, firms are more likely to choose traditional integrated organizational forms. These conditions are further exacerbated in cases where only a small number of suppliers exist that can satisfy a firm's needs, increasing the potential for opportunism.. The small numbers problem is more likely to arise when business processes require highly specific inputs, and as a result, do not have a large number of alternative suppliers with which to do business.

A traditional, integrated organization has the advantage of lowering the costs of certain forms of governance. Routine activities are formalized in standard operating procedures, which are known throughout the organization. Formalized communication paths and assigned roles reduce the need to search and negotiate agreements. Employment agreements and hierarchical management free firms from the need to specify future contingencies in contracts. Thus, Williamson (1975) argues that traditional organization forms exist, despite production inefficiencies, because they provide lower-cost mechanisms for governing the execution of complex business processes.

In summary, the transaction cost literature proposes that features of the goods or services being procured and of the potential suppliers directly influence whether a firm makes the good or service in-house or outsources it, thus virtualizing its production process. These features include: (a) features that make a firm vulnerable to opportunism, including the relative power of potential suppliers vis-a-vis the focal firm and the uniqueness of the object being acquired, and (b) the complexity of the transaction, including the difficulty of specifying the object to be acquired, uncertainty because the object itself changes from order to order and unpredictability in the quantity and quality of objects a firm will need in a particular order.

Coordination Mechanisms

Electronic networks and organizational forms. Many distinct mechanisms can be applied to the governance and coordination of business transactions between a firm and its suppliers, including formal legal contracts, information technologies, and informal personal relationships. The focus in transaction cost theory on opportunism and contract-based mechanisms to guard against opportunism has led to an emphasis on governance processes, such as bargaining, negotiation, and enforcement (Williamson, 1985; 1991).

Throughout this article, we use the term "electronic network" broadly, to designate any type of computer or data network that allows companies to exchange information between computers. The information that flows over these networks varies, and can include cases where one computer directly controls other machines involved in production (e.g., a page-layout computer in New York that directly controls type setting machines in printing plants across the country); electronic document interchanges or funds transfers, delivering invoice or payment information; or electronic mail networks, where people exchange messages via computers. We use the term to apply both to data networks deployed within a single firm and inter-organizational networks, deployed between firms.

In a widely cited article, Malone, Yates and Benjamin (1987) argued that firms can use electronic networks to reduce governance costs and thus moderate the effects of specificity and complexity that normally lead to market failures. The argument is that if electronic networks were used to expand the pool of potential suppliers inexpensively--what Malone et al (1987) refer to as electronic brokerage effects-- they would reduce firms’ vulnerability to opportunism. With more potential suppliers, firms are less subject to threats of opportunism from any particular one. Electronic networks could also lead to less vulnerability to opportunism if firms used them to monitor a supplier's internal production processes inexpensively, as has happened among organizations that use electronic silicon foundries (e.g., Hart & Estrin, 1991).

When firms execute complex business activities, they must do more than guard against opportunism. They must also solve significant logistical and communication problems, i.e., problems of coordination. Malone and colleagues (1987) argue that because electronic networks reduce the costs of coordinating business processes between firms, these networks will cause firms to conduct more transactions across organizational boundaries and thus become more virtual. Their argument is that anything that lowers the cost of inter-firm coordination allows firms to exploit the lower production costs they can achieve through outsourcing. A number of authors have followed Malone et al. in hypothesizing that by lowering coordination costs, electronic networks enable virtual organizing, leading to smaller firms that outsource more elements of the production process (cf. Bradley, Hausman & Nolan, 1993; Clemons, 1993; Miller, Clemons & Row, 1993). These arguments about both governance and coordination costs lead to the prediction that the use of electronic networks will allow firms to outsource more and thus cause them to become more virtual.

Except for a series of widely cited case studies (e.g. Malone & Rockart, 1993), there are few empirical tests of the hypothesis that greater use of inter-organizational networks leads to greater outsourcing, and most of the data speak to the issue only indirectly. Consistent with this hypothesis is evidence at the industry level that increases in investment in information technology are associated with declines in average firm size and increases in the number of firms (Brynjolffson, Malone, Gurbaxani, & Kambil, 1994). Kambil (1991) shows that industries investing more of their capital stock in information technology also contract out more of the value of the goods and services they produce to external suppliers (i.e., a higher buy/make ratio in production), with a two-year lag. Limitations in the data, however, mean that the analyses are only suggestive. Most importantly, the prior studies use of investment in general information technology as the independent variable, obscures the unique role that inter-organizational computer and communication networks are hypothesized to play.

Two assumptions underlie much of the argument that electronic networks increase virtualization. The first is that governance and coordination are more difficult between firms than within a single firm. The second is that the use of electronic networks decreases this gap, i.e., that computer networks reduce the costs of between-firm governance and coordination more than they reduce the costs within a single firm. The second of these assumptions, however, is probably wrong. We believe that much of the discussion about the effects of electronic networks underestimates benefits that can accrue to firms adopting computer and communication networks for internal use. In deploying networks internally, firms have greater knowledge of the business processes that could benefit and greater control over the implementation process. Historically, firms first adopt networks for use within the firm. When firms deploy networks internally, they are less vulnerable to intrusion, can enforce technical standards to ensure compatibility, and can control many other factors that can lower the cost of implementation. For example, they have a greater likelihood of enforcing standards so that computers deployed at different points in the production process can communicate with each other. If firms achieve greater net governance and coordination benefits when they deploy within-firm networks than when they deploy inter-organizational networks, then network use should lead to greater internalization of production, not outsourcing. One of the goals of our research is to determine whether greater use of electronic networks is associated with more or less virtualization.

Personal relationships. Although the literature on virtual organizations is dominated by discussion of organizational form and information technology, in many cases personal relationships, not ownership or technology, may be the key mechanism for coordinating complex business processes. Sociologists like Granovetter (1985) have argued that many economic exchanges are mediated by the social relationships among the trading partners, and at times these social relationships can be more efficient than market mechanisms for conducting transactions (c.f., Lawrence & Lorsch, 1967; Tushman & Nadler, 1978). Recent work by Uzzi (1997) provides a detailed analysis of the ways in which coordination via personal relationships contrasts with market or arms-length mechanisms for coordination. People generally rely on personal relationships to resolve problems and deal with unusual situations (Krackhardt, 1992). Personal relationships also serve as a valuable governance mechanism. For example, personal relationships can lead to trust between parties involved in an economic exchange, which in turn reduces the likelihood of opportunistic behavior (Granovetter, 1985; Uzzi, 1997; Zucker, 1986).

The exploratory interviews we conducted in starting our research highlighted the importance of personal relationships between buyers and suppliers. Personal relationships were frequently called upon when firms were dealing with exceptions to routine, such as when an advertising firm needed to deliver an updated ad copy after a magazine's publication deadline or when a supplier needed to provide substitute fabric to a suit manufacturer. Even for more routine tasks, such as in trying to identify a new supplier, purchasers typically considered contacts with people in other firms to be more useful than structured information resources like directories and databases. For example, when a manager in a consumer pharmaceuticals firm needed to find a supplier for an unusual plastic container, he spent most of the search time on the telephone getting referrals from current suppliers and professional colleagues.

Although this discussion implies that personal relationships will be an important mechanism for inter-firm coordination, it is unclear how coordination based on personal relationships and that based on information technology will interact. Studies of electronic networks often imply that information technology and interpersonal relationships compete and that firms derive more benefit from using information technology to coordinate when it replaces person-to-person contact. Yet, most research that finds substitution of electronic coordination for personal coordination focuses on the coordination of routine activities (e.g., Kekre & Mukhopadhyay, 1992). It is likely that personal relationships are most valuable when dealing with non-routine transactions. For example, while electronic data interchange systems may be very effective for arranging routine orders for standardized products, they may be less suited for supporting the negotiation surrounding the acquisition of new goods or services, or when dealing with unusual situations which fall outside of standard procedure. Chan (1997), for example, found that electronic networks are perceived to be more important for routine activities (such as ordering or getting price information) than non-routine interactions with suppliers, such as contract negotiations and after-sale problem-solving.

Because coordination using electronic networks may be more fragile than coordination based on personal contacts, it may require interpersonal backup to work successfully (e.g., Suchman & Wynn, 1984). Kling and Scacchi’s (1982) description of a social web of computing was an early recognition of the importance of social relationships in supporting successful computer use. More recent research (e.g., Hart & Estrin, 1991; Saunders & Hart, 1993) has also stressed how the introduction and effective use of electronic networks for electronic integration requires prior personal acquaintanceship and trust among the parties that will be sharing data.

Effects of coordination mechanisms on coordination success. Empirical research on organizational use of inter-organizational electronic networks finds that they positively influence business process outcomes (Kekre & Mukhopadhyay, 1992; Streeter et al., 1996). It is not clear, however, whether these networks are more useful for supporting virtualization or for in-house production. Similarly, it is not clear how the use of electronic networks compares to the use of personal relationships or in-house production. Coordination through personal relationships may be less efficient, because it relies on costly and error-prone human behavior. On the other hand, if personal relationships lead to trust, thus allowing firms to avoid the need for costly monitoring efforts, or if it leads to the exchange of favors, then coordination through personal relationships may be associated with successful coordination outcomes. For example, Steinfield, Caby & Jaeger (1996) found that an electronic network created to improve the efficiency of selling advertising-time in the French television industry failed because it was perceived as decreasing the quality of outcomes. The network reduced sales representatives’ flexibility, preventing them from using their existing personal relationships to offer better times and rates to preferred and high volume customers. They thus began to bypass the system, leading to eventual failure of the electronic market.

Research questions

We can summarize this discussion with several research questions. Our major goals in this research are to understand how the use of electronic networks influences virtualization and how the use of electronic networks and interpersonal relationships for coordination changes production outcomes. However, to examine the influence of electronic networks, it is necessary to control for other factors that could make electronic networks useful.

RQ1: Under what conditions will firms use electronic networks?

Transaction attributes. First, we expected that when products or services (such as software or information) themselves could be transportable over electronic networks, the cost advantages of using a network would be greater. Second, we expected that the speed advantages of networks would be most beneficial when firms were under greater time pressures, or they could not predict the timing of their needs. Third, we expected that when objects were easier to describe, it would be easier to order over an electronic network. Finally, reductions in labor costs through automation of transactions would motivate firms to employ electronic networks, but only if there were a sufficient volume of transactions for this payoff to be meaningful given the costs associated with network implementation (Keen & Cummins, 1993). Thus, we expected that firms would use electronic networks more when they:

a) were under greater time pressure,
b) had greater unpredictability about their inputs,
c) placed large numbers of orders per year,
d) were acquiring intangible inputs, and
e) were acquiring inputs that were easy to describe.

Personal relationships. While predicting the influence of some production variables, like time pressure, on use of electronic networks is straightforward, predicting the influence of personal relationships is more difficult. Our review of theory suggests that electronic networks can either be used to replace personal relationships as a mechanism for coordination between a firm and its suppliers or can supplement it.

RQ2: What effects do electronic networks have on firms' degree of virtualization (i.e., outsourcing of key components of production)?

Improved coordination. The conventional view among information systems researchers is that inter-organizational computer networks enable efficient outsourcing. While use of electronic networks can facilitate coordination both within firms and between them, many information systems researchers believe that in the long run these networks will lead to greater outsourcing (e.g., Malone et al, 1987). Thus one might expect a positive association between use of electronic networks and virtualization.

Vulnerability to opportunism. Malone and his colleagues reason that inter-organizational computer networks will lead to greater virtualization in part because they can reduce potential opportunism when outsourcing. If this is correct, one would expect interactions between the degree of electronic network use and the effects of the opportunism variables (object specificity, variability, unpredictability, and complexity of product description) on outsourcing. That is, firms will outsource less when procuring highly specific products (or other products subject to opportunism), but this relationship will be reduced among firms that use electronic networks.

However, if the benefits of using electronic networks inside the firm exceed the benefits of using networks to connect to external suppliers, or if the use of electronic linkages to external suppliers follows use within the firm, one might expect to see a negative association of electronic network use with virtualization and no interactions between electronic network use and the opportunism variables.

RQ3: What effects does the mode of coordination have on the outcomes of transactions with suppliers?

Electronic network use and coordination success: Firms are motivated to use electronic networks for transactions because of putative gains in efficiency. Empirical evidence on this topic is limited, but, based on the information systems and transaction cost literature we expected to find that greater use of electronic networks would be associated with fewer errors in orders, greater efficiency in orders, and more satisfaction with the suppliers with which they are used.

Personal relationships use and coordination success: We expected that use of personal relationships would influence the success firms have in coordination with suppliers (errors in orders, efficiency in orders, and satisfaction with the suppliers), but made no prediction on the direction of the effect. The direction should depend on whether the flexibility associated with personal relationships is worth the cost of providing it.

Interactions between use of electronic networks and personal relationships on coordination success: A negative interaction between the use of electronic networks and the use of personal contacts on the outcome measures would suggest that electronic networks are more effective when they substitute for and replace human contact between the supplier and purchasing firm. On the other hand, a positive interaction would suggest that electronic networks are most effective when they are supplemented by personal contacts.

Virtualization and coordination success: To the extent that use of in-house production improves coordination, it should have positive effects on coordination success.

Interactions between virtualization and use of electronic networks on coordination success: If electronic networks are more beneficial when coordinating with external suppliers than when improving in-house production, as the literature on virtualization organizations suggests, then we would expect to see an interaction between electronic network use and degree of outsourcing on coordination success.

Methods

Sample and interview

This research is based on a telephone survey of managers in a sample of 250 firms from the advertising, magazine publishing, women’s apparel, and pharmaceuticals industries. Firms with at least 20 employees in these industries listed in the April, 1996 Dunn and Bradstreet data base were ranked by number of employees, and a stratified sample of large, medium, and small firms (relative to each industry) was selected from each third of the size distribution. The goal was to sample several industries with differing product and production approaches, to look for effects that generalize across industries. Advertising and publishing were chosen as producers of information products. Apparel and pharmaceuticals were selected as examples of manufacturing industries. In the analyses below, we control for industry by including it as a dummy variable. However, because the industries differ on many unmeasured variables, we do not delve further in trying to explain industry differences. Table 1 shows the average number of employees in each industry-size category and the number of firms sampled in each category.

The telephone survey focused on the way in which the producing firm acquired key inputs from suppliers. Respondents were the senior manager most responsible for acquiring the input under consideration. They were asked to identify their company's principal product or service, and then were asked about a specific input needed to produce that product. Depending on the question, respondents were asked about the way they acquired the input in general or about the way they dealt with the "the most important supplier you've worked with within the past 12 months." We stressed that the suppler could be an employee or department in the respondent's firm or an outside individual or company.

Industry SIZE	Pharmaceuticals	Advertising	Women’s Apparel	Magazine Publishing
Small number of employees	44	24	56	27
(Number of Firms)	(18)	(22)	(24)	(20)
Medium number of employees	84	38	146	35
(Number of Firms)	(21)	(21)	(14)	(23)
Large number of employees	13,262	508	261	759
(Number of Firms)	(20)	(21)	(22)	(24)
Total N = 250 firms	(59)	(64)	(60)	(67)

Table 1. Sample Characteristics

Note: Entries represent mean number of employees in the firms and the number of firms presented in industry-size category.

We identified the set of key inputs for each industry through the exploratory, semi-structured interviews, which we have previously described. Inputs were chosen to vary on whether they were tangible, requiring physical transport, or intangible and suitable for transport over an electronic network. Table 2 shows the four transactions for each industry, and their classification as tangible or intangible.

Respondents were questioned about one input randomly chosen from a list of four for the appropriate industry. If a particular company did not use this input or if the quota for the input had already been filled in the industry-size category, then the interviewer randomly selected another input.

The telephone survey was conducted by a professional survey firm. Phone interviews, typically lasting 30 minutes, took place in July and August, 1996. In reaching the respondent, interviewers used a series of screening questions to identify "the person who is most responsible for arranging and acquiring" the specified input. Interviewees included both managers specializing in procurement (e.g., a VP of Purchasing) as well as those in operational areas (e.g., the senior editor responsible for assigning stories to staffers and freelancers). The survey process produced 250 completed interviews on a survey instrument with 102 items. Excluding firms that no longer existed, that could not be contacted after seven tries, or that failed to meet size or industry definitions, the response rate was slightly over 50 percent.

Measurement

To address the research questions described previously, we created multi-item scales to measure concepts related to uncertainty, opportunism, object specificity, coordination mechanisms, extent of outsourcing and coordination outcomes. Individual items were first examined using principle components factor analyses. Items that did not load as expected, or cross-loaded, were removed and factor analyses rerun. Multi-item scales were only created using items that exhibited both discriminant and convergent validity. Scale items were included if they were: 1) included in the survey to measure a single concept; 2) loaded at .5 or higher on the same factor; and 3) did not cross load (at .4 or higher) on any other factor. When our a priori scale items did not load together, but when there was sufficient theoretical justification for including a construct, single-item measures were used in our data analyses. All multi-item scales were subjected to reliability analyses. (Appendix A lists all the multi-item scales and their reliabilities.) Reliabilities as measured by Cronbach’s alpha were between .65 and .90, sufficient for exploratory analyses.

The six sets of variables included in our analyses are described below:

Control variables

In all analyses we included variables for industry, size of focal firm and size of supplier. We wished to hold these variables constant, because they are likely to have wide ranging effects on how firms organize production and the degree to which firms and their suppliers use technology. Dummy variables were used to represent each of the four industries. The size of focal firm is the average of the number of employees in the firm and the firm's annuals sales, first standardized and then logged (alpha = .91). The data were taken from Dunn and Bradstreet's 1996 listing for the producing firm. Supplier size is the average of the respondent's estimates of the number of employees the supplier has and its market share, rated on a five-point scale. Both were first standardized before averaging (alpha=.73). In the analyses reported below, industry effects are considered as controls, since there is no a priori reason why they should differ in terms of the processes underlying the development of organizational forms, network use and coordination outcomes.

Product and production attributes influencing the potential utility of electronic networks

In order to test the effects of electronic network use on virtualization, we must control for important features of products and production that enable or encourage firms to use networks. We reasoned that firms would get the most coordination benefit from using networks when the goods they were acquiring could be transported over the network, when they were under time pressure, and when they orderd frequently.

Tangibility. An input's tangibility refers to the ability to execute the transaction entirely through an electronic network with currently existing technology. If the completion of the transaction required the transfer of a physical object (e.g., a bolt of fabric) then the transaction was categorized as tangible. If, on the other hand, it was technologically feasible to completely eliminate the exchange of material objects and still complete the transaction (e.g., a garment design, which could be exchanged as a computer file or faxed message), then the input was considered intangible, regardless of whether a particular firm actually used electronic networks. Table 2 shows the classification of inputs.

Industry	Input
	Tangible	Intangible
Women’s Apparel	Fabric Trim (buttons, zippers, etc.) Cutting Services	Garment Design
Advertising		Artwork & Graphics Television Time Slots Television Ad Distribution Services Market Research Services
Pharmaceuticals	Chemical Ingredients Packaging Materials Packaging Assembly Services	Clinical Trial Management Services
Magazine Publishing	Paper Stock Printing Services	Stories Color Work Services

Table 2. Key inputs studied in each of four industries

Note: Women's apparel manufacturers came from the following SIC codes: 2331, 2335, 2337, 2339, 2342, 2384 and 2389). Magazine publishers' SIC code was 2721. Pharmaceutical manufacturers SIC code was 2833 and 2834. Advertising agencies were from SIC code 1591.

Time pressure is the respondent's judgment of the extent to which production in the firm was rushed and subject to rigid deadlines (e.g., "To meet schedules related to this product or service, we need to use every available minute efficiently."). (Two items; alpha = .68).

Order frequency is a single item measuring the ordering cycle for the key input, from yearly to monthly.

Product and production attributes influencing vulnerability to opportunism¹

Transaction cost theory proposes that firms will be less likely to outsource production if they are vulnerable to exploitation by external suppliers. At a conceptual level, the key variables are specificity of the required assets and complexity of the inputs sought. We operationalized asset specificity as the degree to which the input was generic to the industry or unique to the acquiring firm (generic versus specific object). We operationalized complexity with three variables: change versus stability in the input acquired (object certainty), unpredictability versus predictability in the quality and quantity of input needed (object predictability), and the ease of identifying the input (ease of description).

Generic versus firm-specific object is the degree to which the key input itself is usable by any producing firm, rather than being custom made for the respondent’s firm. At one extreme are inputs like newsprint or shirt buttons, which are used by many manufacturers, while at the other are news stories or specialized fabrics, which are specific to a particular manufacturer. This dimension was measured by the extent to which respondents agreed on a 5-point Likert scale with four statements, (e.g., "My firm is the only one that uses the [input], " (reversed), and "The [input] we get from this supplier is fairly standard for the industry"). (alpha=.64).

Object certainty is the extent to which the item being ordered, its availability, and its price are stable over time. (e.g., "How much does the [input] your company needs change from order to order?" (reversed) (Three items; alpha=.68).

Object predictability. Predictability is the extent to which respondents know in advance the specific input they would need and its quantity. (For a typical order, how far in advance can you predict what [input] your company will need?"). (Two items; alpha=.79).

Ease of description. Respondents were asked to rate how easily the key input could be described so that it could be ordered ("It is difficult to describe the [input] we routinely acquire (reversed)".

Coordination mechanisms

Electronic networks. The survey asked about the importance of using electronic networks to acquire the input from a major supplier: the importance of using "any type of computer communication that allows you to exchange information with this supplier" in the acquisition process. Measured on a 5-point scale, we asked about the importance of using electronic links for each of six separable stages in the acquisition process (Johnston & Vitale, 1988; Kambil, 1993), including: 1) searching for and selecting a supplier; 2) developing the specifications of the key input; 3) negotiating the terms of the acquisition such as price, delivery date, and so on; 4) ordering the input; 5) monitoring the quality of the good or service; and 6) fixing problems after the order. Respondents made their judgments on 5-point Likert scales, from not at all important to very important. The composite scale has high reliability. (alpha = .94).

Personal relationships. We measured the importance of personal relationships in acquiring the key input for each of the six stages of acquisition. Personal relations were defined as including "any type of personal connection or personal knowledge between people in your firm and people in the supplier's organization." Respondents made their judgments on a 5-point Likert scale of importance. (alpha = .85).

Organizational Form

Outsourced production is a composite index that estimates the extent to which a firm outsourced the key input or produced it in-house. It is based on three items: (1) whether the major supplier for the input was external or in-house; (2) the nature of the relation between the major supplier and the producer firm, ranging from "departments within your firm" to "owned subsidiary" to "a joint venture firm" to an "outsider" firm; and (3) the percentage of the key input which was produced in-house, ranging from 0 to 100. If the index is high, then the firm outsources more of its production and is more of a virtual organization. (alpha=.86).

Outcome variables

Overall order quality is a six-item composite index consisting of two subscales--the extent to which orders were free from error and the extent to which they were executed efficiently. The error items included the percentage of orders arriving after the agreed upon delivery date, the percentage of orders containing any sort of error, and the percentage of orders that failed to meet a firm's quality standards. The efficiency items included the average elapsed time between ordering and receiving the input, the number of different people who are involved in handling the order, and the number of communications with the supplier during a typical order cycle. Because these items were non-normally distributed, estimates were logged and then standardized before averaging. (6 items; alpha=.65).

Satisfaction with a supplier is an index in which respondents indicated their overall satisfaction with the supplier, their satisfaction with six aspects of the ordering process, and their satisfaction with the supplier's ability to handle exceptional orders. (9 items; alpha=.89).

Missing data

Most variables had only a few missing data values. In the regression models that follow, we imputed missing values by replacing them with the industry mean. When imputing firm size we replaced the missing value with the average number of employees in the other firms in the appropriate size-industry category. Since attributes of the suppliers varied with the input being acquired, we imputed supplier size from the average of the non-missing values from firms providing the same input (e.g., from other fabric suppliers).

Analysis

Our research questions were investigated through a series of regression analyses. First, in order to examine research question 1, which asked about the conditions under which firms acquire inputs from suppliers through electronic networks, the variables that indicated a need or an opportunity for coordination efficiencies (such as order frequency, time pressures, tangibility) were entered into a regression equation predicting the importance of electronic network use (Model 1: Predicting Network Use). We entered personal relationships into this equation to examine whether these are complementary or contradictory modes of coordination. Control variables as well as other independent variables reflecting vulnerability to opportunism were also entered into this equation. We next examined research question 2.1, which investigates the effects of electronic linkages on organizational form, by predicting the extent of outsourcing from the importance of electronic network use, controlling for the full set of control and independent variables (Model 2: Predicting Outsourcing (all suppliers). A more conservative test of research question 2.1 is also provided in the results section, by examining these relationships only for that subset of the sample of firms that used external suppliers as their primary source of supply (Model 3: Predicting outsourcing (external suppliers). As part of our analysis of research question 2, we explored the argument that electronic networks permitted greater outsourcing by moderating the factors that cause vulnerability to opportunism (research question 2.2). We did this by testing the interaction of network use with the opportunism variables (Model 4: Networks moderating the effects of opportunism). Our final set of analyses, based on research question 3, asked about the effects of coordination mode on transaction outcomes. To examine this, we explored whether electronic networks and personal relationships predicted perceived order quality (Model 5: Predicting order quality) and satisfaction with supplier (Model 6: Predicting satisfaction), controlling for all other predictor variables and the extent of outsourced production. We included the interaction of electronic networks with personal relationships to determine if the use of these two coordination mechanisms substitute for or complement each other. We included the interaction of electronic networks with outsourced production, to determine if outsourcing was more beneficial when used for in-house coordination or for coordination with outside suppliers.

 

Results

Research Question 1: Predicting use of electronic networks

Model 1 in Table 3 shows standardized beta coefficients from a regression analysis predicting use of electronic networks from control variables, production efficiency variables, and opportunism variables. As expected, firms were less likely to use electronic networks when they were working with tangible goods (ß = -.26, p< .002). However, the relationships between use of networks and greater time pressure (ß = .10, p < .11) and order frequency (p < .40) were not significant.

With the exception of ease of input description, we made no predictions about whether vulnerability to opportunism would be associated with the use of electronic networks. Contrary to the expectations of Malone and colleagues, firms use networks more when they were acquiring items that were more difficult to describe (ß = -.23, p < .001). Networks were also used more to acquire inputs that were more tailored to their firm rather than generic (ß = -.16, p < .02). Both findings suggest that firms use electronic networks when most exposed to potential opportunism from suppliers. The findings do not, however, mean that the networks were being used to outsource more under these conditions. Rather, as we show in later analyses, firms seem to use networks when they are vulnerable to opportunism to support in-house production.

Finally, our data clearly support the contention that personal relationships and electronic networks are used as complementary rather than substitutable coordination mechanisms. Even controlling for other factors, firms that use personal relationships to coordinate with a supplier were also more likely to use electronic networks (ß = .16, p < .01).

Note. Entries are standardized beta coefficients from a multiple regression analysis.
† : p < 0.10, *: p < 0.05, ** : p < 0.01, ***: p < 0.001

Table 3. Predicting use of electronic networks and outsourcing

 

Research Question 2: Does the use of electronic networks predict virtualization?

Model 2 in Table 3 shows results of a regression analysis predicting virtualization (i.e., outsourcing of a key production input) with control, production efficiency, and opportunism variables.

Transaction cost predictors of virtualization. Transaction cost theory posits that organizations outsource most when inputs are more generic, stable, predictable, and easier to describe (and thereby find). Results only weakly supported these predictions. Firms were more likely to outsource more generic inputs (ß = .10, p < .10). In addition, they outsourced tangible inputs more than intangible ones (ß = .31, p < .001). Specifically, the inputs of fabric, trim, chemical raw materials, packaging materials, paper, and the time slots for advertisements² were more likely than average to be acquired externally. Garment designs, cutting services, artwork and graphics, stories, color work, television ad distribution, and packaging assembly were more likely to be accomplished in-house. The findings are consistent with transaction cost theory predictions about asset specificity. That is, highly customizable inputs, most of which were intangible (e.g., garment designs and stories) were more likely to be produced in-house. More commodity-like inputs or those that could be supplied to multiple producers (e.g., fabric and paper) were more likely to be outsourced. While transaction cost predictions about asset specificity were supported, other predictions derived from transaction cost theory were not. In particular, order certainty, order predictability, and ease of order description were not associated with outsourced production.

Firms were more likely to outsource when suppliers were larger (ß = .19, p < .01). However, this seems to be an artifact created by the roughly one quarter of the sample for which an in-house division was the major supplier. Presumably, when a firm does in-house production, the in-house division is small relative to other firms in the industry. The relationship between in-house production and supplier size disappears when we include in the analysis only the subset of the sample whose major supplier is an external organization (see Model 3 in Table 3).

Model 2 in Table 3 shows that the more firms use electronic networks, the more they acquired needed inputs from internal sources rather than external ones (ß = -.28, p < .001). This is inconsistent with the hypothesis that network usage leads to greater outsourcing.

To determine whether this result was an artifact of the minority of firms whose main supplier was a division of their own company, we redid the analysis using the subset of the sample whose major supplier was a separate company (see Model 3 in Table 3). Here again, the results show that the more heavily firms used electronic networks to coordinate with their main external supplier, the less likely they were to outsource production (ß = -.07, p < .01). This finding again suggests that the more firms relied on electronic networks to coordinate the acquisition of important inputs, the more integrated they were with their suppliers.

Figure 1. Influence of use of electronic networks and object specificity on degree of outsourcing.

Note. Plot from Model 4 in Table 3, showing the influence of increasing the degree of network use and object specificity by one standard deviation on extent of outsourcing (in standard deviation units). Scores are averaged across the four industries, with all other variables set to their mean levels.

Vulnerability from opportunism. We also tested to see if network usage influenced virtualization by moderating the effects of vulnerability to opportunism. The "protection from vulnerability" hypothesis holds that by using electronic networks, firms can reduce their propensity to produce in-house when they are vulnerable to opportunism, because the networks help them lower the costs of identifying and acquiring firm-specific assets in the market and monitoring transactions involving them. Model 4 (Table 3) shows the results of the relevant analysis, predicting the degree of outsourced production from the interaction of electronic networks with each of the independent variables intended to measure vulnerability to opportunism (i.e., ease of input description, object certainty, order predictability, and generic object). The analysis used the full sample of firms.

Only the interaction between electronic network use and object specificity reached significance (ß = -.15, p < .02). Figure 1 illustrates the interaction. It shows that when firms fail to use electronic networks, they produced firm-specific inputs in-house and outsourced the generic ones, as transaction cost theory would predict. When they used electronic networks, this gap in outsourcing was reduced. However, the reduction was not consistent with the hypothesis that electronic networks protect firms from opportunism. Firms did not use networks to outsource firm-specific assets. Instead, their use was associated with more in-house production overall, and especially for the generic objects they had previously outsourced.

Research Question 3: Effects of Coordination Mode on Transaction Outcomes

Table 4 shows the association of the coordination mechanisms with respondents' judgment of the outcomes of ordering from their primary supplier, controlling for a number of other factors that might lead to use of the coordination mechanisms or directly change business process outcomes. Model 5 in Table 4 shows the analysis for order quality--efficiency of performing orders and their freedom from various types of errors. Model 6 shows the analysis for respondents' satisfaction with their supplier. As can be observed, the significant predictors of the two outcome variables are different.

Table 4. Predicting coordination success

Note. Entries are standardized beta coefficients from a multiple regression analysis.
† : p < 0.10, *: p < 0.05, ** : p < 0.01, ***: p < 0.001

Order quality. On the overall order quality outcome variable, large firms performed more poorly (ß = -.19, p < .05), especially on items measuring the efficiency of ordering. When firms were acquiring inputs that were more stable and did not change from order to order, (e.g., trim in the apparel industry or printing services in magazine publishing), working with the supplier was both more efficient and error-free (ß = .13, p < .05). In contrast, firms had more trouble when orders were predictable far in advance (ß = -.16, p < .05) (e.g., fabric in the apparel industry). It is unclear whether this occurs because time lags lead to problems or because firms allow more lead time when they anticipate problems.

Surprisingly, the more firms used electronic networks for coordination with their main supplier, the poorer was the overall quality of the ordering process with that supplier (ß = -.17, p < .05)³. In contrast, the more they used personal relationships, the better the outcomes were (ß = .15, p < .05). Of particular importance is the finding that these two modes of coordination had a statistically significant interaction on overall order quality (ß = .17, p < .01). Their interaction, plotted in Figure 2, shows that using electronic networks actually degraded the overall quality of the order process when firms failed to supplement network use with personal relationships. However, when personal relationships were also used for coordination, the negative effect of electronic networks was reduced.

Figure 2. Influence of use of electronic networks and personal relationships on order quality.

Note. Plot from Model 5 in Table 4, showing the influence on order quality (in standard deviation units) of increasing the degree of network use and use of personal relationships by one standard deviation. Scores are averaged across the four industries, with all other variables set to their mean levels.

Supplier satisfaction. In terms of the other outcome measure, satisfaction with supplier, respondents were more satisfied with their primary supplier when the supplier was smaller (ß = -.15, p < .05), and if they acquired inputs from them more frequently (ß = .11, p < .10). They were more satisfied with suppliers who delivered orders more efficiently and with fewer errors (ß = .37, p<001). Holding constant the overall quality of the ordering processes, the use of personal contacts was again associated with better outcomes (ß = .20, p < .001). Respondents were more satisfied with the supplier the more they used personal relationships as a coordination mechanism, while use of electronic networks was unrelated to respondents' satisfaction with suppliers.

Discussion

This article focuses on the choices that firms make when coordinating production: outsourcing key components or producing them in-house and coordinating with their suppliers through personal relationships and electronic networks. Much of the literature on virtual organizations has argued that (a) electronic networks substitute for personal relationships in coordinating production, (b) electronic data networks are a prerequisite for virtualization of production and that the availability of electronic networks may lead firms to outsource more components of production than they would have otherwise, and (c) use of these networks generally has a positive effect on coordinating production. Our results, summarized in Figure 3,⁴ challenge all three assumptions.

Figure 3: Relationships among product, coordination and outcome variables

First, our data provide evidence that the use of personal relationships and electronic networks are complementary methods of coordination with suppliers rather than competing mechanisms. Firms use personal relationships and electronic networks concurrently to coordinate. The same firms that report using electronic networks heavily also report heavily using personal relationships for coordination. In multivariate regressions, controlling for firm and product characteristics, the existence of personal relationships between a focal firm and a potential supplier is a predictor of their use of electronic networks to coordinate production.

Our regression results further suggest that firms were more likely to use networks to acquire complex, specific, and intangible inputs. Firms appear to be using networks as a means of reducing their vulnerability to opportunism, although not, as discussed below, by coupling network use with greater outsourcing, but by using it to support internal production.

Indeed, our second set of findings are inconsistent with the hypothesis that increased use of electronic networks as a coordination mechanism leads to greater outsourcing of production. Instead we find that the more firms use electronic networks with an external supplier the more they produce their key inputs in-house. That is, increased use of electronic networks is associated with less virtualization and with less use of the market to acquire key production components. This finding is consistent with the growing research literature showing that inter-organizational networks are associated more with hierarchical relations than market-based ones (Brousseau, 1990; Hart & Estrin, 1991; Keen, 1988; Malone & Rockart, 1993; Steinfield, Kraut & Plummer, 1995; Streeter, Kraut, Lucas & Caby, 1996).

We are making no claim here that the use of networks leads to in-house production. Indeed, the most plausible interpretation of the findings is that the control and trust that results from ownership encourages firms to invest in greater internal network use. This account of the evolution of firms' use of networks is consistent with trade press reports that firms are using intranets rather than placing strategic company business on the public Internet; if this is indeed a common occurrence, then this use of electronic networks illustrates that firms desire a great degree of control and trust before they are willing to make such investments or share sensitive data. To tease apart causal direction will require longitudinal data.

Third, we were surprised to see that greater use of electronic networks was associated with poorer outcomes when working with suppliers. This finding runs counter to the expectation that the use of information technology is associated with increased efficiency and quality of inter-organizational transactions (Davidow & Malone, 1992; Malone & Rockart, 1993) and with several quantitative case studies that show strong positive effects of using EDI and other inter-organizational computer networks (e.g., Kekre & Mudhopadhyay, 1992). It is likely, however, that the case studies have overstated the positive impact of using networks, by focusing on large firms dealing with high volumes of very routinized transactions. The positive impacts found in these contexts may not generalize to a wider and more heterogeneous sample. The data are consistent with analyses challenging the value of information technology in improving business productivity (e.g., Attewell, 1995; Landauer, 1996).

It is possible that the relationships we observed in this study between use of electronic networks and errors and inefficiencies in ordering may be transient. Many firms have had little experience using electronic networks with suppliers and are relying on proprietary and ad hoc applications to share data. Because of lack of standards, inexperience and problems with security, applications that allow data sharing with one supplier do not necessarily allow sharing with others. For example, an international advertising firm that shares files with other offices in their own firm over a proprietary data network were unwilling to send the files to outsiders electronically. They are worried both about the security of their corporate network and about compatibility standards with outsiders. Thus, they use courier services to get removable digital media to clients and color separation services they work with.

Another reason that network use was associated with poorer outcomes such as late deliveries may, paradoxically, be a result of their speed. The instantaneous communications networks allow may encourage firms to wait until the latest possible moment before shipping, increasing the chances of late delivery and last minute errors. In our interviews in the advertising and magazine publishing industries, precisely this behavior was described by managers. Magazine publishers noted, for example, that one effect of networks was that it allowed advertisers to transmit their copy much closer to scheduled publication. This just-in-time effect increased the likelihood of missed deadlines and mistakes.

In contrast to the poor outcomes associated with use of electronic networks for coordination, greater use of personal relationships was associated with better quality and efficiency in executing transactions. As we have described previously, these personal relationships seemed to be especially important for dealing with non-routine events. Moreover, it was when firms supplemented use of electronic networks with personal relationships that they achieved the best results with the electronic networks, or at least mitigated some of the potential negative consequences of using networks for coordination. For example, one pharmaceutical company reported heavy use of electronic mail to keep in touch with researchers with whom they had contracted out clinical trials. But they reported that the electronic mail worked only because it was supplemented by periodic face-to-face meetings, in which the principals worked out strategy and monitored performance. Other interviewees reported that it was crucial to know people in the other firms when dealing with the problems that arose in unusual situations (such as rush or special orders) and for dealing with problems that were created by computerized ordering systems (such as orders that had to be delivered by a particular date, even when a customer had failed to schedule a warehouse receiving slot until after the date).

Greater use of personal relationships was associated with greater satisfaction with the client, even after controlling for the relatively objective performance outcomes. To the extent that satisfaction is a proxy for the likelihood of re-using a supplier, personal contact is likely to be associated with repeat business.

Directions for future research

Our conclusions must be tempered by several limitations constraining our ability to interpret these data. The major source of ambiguity is the cross-sectional design of this research which prevents us from determining the causal direction of many of the relationships identified here. Research in this area would benefit from longitudinal data collection.

Given the cross-sectional design, we are unable to offer any quantitative evidence for the direction of causation. This is a particular problem in trying to determine the extent to which electronic network usage influences firms' make-buy decisions. Our finding of a positive association between network usage and in-house production is generally inconsistent with an electronic markets hypothesis. We believe that the most plausible interpretation for this finding is that firms first develop an in-house capability for producing needed inputs, and then subsequently interconnect the related units with intra-company local and wide area networks.

The same problem exists in interpreting the positive association between electronic network usage and personal relationships as coordination modes. It may be that the very act of putting in and working with electronic networks causes a greater need for personal coordination. Alternately, pre-existing personal relationships used for organizational coordination may also help firms to coordinate electronically.

We do not know the extent to which the results reported here are stable or will change as organizations become more sophisticated in outsourcing and in using electronic networks. Over time, it may be that successful internal network use will encourage firms to extend electronic transactions across boundaries to their closest trading partners--a pattern suggested by the popularity of intranets and extranets as the basis for enterprise networks. It may be that as external networks become more widespread (as they are becoming with the growth of the Internet), the quality and satisfactions problems associated with their use will disappear.

Follow-up research would also benefit from a more detailed examination of the mechanisms by which use of electronic networks and personal relationships influence virtualization and coordination success. In particular, it would be useful to understand better the way that effects of these coordination mechanisms are mediated by flexibility and trust.

Summary

Despite these limitations, the research reported here advances our understanding of the causes and consequences of virtual organization. We have tried to make the case that virtual organization is a matter of degree, rather than a unique type of organization. Ours is one of the few empirical studies of virtual organization and electronic networks that has examined causes and consequences across a broad range of firms in several industries. Contrary to much recent speculation, the research did not find that use of electronic networks for transactions was associated with increased outsourcing, but rather with greater dependence on internal production. Moreover, the use of interpersonal relationships for coordination, which many think of as an alternative to electronic network use, was associated with greater network use. Surprisingly, use of electronic networks was negatively associated with such outcomes as order quality and efficiency, and satisfaction with suppliers, while more reliance on personal linkages was associated with better outcomes and mitigated the negative consequence of using electronic networks.

Acknowledgments

The authors are grateful to the National Science Foundation for the grant (IRI-9408271) which funded this research. We would also like to acknowledge the helpful comments of several anonymous reviewers.

Footnotes

¹We included the relative power of the focal firm vis-a-vis the supplier in preliminary analyses. This variable consisted of the ratio of the standardized focal firm size divided by the standardized supplier firm size. However, because this variable did not approach signifiance in any analysis and because we include focal and supplier firm size as control variables, we have excluded this ratio from the results reported below.

² In retrospect, the selection of broadcast time slots as an input was a poor choice. While airtime is a key input into a broadcast advertising campaign, since no advertising agencies own radio or TV stations, they must by necessity buy this input from an external organization. However, our conclusions remained the same when this input was eliminated from analyses.

³ This negative association with electronic network usage was also evident when examining the component factors of transaction outcomes - ordering efficiency and percentage of orders with some type of error - separately.

⁴ Figure 3 summarizes the path analysis implied by the coeffiencients in Tables 3 and 4 . We do not wish, however, to present a full, structural equation model with coefficients for all variables in the model for two reasons. First, we make no strong claims for the causal order of variables represented in Figure 3. Second, we make no strong claim that we have captured all of the important variables that influence the decision to outsource and order quality.

References

Attewell, P. (1995). Information technology and the productivity paradox. In D. H. Harris, (Ed.), Organizational linkages: Understanding the productivity paradox. Washington, DC: National Academy Press.

Bakos, Y. (1991). A strategic analysis of electronic marketplaces. MIS Quarterly, September, 295-310.

Bakos, Y., & Brynjolffson, E. (1992, October). When quality matters: Information technology and buyer-supplier relationships. Unpublished manuscript. Irvine, CA: University of California.

Baura, A., Kriebel, C., & Mukhopadhyay, T. (1995) Information technologies and business value: An analytic and empirical investigation. Information Systems Research, 6(1), 3-23.

Bradley, S. (1993). The role of IT networking in sustaining competitive advantage. In S. Bradley, J. Hausman, & R. Nolan (Eds.), Globalization, technology and competition: The fusion of computers and telecommunications in the 1990s, (113-142). Boston, MA: Harvard University School Press.

Bradley, S., Hausman, J., & Nolan, R. (Eds.). (1993, June). Globalization, technology and competition: The fusion of computers and telecommunications in the 1990s. Boston, MA: Harvard University School Press.

Brousseau, E. (1990, June). Information technologies and inter-firm relationships: The spread of inter-organizational telematics systems and its impact on economic structure. Paper presented to the International Telecommunications Society, Venice.

Brynjolfsson, E., Malone, T., Gurbaxani, V., & Kambil, A. (1994). Does information technology lead to smaller firms? Management Science, 40(12), 1628-1644.

Cash, J. I., & Konsynski, B. R. (1985). IS redraws competitive boundaries. Harvard Business Review. March-April, 134-142.

Chan, A. P. (1997). Coordination and control of retailer-supplier transactions: Factors influencing organizational adoption and use of electronic information networks. Unpublished Doctoral Dissertation. East Lansing, MI: Michigan State University.

Choi, S. Y., Stahl, D., & Whinston, A. (1997). The economics of electronic commerce. Indianapolis, IN: Macmillan Technical Publishing.

Clemons, E. (1993). Information technology and the boundary of the firm: Who wins, who loses, who has to change. In S. Bradley, J. Hausman, & R. Nolan (Eds.), Globalization, technology and competition: The fusion of computers and telecommunications in the 1990s (219-242). Boston, MA: Harvard University School Press.

Davidow, W., & Malone, M. (1992). The virtual corporation. New York: Harper.

Granovetter, M. (1985). Economic action and social structure: The problem of embeddedness. American Journal of Sociology, 91(3), 481-510.

Gurbaxani, V., & Wang, S. (1991). The impact of information systems on organizations and markets. Communications of the ACM, 34(1), 59-73.

Hart, P., & Estrin, D. (1991) Inter-organizational networks, computer integration, and shifts in interdependence: The case of the semiconductor industry. ACM Transactions on Information Systems, 9(4), 370-398.

Hennart, J. (1993). Explaining the swollen middle: Why most transactions are a mix of "market" and "hierarchy." Organization Science, 4(4), 529-547.

Johanson, J., & Mattson, L.G. (1987). Inter-organizational relations in industrial systems: a network approach compared with the transactions-cost approach. International Studies of Management and Organization, 17(1), 34-48.

Johnston, H. R., & Vitale, M. R. (1988). Creating competitive advantage with inter-organizational information systems. MIS Quarterly, June, 153-165.

Johnston, R., & Lawrence, P. (1988). Beyond vertical integration: The rise of the value-adding partnership. Harvard Business Review, July-August, 94-101.

Kalakota, R., & Whinston, A. (1997). Electronic commerce: A manager's guide. Reading, MA: Addison-Wesley.

Kambil, A. (1991). Information technology and vertical integration: Evidence from the manufacturing sector. In M. Guerin-Calvert & S. Wildman (Eds.), Electronic services networks: A business and public policy challenge (22-38), New York: Praeger.

Keen, P., & Cummins, M. (1993). Networks in action: Business choices and telecommunications decisions. Belmont, CA: Wadsworth.

Keen, P. (1988). Competing in time: Using telecommunications for competitive advantage. Cambridge, MA: Ballinger Press.

Kekre, S., & Mudhopadhyay, T. (1992). Impact of electronic data interchange technology on quality improvement and inventory reduction programs: A field study. International Journal of Production Economics, 28, 265-282.

Kling, R., & Scacchi, W. (1982) The web of computing: Computer technology as social organization. Advances in Computers, 21, 1-90.

Krackhardt, D. (1992) The strength of strong ties: The importance of philos. In N. Nohria & R. Eccles (Eds.), Networks and organizations: Structure, form, and action (216-239). Boston: Harvard Business School Press.

Kreps, D. M. (1990) Corporate culture and economic theory. In J. E. Alt & K. A. Shesple (Eds.), Perspectives on positive political economy (90-143). New York: Cambridge University Press.

Landauer, (1996). The trouble with computers. Cambridge, MA: MIT Press.

Lawrence, P., & Lorsch, J. (1967). Organizations and environment. Boston: Harvard University.

Lorenz, E. (1989). Neither friends nor strangers: Informal networks of subcontracting in French industry. In D. Gambetta (Ed.), Trust: Making and breaking of cooperative relations (194-210). Oxford: Basil Blackwell.

Malone, T., & Rockart, J. (1993). How will information technology reshape organizations? Computers as coordination technology. In S. Bradley, J. Hausman, & R. Nolan (Eds.), Globalization, technology and competition: The fusion of computers and telecommunications in the 1990s (37-56). Boston, MA: Harvard University School Press.

Malone, T. (1987). Modeling coordination in organizations and markets. Management Science, 33, 1317-1332.

Malone, T., Yates, J., & Benjamin, R. (1987). Electronic markets and electronic hierarchies: Effects of information technology on market structure and corporate strategies. Communications of the ACM, 30, 6, 484-497.

Malone, T., Yates. J., & Benjamin, R. (1989). The logic of electronic markets. Harvard Business Review, May-June, 166-171.

Miller, D., Clemons, E., & Row, M. (1993). Information technology and the global virtual corporation. In S. Bradley, J. Hausman, & R. Nolan (Eds.), Globalization, technology and competition: The fusion of computers and telecommunications in the 1990s (283-307). Boston, MA: Harvard University School Press.

Ouchi, W. G. (1980). Markets, bureaucracies, and clans. Administrative Science Quarterly, 25, 129-141.

Perrow, C. (1986). Complex organizations: A critical essay (3rd. Ed.). New York: Random House.

Porter, M. E., & Millar, V. E. (1985). How information gives you competitive advantage. Harvard Business Review, May-June, 149-160.

Porter, M. E. (1980). Competitive strategy: Techniques for analyzing industries and competitors. New York: The Free Press.

Powell, W. (1987). Hybrid organizational arrangements: New form or transitional development. California Management Review, 30, 66-87.

Powell, W. (1990). Neither market nor hierarchy: Networked forms of organization. In B. Staw & L. Cummings, (Eds.), Research in organizational behavior (Vol. 12, 295-336). Greenwich, CT: JAI Press.

Saunders, C., & Hart, P. (1993). Electronic data interchange across organizational boundaries: Building a theory of motivation and implementation, Working Paper #7-93. Decision and Information Systems, Florida Atlantic University, Boca Raton, FL.

Steinfield, C., & Caby, L. (1993). Strategic organizational applications of videotex among varying network configurations. Telematics and Informatics, 10, 2, 119-129.

Steinfield, C., Caby, L., & Vialle, P. (1993). Internationalization of the firm and the impacts of videotex networks. Journal of Information Technology, 7, 213-222.

Steinfield, C., Kraut, R., & Plummer, A. (1995). The impact of interorganizational networks on buyer-seller relationships. Journal of Computer Mediated Communication [Online], 1(3). Available http://www.ascusc.org/jcmc/vol1/issue3/steinfld.html.

Stinchcombe, A. (1990). Information and organizations. Berkeley: University of California Press.

Streeter, L. A., Kraut, R. E., Lucas, H. C., & Caby, L. (1996). The impact of national data networks on firm performance and market structure. Communications of the ACM.

Suchman, L., & Wynn, E. (1984). Procedures and problems in the office. Office: Technology and People, 2(2), 133-154.

Tornatsky, L., & Klein, K. (1982). Innovation characteristics and innovation adopting-implementation: A meta-analysis of findings. IEEE Transactions on Engineering Management, EM-29, 28-45.

Tushman, M., & Nadler, D. (1978). Information processing as an integrating concept in organizational design. Academy of Management Review, 613-624.

Uzzi, B. (1997) Social structure and competition in interfirm networks: The paradox of embeddedness. Administrative Science Quarterly, 42 (March), 35-67.

Weber, J. (1995, March 6). Just get it to the stores on time. Business Week, 66-67

Wildman, S., & Guerin-Calvert, M. (1991). Electronic services networks: Functions, structures, and public policy. In M. Guerin-Calvert & S. Wildman (Eds.), Electronic services networks: A business and public policy challenge (3-21). New York: Praeger.

Williamson, O. (1975). Markets and hierarchies: Analysis and antitrust implications. New York: Free Press.

Williamson, O. (1985). The economic institutions of capitalism. New York: Free Press.

Williamson, O. (1991) Comparative economic organization: The analysis of discrete structural alternatives. Administrative Science Quarterly, 36(2), 269-296.

Williamson, O. (1996) Economic organization: The case for candor. Academy of Management Review, 21(1), 48-57.

Zucker, L. (1986). Production of trust: Institutional sources of economic structure: 1840-1920. In B. Staw & L. Cummings, (Eds.), Research in organizational behavior (Volume 8, 53-111). Greenwich, CT: JAI Press.

 

 

Appendix A: Scale Items

Supplier Size (alpha = 0.73)

Approximately how many employees does this supplier have? 1. Fewer than 20 2. Between 20 and 99 3. Between 100 and 499 4. 500 or more

Would you say that this supplier is one of the ten largest sources in the United States for [this product or service], in terms of volume, a major source but not in the top-ten sources, an average source, or a smaller than average source? (reversed) 1 Top ten 2. Major, but not top-ten 3. Average 4. Smaller than average 5. Not in this country

Time Pressure   (alpha = 0.68 )

To meet schedules related to this product or service, we need to use every available minute effectively.

In developing, producing, and distributing this product or service we have to meet tight deadlines.

Ease of Description

It is difficult to describe the [product or service] we routinely acquire. *

Object certainty (alpha = 0.68 )

How much does the availability of the [product or service] change over the course of a year?

How much does the [product or service] your company needs change from order to order?

How much does the unit price of the [product or service] change from order to order?

Object predictability (alpha = 0.79)

For a typical [product or service], how far in advance can you predict what [product or service] your company will need? 1. Within a week of when you’ll need it 2. Within a month 3. Within three months 4. Within six months 5. More than six months from when you’ll need it

How far in advance can your predict the quantity of [product or service] your company will need? 1. Within a week of when you’ll need it 2. Within a month 3. Within three months 4. Within six months 5. More than six months from when you’ll need it

Order Frequency

How often do you acquire [the product or service] from this supplier? 1. At least monthly 2. At least every three months 3. At least every six months 4. At least every 12 months 5. Less frequently than every 12 months

 

 

Generic Object (alpha = 0.65)

My firm is about the only one that uses [this particular product or service]. 1 : Strongly Agree - 5 : Strongly Disagree

The [product or service] we get from this supplier have certain features that are only useful to us. 1 : Strongly Agree - 5 : Strongly Disagree

Many other companies use the same types of [products or services] when they develop, make and distribute [their products or services]. 1 : Strongly Agree - 5 : Strongly Disagree

The [products or services] we get from this supplier are fairly standard for the industry. 1 : Strongly Agree - 5 : Strongly Disagree

Outsourced production (alpha=.86)

The next set of questions asks about your company's dealings with one supplier of [good or service]. Think about the most important supplier you've worked with in the past 12 months. This supplier can be a person or department in your own firm or an outside individual or another company. Is this supplier: An individual or departments within your firm A partly or fully owned subsidiary of your firm Another firm which is your partner in a joint venture Outsiders (individuals you don't employ or suppliers in which your firm has no ownership stake)

What percentage of [the input] does your firm get from in-house sources?

 

Coordination through electronic networks (alpha = 0.93)

An electronic connection is one that connects your computers with each other and with computers from outside your company. When we refer to electronic connections here we mean any type of computer communication which allows you to exchange information with this supplier. We include modem connections, local area networks, online databases, electronic mail, electronic data interchange, Lotus Notes, or the Internet. Using a five point scale where "1" is very important and "5" is not at all important, please tell me how important electronic connections are to the following items, or tell me if you don't use electronic connections at all for that purpose. How about...?

a. In selecting this supplier

b. Developing the specification of the [product or service] your company orders

c. The negotiation of agreements to acquire [the product or service]

d. Getting [the product or service]

e. Monitoring the quality of the [products and services] you receive

f. Fixing problems after you have ordered [the product or service]

Coordination through personal relationships (alpha = 0.85)

Now I’m going to ask you about the importance of personal relationships in selecting and working with this supplier. When I refer to "personal relationships" I am including any type of personal connection or personal knowledge between people in your firm and people in the supplier's organization. Using a five-point scale where "1" is very important and "5" is not at all important, please tell me how important are personal relationships in...?

a. Helping your company select this supplier

b. Developing the specification of the [product or service] your company order[s] from this supplier

c. The negotiation of agreements to acquire [the product or service]

d. Getting [the product or service]

e. Monitoring the quality of the [products and services] you receive

f. Fixing problems after you have ordered [the product or service]

Satisfaction with supplier (alpha = 0.89)

Using a five-point scale where "1" is extremely satisfied and "5" is not at all satisfied, how satisfied are you with the follow aspects?

a. The supplier’s ability to handle a rush [order]

b. The supplier’s ability to handle an out-of-the-ordinary [order]

c. That you have identified the best supplier

d. That your company can easily specify exactly what you want from your supplier

e. The terms your company is able to negotiate

f. The efficiency with which your company gets [the product or service]

g. Your company’s ability to monitor and evaluate [the product or service] quality

h. Your company’s ability to fix problems with this supplier

Using the same five-point scale, please tell me how satisfied are you with this supplier overall?

Overall Order quality (alpha = 0.65)

Subscale: Order quality (alpha = 0.72)

Now I would like to ask you about the quality of the orders you have placed with this supplier over the past 12 months.

What percentage of these [orders] arrived after your targeted delivery date?

What percentage of these [orders] had an error of any sort?

What percentage of these [orders] did not met your quality standards?

Subscale: Order efficiency (alpha = 0.48)

When you are processing a typical [order] for [this product or service], approximately how many different people in your firm would be involved with it, from the time it is placed to the time it is actually received?

During a typical cycle, from the time you think you need [the product or service] to the time the order is complete, how many times would someone in your firm communicate with the supplier?

Approximately how much time would elapse from when your company places a typical [order] for [this product or service] to when you would actually receive it?

Note: For attitude statements, most items were represented as a 5-point Likert scale, where 5 meant strongly agree with the statement and 1 meant strongly disagree with the statement.

About the Authors

Robert Kraut (Ph.D. Department of Psychology, Yale University) is a Professor of Social Psychology and Human Computer Interaction at Carnegie Mellon University. He previously directed the Interpersonal Communications Research Program at Bellcore, was a Member of Technical Staff at AT&T Bell Laboratories, and was a member of the faculty at Cornell University and the University of Pennsylvania. He has broad interests in the design and social impact of computing. He has conducted empiricial research on office automation and employment quality, technology and home-based employment, the communication needs of collaborating scientists, the design of information technology for small-group intellectual work, and the impact of national information networks on organizations and families. He was instrumental in the design and testing of several new information technologies including video telephony systems and software for collaborative work teams.
Address: Human Computer Interaction Institute, Carnegie Mellon University, 5000 Forbes Ave., Pittsburgh, PA 15213.

Address: Carnegie Mellon University, Pittsburgh, PA 15213 USA.

Charles Steinfield is Professor of Telecommunication at Michigan State University.
Address: Room 436, Communication Arts Building, Michigan State University, East Lansing, MI 48824-1212 USA.

Alice P. Chan (Ph.D., Michigan State University, M.A. and B.A., University of Hawaii at Manoa) is an Assistant Professor in the Department Communication and a Faculty Research Associate of the Interactive Multimedia Group (IMG) at Cornell University. Dr. Chan's research centers on the uses and impacts of new communication and information technologies, concentrating in organizational communication, inter-firm relationships and electronic commerce. She is a contributor to a previous issue of JCMC and has given numerous presentations in several countries on electronic network use and coordination of buyer-seller transactions. (Some of these writings were under Chan's former last name, Plummer).
Address: 331 Kennedy Hall, Cornell University, Ithaca, NY 14853 USA.

Brian Butler holds a B.S. in Computer Science and a M.S. in Information Systems from Carnegie Mellon University. He is completing his Ph.D. in Information Systems at the Graduate School of Industrial Administration at Carnegie Mellon University. His work has examined the role of social and power structures in the adoption of information technologies within organizations, the impact of developing telecommunication technologies on the structure of retail and business-to-business commerce, and the consequences of communication activity for the sustainability of long-term electronic groups.
Address: Graduate School of Industrial Administration, Carnegie Mellon University, 5000 Forbes Ave., Pittsburgh, PA 15213 USA.

Anne Hoag (Ph.D., Michigan State University; BA, University of Michigan) is an Assistant Professor in the College of Communications at Penn State University. Hoag's research interests include electronic commerce and the social effects of new communications technologies.
Address: College of Communications, Penn State University, University Park, PA 16802 USA