The Impact of Interorganizational Networks on Buyer-Seller Relationships

Abstract

Interorganizational data networks can have two opposing effects on buyer-seller relationships. On the one hand, networks may be used to foster electronic marketplaces characterized by more ephemeral transactions between buyers and sellers. Also plausible, however, is the use of networks to strengthen existing commercial relationships and lock in partners by increasing the costs of switching to new trading partners. Our review of the literature suggests that this latter tendency toward what have been called electronic hierarchies is more prevalent. This paper examines the theoretical rationales behind these competing effects and presents some evidence to show the conditions under which electronic marketplaces or electronic hierarchies are likely to prevail.

Introduction

During the past decade, organizational theorists, business consultants, and telecommunications managers and vendors have directed our attention to the strategic role that information can play in the competitive strategy of firms (see for example, Bradley, Hausman & Nolan, 1993; Keen, 1988; Porter & Millar 1985). Throughout the 1980s, widely discussed case examples demonstrated how the use of telecommunications networks to link firms to their suppliers and distribution chains conveyed important first mover advantages to such firms as American Hospital Supply and McKesson. The reported benefits to the firms deploying such interorganizational networks included increased efficiency of order processing, reduced costs due to just-in-time inventory management, locking in trading partners because of the difficulties competitors faced once a network is in place, and greater ability to customize products and services based upon information arising from the transactions carried by the network (Cash & Konsynski, 1985; Johnston & Vitale, 1988).

In the past, these networks were typically put into place by a dominant firm in a value chain, and were built upon proprietary applications running over private networks. Chrysler, for example, required its parts suppliers to participate in its Electronic Data Interchange (EDI) network. Hence it should not be surprising that they were often implemented with the most important existing trading partners. In fact, on the upstream side, a typical goal for such applications as EDI, was to reduce the total number of suppliers and enhance the quality and efficiency of the overall purchasing function (Kekre & Mudhopadhyay, 1992).

The increasing standardization of such applications as EDI, as well as the availability of lower cost public network infrastructures, has convinced several researchers that interorganizational networks will not only proliferate, but will be applied in qualitatively different ways. In particular, as the barriers to participate in such electronic transactions diminish, some researchers now believe that the conditions are ripe for the establishment of electronic marketplaces. Rather than having networks only link existing trading partners in a tightly coupled arrangement, such new electronic markets could conceivably include larger numbers of buyers and sellers (Malone, Yates & Benjamin, 1987; Wildman & Guerin-Calvert, 1991).

The popular press is inundated with reports of the ongoing commercialization of the Internet, and the rush of firms to establish a presence in this new virtual marketplace (e.g. Business Week, 1994). Because this is a near ubiquitous and public infrastructure, built upon a common set of standards, it has the characteristics that many feel would be necessary for the growth of electronic marketplaces. In much the same way, years before, France established the Teletel infrastructure that permitted firms to establish linkages to their trading partners at a lower cost than if they used private networks. Although not meant for high volume transactions (e.g. Teletel required the establishment of a dial-up connection for each transaction) it could easily permit smaller trading partners to share data and automate transactions with each other (Steinfield & Caby, 1993).

The purpose of this paper is to examine the prevailing theoretical expectations for the effect of networks on interorganizational relationships, and develop propositions that are relevant for the new infrastructure environment characterized by public and ubiquitous data networks like the Internet. Our conclusions are based upon a critical review of previous literature on interorganizational networks, and a reexamination of empirical studies that specifically look at the uses of public and ubiquitous networks.

The paper is organized into the following sections. Section II provides a review of the literature examining the likely impacts of interorganizational networks on the ways that firms coordinate with each other. Here we show that the prevailing logic, adapted from Malone, Yates, & Benjamin (1987), is that a reduction in coordination costs enabled by information technology is likely to produce large changes in the ways that firms select and maintain relationships with trading partners. Malone and colleagues emphasize that networks can facilitate tighter coupling through the integration of production across firms as well as reduced search costs that might encourage more market-like relations across firm. However, their essential argument, that reduced coordination costs afforded by networks will lead to more market-like arrangements, is somewhat ambiguous, since improved coordination is necessary for both types of outcomes. Hence, their analysis makes it difficult to hypothesize about the ultimate effects of interorganizational networks on the relations between firms. We explicitly focus on two variables not well developed in the previous literature: the relative openness of the network infrastructure, which has its effects by reducing the costs of adding new parties to an interorganizational network, and the locus of control of the electronic network service, which influences the kinds of incentives participants face vis-a-vis the addition of new buyers or sellers.

Section III then examines the empirical evidence to date, in particular by examining survey research and case studies where firms did, in fact, use public (hence relatively open in our view) data infrastructures as a platform for interorganizational networks. We show that the reduced cost of adding new subscribers does appear to alter the ways in which interorganizational networks are used, but not necessarily in a fashion consistent with the expectations of Malone et al. (1987). We further show that the locus of the control of an electronic network service does appear to explain different structural outcomes. Taken together, both the theoretical arguments and the empirical evidence lead us to believe that firms will use interorganizational networks to build tight relationships with their trading partners, rather than to select suppliers on a transaction by transaction basis from a large pool. We further believe that for electronic marketplaces to emerge, where there are both multiple buyers and multiple sellers and the network performs search and product/service matching functions, some form of third party market maker is needed.

Finally, Section IV discusses the limitations of the empirical work to date, noting that it lacks measurement of the processes by which networks are hypothesized to have their effects on interorganizational relations. In particular, without specific measurements on the costs and benefits of network usage for acquiring goods in an electronic marketplace vs. usage for more tightly integrated production that implies foregoing the advantages of the market as a coordination mechanism, we cannot fully explain the prevalence of electronic hierarchies in the empirical studies. We suggest that alternative explanations, based upon such variables as the existence of interpersonal relationships among members of a firm, and the degree of trust between firms. Without further empirical investigation, our conclusions remain speculative.

Electronic Interorganization Networks and Organizational Forms

Out-sourcing versus In-house Production

The production of almost any complex product consists of acquiring various raw materials and other components and crafting them, with each step presumably adding value as the product wends its way towards its ultimate consumer. The value chain consists of the movement of the components through various stages of production and distribution as they are transformed into final products (Porter, 1980). To make and sell a news magazine, for example, one must integrate reporting, writing, editing, photographs, typography, layout, paper, printing, subscription services, mailing labels, and physical distribution. A firm must decide to produce each of these goods or services in-house or to out-source it. For example, most news magazines do not own either their own paper mills or printing presses, but they do hire editors as salaried employees. They may instruct employee-reporters to write a story or they may commission one from a stringer or other freelancer. Photographs may come from their own photographers or be purchased from a stock house. Explaining whether a firm would produce a good or service in-house or buy it on the open market has been a traditional question in economics and organizational sociology.

Most economists believe that were it not for the costs of coordination markets would generally be more efficient mechanisms for production than hierarchies. By being able to sell to many customers in a market, a producer could acquire more experience, level the load of production across many customers, and capitalize on economies of scale, all of which generally lead to more efficient production. Similarly, in a market, customers can shop around for the best combination of price, quality, or other desirable attributes and because they have choices, are not held hostage to the opportunistic behavior of any single supplier. Competition among suppliers would also lead to efficiencies.

The rub is that when purchasing goods and services on the open market, costs of coordination can be high. Economists point to the extra costs that out-sourcing imposes on coordinating the economic transactions themselves. In particular, when out-sourcing, firms have additional costs as they search for appropriate suppliers, specify contracts, enforce the contracts, and handle financial settlement (Williamson, 1975). For example, customers can have difficulty specifying what they want and searching through the many alternatives to find the best suppliers and best wares, and suppliers incur costs in advertising the availability of their goods and services to potential customers (Malone et al, 1987). Once an economic transaction has been agreed upon and goods produced, the customer needs to insure that suppliers are meeting the terms of the contract, for example, by continuing to supply high quality goods according to an agreed-upon schedule. Monitoring through various quality-control programs is expensive. Once the goods have been delivered, the customer needs to receive an invoice and producers need to be paid. Arranging billing and payment between large numbers of customers and suppliers can be expensive.

Williamson argues firms exist precisely to reduce the costs of negotiating, monitoring and executing transactions that are necessary when acquiring goods and services in the open market.1 His term transaction costs refers to these costs.2 He distinguishes between markets and hierarchies in terms of the economic location of production. The term hierarchy is used to refer to the case of in-house production, with economic decisions made by managerial fiat, while the term market refers to cases of out-sourcing, with the acquisition of goods is subject to the laws of supply and demand.

Williamson further explains a firm's choice of in-house (hierarchy) vs. out-sourcing (market) in order to obtain some good by examining factors that can influence both opportunistic behavior of other firms and the costs of search in the marketplace. A variety of forms of asset specificity can make a firm vulnerable to opportunism. When the assets required to produce a good are specific, they are only of value to the firm needing them.3 If provided by another firm, this firm may behave opportunistically (i.e. extract a higher price than necessary) since it realizes the buyer is dependent and cannot easily obtain the asset elsewhere. Under these conditions, a firm is more likely to choose hierarchy as a form of governance.

Complexity of product information has also been related to opportunism and search costs (Malone et al., 1987). When a good is highly complex, it may be harder for a buying firm to know whether or not the selling firm is engaging in opportunistic behavior. Moreover, it may be harder to communicate the good's features in the marketplace with enough precision to enable adequate matching of needs. Hence, this feature too should encourage a firm to produce the required good in-house.

In addition to the cost of coordinating the economic transactions, out-sourcing often leads to higher costs of coordinating production itself. Because external suppliers of raw materials or services are likely to be further away than an in-house supplier, the difficulties of coordination across space are likely to be higher. In addition, because external suppliers are likely to have worked with a firm for a shorter duration than in-house suppliers, the external suppliers will have lost some advantages due to learning. In the case of a news magazine, for example, it may be more difficult for editors and photographers to decide which one in a set of photographs is appropriate for an article and to determine how to crop it, if the photographer is located far away. By working with the same photographer over an extended period, the editor is likely to have learned the photographer's tastes and vocabulary, making editorial direction and decision making easier.

Malone and colleagues (1987) argue that the use of electronic communication links between firms can reduce both the costs of coordinating economic transactions and the costs of coordinating production. Coordination of both sorts consists of communicating information and processing it. Because modern information technology lowers the costs of both communication and information processing, Malone and his colleagues hypothesize that “the result of reducing coordination costs without changing anything else should be an increase in the proportion of economic activity coordinated by markets. (p. 489).” That is, they expect that these lowered coordination costs would encourage more out-sourcing by enabling firms to buy goods and services less expensively than to produce them in-house (Malone, 1987; Malone et al., 1987; Malone et al., 1989). To the extent that the cost of communication and information processing are reduced, the cost disadvantage of out-sourcing a production process is also reduced. Returning to our magazine example, if an editor could as easily identify and vet a distant, freelance photographer as a nearby employee and could as easily share information with the freelancer, the editor would benefit from out-sourcing photography. If, for example, the editor can easily search through photo archives over the World Wide Web, or share new photographs with a freelancer through multimedia electronic mail or through screen sharing software, the costs of working with external sources of photography would be reduced, but the advantages of having a larger pool of photographs or photographers would remain.

Empirical tests of the hypothesis that greater use of interorganizational networks leads to greater outsourcing are only starting. Consistent with this hypothesis is evidence at the industry level that increases in investment in information technology was associated with a decline in average firm size and rise in the number of firms (Brynjolffson, Malone, Gurbaxani, & Kambil, 1993). 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 produced 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. For example, analyses conducted at the industry level (2-digit SIC codes) don't necessarily speak to the way particular firms deploy information technology. By using general information technology as the independent variable obscures the unique role of that interorganizational networks might play.

Electronic Markets versus Electronic Hierarchies

Malone et al (1987) make an additional prediction about the effects of interorganizational data networks that has not yet been tested. They argue that even when production is out-sourced, networks will encourage market-like rather than hierarchy-like control of production processes. When one considers markets and hierarchies as control mechanisms rather than as locations of production, the distinction between them is continuous rather than dichotomous. Control here refers to the means by which the behavior of employees or trading partner firms is governed (hence the term governance is often applied), and opportunistic behavior avoided. Even when out-sourcing, firms can have more or less hierarchical relationships with an economic partner. Under a hierarchical mechanism, value chain activities are controlled and directed by management decisions either within a single firm or across several interacting firms. Product and service attributes are set by managerial decisions, rather than selection in the open market. A buyer does not choose from among many potential sellers, but instead procures goods and services from a predetermined supplier. There are many different forms of hierarchical control, including partnerships, alliances, or simply stable, long term buyer-supplier relationships. If a single firm serves as the sole supplier to many buyers, the former's relationship with each of the latter is considered hierarchical. Similarly, long term relationships in which trading partners adopt common values, operate on the basis of trust rather than contract, and make accommodations to each other also represent hierarchical relationships.4

In contrast, when a market mechanism is at work, the flow of materials and services through the value chain is coordinated by a decentralized price system. Relationships among firms can be short-lived, since price and net value received on a transaction by transaction basis determine the exchange of economic resources. The availability of goods and services varying in price, quantity produced, quality standards, delivery schedules and other attributes is not determined by explicit managerial direction in a dominant supplier or buyer firm or a cluster of them. Assuming perfect market information, buyers purchase from whichever supplier can offer the best combination of these attributes.

Again, consider the editor of a news magazine looking for a photograph to run with a story about eastern Europe. If the editor sent an employee to take the pictures, this is clearly an example of a hierarchical relationship, as the term is used by Williamson. If the editor searched through several stock houses to select an already existing photograph, this would be an example of a market-based purchase. However, if the editor gave the photography assignment to a freelance photographer with whom he had worked extensively in the past and who received a substantial portion of his income from the news magazine, most observers would agree that the relationship between news magazine and photographer is hierarchical, even though the editor out-sourced the photography.

According to Malone et al. (1987) interorganizational electronic networks can improve coordination between firms in two contrasting ways. By using electronic networks to reduce the costs of searching for appropriate goods and services, firms can achieve an electronic brokerage effect. Examples of using electronic interorganizational electronic networks to reduce search costs include the NASDAQ system which creates an electronic market for over-the-counter stocks, the EasySabre airline reservation system, which allows consumers to search for and compare ticket prices and availability before ordering, on-line multiple listing services which help real estate agents and customers to narrow down they houses they visit, and CommerceNet which allows firms in Silicon valley to order computer supplies on the World Wide Web. These services all connect different buyers and sellers through a shared information resource (like a centrally maintained database or the decentralized World Wide Web) and provide some tools for searching the data. They help the buyers to quickly, conveniently and inexpensively evaluate the offerings of various suppliers. The electronic brokerage effect can increase the number of alternatives as well as the quality of the alternative ultimately selected, while decreasing the cost of the selection process.

By using electronic networks to reduce the costs of tightly integrating a particular buyer and seller, firms can achieve an electronic integration effect. An example would be an electronic data interchange system that connects a retailer's point of sale terminals to a supplier's delivery system, decreasing the likelihood of the retailer going out of stock on popular goods (Weber, 1995). Another example is the integration of CAD/CAM systems between a computer chip design firm and a silicon foundry, that allows the designers of a chip to monitor the manufacturing process and to have more flexibility in changing their designs (Hart & Estrin, 1991). This effect is manifested when technology is used not only to facilitate communication but also to tightly couple processes at the interface between stages of the value chain.

Malone et al hypothesize that using electronic interorganizational networks for these different effects will have consequences for the control mechanisms regulating firms transactions with their trading partners. To the extent that firms use electronic networks for their electronic brokerage effects (i.e., to lower search costs), they will tend to develop electronic market relationships with their trading partners. Since the technology makes it easy for firms to search through many possible suppliers, firms will use these networks to develop a relatively large pool of suppliers, with relatively little loyalty over different transactions to particular suppliers. On the other hand, to the extent that firms deploy electronic networks for their electronic integration effect (i.e., to tightly couple with trading partners), they will tend to develop long-term, hierarchical relationships with their trading partners.

Malone et al. (1987; 1989) propose that over time electronic brokerage effects will dominate electronic integration effects, as competition, regulation and improving public infrastructure draw additional firms onto networks and as search costs drop more rapidly than integration costs. As a result, they claim that one of the major effects of inter-organizational networks would be a shift from hierarchical to market relationships. “Some of the initial providers of electronic markets have attempted … to capture customers in a system biased towards a particular supplier. We believe that, in the long run, the significant additional benefits to buyers possible from the electronic brokerage effect will drive almost all electronic markets toward being unbiased channels for products from many suppliers (Malone et al, 1987, P. 492.).”

Malone et al base their predictions about the dominance of electronic markets on the relative benefits they anticipate firms will receive from electronic brokerage and electronic integration. These benefits do no invariably favor electronic brokerage over integrate. Which they favor is likely to be a complex function of the attributes of the products being exchanged, the information networks being used, and the business environment in which the networks are deployed. In addition, whether firms deploy electronic networks for brokerage or integration itself is also likely to be influenced by the pre-existing relationships among particular suppliers and buyers, and these starting conditions will therefore shape whether market or hierarchical relationships develop over the long term. The following section briefly discusses some product attributes, network infrastructure, business environment and interfirm relationships that are likely to influence how interorganizational electronic networks are deployed.

Product attributes.

The value of external networks for product search increases as geographic, temporal, or cognitive constraints make product location and selection difficult. Thus, use of information networks should lead to more market-like relationships between buyers and sellers, when there are large numbers of potential suppliers, when suppliers are geographically separated, when there are many comparable products in the industry, and when prices or products change rapidly.

As Malone et al (1987) note, products low in complexity are easier to describe and search for than are products higher in complexity. As a result, they hypothesize that electronic markets are more likely to develop among firms exchanging simpler products. Product complexity is partially an intrinsic characteristic of the product, but can be the outcome of prior attempts to apply information technology. For example Malone et al would expect to see an electronic market develop for books, because books can be easily searched through various information retrieval techniques (e.g., Salton & McGill, 1979) and because prior applications of information technology have led to the construction of the ten digit ISBN number, which for some purposes completely describes a book. On the other hand, an electronic market for photographs, for which no conventional product descriptors or search techniques exist, would be less likely to develop.

Product complexity may interact with network attributes. Text descriptions may limit the complexity of product descriptions on low-bandwidth telecommunications networks, while graphical browsing may allow richer product descriptions on high bandwidth networks.

When product search is difficult, networks can provide additional value to the extent to which they facilitate searches. Thus one would expect that network facilities such as directory services, which make locating a large number of potential suppliers easier, and intelligent comparison programs would lead to more market-like relationships between suppliers and buyers.

Network infrastructure.

The nature of the network infrastructure can directly influence the type of relationships developed among firms. The openness of a network is perhaps the most important network attribute to consider. Networks are more open to the extent they allow easy communication with a new customer, supplier or other entity over the network. Networks become more open if they use public protocols already adopted by major equipment and software providers, if many individuals and firms already use the networks, and if the financial and behavioral costs of acquiring, deploying, and using the hardware and software needed to communicate over the network are low. The telephone network in the United States is prime example of an open network. Over ninety percent of individuals and almost all business have telephone service, the open protocols governing telephone interconnections means that interconnection is easy, and because of economies of scale and various subsidies, the costs of both terminals (i.e., phones) and monthly service is relatively cheap. The Teletel network in France and the Internet/World Wide Web network in the United States are examples of open data networks. Through the use of any of these networks, the cost to a customer to search for potential suppliers (via an 800 number call or a search service on the World Wide Web) is low. Conversely, the cost for a supplier to add an additional customer is also very low.

It is generally the case that large firms most aggressively take advantage of innovations (Moch & Morse, 1977), including interorganizational electronic networks (Streeter, Kraut, & Lucas, In press). One consequence of the low cost of open networks, however, is that the advantages of both electronic brokerage and electronic integration may be extended to small firm and even individuals.

Open telecommunications networks are a prerequisite for market-like relationships between suppliers and buyers. Both multiple suppliers and multiple buyers need to be joined, and this can happen only if standards or other network attributes permit interconnection by multiple parties, without substantial network-specific investment on the part of either buyers or sellers. Thus one would expect more market-like relationships between buyers and sellers when the communication networks they use are more open. While open telecommunication networks are a prerequisite for electronic markets, they do not guarantee it. Simply by encrypting some data or by allowing only individuals with passwords to access a service, one can convert an open application to a proprietary one.

Business environment and network control.

We have argued that open networks are a prerequisite for using external networks to achieve electronic markets. But whether firms will agree to subscribe to industry-wide standards, to have competitors join trading consortia, or to make their data and other internal processes open to a large number of outsiders depends in part on the entities that control the networks and on the industry structure. Networks run by third parties who are neither buyers nor sellers on the network are more likely to lead to market-like relationships between buyers and sellers. Since the third party's return is based on the volume of network transactions, the network provider has little reason to exclude either buyer or seller. On the other hand, on a seller controlled network, sellers will be motivated to exclude other sellers in order to retain customer loyalty (e.g., American Hospital Supply). Similarly, on a buyer controlled network, the buyer may be motivated to exclude other buyers to capture the benefits from efficiency that the network provides (e.g., General Motors's EDI system). Moreover, under some circumstances, buyers may wish to exclude some sellers to reduce their search costs and to convince a restricted number of sellers to engage in relationship-specific investments (Bakos & Brynjolffson, 1992).

Network control will have its effects primarily when the power relationships among potential sellers, among potential buyers, and between buyers and sellers are unequal. If one seller dominates a market, then the seller will perceive opening their networks as threatening its market share. On the other hand, if no seller has dominant market share, then the network externalities associated with increasing the size of the market overall are likely to prevail. In particular, since having a mass of small and roughly equal sellers on the network would encourage additional buyers to use the network, these opportunities to increase the customer base would encourage sellers to tolerate other sellers on the network. Similarly, if no one buyer has enough power to dominate an industry or dominate a network of suppliers, then the benefits of network externalities will overwhelm the disadvantages of sharing efficiency gains (e.g., mass market customers). These benefits include both the increased numbers of suppliers to choose from, and the cost reduction in using the network resulting from economies of scale, greater reliability and ease of use of the network, etc.

Network control and the business context will also influence the evolution of services on a network. In a seller-initiated network, when one seller dominates a market, not only will any efforts to open the network to other sellers be resisted, but a service evolution strategy that progressively locks in buyers by raising switching costs will be followed. To implement this strategy, the supplier may provide the buyer with value-added information such as purchase histories or other useful management information. Such a service evolution strategy will further result in more tightly coupled, hierarchical relations between the seller and its buyers. If buyers are very numerous, network services will increasingly be used both to reduce sellers' share of transaction costs and to enhance revenue. The seller's costs can be reduced by imposing on the buyer both labor costs (e.g., having buyers key in orders) and capital and communications costs (e. g., using buyers' terminals and requiring them to pay their own telecommunications charges) for the seller's. The more numerous (i.e. mass market) the buyers, the greater will be the tendency to derive revenue directly from network service provision itself, mostly by charging communications and information service usage fees to buyers. However, in cases where networks become highly successful, sellers may chose to allow in competitors on a fee basis, as in the SABRE system of American Airlines. In cases where no one seller is dominant, and buyers are numerous, seller provided network services will mainly focus on increasing the share of basic transactions that are conducted electronically. To the degree that an electronic market develops, new services providing a richer set of functions will be offered, often for a fee. However, these features will not be as custom tailored to the needs of any particular buyer as in the case of a seller network that is attempting to lock in buyers.

Empirical Investigations of Electronic Networks and Interorganizational Relationships

The previous section reviewed many of the predictions that a transaction cost perspective would make about the impact of interorganizational electronic networks on relationships among firms, and suggested some technological and business-environment issues that must be considered in specifying predictions. While Malone and his colleagues (1987) predicted that in the steady state, greater use of interorganizational networks would lead to more market-like relationships among firms, the empirical evidence reviewed below does not support this expectation. At this time it is not clear whether this failure of prediction results from 1) insufficient openness of the telecommunications networks, 2) business environments in which interorganizational networks have been first deployed, or 3) a misspecification of the electronic brokerage versus integration gains afforded by the networks. To date, the empirical evidence has been too gross to identify the mechanisms through which the use of interorganizational networks is affecting the relationships among buyers and sellers.

Most research on interorganizational networks has used case studies of specific private network-based systems to examine the conditions under which firms use external data networks and the strategic advantages they seek to achieve. Well known examples include McKesson's use of their network to link to independent pharmacies and American Hospital Supply's use of their network to take orders from hospitals (Keen, 1988; Malone & Rockart, 1993). In both of these cases, the systems were interorganizational, but were not open to other suppliers. The networks and the application were proprietary and under the control of the supplier, with private networks used to provide the infrastructure. The networks were primarily put into place with existing trading partners, and were not accessible by those not included in the system. Particularly in the case of McKesson, which extended the applications supported by their system from simple order entry to include such value-added services as insurance claims processing for pharmacists, the use of the network more tightly bound customers to the supplying firm and enhanced an already existing hierarchical relationship.

A multi-organization study by Brousseau (1990) also concluded that the modal use of business-to-business external networks has been to support electronic hierarchies, not electronic markets. Brousseau reviewed 26 inter-organizational networks, finding that most were used to reduce production or distribution costs and served to reinforce already existing hierarchical relationships among firms. Only in two–the petroleum business and textiles–was the use of inter-organizational networks associated with buyers gaining advantage by having more suppliers from which to choose. Taken collectively, previous case study research illustrates that interfirm networks are often used to support electronic hierarchies, rather than electronic marketplaces.

One hypothesis explaining the prevalence of electronic hierarchies is that the high cost of extending private networks to trading partners encourages linkages only when a meaningful volume of transactions is expected to occur. The firm controlling the network application is further encouraged to add new types of transactions to fully load the network, justifying the dedicated capacity to each trading partner. We would thus anticipate that where data networks are proprietary and fragmented, as in the United States before the explosive growth of the Internet, electronic hierarchies would be more common than electronic marketplaces. This further suggests that when public and ubiquitous data networks are available, the same constraints on connecting low volume trading partners, or potential partners for whom anticipated volume is uncertain, does not exist. Under such conditions, electronic marketplaces, where more fleeting trading relationships can economically be supported by network-based transactions, should be more likely. It is useful, then, to explore the outcome of earlier experiences with ubiquitous data networks, in order to better understand the dynamics of electronic commerce on the Internet and other future electronic commerce platforms.

Open Networks as a Platform for Electronic Commerce

Before the opening of the Internet to commercial and non-academic traffic, public data networks in the United States were mainly offered by on-line information service providers such as CompuServe. At one point CompuServe even marketed a business information service, enabling firms to use CompuServe's extensive network for electronic mail and the provision of company data to employees and others to whom a company wished to provide access.

Outside of the United States, one of the oldest examples of a truly ubiquitous, open, public data network is the Teletel network created by France Telecom starting in 1982 (Lucas, Levecq, Kraut, & Streeter, 1995). Teletel consists of the widespread deployment of the Minitel data terminal, an electronic telephone directory, and almost 25,000 other information and communication services and business applications, used by businesses and residential consumers. In 1995, there were over 6.5 million Minitel terminals in France with a growth rate of approximately 10% per year. About 20% of French households have Minitels and about 80% of businesses have at least one Minitel. Altogether about 40% of the non-retired French population had access to Teletel either at work or at home in 1992 (France Telecom, 1992). It thus has the basic characteristics to function as an electronic commerce platform, and has supported the emergence of an entirely new electronic information services industry in France, despite the comparatively low penetration of home computers.

Case studies conducted on firms using the Teletel and other broadly available electronic service networks as business networking platforms (i.e. where the network infrastructure itself is not under the control of the firm offering the application) illustrate the motivations of suppliers when they develop applications for electronic commerce. Steinfield, working with several collaborators has conducted qualitative case studies of the way firms in the United States and France use inter-organizational videotex networks (Steinfield & Caby, 1993; Steinfield, Caby, & Vialle, 1993). These studies have documented the ways in which the openness provided by the network in large measure determines the kinds of applications that can be deployed, but not the motivation for deploying them. Unlike internal data networks, which are primarily used to cut operational costs, external networks allow firms to create new differentiating services or revenue-producing products. The ability to use networks for revenue production increases as the networks become more ubiquitous. Surprisingly, however, the modal use of even open networks is to develop hierarchical, long-term relationships among firms. The low cost of the open networks allows these hierarchical relationships to extend to even tiny business partners. It is primarily, however, when the open networks extend to the end consumer that these networks are used to support electronic markets.

The case of a large US manufacturer illustrates the use of an external network with a limited subscriber base to offer EDI-like capabilities, resulting in cost savings plus an ability to differentiate its service from those of other manufacturers (Steinfield & Caby, 1993). This manufacturer initially used CompuServe to communicate with its mobile sales force. Once its independent dealers saw sales people communicating by electronic mail, they requested access to the system, and soon two thirds of the seventy or so dealers had accounts. The marketing staff, realizing that they now had an electronic link to their immediate customer base, deployed applications that provided product and sales-support information directly to dealers. Essentially, private company data was made accessible to dealers. The system was expanded to facilitate completion of transactions as well, enabling dealers to order new trucks on-line. Advantages to dealers included fewer errors in ordering, the ability to check on delivery dates, and the more rapid response from marketing and sales support personnel. In line with our propositions above, the service, developed and under the control of the manufacturer, was clearly used to strengthen preexisting ties to the dealers, rather than to permit currently unaffiliated dealers to order products. Hence, the ubiquity was merely among a “closed user group” rather than among the set of all truck dealers. As suggested by such authors as Cash & Konsynski (1985), the network raised switching costs, rather than lowering them, despite the fact that CompuServe's network was widely available and could have easily supported “switched” services (i.e. a dealer dialing up another “service”).

Once networks become more ubiquitous, companies can use them to create revenue producing products, new lines of business, and even new markets. A central point is that with these widely available data networks, service providers can assume a large potential customer base, with whom it pays to have transactions even on an infrequent and casual basis. In France, the Teletel videotex system provides such ubiquity. But, even so, it does not transform business trading relationships into electronic marketplaces. On the other hand, case studies again illustrate the role that such networks play in enabling suppliers to reinforce existing trading relationships, capture new business, and increase the costs their customers incur if they wish to switch to new suppliers.

Steinfield, Caby, & Vialle (1993) describe a large, multinational, electrical appliance and consumer electronics manufacturer that used Teletel for EDI-like connections to approximately 10,000 separate retailers and independent repair people throughout France. The after-sales service subsidiary of this manufacturer provided replacement parts and training to its widely dispersed customer base. The Teletel system permitted electronic transactions even with the smallest trading partners. By permitting on-line ordering, coupled with courier service for rapid delivery, the firm was able to eliminate regional parts warehouses, and reduce the average time for repairing defective appliances from two weeks to two days. This savings occurred because repair people used to wait until they had a sufficient need for parts before driving to a regional warehouse. After the system, they essentially used the equivalent of a “just-in-time” stocking practice for replacement parts. Moving to a centralized warehouse near Paris reduced the need for replicated inventories, and extra personnel around the country and created substantial savings. Moreover, repair people were further bound to the supplying firm after the introduction of a revenue producing expert system-based training application. Technicians connected to the expert system which proceeded to ask a series of questions designed to diagnose the fault and indicate the repairs needed. This “just-in-time” training service meant that repair people did not need expensive and lengthy in-person training, a difficult task given the short life cycle of new electronics products. They were charged a fee for connecting to the service, but it clearly helped to further promote their dependence on the supplying firm. The expert system also accumulated data on repair problems and provided feedback to the design and manufacturing divisions of the company in order to help detect and correct potential structural flaws in their products. As in the earlier examples, a primary motivation for this service was to prevent repair people from obtaining parts and services from other suppliers. Ubiquity merely made it possible for a supplier to manage relationships with a very large set of buyers, but did not drive the supplier to open their service to other suppliers.

Other case examples from Steinfield et al., (1993) show similar patterns of use of electronic service networks. In one, a clothing manufacturer specializing in men's suits used the Teletel network to link to boutiques and other clothing retailers. They created an application that allows a sales person to input a customer's measurements electronically using their Minitel terminal. The measurements are sent to a computer-controlled cutting machine at the factory, and the equivalent of a custom-made suit is then produced. Hence, retailers could offer a specialty product heretofore only available from tailors, with more rapid turnaround. At the same time, of course, the application was designed to offer retailers a wide range of standard-sized suits produced by this manufacturer. The custom-made suit application helped to attract retailers into a trading relationship, which then was broadened to include many other product lines. This application, as the others noted above, did not offer boutiques any possibility of acquiring products or services from other clothing manufacturers.

Were there examples of electronic marketplaces on the Teletel system? Indeed there were, although case study techniques do not tell us anything about their prevalence. Perhaps the most famous electronic marketplace using the Teletel network is the service created by Lamy, a French firm that publishes various industry reference guides (Steinfield et al, 1993). Lamy originally published the government tariff regulations for the French trucking industry. They then used the Teletel network to provide an application that assisted truckers in calculating the appropriate tariffs for carrying freight between French destinations. After the deregulation of the trucking industry tariffs in France, Lamy used their relationship with truckers to offer a new service that created a spot market in the industry. Freight forwarders subscribed to the service, entitling them to post offers in real time for any immediate hauling needs they had. Truckers could then dial up the service at any time, input their originating and destination points and available capacity, and receive all matching offers. Truckers then contacted the freight forwarder to arrange to transport the goods. In this way, truckers could fill up excess capacity that remained during a normally scheduled trip, or fill up their trucks for a return trip home. France's trucking industry became the most efficient in Europe following the industry of this service.

Lamy essentially played the role of the market-maker in this example. They were not themselves in the freight-forwarding or transport business, but brought the two together to create a real time exchange. The usage was impressive - with France's 25,000 independent truckers regularly searching the database for a small fee. Posted offers arrived constantly; an informant reported during the interview for the case study, that dozens of offers arrived in a several minute period. Lamy extracted a fee for their market-making service, however, based on the pay-per-call charges to truckers and the subscription fees to freight forwarders. Their incentive was to attract as many freight forwarders and truckers as possible. Therefore, as proposed above, an electronic marketplace emerged only with the help of a third party player who functioned as an exchange.

One quantitative study of interorganizational networks in France and the United States provides more compelling evidence that even open networks are typically used to support hierarchical relationships among firms (Streeter, Kraut, Lucas & Caby, In press). Their survey of over 600 firms in the two countries compared France, with its nationally interconnected and relatively ubiquitous Teletel network, to the United States, with its collection of private data networks, most of which were inaccessible to the general public.5 The research showed that firms that extensively used data networks connecting them to trading partners were more efficient in order processing. Results also showed that compared to small firms, larger firms, (i.e., ones with greater sales and more employees), are more likely to use the data networks for a wide variety of corporate communication, even in France. The survey measured use of any form of data network; not just the Teletel. However, among firms in France that used Teletel for their business applications, the association of firm size and network use disappeared. This result suggests that ubiquity makes the benefits of networking available even to smaller firms.

Both in France and the United States firms seemed to use the interorganizational data networks to bind customer and supplier firms. In support of this, the survey found that the more firms used external data networks of any sort, the more stable and long-lived were their the trading relationships with their customers and the more frequent were their transactions with them. So, even though the Teletel network made connections to low volume trading partners financially practical, it did not transform the relationships among trading partners into electronic marketplaces with short-lived relationships among firms. Rather, the quantitative evidence suggests that open networks reinforced existing trading patterns. One important exception to this pattern was that while the electronic hierarchy structure was prevalent with business trading partners, firms using Teletel were more likely to pursue a mass market strategy with the general public.

Discussion and Conclusions

Our review of the empirical literature shows that both electronic hierarchies and markets have been observed in practice, but the former are, in fact, more commonly observed in business to business networks. In general, the more extensively firms used inter-organizational networks, the more hierarchical were their relationships with trading partners, even when using highly open and ubiquitous public data network infrastructures. The relative openness of the network did not appear to predispose firms to relate to their business trading partners on a transaction-by-transaction basis, although the reduced cost of adding new partners did appear to make it easier for electronic service providers to include trading partners that generate low volumes of transactions. Opening the network even extended to individual customers. The case study findings also were consistent with our expectation that third party market makers, who provide such brokerage services as product search, comparison, and evaluation, are likely to be necessary for real electronic marketplaces to evolve.

Although the survey and case study findings presented are broadly consistent with the expectations developed earlier in the paper, they cannot distinguish between two competing explanations for the prevalence of electronic hierarchies over markets. If the need for third parties to set up an exchange were the only reason, then why aren't there many more examples like Lamy?

The economic explanation for the prevalence of electronic hierarchies is that the benefits derived from heightened integration of production activities outweighs the advantages the network might afford in finding new sources of supply. The survey and case studies tell us little about the specific costs firms incur in their transactions – costs of search and advertising, trade execution, settlement, monitoring, or service after the trade. We do not know if other suppliers were available, and if there is sufficient variability in price, quality, or service to make search for new suppliers worthwhile. We do not know if acquired goods were standard, and therefore obtainable elsewhere, or customized, and asset specific. To see where electronic networks provide benefit, and whether that benefit is greater in reducing search costs (hence encouraging electronic markets) or integration costs (hence encouraging electronic hierarchies), the unit of analysis must be shifted to the level of the transaction itself. Only then can the costs and benefits of using a network in support of transactions with external trading partners be examined, and a comparison of electronic search versus integration be made.

Such an approach will also permit us to contrast the above economic explanation with one based more on sociological factors. The empirical relationship frequently observed between the use of external data networks and electronic hierarchies may result from firms opening their networks only to those with which they already have a long history of successful commercial transactions, emphasizing the importance of interpersonal relations and organizational trust (see Zucker, 1986).6Mansell & Jenkins (1992) have found that even in industries with EDI standards, participation was generally confined to pre-existing business relationships. Granovetter (1985) has provided a compelling argument for expecting social relationships to form the basis for much economic activity. This view is consistent with the recent competitive strategy and organization theory literature that emphasizes cooperative relations among firms in a value chain, network, or cluster (Johnston & Lawrence, 1988; Powell, 1990). By empirically examining transactions among firms, noting which are electronic and what the relational characteristics of the trading partners are, this source of explanation can be tested.

Given these limitations, and the directions for future research, does our review have some implications for the future of electronic commerce on the Internet? The Internet permits information rich communication, is open, is based on a set of common standards, and is increasingly ubiquitous. With the soaring popularity of the Internet, which is considered by many industry observers to be the most viable national (and global) information infrastructure, the U.S. may soon have the open, ubiquitous network ideal for electronically mediated commercial transactions. Fledging attempts at creating electronic marketplaces at the business-to-business level are in evidence, such as CommerceNet. Many electronic shopping malls have appeared as well, although these seem to be targeting the end consumer rather than the upstream business market.

Our review suggests that network attributes - in particular, openness - are important, but do not operate in a vacuum. Other factors, such as the locus of control of the electronic service application and the business environment, are likely to interact with the use of the Internet to promote both hierarchical and market-like structures. Given the Teletel experiences, we should see many supplier organizations developing Internet applications that promote electronic transactions and information flows among trading partners. If previous trends continue, at the business-to-business level, these will most likely map onto existing trading relationships, capitalizing on the benefits of electronic integration. The logistics of dealing with more trading partners at this intermediate level will likely preclude market-like arrangements. However, as with Teletel, many firms will exploit the reach of the network to find new customers, and will be encouraged to move to a mass market strategy if they are not already doing so. Our review suggests, however, that true electronic markets, permitting the easy comparison and acquisition of products and services across many suppliers on a spot market basis will be less common, and dependent upon market making intermediaries. The directory, search, and intelligent agent services appearing now on the World Wide Web are evidence of this need (see Sarkar, Butler, and Steinfield, this volume).

Footnotes

References