INTERNET DEMAND AND PRICING
Internet access and traffic services are provided as inputs to the production of goods and services and to final users. Given the large number of players in the Internet system and the historical reliance on decentralized forms of coordination, numerous non-market and market transactions take place.
A wide range of prices and pricing models exists for transit, peering, interconnection at Internet exchanges and other connection points, access of content providers to ISPs, CDNs, arrangements between content provider and access providers, hosting, and many more. Due to the multi-sided nature of many market relations, prices between any two players may be positive, with payments flowing in either direction, or zero (as in traditional peering). As many of the prices at the wholesale level are negotiated in private contracts and agreements, only limited information is available in the public domain. Aspects of interconnection and peering are discussed by Clark, Lehr and Bauer in Chapter 16 in this volume and will not be explored further here. Rather, we will focus on the pricing of access for consumers and (by analogy) business users.The literature on network pricing has historically differentiated prices of access and prices of usage. In the widespread unlimited usage models the latter is zero. Like in other communication markets, such as mobile services, Internet access services have been priced in response to demand- and supply-side conditions as well as the institutional environment in which Internet services developed. These choices had clear implications for the adoption and use of services. In the USA, local telephone service was historically offered at both flat and measured prices, with most subscribers opting for flat rates (even if measured rates would have been associated with lower total spending). This allowed early dial-up users to go online without concerns about high phone bills.
After a brief period during which ISPs such as AOL charged measured prices they also adopted flat pricing. Unlimited use probably had a positive effect on the larger Internet ecosystem, stimulating both demand and supply in a virtuous cycle. It also led to congestion and the (temporary) dubbing of the World Wide Web as the ‘World Wide Wait’.During the late 1990s, an increasing number of experts argued in favor of measured rates or some other forms of congestion pricing (e.g., MacKie-Mason and Varian, 1995a, 1995b; MacKie-Mason et al., 1997; Cremer and Hariton, 1999) but as the capacity of networks and routers was expanded and with the migration to broadband services that were initially less capacity constrained, flat unlimited pricing remained the dominant model. Beginning in the late 1990s, mobile voice service providers adopted multi-part self-selection pricing and gradually designed more and more complex bundles of services. With the growing importance of video traffic, network operators in high-income countries claim to face capacity constraints and are increasingly relying on volume-based multi-part pricing approaches in which the lower service tiers come with data caps. This may again be a short-lived phenomenon. Odlyzko (2014) argues that multi-part and volume-based pricing are best understood as attempts by network operators to extract rents while flat rates would be better suited to help fill the available network capacity.
Consumers typically pay for access to the Internet and they may pay for usage depending on the specific plan they subscribe to. Unlimited plans do not have any incremental usage charge but plans with data caps may require overage fees. Some network operators have implemented other responses, such as degrading service or even suspending it. While multi-part pricing and especially data caps are rather unpopular among avid Internet users, they may actually be welfare enhancing (e.g., Bauer and Wildman, 2012). Such pricing schemes are a workable approach to recovering high fixed and shared costs of network access services in ways that are responsive to demand.
Initial estimates of the price elasticity of broadband demand suggest that it is inelastic (Rappoport et al., 2003; Rosston et al., 2011). Price differentiation also allows introducing low-priced plans or even zero-priced options (‘sponsored data’) and thus to expand adoption to additional user groups. However, to assure a workable market such models require sufficient price and quality transparency.Consumers also pay for content and other online services, but often indirectly and in somewhat disguised form. An increasing number of OTT services, such as Netflix, Amazon Prime, or providers of online games charge a flat or per-use price. Likewise, many apps, music, and videos can be bought from online stores. In other cases, transactions are organized in multi-sided markets and the flows of payments are more complex. For example, music streaming services such as Spotify, Deezer, or Pandora offer free tiers that periodically interrupt the service for commercials or require participating in sponsor activities. In these cases advertisers and the companies enlisting their services generate the revenue flow. Online services such as Google Search or Facebook monetize aspects of their users’ online behavior by harvesting massive amounts of online data from users, sometimes without their explicit knowledge and consent. The pervasive nature of multisided markets raises new and challenging competition issues for which antitrust does not yet have clear answers (see Haucap and Stuhmeier, Chapter 9 in this volume).
Internet traffic service providers are platform providers serving end users as well as application providers. Congestion and quality of service differentiation not only play a role in access networks but also in the markets for traffic capacity in the entire Internet. They are a response to heterogeneous demand for traffic qualities such as prioritization offered to application service providers as discussed in the following section.
2.6