TECHNOLOGY AND COSTS

Technology and cost characteristics can be observed in each of the complementary market segments: infrastructure capacity, Internet traffic, as well as Internet applica­tions and services.

2.4.1 Specialized Technologies Versus General Purpose Technologies (GPTs)

The Internet historically resided on the specialized network technologies that were in use to provide telecommunications services. Although fixed and wireless networks both exhibit economies of scale and scope as well as network effects, they differ in their unique cost structures. Historically, wireless networks exhausted economies at a lower scale of operations than fixed networks. Moreover, the regulatory framework influenced technol­ogy choices and the extent of scale economies. Networks built in an era of monopoly regulation were designed to reach minimal costs at a much higher scale than networks rolled out in an era of competition. As existing networks are gradually integrated into all-IP platforms and new IP-based networks are deployed, old and new networks become GPTs for Internet traffic. The transition from specialized traffic networks (ISDN voice and best-effort Internet) towards multi-purpose QoS Internet with active traffic manage­ment is a time-consuming process. While GPTs will handle the bulk of traffic, specialized networks built for specific services such as financial transactions will likely continue to coexist with multi-purpose technologies.

2.4.2 Economies of Scale, Scope, and Density

Important economic characteristics such as economies of scale, economies of scope, and economies of density are visible at the level of individual ISPs and firms in the higher layers.

2.4.2.1 Economies of scale and economies of scope on the supply side

Economics of scale are typical for the supply side of Internet markets, including markets for infrastructure capacity, Internet traffic (where multiplexing can achieve high econo­mies of scale), and Internet application services (e.g., software markets). Economies of scale prevail if a proportional increase of all input factors causes a disproportionally higher increase of output. In the single-product case economies of scale manifest them­selves in declining average costs. In digital markets such as software, declining average costs are mainly driven by the high development costs. This large intellectual property component of new digital products contributes to large upfront costs. Consequently, the average costs of production, which include the development costs, are then significantly higher than the marginal costs. Markets with this unique cost structure are typically char­acterized by high market concentration. Forms of price differentiation will be needed to recover total costs (Katz and Shapiro, 1998, p. 6).

The markets for Internet traffic consist of a combination of Internet logistics and infrastructure capacities. These traffic service markets can realize economies of scale and scope by bundling different types of traffic. In the past these cost characteristics of networks have been of particular relevance in local telecommunications access networks (local loops), which were regarded as ‘natural’ monopolies with typically only one active provider. After market liberalization an increasing number of formerly monopolistic market segments, such as long-distance and wireless communications, developed effective competition with several active network providers.

2.4.2.2 Network effects and positive externalities

Network effects exist if the utility of participating in a network is dependent on the number of users. They can be direct (e.g., the benefits of having more communication options in larger networks) or indirect (e.g., the benefits of the availability of more complementary goods and services in networks with a larger user base). Network effects can be positive and negative (e.g., the disutility of congestion). Positive network effects may also be characterized as economies of scale on the demand side: as an increasing number of users choose a technology its benefits increase. Suppliers often take such network effects into account when pricing their services. However, if interdependencies are not reflected in market transactions network externalities may prevail (Liebowitz and Margolis, 2003). Examples are positive effects of network size on innovation or negative effects on other users of insufficient protection against malware.

The theory of compatibility standards has been developed during the past decades, using standards competition as a metaphor for rivalry between competing technolo­gies (Liebowitz and Margolis, 1996, p. 290). Katz and Shapiro (1994, p. 100) point out that positive adoption externalities are not limited to communication networks but also affect hardware and software, These effects may be (fully or partly) internalized through network sponsorship, integration, or contracts. Taking into account heterogeneous pref­erences for alternative technologies, a trade-off between network externalities and variety has been considered in analogy to the traditional theory of monopolistic competition (Farrell and Saloner, 1986). Free entry within a dynamic world leads to an important additional dimension, namely, the search for new technologies.

2.4.3 Path Dependence and Critical Mass Phenomena

The analysis of the conditions under which technologies reach a critical mass of users and become viable has a long tradition in telecommunications (Rohlfs, 1974; Oren and Smith, 1981). A key question is whether such a critical mass of users can be reached via market strategies or whether public policy intervention is required. In evolutionary economics, the notion of path dependence refers to a situation in which earlier decisions constrain later decisions. Technology upgrade paths therefore may show signs of path dependence. In a more narrow sense, the question has been examined of whether an already implemented technology can prevail even if superior technologies become available. In this context a highly controversial debate has unfolded: Arthur (1989) and David (1985) argue and illus­trate with the historical example of the QWERTY keyboard that a specific standard may persist due to positive network effects and externalities, even if it is technically inferior.

Liebowitz and Margolis (1990) subsequently raised doubts whether specific histori­cal examples can ‘prove’ the survival of inferior technologies as a consequence of path dependence.

2.4.4 Standardization and Path-dependent Technology Choice

Many studies on compatibility standards focus on standards for single products. Within large technical systems such as the Internet, standards for complementary components and layers gain particular relevance. Considering the layered technical architecture of the Internet, a logical relation between standards can be established in which basic stand­ards form preconditions for more specific standards (Blankart and Knieps, 1993, p. 40). Applied to the Internet, the all-IP physical infrastructure layer consists of alternative network infrastructures (including mobile, fixed telecom, and cable access networks) (Knieps and Zenhausern, 2015). Based on such all-IP infrastructures, a variety of het­erogeneous standards for Internet traffic management have evolved in addition to the best-effort TCP/IP, which has historically been dominant. For applications and services, a proliferating variety of standards for content delivery, video streaming, VoIP, and so on, are in use and being developed.

Path-dependent technology choice can become relevant on all physical network infra­structures. Depending on the existing stock of sunk investment, businesses face decisions to upgrade established networks, build new network infrastructures, or lease access from other players (Bourreau and Dogan, 2006).

Path dependence is also relevant in markets for network services. Under competi­tive conditions consumers can choose between different providers of network services and thereby reveal their preferences for network effects, variety and new technologies. Consumers with strong preferences for an established technology and a large user base are rather prepared to adhere to the established technology, whereas consumers with weak preferences for network effects and strong preferences for novelty will choose the advantages of a new technology. If firms were to neglect network effects of their services when making their price and product policies, they would not survive competitive pres­sures. Path dependence has also played an important role in the history of the public Internet. One example is the durability of the best-effort TCP that does not provide for active traffic management. With the transition to all-IP networks, the importance of variety and search for innovative traffic architectures in markets for Internet traffic services greatly increases. Although traffic networks are software defined and can be configured easily, experience suggests that it is difficult to initiate a bandwagon effect to introduce active traffic management more widely.

2.5