This chapter examines the economic characteristics of e-mail service and the regulatory and policy issues that are raised by universal access. Economic factors strongly condition the unregulated development of communication services. Understanding their influence and direction is essential in determining the potential need for public intervention and formulating appropriate policies to promote effective, widespread e-mail service.
Our approach is to examine e-mail as part of a larger communications market. For this purpose, e-mail is considered to be a service that allows any location with a valid e-mail address to exchange information, in accordance with a common e-mail standard, with any other location with a valid e-mail address. The key elements defining e-mail are the set of reachable addresses and the permitted forms of information exchange; see also our definition of e-mail and universal access in Chapter Three.
The e-mail market is young, fast-changing, and growing rapidly. In the first section, we examine this market's basic characteristics, comparing them with more mature communications services and the universal service issues they have faced. In the next sections, we take up the particular demand and supply features of e-mail service. Although the e-mail market is evolving rapidly, we examine recent trends to venture tentative predictions of the emerging equilibrium, industry structure, and range of supply prices for basic service. This leads to considering potential market failures that may need to be addressed by policymakers.
To focus on the distinctive characteristics of e-mail, we first review the development of two communications technologies that have achieved nearly universal levels of service--the telephone and (within business) facsimile.
Alexander Graham Bell was granted his first patent in March 1876, and the first Bell Telephone Company was formed the next year. From this beginning, a worldwide communications system made up of public telephone networks has developed in four distinguishable phases.
Launching a New Service. In its initial years, local exchange businesses linked a community of business subscribers within small central-city areas. From a single telephone, only subscribers to that local network were reachable. Network operators competed side by side for subscribers, and businesses with broad communities of interest had to maintain separate telephones for each network to reach most of their correspondents.
Widening the Network. Local telephone exchanges began to interconnect with distant cities, extending their reach by long- distance lines. Competition between local networks disappeared as exchanges were merged or acquired by AT&T.
Deepening the Network. With nationwide connectivity achieved, federal policy (in the Communications Act of 1934) proclaimed the goal of universal service and telephone companies emphasized expanding the number of subscribers and extending service to high- cost and remote areas.[2] Monopoly supply, regulated by government commissions, guided developments. Policies, such as geographic averaging of rates and the residual pricing of the monthly charge paid by residential consumers, were used to promote universal service.
Restructuring for Economic Efficiency. Competition policy led to the breaking up of AT&T into separate long-distance and local businesses in 1984. Price structures that favored broad customer classes were replaced by programs to assist marginal subscribers, and barriers to entry of competitors and new technologies were dismantled.
The costs of telephone service in the unified Bell system were almost entirely in the network infrastructure, the distribution cables and local switching facilities in particular. With these large fixed costs and other large network costs common to supplying both business and residence service, the industry had many elements of a natural monopoly--wherein the costs of production by a single firm fall below the total costs of two or more firms. At the same time, customers placed widely different values on obtaining service. These factors together allowed telephone companies to charge systematically higher prices for business service and for long-distance calls to keep basic residential rates low. For decades this practice was the cornerstone of the broadly successful universal service policy guiding the operation of the telephone network.
Modern facsimile service provides a different example of a widely used communications service that is nearly universal in the business sector.[3] Fax is a value-added service provided via the telephone infrastructure. No additional network infrastructure or network service is required. Instead, the service depends on users having compatible fax terminal equipment from which to transmit and receive images over regular telephone lines. With the telephone network in place, the development of widespread fax service depends only on acquisition of fax terminals by consumers and standardization of manufacturing protocols. The importance of the telephone network as a critical factor in the success of fax cannot be overstated. Alexander Bain invented the fax machine in 1843, one year before telegraph services were deployed, but the service was not widely used. According to one observer: "Since the phone wasn't invented until 33 years later, the system proved to have limited appeal" (EMMS, 1994, p. 3).
Common protocol usage has evolved through market experience. Standardization, while assisted by international technical organizations, is not regulated. Elements of the fax service continue to evolve. Terminals and protocols provide for higher- speed and finer-quality image transmission, while maintaining compatibility with the most widely used basic standards. Service providers develop mailbox and broadcast capabilities. Manufacturers have added elemental scanning and transmission capabilities to personal computers, printers, and wireless notebooks. These developments are simultaneously extending the fax network's market penetration and broadening its usefulness and value to consumers.
Both telephone and fax are also platforms for additional services. The telephone system, under development for more than 100 years, now supports data transmission, fax, and switched video. The fax network has also expanded, from user-attended bilateral exchange of single sheets to unattended document exchange and broadcast distribution. Fax can be transported over data networks as well as voice telephone networks.
What are some salient economic characteristics of e-mail?[4]
The addressing and message protocols of basic e-mail service are developing into a platform for additional communications services (see Chapter Three). E-mail "user-agent" programs can automate the search for and retrieval of information from remote databases. Besides point-to-point communication, e-mail is well suited to broadcast distribution of messages and electronic conferencing. Multimedia (audio, video) components can be included in, or appended to, e-mail messages.
The supply of a network communications service is compatible with several quite different organizations of the market. Public ownership and operation of the network has been the rule for postal services and was widely used outside the United States for telephone services until quite recently. In the United States, telephone service was begun by competing private companies, competition that eventually proved unsustainable. The emergent territorial monopoly structure, dominated by AT&T, realized the advantages of economies of scale and the balance of political power. Then in the 1970s and 1980s, technological changes, procompetitive public policies, growth in demand, and almost full penetration again made multiple networks possible. Today, the U.S. telecommunications sector is firmly on the path to open competition.
Technological and regulatory changes have driven the transformation from monopoly to competition. The cost structure of much of the pre-1970s telephone network had strong elements of natural monopoly. However, the existence of substantial economies of scale for producing telephone terminal equipment was doubtful and when that market was deregulated, the successful entry of a variety of suppliers marked the first step toward disintegration of the broader monopoly of network infrastructure.
Beginning in the 1950s, microwave radio proved a superior, lower- cost technology for long-distance transmission and greatly reduced the incumbent's scale and right-of-way advantages in intercity markets. By exploiting microwave technology, the newly formed Microwave Communications, Inc. (MCI) company became a powerful national competitor. More recently, fiber optics and digital switching have enabled new competitors to enter formerly monopoly markets. Competitive access providers such as Teleport and MFS connect large business locations in urban areas with the facilities of interexchange carriers, using fiber rings for high- speed data and telephone service. New wireless technologies, such as those envisioned by the newly licensed Personal Communications System (PCS) operators, and the development of telephony over cable television networks (described in Carroll, 1995) are likely to result in additional competition in the local exchange.
Procompetitive public policy developments have grown as technology has changed. Policy increasingly has recognized the economic benefits of competition when alternative communications technologies are available. As political power shifted from states to the federal government, the regulatory barriers facing new entrants have been lowered. Regulators have explicitly adopted competitive instruments in important cases. In 1995, the federal government conducted the first use of auctions to allocate radio frequency spectrum for personal communications systems.[5] Competitive procedures also have been used to encourage innovative approaches to common standards for advanced television equipment and protocols.
Telecommunications firms use many communication technologies to supply markets. Economies of scope and scale lead firms to integrate "vertically" in two separate dimensions--over two or more service layers (transport, switching, and provision of information services) and over hierarchical levels of network switching and control (local, long-distance, and international service). Similar economies provide incentives to integrate "horizontally" across geographic areas as well as across nonsubstitutable services such as cable TV distribution and switched-voice telephone services. The landscape of the local telecommunications sector is changing rapidly. Local and long- distance companies are racing to enter each others' markets, cable television operators are experimenting with switched telephone service and Internet access, and wireless suppliers are expanding into mobile data and fax messaging services (see Vogelsang and Mitchell, 1996).
Rules governing interconnection can affect the ways in which network service providers compete with one another. Consider, first, an entrant who obtains an interconnection agreement with the incumbent and supplies a technically comparable service. The incumbent and the entrant will offer the same reachability, and competition will likely focus on price and service quality (as measured, for example, by billing accuracy and refunds for billing errors). Next, consider a new supplier whose network is not connected to the incumbent's network. In this case, the entrant will have significantly less reach than the incumbent, and must offer a different product that will induce at least some consumers to buy its services in addition to the services of the incumbent. Producers of software that permits individuals with multimedia computers to talk to each other over the Internet fall in this category: Their customers can talk to each other but not to individuals who possess an ordinary phone. Although telephony over the Internet has begun to grow, it offers considerably less reach than the public switched telephone network. The main attraction of the software is that a few calls placed over the Internet (which charges most users a usage-independent fee) saves enough in phone charges to recover the cost of the software. In addition, the interface provided by computer-based telephony is preferred by some consumers to using the standard telephone.
In the case of e-mail and other services provided over virtual networks, some degree of competition is possible without interconnection. By subscribing to multiple networks, consumers can communicate with correspondents who are not members of the same e-mail system. In effect, users with multiple subscriptions can serve as e-mail gateways between unconnected systems. Where e-mail network providers offer service at low monthly fees and usage charges, these arrangements could persist in an equilibrium of less than fully interconnected systems.
It is clear that the terms and conditions of interconnection dictate the nature of competition and play a central role in the evaluation of the telecommunications market. Incumbent carriers initially argued that they could not supply the same quality of access to outsiders that they provide within their own network. Thus, MCI's first customers had to dial many extra digits to reach their supplier over the old Bell system, and it took over a decade and substantial investment in upgraded switching equipment to achieve equal access. Furthermore, after providing interconnection, incumbent carriers have often successfully charged high access prices that replace contributions lost to the entrants in the retail markets.
Does competition hinder universal service? In the telephone network, entry of new competitors has largely removed the opportunity to support low-price network access and subscription services by contributions from high-value services elsewhere in the network. If some services had made disproportionate contributions before competition had arrived, market entry would have forced a rebalancing of rates, lowering the prices of competitive services. The final effect could increase or decrease overall penetration.
Policies to promote universal telephone service have shifted from maintaining regulated monopolies and disproportionate contributions from some services to developing broader-based industry funds for use in targeted subsidy programs for defined classes of users.
To summarize this section: U.S. communications networks will be supplied by competing providers. Public policies to promote universal service, e-mail or otherwise, cannot be based on a monopoly model of a network.
In modern usage, universal telephone service has come to mean maximum penetration of the potential market of all subscribers.[6] As discussed in Chapter One, this is the meaning we use throughout this report. This goal is often attributed to congressional intent in enacting the 1934 Communications Act, whose purpose is to
make available, so far as possible, to all the people of the United States a rapid, efficient, Nation-wide, and world- wide wire and radio communication service with adequate facilities at reasonable charges.
Although the words "universal service" do not appear in the act, since 1934 federal and state regulators increasingly invoked the goal of universal service to justify policies that restricted competition and encouraged price structures that promoted maximum penetration (Mueller, 1993).
Where does e-mail fit in this picture? E-mail connectivity for basic text messaging is growing rapidly (see the section, below) and the transport infrastructures on which it travels are effectively universal. E-mail penetration is at a middle level for business users and quite small for households. The state of e-mail penetration in the United States is discussed more fully in Chapter Two.
This section first discusses the major segments of the e-mail market. We then take up two central features of the demand for e- mail: the network externality that affects demand and the relationship (complementarity and substitutability) between e-mail and other messaging and communications services.
Small businesses, large businesses, and households use e-mail to satisfy a range of messaging and communications needs. Until recently, large and medium-sized businesses with local area network (LAN)-based computing environments used e-mail primarily as a tool for internal communications and placed a premium on reliability, security, and confidentiality. According to one study, about 70 percent of all e-mail remains within a given work group, 20 percent is between different company locations, and only 10 percent is between different companies or individuals (EMMS, 1994, p. 79). Households continue to use e-mail as an adjunct to information services and not as a ubiquitous communications service. The needs of the different market segments vary, as do their willingness to pay. These differences are clearly revealed by data on subscription rates and prices.
Traditionally, most households have obtained e-mail connectivity from an on-line service provider. The four major on-line services--GEnie, America Online, Prodigy, and CompuServe--offer an e-mail box as part of their basic packages. In addition, all four now exchange e-mail with the Internet. Worldwide on-line subscriptions for these providers are summarized in Table 4.1.[7]
Worldwide Subscriptions for Key E-Mail Providers
(in
millions)
____________________________________
As of
___________________
Provider 12/31/94 3/31/95
____________________________________
CompuServe 2.66 2.8
America Online 1.5 2.2
Prodigy 1.2 1.3
GEnie 0.1 0.1
____________________________________
SOURCE: MacWeek, July 10, 1995.
Business users subscribe to some of these e-mail services (particularly CompuServe). The number of residential users was estimated at about 2.7 million in 1993 (Winther, 1994). As of March 1995, it is estimated that there were 6.7 million consumer mailboxes, with CompuServe, America Online, and Prodigy being the only providers with more than 1 million mailboxes (EMMS, April 3, 1995, p. 3). Each of those three services currently charge $9.95/month for the first five hours of use, then $2.95 for each additional hour, with some discounts for frequent users (Clark, 1995).
Six major providers of e-mail services target businesses: GE Information Services, IBM Mail, SprintMail, MCIMail, AT&TMail and CompuServe. The worldwide annual revenue from e-mail (almost all from the companies listed above) is estimated to be about $568 million per year in 1993 and growing at 17 percent per year (Michel, 1994, p. 23).
The computed price for representative e-mail messages varies from a high of 27.5 cents per thousand characters (GE Information Services) to 2 cents per thousand characters (CompuServe) (EMMS, 1994, p. 183). The average mailbox generates about $20 per month (Michel, 1994, p. 24).
E-mail shares an important economic characteristic with a larger family of electronic messaging and communications services: Call and network externalities are an important determinant of demand (Mitchell and Vogelsang, 1991, pp. 55-61). Call externalities arise because a communication between two parties is typically initiated by one party, whose decision is guided by its private benefit. But benefits (which in some cases could be negative) are conferred (or inflicted) on the other party. In this case an individual's decision about using a communications service may not be consistent with maximum social welfare because the individual benefits of decisionmakers do not coincide with social benefits. Network externalities arise because an individual's decision to join a network is typically guided only by the benefit to that individual of communicating with others on the network. This private benefit differs from the social benefit, which includes the benefits others receive from being able to communicate with the individual in question. The traditional view is that these externalities will result in networks that are too small in comparison to the optimal (social welfare-maximizing) network (Mitchell and Vogelsang, 1991, p. 55). The lack of interconnection among early e-mail services (discussed below) likely resulted in virtual networks that were too small.
Until quite recently, neither the interoperability of LAN-based e- mail offered by different vendors nor the interconnectivity of wide area network (WAN)-based e-mail systems could be assumed. E- mail networks were disconnected islands, largely because important users did not conceive of e-mail as the basis for ubiquitous public switched telephone network (PSTN)-like connectivity. As a consequence, the addresses (or people) that could be reached by any individual depended on the address/service provider of that individual: CompuServe's customers could exchange e-mail with one another, but not with the customers of Prodigy.
Early interconnections among public e-mail providers resulted from a patchwork of bilateral agreements. One of the first was the link between MCIMail and CompuServe, established in February 1986 (EMMS, 1994, p. 207). In February 1993, the original proprietary interconnection protocol (MEP2) was replaced with an X.400 gateway, and CompuServe eliminated its surcharge for messages sent to MCIMail users. While details of interconnection continued to evolve, the fact of interconnection soon became unremarkable. In March 1993, according to one report: "AT&T EasyLink . . . interconnects with seven additional Administrative Management Domains (ADMDs), and it does not even bother to ask for press attention." (EMMS, 1994, p. 196). The proliferation of interconnection arrangements increases the value of e-mail to the network service providers' customers through an increase in the reach of their networks.
The e-mail market has continued moving from a fragmented to an integrated model. The primary force responsible for this change has been the growth of the Internet, which clearly demonstrated the benefits of the integrated model. The Internet concept originated independently at RAND (Baran et al., 1964) and at Bolt, Beranek and Newman (BBN) in the early 1960s, took seed as an Advanced Research Projects Agency (ARPA) project in the late 1960s, and transitioned to commercial status by the early 1990s. As mentioned in Chapter One, Lynch and Rose (1993) provide a brief history of the Internet.
The key feature of the Internet is openness. This is partly a matter of technological capability and partly a matter of business necessity. On the technical side, the Internet was designed to run over a variety of underlying network substrates and from the outset served an important "gateway function" (as described by Bunn and David, 1988) that allowed communication across incompatible lower-level protocols. On the business side, the relatively low cost of entry (discussed below) together with a liberal interconnection policy developed under the National Science Foundation's (NSF's) leadership ensured that no one firm could capture a large market share. As all network users were accustomed to full connectivity, at least among university sites that dominated the early Internet, competing Internet service providers (ISPs) could not disconnect from other ISPs and retain customers.
The interconnection of commercial e-mail services with the Internet was initially accomplished by Commercial Mail Relay (CMR), a gateway operated by the Information Sciences Institute in Marina del Rey, California. CMR interconnected the Internet with MCIMail, Tele-mail and DIALCOM (NNSC Report on the Internet, Section 5.20). Currently, as discussed in Chapter Three, the Internet has become the de facto means of interconnecting a variety of e-mail systems.
To summarize: E-mail services have moved quite rapidly to maximize connectivity. Interconnection to the Internet is now a requirement for any service provider who wishes to serve the mass market. Thus, the market has largely solved the problem of universal service, when it is defined as maximum reach/connectivity. As a consequence, consumers do not currently have to ask: "How many e-mail networks should I join?" Instead, they can ask: "Should I get e-mail service, and if so, which provider should I use?" This latter decision may still be influenced by network externalities, and hence the free market may form networks that are too small. It is also possible that future entrants will not be fully interconnected, and public policy may need to address the fact that universal service may require subscription to multiple networks. Thus, unregulated markets may on their own produce insufficient penetration. Below we address the policy issue of public interventions to extend service to additional users.
Electronic mail is part of a spectrum of complementary and competing communications and messaging services, but with some unique features. These services include telephone calls, faxes, voice mail, and letters (regular mail and courier services). The market for electronic services in this group is growing substantially.
Fax phone calls in the United States accounted for $11 billion[8] per year in 1993 and $25 billion worldwide (EMMS, 1994). In the United States, nearly 20 percent of all business telephone lines are connected to a fax machine. Enhanced fax services (Fax Mailbox, Automatic Retry, Never Busy Fax, and Fax Broadcasting) are small by comparison to basic fax calls and accounted for $15 million to small businesses and residences in 1993 (Winther, 1994).
Voice messaging services to small businesses and residences generated about $220 million in 1993 and are projected to generate about $1 billion in 1998 (Winther, 1994, p. 2).
There is some evidence that patterns of substitution may be unexpected. It might be thought that e-mail, voice mail and fax are close substitutes: All three are noninteractive, store-and- forward applications where real-time response is not critical. However, an early (1986) study cited by Huber (1987, p. 11.2) suggested that 55 percent of e-mail is diverted from the telephone, 10 percent from Telex, and 5 percent each from first class mail, courier services, and other electronic transmission techniques. The remaining 20 percent represents new demand. Whether these patterns are found in the current market is not known, but the growth of answering machines and voice mail suggest that telephone calls often turn out to be store-and-forward messages, for which e-mail might be an acceptable substitute.
While we did not discover any empirical studies that directly address the substitutability and/or complementarity of voice, e- mail, and fax services, commercial initiatives over the past few years suggest a strong complementarity between the services (Markus, Bikson, El-Shinnaway, and Soe, 1992; Sproull and Kiesler, 1991b; Bikson and Eveland, 1990; and Culnan and Markus, 1987). Traditionally, fax machines communicated with other fax machines, telephones communicated with other telephones, and e-mail users communicated with e-mail boxes. Each service had a reachability determined by the penetration rate of that service. Interoperability across services would increase the reachability of each service and increase the value of each service to every user. This is a natural consequence of the network and call externalities described above. Subsequently, several vendors have announced initiatives to provide integrated service. For example, AT&T and Lotus have formed a strategic alliance to integrate AT&T's INTUITY AUDIX voice messaging system with both cc:Mail and Lotus Notes. The system will contain a text-to-speech synthesizer to permit e-mail to be read to the user by a mechanical voice (Probe Research, 1994). Lotus is also developing an integrated voice/e-mail messaging system with Siemens; the joint effort will be based on Siemens' Phone-mail, Lotus Notes, and cc:Mail (EMMS, February 6, 1995, p. 4).
Gateways connecting fax and e-mail services have long been a staple feature of the high-end e-mail systems (such as SprintMail and MCIMail) aimed at business users. More recently, fax/e-mail gateways have been demonstrated on the Internet. Carl Malamud and Marshall Rose were responsible for the first implementation ( http://www.town.hall.org). A current version of their approach can be found at http://bond.edu.au/Fax/faq.html (Bond University in Australia). Numerous commercial implementations also exist, including InterFax and FAXiNET (http://www.awa.com/faxinet/).
Further off in the future is unified messaging, which uses a single mailbox for voice, fax, data, and video; manages all messages with common tools; and permits retrieval from the user's terminal of choice. Applied Voice Technology (AVT) has described its vision of unified messaging, which allows for seamless integration of the four major media (EMMS, 1994, p. 361). AVT currently sells products that offer limited integration of voice mail, fax, and e-mail through a single mailbox accessible through a graphic user interface (GUI) on a PC.
Developments in e-mail technology reinforce this trend toward convergence. The initial standard for Internet e-mail was the SMTP, which supports aliasing and mailing lists, both of which can be used to provide multicast service to the users. (See also Chapter Three.) Although SMTP limits the content of the message to simple American Standards for Information Interchange (ASCII) characters, this restriction may not be as prohibitive as it appears: uuencode and binhex formats, and Postscript files, can be included in SMTP messages. (These formats permit binary files including images to be sent as SMTP mail.) In addition, the MIME standard permits Internet mail to contain multimedia data including image, audio, and video.[9] Although not all gateways among e-mail systems currently support various encoding and MIME protocols, the SMTP and MIME protocols form an appropriate basis for universal e-mail systems supporting multimedia messages. See Chapter Three for more details on these protocols and standards.
In sum: E-mail has evolved from a limited role as an adjunct to information services for households and an extension of LAN-based intracorporate communications to a broadly accessible messaging and communications service that is more like the public telephone network. Technological developments point toward the convergence of previously distinct messaging and communications services into a closely integrated communications environment. The implications for universal service are twofold: Electronic messaging service will likely be supplied by several technologies, each capable of providing widely available, addressable message exchange. A universal service policy can be geared to a common-denominator service: plain text message exchange to and from the Internet (or possibly Web browsing as a basic standard) in the near to mid term, and access to a ubiquitous information infrastructure offering voice, video, and data services in the longer term.
E-mail is a communication and computer service provided over two large infrastructures that support many communications services and computer applications. The communications infrastructure includes the public switched telephone network, public data networks (such as the Internet), and cable television networks. Other services that use this infrastructure include voice, fax, and video communications and broadcast entertainment. The computer infrastructure that is used to support e-mail consists of many terminal devices, including individual personal computers and LANs connecting members of organizations. This infrastructure is also used for database, spreadsheet, and word processing applications and for computer games.
The two infrastructures vary in some important dimensions. Many users share large portions of the communications network, which spreads the cost of infrastructure investment. In contrast, many elements of the computing infrastructure do not possess this property, and the cost of an individual user's computer hardware and software is high. However, households use PCs for a variety of purposes (such as computing, communications, education, and entertainment), and users may find that the incremental cost of any one application is small.
E-mail, as it is currently constituted, also requires a service infrastructure. Individuals need to have mail stored for them when they are not on-line, they must have their outgoing messages correctly routed, and they need access to a directory of addresses. This infrastructure, which is supplied by the firms described above, is the focus of this section. Brief consideration is given to terminal devices, since their cost may be an important barrier to widespread penetration of e-mail.
Given the joint use of the communications, computer, and e-mail service infrastructures, the difference between the total cost of supporting all the communications and computing needs of an end user and the incremental costs of providing e-mail can be substantial. While the fixed cost associated with the communications and computing infrastructure is extremely large, the incremental cost of providing e-mail (the service infrastructure) may be relatively small.
The relative importance of fixed and variable costs of e-mail provision vary among service providers and vary with the type of customer being served. These variations are captured in some of the published data reviewed below.
In 1987, e-mail was used primarily by business customers. "While entering the office automation market would require heavy capital expenditures, adding software to provide an electronic mail capability in existing computer products would require only modest expenditures" (Huber, 1987, p. 11.1). Huber's study supports the notion that there are substantial economies of scope between e- mail provision and other computer-based services. It reports that a typical e-mail service bureau would have the cost structure shown in Table 4.2, at two different information system capacity levels.
Cost Breakdown in a Typical E-Mail Service Bureau
_____________________________________________
Message Volume at
_____________________________________________
25% of 100% of
Input Capacity Capacity
_____________________________________________
Computer hardware
and maintenance 59% 40%
Software 12% 8%
Packet networks 10% 22%
Sales/marketing 10% 21%
Local access lines 5% 2%
Support/other 3% 7%
____________________________________________
SOURCE: Huber (1987), Table EM.6, p. 11.16.
The computer hardware, access lines, underlying packet transport, and sales and marketing, which account for more than 85 percent of all service bureau costs, potentially can be shared with other services provided by that firm, generating scope economies for the e-mail provider. In the short run, the legacy of past investment and integration decisions will be an important factor in determining scope economies. Over time, the fundamental economies of scope will determine the optimal form of investment and the degree of integration. Huber points out that companies like GEISCO, that began as computer time-sharing firms, use the same hardware to provide e-mail and generate economies of scope through the efficient sharing of computer hardware. AT&TMail and MCIMail, which started out as e-mail services, used dedicated computer hardware and did not obtain these economies of scope at that time. Thus, different firms realized different degrees of scope economies. However, fundamental economies of scope (deriving from the joint use of hardware and other inputs) has led to greater integration over time. AT&T EasyLink, for example, currently offers a variety of enhanced messaging services, electronic data interchange (EDI) services and information services (Gonzalez, 1995, p. 179).
For a provider that owns the underlying network (e.g., MCI), the incremental cost of the packet transport would be close to zero if the packet network were underused, and the sunk cost would be substantial. For an unintegrated provider of e-mail (such as CompuServe) that leases facilities on which it built its packet transport network, the incremental cost of transport would be determined by the prices of leased lines and packet switches. To the extent that lines can be leased at large volume and term discounts, the cost of transport may look more like a fixed cost than a variable cost. In contrast, e-mail providers who purchase packet transport from value-added networks (VANs) will have a cost structure that reflects the price structure of packet transport. These prices have traditionally not offered steep volume and term discounts, and so e-mail providers in this category have a greater proportion of variable costs, compared with providers in the other two categories. This suggests that owners of facilities-based networks may have greater economies of scale (high fixed and low incremental costs) than owners of VANs, and VAN owners may have greater economies of scale than purchasers of packet transport. These cost advantages of facilities-based networks have led to expansion through greater vertical integration, often through acquisitions. For example, AT&T acquired the largest provider of e-mail services, EasyLink, from Western Union in 1991.
The economies of scope in marketing and sales costs arise when providers bundle distinct services into packages that are developed for specific market segments. For example:
In a number of announcements over the past two years, [network service providers] have begun to target new and potentially less technologically advanced users by offering simplified access to multiple services . . . . [T]he move will lead to new categories of services--of which mail services are just a part (Michel, 1994, p. 28).
A recent example of this approach is MCI's decision to offer members of its Friends & Family Program a free electronic mailbox (Wall Street Journal, November 11, 1994, p. B3).
In sum, traditional providers of e-mail appear to have considerable economies of scale and scope. These economies have consequences that are discussed below.
ISPs vary greatly in many dimensions, including target market served, area of coverage, and organizational form (nonprofit versus for profit). The costs of ISPs vary with their positions in each of these dimensions. There is some scattered information on the costs of providing Internet access (MacKie-Mason and Varian, 1993, and "Internet Costs . . . ," 1995).
BARRNet,[10] a regional, university-based, ISP located in the San Francisco Bay area, posted its budget for 1993 on its FTP server. Its expected cost allocation for 1993 was
Personnel (mainly for network management, consultants) 54% Equipment and maintenance 18% Office/general expenses 12% Backbone data circuits 16%
These costs reflect two important strategic decisions. First, its decision to offer services primarily in the Bay Area meant that its customers were quite close to one another. BARRNet did not need to purchase long backbone links to establish nodes near its customers. Second, BARRNet did not provide a great deal of customer support or market its services actively. It described its early business approach in the following terms:
BARRNet was conceived and implemented as a network of networks. It connects "sites" or "campuses" rather than individual computers. Our assumption has been generally that our member sites operate their own networks, and support their own users. BARRNet is then more a provider of "wholesale" network service than "retail" service . . . .
Thus, customer support was not a significant portion of BARRNet's expenses.
A broader examination of publicly available information on other large ISPs suggests that approximately 25 percent to 40 percent of the expenditures related to Internet access flow through to providers of underlying transport ("Internet Costs . . . ," 1995). There is little information on the relative shares of hardware, software, network management, customer support, general administrative expenses, and overhead. However, all of these major inputs can be shared by e-mail, newsgroups, browsing tools such as gopher and Netscape, and other Internet applications. In consequence, e-mail is rarely sold as a stand-alone service by ISPs.
There has been a consolidation of mid-sized regional ISPs such as BARRNet. Recently, in addition to its BARRNet purchase, Bolt, Beranek and Newman (which designed the first packet switches on the ARPAnet) acquired NEARNET (a regional provider based in Boston) and SURAnet (a regional provider covering the Southeastern United States). Although this development may be consistent with the presence of large economies of scale in the provision of Internet access, an explosion in the growth of agile, small resellers calls into question the advantages of size. According to Joel Maloff, these resellers represent the fastest growing segment of the ISP market. (Maloff Company, 1995). This latter development has been driven largely by the low entry costs. According to one recent entrant, Smoot Carl-Mitchell: "The total cost of the hardware to get into the business was less than $15,000 . . . . Maybe you should not let that out . . . or everybody and his dog will be setting up as Internet service providers in their kitchens." (Lewis, 1994a, p. D1). The article talks about a 15-year-old high school student who sells Internet access out of a spare room at home.
The simultaneous growth of large and small firms, and the vanishing middle, may initially be puzzling, as they seem to suggest opposite conclusions regarding underlying economies of scale. However, a more careful analysis of the record suggests that small entrants face major problems in scaling up their enterprises and may be viable only as small niche players. The Inet Access FAQ[11] by David Dennis (http://amazing.cinenet.net/faq.html) provides a partial description of the economics of entry by small providers: "Finally, if your only reason is to make money, you probably shouldn't do it. You'll be beaten out by those of us who love the net and who are willing to work utterly ridiculous hours to make our system a success."
The economic viability of this approach is not assured. As the FAQ points out, a reseller with a 56 KB[12] connection to the Internet can support 100 users and make a profit of $500 per month. Although a reseller with a T1 connection can support many more dial-up customers and be more profitable, the ability of the entrepreneur to deal with routine business activities will be strained.
Thus, although there appears to be rapid growth among resellers, there is reason to believe that the enthusiasm of net-lovers is not supported by the fundamental economics. Economies of scope and scale will favor larger providers with distinct cost advantages. As the Internet matures and the explosive rates of growth fall, ISPs will focus their marketing efforts on each others' customers, and economies of scale and marketing clout will count. The growth of small ISP resellers may well be a transitory phase. If bandwidth-intensive uses of e-mail (via MIME or other standards) become very popular, then facilities-based carriers may have a decisive cost advantage, depending on the equilibrium that emerges in the market for raw transport, to be discussed below.
In sum: There are considerable economies of scope at all levels of output, and there are considerable economies of scale. In the long run, the industry will move to a few large providers who offer packages of computer and other communications services. The potential oligopolistic structure, and the attendant dangers of inefficient pricing, may be a public policy concern. However, the continued opportunity for niche-player entry may temper this concern, and the potential role of niche players in introducing new products to the market may provide for dynamic efficiency. A continuing fringe of small and innovative players at the leading edge of the market may alleviate the concerns usually associated with relatively high degrees of concentration.
Chapter Three describes different devices that individuals use to send and receive e-mail. Personal computers, game machines, set- top boxes, wireless handheld access devices, screen-based telephones, and dedicated devices were considered and their advantages and disadvantages described. An important determinant of the appropriate choice of access device is the consumer's ability to use it for multiple purposes. Consumer valuation of a multipurpose device would be determined by the sum of the values placed on all uses of the device. A heterogeneous population may value a particular application, such as e-mail, in quite different ways, resulting in small adoption rates for special-purpose devices. However, the sum of values placed on a variety of uses is more likely to result in a greater proportion of the population with valuations above a critical value. This can combine with the economies of scope present in the use of major hardware and software components to result in deeper penetration rates for e- mail service.
A multipurpose device, such as a PC, may also be compatible with a greater variety of network substrates and support a greater variety of access modes and speeds. PCs are currently used over the phone network, cable TV networks, and wireless data networks, and significant economies of scale and cumulative production continue to drive prices down. Devices that run only over a single network (such as a wireless network) may never achieve the scale economies of PCs.
The history of computer and telecommunications terminals is one of rapid and nearly continuous cost reduction and expansion of features, driven by fast-paced technological improvements in semiconductors, fabrication and manufacturing, and economies of large-scale production. The markets supplying terminal equipment are highly competitive, ensuring that these advances are soon translated into lower prices and products with differentiated features to more closely match users' preferences. (But see the discussion in Chapter Three of reasons why PC prices may remain relatively high.)
Basic e-mail capability can be added to a general-purpose terminal at small incremental cost. The additional components needed in a PC are a modem and interface software; in a TV set-top controller, some interface software, plus possibly a keyboard and additional memory; in a fax machine, a keyboard and software. The upgrade needed for a voice telephone could be minimal, if voice messaging is handled centrally by a network, or it could require a display screen, keyboard, and software. The capabilities of different types of terminals will affect the costs and architectural requirements of the e-mail service provider. When a PC is the e- mail access device, it can readily provide message storage, filtering, and editing functions. Voice and portable data access devices, in contrast, will need to have these services supplied by a network service provider or possibly a supplementary piece of terminal equipment.
In sum: The costs of basic e-mail terminal capabilities are dropping rapidly, allowing a larger number of potential users to economically add e-mail features to equipment they may acquire for entertainment, computation, or other communications uses. Lower costs, combined with growth in network services and the rapidly growing network community that can be reached, will lead to rapidly rising e-mail penetration among households. Nevertheless, the costs of acquiring a basic terminal will remain a deterrent for some potential users with limited resources and for others who expect only small benefits from an e-mail connection. As mentioned in Chapter Three, the cost may be offset or covered by various advertising offerings, but this option remains quite speculative. Public actions to encourage universal access will need to address this barrier. We examine possible strategies below.
The Emerging Equilibrium Market Structure
This section reviews trends in messaging services and network infrastructure markets. We also identify important economic factors that are leading toward an equilibrium market structure composed of predominantly national, vertically integrated suppliers with resellers and niche suppliers occupying secondary roles.
Economies of scope in the provision of electronic messaging services suggest that providers may offer a range of messaging alternatives. These alternatives likely will be packaged in various ways to meet the needs of different market segments. This means that revenues from each customer will be relatively constant even if the mix of e-mail, fax, and voice mail used changes dramatically. Customers will benefit from less variable bills and the greater convenience and reachability of messaging services that interoperate.
The trend toward bundling messaging services may be strengthened by other fundamental economic forces in the communications industry. These include the large sunk costs and excess capacity in underlying transmission links, and oligopolistic competition among a few large companies that invest in the underlying physical communications infrastructure. AT&T, MCI, Sprint, and Wiltel each owns a national fiber optic network that connects all major cities. Each has substantial excess capacity. The cost to provide transmission links include high sunk costs of construction and the variable costs of digital cross connects, multiplexers, and the optoelectronics needed to operate the fiber. Providing switched service requires additional investments in switch hardware and software. In addition, firms incur costs associated with customer acquisition (marketing and sales), turning service on and off (provisioning, credit checks, setting up a billing account, etc.), maintaining and monitoring the network to ensure that customer expectations for service are met, terminating customers, and general administration.
The incremental cost of carrying traffic over the infrastructure network is zero, as long as there is excess transmission and switching capacity. During periods of congestion, when additional traffic is offered it imposes a social cost that can be measured by the degradation in quality of service. In traditional telephony, congestion can result in delay in getting dial tone, a fast busy signal, or a recorded announcement requesting that the user try placing the call later. Delayed dial tone arises when the originating switch is congested and the fast busy signal denotes the unavailability of trunks.
Reduced service quality can be measured by the delay imposed on the user. Economically efficient approaches to dealing with congestion for local calls are discussed in Park and Mitchell (1987) and Koschat, Srinagesh, and Uhler (1995). One finding of quantitative studies is that the benefits of peak-load pricing[13] are quite small when there is sufficient capacity in the network to keep blocking probabilities low.
The economic analysis of congestion in data networks is quite similar. Congestion at the packet level may result in slower throughput at the application layer, and this may not be perceptible to the consumer, just as a small delay in getting dial tone may be imperceptible. Congestion can be reduced by overprovisioning capacity, using real-time pricing mechanisms (MacKie-Mason and Varian, 1993), offering multiple qualities of service (Cocchi et al., 1993), or using peak-load prices (Park and Mitchell, 1987). The application of congestion pricing to data networks is a topic of ongoing research, and no definitive answers regarding optimal price structures are available. Present practice is to offer users free e-mail until a fairly generous limit is reached, at which point use charges begin to apply. Current prices are virtually unrelated to congestion. The widespread use of multimedia mail, which can more easily congest the network, may move price structures toward greater reliance on a mix of use charges and multiple grades of service (e.g., priority, standard, economy).
The history of leased-line prices in recent years reveals a strong downward trend. According to Business Week, private-line prices fell by 80 percent between 1989 and 1994, and this is consistent with competition that drives prices to marginal costs (Business Week, September 12, 1994, p. 90). During the same period, the use of term and volume discounts increased dramatically. AT&T offers customers a standard month-to-month tariff for T1 service and charges a nonrecurring fee, a fixed monthly fee, and a monthly rate per mile. Customers who are willing to sign a five-year contract and commit to spending $1 million per month are offered 57 percent off the standard month-to-month rates. Smaller discounts apply to customers who choose shorter terms and lower commitment volumes: A one-year term commitment to spend $2,000 per month obtains a discount of 18 percent.
The overall trend toward lower prices masks a more complex reality.
There are two types of tariffs: "front of the book" rates, which are paid by smaller and uninformed large customers, and "back of the book" rates, which are offered to the customers who are ready to defect to another carrier and to customers who know enough to ask for them. The "front of the book" rates continue their relentless 5 to 7 percent annual increases (Hills, 1995, p. 32).
In 1994, AT&T filed over 1,200 special ("back of the book") contracts, and MCI filed over 400.
Nonstandard commercial contracting is one means for sharing risk between the producer and consumers (Williamson, 1988, pp. 159- 161). In addition to risk reduction, long-term contracts reduce customer churn, which often ranges from 20 to 50 percent per year in competitive telecommunications markets. According to data submitted by AT&T, "consumer churn is running at the remarkable annual rate of 30 million" in 1995.[14] Since service activation and termination costs can be high, reduction of churn can be an effective cost saving measure.
Some tariffs for fast packet services filed by local exchange carriers offer term and volume discounts, and the economic forces that give rise to term/volume commitments for private lines have probably resulted in term/volume commitments for long distance fast packet services as well. There is little published information on long distance fast packet prices. According to one source, none of the long distance carriers or enhanced service providers (e.g., CompuServe) tariff their frame relay offerings (Toth, 1995, pp. 23-24).
The effect of term/volume commitments in private lines and fast packet services affects the cost structures of VANs that do not own their own transport infrastructure. It may be expected that large VANs who lease their transport infrastructures will sign multiyear contracts, possibly on an exclusive basis, with a single carrier. These VANs will have sunk costs. Competition among VANs will therefore be similar to competition at the lower level, and we may expect to see term/volume contracts emerge for messaging services offered by VANS, including Internet Service Providers. A quick survey of Internet providers shows this to be the case. For example, in January 1995, AlterNet offered customers with a T1 connection a 10 percent discount if they committed to a two-year term. Global Connect, an ISP in Virginia, offers customers an annual rate that is 10 times the monthly rate, amounting to a 17 percent discount for a one-year commitment. There are many other examples of this sort on the Internet.
The Internet is beginning to resemble the private-line market in one other aspect: Prices are increasingly being viewed as proprietary. ISPs that used to post prices on their FTP (file transfer protocol) or Web servers now ask potential customers to call for quotes. Presumably, prices are determined after negotiation.
To summarize: The market will move increasingly toward vertically integrated or allied firms, tied together by long-term, exclusive contracts, together with a competitive fringe of resellers/innovators who offer leading-edge solutions. The distinctions between e-mail, voice mail, and fax may not always be discernible to customers. Customers will face a variety of optional price structures and will be encouraged to purchase bundles of services with inducements to stay loyal (term commitments, frequent flyer tie-ins, cash back every quarter, etc.). Consumers without stable jobs, incomes, or housing, and therefore not qualified for special discounts tied to long-term contracts, may face relatively high prices.
As described in this and previous chapters, e-mail is supplied largely through conventional market arrangements that use well- established communications infrastructures to physically transport and route messages and administrative traffic. These infrastructures--the public switched telephone network, packet- switched data networks interconnected to the Internet, and cable television distribution systems--have nearly ubiquitous coverage across the United States.
Universal connectivity of e-mail systems, while not as far advanced, is within sight. User interest in widely reachable messaging has impelled commercial service providers to establish gateways and develop protocol conversion software to allow proprietary messaging systems to exchange plain-text e-mail messages. These forces will be sufficient to ensure quite widespread connectivity for basic messaging.
What remains to be achieved, then, is widespread subscription to and use of e-mail service. In the business sector, universal penetration will be approached when the messaging services provide high-quality support for financial transactions and sensitive commercial communications, which in turn require e-mail privacy and integrity and the ability to verify identity and authenticity of correspondents (see Chapter Three for a technical discussion of these issues).
The extent of residential penetration will depend to a great degree on quite different factors. Demand for e-mail is likely to be quite price-elastic for many consumers who have relatively good substitutes (postal mail and telephone service) for many messages. For other messaging, as emphasized in Chapter Five, such as participation in school and civic activities and interest-group discussions, e-mail may evolve into the dominant communications medium. Low first-time costs, especially for a user terminal, and low recurring charges for e-mail service could substantially expand residential penetration. Because most users may obtain e- mail service as a by-product of a multipurpose terminal device, the market success of set-top television devices and home computers will be an important determinant of penetration. Equally important may be new mass-market applications, such as financial transactions and home shopping.
To summarize: Many of the economic prerequisites for universal e- mail service are either in place or are developing rapidly through market processes. The principal gaps in achieving a universal standard of service are a limited number of issues concerning interoperability of e-mail systems and common addressing, as outlined in Chapter Three, and a program to provide financial support to marginal consumers who would not otherwise subscribe to an e-mail service. We turn to these matters in the next subsections.
Without public intervention, market forces are rapidly extending the availability of e-mail to broad segments of American business and consumers. Nevertheless, the market structure that is now developing raises concerns in the following areas.
Standards, Interoperability, and Interconnection. The convergence of e-mail, fax, and voice messaging may be hampered by difficult technical and business issues.
On the technical side, standards or gateways may be necessary to permit different systems to interoperate, at least for a core set of functionalities. The industry has taken various approaches to interoperability in different messaging markets. E-mail interoperability appears to be evolving rapidly through the use of gateways. A discussion of specific gateway programs that enable interoperability between LAN e-mail systems and the Internet is contained in Lipschutz (1995). Fax interoperability has been achieved through the relatively wide acceptance of a common set of standards (G3 and G4 Fax). In contrast, voice mail interoperability does not exist; with few exceptions, voice mail saved on one system cannot be forwarded to a voice mailbox on another system.
Interoperability of e-mail systems, fax machines, and e-mail/fax gateways appears to be developing in the market in response to competitive pressures. Voice mail interoperability has not developed in a like manner. However, recent announcements of integrated messaging may result in the development of a de facto standard. These solutions may have to compete with alternative approaches, such as MIME and X.400 e-mail standards. The market has not yet settled on a de facto standard for the converging electronic messaging industry.
As providers seek to differentiate their service, the market may settle on multiple standards and incompatible service definitions. Gateways between the different services may enable interoperation of only basic features (such as ASCII-based plain text mail). By subscribing to multiple services, businesses that rely on electronic messaging and some households may be able to communicate with others using advanced features all or most of the time. But consumers who can afford only one software interface or one service provider may find that advanced features often do not work, since most of their messages go through gateways that translate only core functionalities. Will these core functionalities provide capabilities that are envisioned by current definitions of universal e-mail? At this writing, the appropriate government role may be to wait and see what the market does.
Longer-run policy options cover a wide spectrum. At one end, the government may decide to adopt a laissez faire attitude. The argument for this option is that government regulations, and the proceedings necessary to develop and enforce them, cannot keep pace with rapid technological change in the computing and communications industry. At the other end of the spectrum, the government could establish requirements governing the degree of interoperability among e-mail networks. A light-handed regulatory approach could specify minimum capabilities that will be transferred across networks (perhaps plain text mail and some basic directory capabilities), while leaving providers free to offer optional enhancements to their customers.
In between these two extremes, the government could attempt to promote open standards by using its clout as a large purchaser. For example, the federal government could announce that its contract for electronic messaging services would be awarded to multiple firms (just as FTS2000 was awarded to two firms) and require all bidders to provide services that interoperate with each other for a specified set of functionalities. However, the federal government's record in setting standards that are then extended into general commercial practice is poor. The use of the ADA computer language and standardized protocol layers for computer communication have not been widely adopted. We recommend that instead of attempting to mandate economywide e-mail standards through government procurement, the more effective role is to facilitate cooperative approaches to standard setting, as federal agencies did with the development of high-definition television specifications.
Terms of Interconnection. The business issues that stand in the way of rapid convergence of different messaging technologies arise from difficulties in formulating the proper mix of competition and cooperation among competitive networks. Seamlessly interconnected networks are of greater value to consumers than fragmented or partially interconnected networks. Cooperative interconnection arrangements may therefore be attractive to all network providers. However, if networks are seamlessly interconnected, large network providers with substantial investments in international facilities may look no different to the consumer than a small local reseller of e-mail service. If the terms of interconnection favor the smaller network, the small reseller, with almost no sunk cost, may be able to consistently undersell the large provider from whom it obtains interconnectivity. But, if interconnection terms are too unfavorable, new entrants may not gain a toehold, and effective competition may not develop. The terms and conditions of interconnection present a difficult bargaining problem. A regulatory framework may be necessary to ensure that interconnection arrangements are fair to all competing networks. (The economics of interconnection and proposed regulatory frameworks are discussed in Arnbak et al., 1994.)
The convergence of electronic messaging systems may strain existing interconnection arrangements. Currently, a consumer who makes a long distance telephone call and then leaves a voice mail message will pay for the call. The price for the call typically varies with the distance, the time, and the duration of the call. A significant component of that price is the access charge paid by the long distance carrier to the local exchange carriers who originate and terminate the call. The framework of access charges is set by regulators as part of the terms and conditions of interconnection between long distance and local telephone companies.
E-mail (including e-mail that contains a voice message in digital form) escapes the access charge when the originator calls a local number to reach his e-mail provider and only local service charges apply. Even though the e-mail message is destined for a distant location, and may travel over facilities of the long distance carrier who carried the voice mail, the suppliers of e-mail, fax, and voice mail as well as other enhanced service providers (ESPs) are exempt from access charges. The exemption creates an artificial cost differential between traditional voice messaging (which imposes a high cost on the sender) and e-mail (which imposes a smaller, often zero, cost). The cost differential is based on an increasingly arbitrary distinction between interexchange carriers and ESPs. Ideally, a consumer's choice of messaging service should be based on economic costs and not on artificial regulatory distinctions; this is not the case today.
These distortions have several implications for universal service. In general, as messaging technologies become closer substitutes for each other, regulatory policies that systematically advantage a particular type of service should be closely examined. Pricing distortions are usually undesirable public policy, because they lead to higher-cost methods of production when consumers and product developers choose among alternatives using prices that are not based on costs. Pricing that advantages one messaging technology over another will artificially promote the development of the favored service. To be sure, such pricing distortions will promote more widespread use of the preferred services, and to that degree perhaps contribute to universal service for that type of messaging technology. However, universal telephone service is partially funded by toll and access services that contribute toward the sunk costs of network facilities. A policy that artificially depresses the price of e-mail (as the ESP exemption does) may indirectly weaken the funding of universal telephone service. This will have a negative effect on universal e-mail to the extent that e-mail rides over the telephone network: A household without telephone service will have fewer alternatives for access to its e-mail provider. The tradeoff between preserving universal telephone service and achieving universal e- mail service needs more careful analysis.
Number and Address Portability with Bundling. The trend toward bundling communications and messaging services highlights other inconsistencies in the regulatory treatment of traditional telephony and ESPs. Different rules govern number portability and address portability. "Equal access" requirements now in place for telephone carriers permit virtually all consumers to select a preferred long distance carrier while retaining "1+" dialing capability. Number portability extends to 800 numbers. Customers can retain their 800 number when they switch providers. Number portability is being considered for customers who are served by competing local exchange carriers; trials are now under way, and a request for proposal for the required database technology has been issued by a group that includes NYNEX, Rochester Tel, AT&T, MFS Intelenet, Teleport Communications Group, and Sprint (Vittoro, 1995). Direct inward dialing trunks and call forwarding are being used as interim solutions.
The rationale for number portability is to reduce customer switching costs and reduce the barriers facing potential entrants. Without number portability, businesses would be reluctant to switch carriers and change their number, as they would have to replace their existing stationery and their employees' business cards, change their advertisements, and undertake costly efforts to ensure that their customers and suppliers knew about the change. Residential customers may have smaller absolute switching costs, yet the costs might be substantial in comparison to their potential savings from changing providers. Switching costs can pose a significant barrier to competition. A firm seeking to enter the market must offer equal service at a sufficiently lower price to offset the potential customer's switching cost, or offer a sufficiently higher-quality product to obtain the business at the same price.
While telephone number portability is driven by a combination of industry efforts and regulatory requirements, there is no similar initiative to make e-mail addresses portable. Thus, a consumer who obtains e-mail from Prodigy might have an e-mail address of the form jane@prodigy.com. If the consumer switched to another provider (e.g., to PSI), her e-mail address would also switch (e.g., to jane@psi.com). Consumers can avoid this problem by purchasing their own "vanity" domain name (e.g., jane@joe.com), but this service costs between $100-$200 per year, which may exceed the switching costs for many consumers (Berlin and Kantor, 1995). Alternatively, a user can maintain a single e-mail box for all incoming correspondence and redirect that mail to a service- provider mailbox. (One service providing such mail redirection services is "pobox.com," described at http://www.tgc.com/70011.html.)
The consequences of no address portability can be larger than expected in a market where communications and messaging services are bundled. Consider, for example, MCI's offer of a free e-mail box to members of its Friends & Family program (Wall Street Journal, November 11, 1994, p. B3). Customers who join the program receive an MCI mailbox. If the address is of the form jane@mcimail.com, customers will not have a portable e-mail address. Since the e-mail account is bundled with the long-distance package, customers will incur a switching cost if they decide to change long distance providers: The free mailbox will no longer be available, and a monthly charge for e- mail will apply if the consumer continues to obtain e-mail service from the same provider. If the consumer decides to accept a package, including e-mail, from another provider, then switching costs associated with the new e-mail address will be incurred. The friction in the e-mail market will be felt in the voice market as well.
The implications of these supply conditions for universal e-mail are principally that public legislative and regulatory policy should adopt a broad view and aim for consistency across the entire range of communications and messaging services. In this way, resource costs, and not arbitrary regulations, will guide consumer choice and will permit the full benefits of competition to be realized.
If e-mail is to reach the vast majority of the population, marginal consumers--those who would not at current prices subscribe to an e-mail service--will need economic assistance to participate. Two approaches could be considered. The first would place requirements on service providers to offer below-cost service to consumers as a condition of doing business. The alternative approach is to provide funds directly to consumers with which they can purchase e-mail equipment and services.
The regulatory approach of imposing universal service obligations on service providers has been the prevailing practice in the telephone industry (OECD, 1991). As a condition of obtaining a franchise, or license, to supply local telephone service, a carrier is required by its state regulator to construct its network in all areas of its service territory and to offer basic services at standard tariffed rates to all consumers. In areas that are especially costly to serve (e.g., because they are remote, topographically difficult, or thinly populated), the costs of satisfying these obligations exceed the revenues the carrier can expect from the customers it gains. In other areas, rates for the basic access service to residential consumers may be set above the incremental cost of providing access and make a partial contribution toward recovery of fixed or sunk costs. Remaining costs are then recovered through higher contributions in other areas and for other services.
Such price structures, which result in disproportionate contributions from some customers or services, are sustainable so long as the supplier does not face competition. But when the profitable segments of the market are open to entry, other suppliers are able to capture the profitable business with lower prices or better service while avoiding the obligation of producing unremunerative service. With reduced revenues from profitable services, the original monopolist is less able to maintain services to marginal subscribers and high-cost areas. Ultimately, competition is incompatible with maintaining universal service by cross-subsidies.
This experience from the telephone industry indicates that a policy of promoting universal e-mail service that keeps basic subscription prices low by generating high contributions from other services will be infeasible in the openly competitive information services industry.[15] If an obligation to provide below-cost service were imposed on particular e-mail service suppliers, they would find themselves losing profitable markets to other firms that do not have the cost of those obligations. In principle, entry into the e-mail service provision business could be regulated and a service obligation imposed on each licensed entrant. However, regulating entry would go strongly against the dynamic of the innovative computer networking industry. Indeed, erecting successful barriers to entry appears improbable. The history of innovation in digital technology is one of an increasing number of alternative ways of representing and transmitting information. Bootleg messaging systems would likely circumvent any regulatory effort.
The alternative policy approach is to provide purchasing power directly to marginal consumers. A program of e-mail service vouchers would enable consumers to shop for terminal equipment, user training, and e-mail service in competitive markets. A voucher program would require several policy decisions about funding and eligibility. The program could be funded by a "tax" on a defined segment of the communications and information services industry, included as a surcharge in prices that e-mail service providers charge other firms, or from general revenues.
General revenues have been the source of funds for broad social welfare programs, including education and library services. Vouchers have been used in some public programs, notably food- stamps, to put purchasing power directly in the hands of marginal consumers. However, proposals to use vouchers more widely, for energy subsidies and choice of schools, have received only limited support.
What might the costs of basic e-mail service be? The costs of serving low-use telephone customers are instructive. According to AT&T, its "basic schedule rates do not cover the direct costs of serving the one-third of consumers who make under $3 per month in calls. These costs not only include AT&T's network costs, but also universal service costs of $.52 per customer and bill rendering costs ranging from $.33 to $.88 per customer."[17]
This suggests that an unregulated market may charge a minimum monthly fee above $3 for subscription to a basic e-mail package. Current prices appear to be in the range of $10 per month for a range of information services and e-mail, with monthly usage allowances that are quite generous. Individuals who make very limited use of e-mail and other communications/networking services (spending less than $5 per month) may not be able to find a provider who is willing to serve them. The amount of the subsidy necessary to induce the marginal consumer to subscribe (and thus be reachable by others) will depend on the difference between his willingness to pay for local and long distance service, e-mail, and other messaging options, and the price of a bare-bones package of these services. This subsidy could, conceivably, be higher or lower than current subsidies offered to telephone subscribers. If the additional value of e-mail to the marginal subscriber is greater than the incremental cost of providing e-mail service, the necessary subsidy will be smaller. If the additional value of e- mail is less than its incremental cost, the necessary subsidy will be greater.
The potential costs of subsidizing universal e-mail service are considerable. To obtain a sense of the magnitude that might be involved, consider a minimum-cost basic e-mail service that might cost $5 per month, or $60 per year, in competitive markets. Amortizing modest terminal and training costs over five years could add at least $40 per year. Thus, $100 per year would be a conservative estimate of the amount required to underwrite basic service for marginal consumers. In a mature industry, unsubsidized e-mail penetration could perhaps eventually approach the levels achieved in the telephone system. In 1994, about 94 percent of U.S. households had access to a telephone. If e-mail subsidies of $100/year were required for 10 percent of the approximately 100 million U.S. households, this would require an annual budget of $1 billion. (But recall the discussion in Chapter Three of possible "free" e-mail services and terminals becoming available through the bundling of advertising with services and devices, which might materially lower this amount.)
Eligibility for e-mail service subsidies would need to be tightly defined to avoid potentially large cost overruns or unfunded mandates. To the extent that marginal consumers will not purchase e-mail service because of limited resources, eligibility for subsidies should be defined in financial terms. To avoid establishing a separate administrative bureaucracy to determine eligibility, qualification for voucher assistance could be linked to eligibility for other public-assistance programs, although this could make the program subject to considerable variability across the states. To assist other citizens unlikely to subscribe on their own, outreach programs based in community organizations such as libraries and senior centers--exemplified by "wired community" efforts described in Chapter Five--could qualify for per-user subsidy payments. See Johnson (1988) for additional analysis of the costs of lifeline telephone programs per supported subscriber.
The universal-service policies considered here have been consciously confined to basic e-mail service. Multimedia messaging will undoubtedly grow in popularity and evolve in directions yet to be established. But designing policies to support standardization of such services and financing access to them are premature when those technologies are still being discovered.
Several overall conclusions emerge from this chapter's discussion.
[2]The notion of universal service has evolved with changing circumstances from maximum interconnectivity among independent network providers to maximum penetration of the potential market. This evolution is discussed further below.
[3]To date, only 6 percent of households have fax capabilities, a much lower penetration than the 27 percent of households with computers (Times Mirror, 1994).
[4]Cost characteristics and most demand characteristics are discussed separately below. This discussion concentrates on relevant general factors.
[5]This action finally introduced the economic efficiency advocated by Ronald Coase over 35 years ago (Coase, 1959).
[6]In the expansionary phase of local telephone systems at the turn of the century, Theodore Vail first used the term "universal service" with respect to the telephone network in AT&T's 1907 annual report. Vail articulated the goal more explicitly in 1910 as one of maximum connectivity--enabling a telephone user, by subscribing to one network, to reach as many other telephones as possible (Vail, 1910).
[7]For commercial e-mail accounts, monthly payments by subscribers presumably indicate active use or an option value of maintaining service. This is in contrast to noncommercial networks described in Chapter Five in which the number of active users may be considerably smaller than the number of subscriber accounts.
[8]Unless otherwise noted, the dollar amounts in this section represent gross revenue to service providers resulting from these specific services.
[9]MIME is compliant with both the SMTP and X.400 addressing standards.
[10]BARRNet was purchased by Bolt, Beranek, and Newman and is now known as BBN Planet. See Web site http://www.barrnet.net/ for details.
[11]FAQ is a common abbreviation for a file containing answers to frequently asked questions.
[12]"KB" is a unit of communications bandwidth representing thousands of bits per second. Fifty-six KB is the bandwidth of a voice grade telephone line.
[13]Peak-load pricing recognizes that investment in capacity is driven by the need to accommodate peak traffic, and that reductions in peak traffic can result in reduced capacity costs. Premium prices during peak periods are used to reduce peak traffic and realize capacity cost savings.
[14]Ex Parte Presentation in Support of AT&T's Motion for Reclassification as a Nondominant Carrier, CC Docket No. 79-252, April 20, 1995, p. 34.
[15]It is not clear how one would impose equal obligations on all providers. Suppose a company argued that some proportion of its customers were being served below cost, and it was meeting its obligations, and that the existence of unserved customers meant that the other providers were shirking their duty. A regulator would have to do a customer-by-customer analysis of all carriers to find out whether that was true. With only one provider, there can be no shifting of blame. With many providers, the information needed to find out who, if anyone, was shirking their obligation would be hard to get and easy to falsify.
[16]If e-mail subtracts business from "retail" telecommunications services, it may have to make more than a marginal contribution to network costs--becoming more expensive, which may in turn jeopardize universal service.
[17]Ex Parte Presentation in Support of AT&T's Motion for Reclassification as a Nondominant Carrier, CC Docket No. 79-252, April 20, 1995, p. 51.