161 market, within the meaning of the existing case law does not mean that the owner of this facility might not be in a dominant position” (EC, 2002: §81). 9.3 Network neutrality regulation The above overview of telecommunications regulation in the U.S. and in the EU focuses on market power regulation in the physical network layer. There is currently no market power regulation in the logical or the applications layer of the Internet.159 There is, however, a very concrete debate over the possibility of market power regulation of Internet transport services in the near future. This debate over so-called “network neutrality regulation” is being carried out mostly in the United States. Advocates of network neutrality regulation want the U.S. government to pass legislation that would guarantee that Internet service providers transport all content and services of independent applications and content providers without blocking, impairing, degrading, or discriminating these services in favor of own services and content (Litan and Singer, 2007, Sec. I. A).160 Concrete demands call for a codification into law of the traditional “best-effort principle,” a practice by which IP packets are delivered regardless of their source and destination (Ganley and Allgrove, 2006: 456). Historically, the “best-effort principle,” by which all IP packets receive the same priority, resulted from the endto-end conception of the Internet. The intelligence of the network design was generally located at the edges of the network (in attached hosts), whereas the routers which connect networks and transmit data packages within networks were designed as “dumb” terminals. An undifferentiated forwarding of packets according to a simple algorithm was fitting to this network design. More recently technologies have become available that allow routers to use information included in IP packets to discriminate in the forwarding of these packets. Peha (2007) provides an overview of the new technical means available for Quality of Service (QoS) differentiation. There are many benefits to customers from traffic discrimination, including improved security, improved quality of transmission, efficient resource allocation, and decreased infrastructure costs (Peha, 2007: 13). According to Marcus these technologies are already widely implemented within networks and there are no technological constraints to implementing QoS differentia- 159 This does not imply that these layers are entirely free from regulation. In the logical layer there are, for instance, technical regulations regarding the assignment of IP address blocks. In the applications layer regulations which concern in particular the content market play a role. These are part of media law and policy, such as for instance must-carry rules by which cable television providers are obligated to carry specific broadcasts. 160 The discussion over network neutrality is in truth more encompassing, as many demands by network neutrality activists target policy goals closer to media-policy objectives. Hogendorn (2007), for instance, differentiates between open-access regulation on the physical network layer and “openness to content” on the applications layer, which he views as the central aspect of network neutrality. 162 tion “between or among independently managed IP-networks” (Marcus, 2006b: 34). He explains the fact that this traffic discrimination across network boundaries has so far not been implemented widely by economic considerations. Current Internet applications tolerate a level of delay, which is generally not exceeded in today’s large IP-based networks.161 In the past it has also always proved cheaper to add more bandwidth when network congestion occurred rather than to implement expensive traffic control algorithms (Ganley and Algrove, 2006: 461). As long as network performance is satisfactory without QoS differentiation, consumers have little willingness to pay for an increase in quality of service. In future NGN networks, however, applications may require higher bandwidth than is currently available. Congestion may then become a common problem and QoS differentiation across network boundaries might become a more pressing issue. What initiated the network neutrality debate is that Internet transport service providers are arguing that investments into next generation networks require new revenue models for network providers. Network operators want content providers, which require a high quality data transmission for their applications, to pay for better quality of service (Ganley and Algrove, 2006: 455). Today, ISP revenue is mostly generated from end-users, which pay their home network for access to the entire Internet. This revenue is distributed between ISPs through transit payments for forwarding and transmitting IP packets. A further revenue source for ISPs is web-hosting for content providers. If content providers pay for better quality transmission of their IPpackets, then end-users could experience that some content is transmitted faster than other content. For instance, the VoIP service of one provider could have better quality than that of another; or the download speed from, for instance, Google could be better that that from Amazon sites. Advocates of network neutrality regulation fear, that if good quality transmission is available only to those that have the means to pay for it, new and innovative content providers, with little financial strength, could have difficulty establishing themselves. Furthermore, they point out that a vertically integrated provider of both Internet transport services and applications and content could favor own content (such as own video on demand service and own VoIP service) and discriminate competitors. How can this debate be appraised in light of the above discussion on recommended regulation of Internet periphery services? To the extent that advocates of network neutrality endorse regulation of Internet transport services it is important to remember that the preceding analysis has shown that competition is effective in Internet transport services. No provider can benefit from discriminating third-party 161 Marcus (2006b: 34) explains that not even voice and video streaming require assured quality of service because these applications generally use a jitter buffer to smooth over the variability in the transmission delay. Streaming will simply start with a few seconds delay and the jitter buffer stores and forwards the packets in a way, which smoothes out any variance in transmission speed. The problem of delayed transmission gets more difficult in real-time bidirectional communication, as in VoIP. Here the end-to-end delay has to remain below a threshold of 150 to 200 milli-seconds, otherwise users will “collide” by commencing to speak at the same time. 163 content, since the discriminated party can switch away from this provider and still receive high-quality universal connectivity services. If a market power problem exists in Internet transport services, then this must be the outcome of market-power leveraging from monopolistic bottleneck network areas in the local access market. Knieps and Zenhäusern (2008: 121.) correctly argue that to bring light into the network neutrality debate it is crucial to take a disaggregated view of the Internet and differentiate between core Internet markets and Internet periphery markets (see Figure 2.2. in section 2.2 above). There are no monopolistic bottlenecks in Internet core markets. There is therefore no stable market power in the Internet core which would justify network neutrality regulation. The potential for market power leveraging from Internet periphery markets into the Internet core needs to be confined by regulating market power at its roots. Remaining monopolistic bottleneck network elements in the local communications infrastructure should be subject to nondiscriminatory access regulations. This would guarantee competition on the complementary Internet access services market. Internet access services providers could then offer QoS differentiation without suspicion of acting in an anti-competitive manner. The real issue underlying the network neutrality debate is therefore that the regulation obligations in the local access market need to be reviewed with a focus on the leveraging concern. In the United States, the fact that the FCC reclassified broadband Internet access services as information services in the summer of 2005 limited the extent to which competitive local exchange carriers or ISPs have a right to nondiscriminatory access to wholesale high-speed Internet access. Competing Internet access service providers must now vertically integrate into local telecommunications services and offer the complete set of local services to end-users on the basis of the unbundled local loop or own infrastructure. In all areas in which the incumbent is deploying FTTC or FTTH it is not obliged to offer a wholesale access line capable of high-speed data transmission. It is therefore possible, that in some geographic markets the incumbent local exchange carrier has the control over a monopolistic bottleneck access infrastructure and that in these markets competition in Internet access services is prevented. Generally, if the market power on the upstream market is regulated, then the provider of local access infrastructure may have an incentive to vertically integrate into the downstream markets. Answering the question of whether exclusionary practices of integrated ISPs have been observed is difficult. Speta (2000: 76) convincingly argues that high-speed broadband access services of local operators will only be interesting to end-users, if they value applications requiring such bandwidth. From this perspective, local providers have an incentive not to foreclose content markets but rather to allow good quality access to all kinds of Internet content. The problem in identifying discriminatory practices in Internet transport services is, that not all price and quality differentiation can be taken as evidence for anticompetitive discriminatory behavior. Price differences in Internet transport services 164 may guide the efficient allocation of scarce resources. Content providers today already pay more for better quality data transmission by purchasing strategic hosting services at Internet-data centers located closer to the end-user.162 This guarantees faster download speeds. Network operators also offer different guaranteed service levels as defined by characteristics relating to network security, network response time, application performance, etc. to content providers and end-users (Litan and Singer, 2007: Sec. II A). It can be argued that QoS differentiation, which processes IP packets differently depending on information available on an IP header or within the IP packet, is only a further step in the type of QoS differentiation that already exists. Network neutrality regulation of the services and prices that ISPs can offer would stand in the way of efficient bandwidth allocation in the Internet and therefore ultimately harm consumer interests. In conclusion, price differentiation of Internet transport services can be economically efficient and will probably be necessary, when more bandwidth-intensive uses become available in NGN environments. To make sure that QoS differentiation for traffic transmission is not anti-competitive, but economically efficient, nondiscrimination regulation should be applied where market power exists. From the point of view of the disaggregated approach, accounting separation and price-cap regulation should be applied to monopolistic bottleneck network areas in the local communications infrastructure market. There is no justification for regulation in the market for Internet transport services. Network neutrality activists should focus their activities on ensuring competition in last-mile access technologies rather than on demanding regulation of the competitive logical layer and applications layer of Internet service provision.163 In Europe the buzzword “network neutrality” has not reached the same level of significance as in the United States. However, QoS differentiation will become as important here as in the U.S. once applications for NGNs are introduced into the market, which require high and reliable data throughput rates. It is to be expected that rent-seeking market participants will push a debate on Internet transport regulation under the mantle of “network neutrality” just as in the United States. To date, the European Commission has stated that: Open and competitive markets together with the current provisions in the EU regulatory framework, should in principle be sufficient to protect “Net Freedoms” and to offer a suitably open environment for both users and providers. In particular, the existing provisions for NRAs to impose obligations on operators with significant market power, and the powers for NRAs to address access and interconnection issues could be used to prevent any restriction of access to 162 The company Akamai, for instance, operates 20,000 servers in 71 countries (www.akamai.com; site last visited on Feb. 16, 2008). 163 Prof. Christopher Yoo, outspoken opponent of network neutrality regulation, argues in the same vein: “…the major network neutrality proposals advocate regulating the logical layer in a way that promotes competition in the applications and content layers. Instead, the focus of public policy should be to promote competition at the physical layer, which remains the level of production that is currently the most concentrated and the most protected by barriers to entry” (Yoo, 2006: Chapter 2, 27). 165 information society services, and to impose appropriate interoperability requirements (EC, 2006: 32f.).164 This assessment of the European Commission is in line with the above conclusion that market power regulation of monopolistic bottlenecks is sufficient to hinder discriminatory market power leveraging. 9.4 Conclusions The overview of telecommunications regulation in the U.S. and the EU focused on the regulation of access to network elements of the local telecommunications infrastructure. It was shown that the trend to the convergence of telecommunications, media, and Internet services has had a different impact on traditional telecommunications regulations in the United States and the European Union. In the United States, liberalization of local telecommunications services was introduced in 1996 and was at first accelerated by intensive unbundling regulations. When infrastructure-platform competition increased, these regulations were subsequently curtailed dramatically. Today, incumbents are only obliged to offer unbundled access to the local loop for narrowband voice telecommunications services. For broadband access, U.S. regulation currently relies on competition by other infrastructure platforms. In Europe full liberalization of telecommunications services provision was introduced in 1998. Convergence was accounted for in the new EU regulatory framework for electronic communications of March 2002, which equally applies to different electronic communications platforms. The advent of network convergence has not had the effect of reducing regulation of wholesale telecommunications inputs. Rather, market-power regulation was extended to sectors previously not included. In addition, the analysis showed that the importance of competitive broadbandaccess markets is increasing with the advent of NGN services. The fact that broadband-access regulation was curtailed so significantly in the U.S. is partly responsible for the fear of network neutrality activists that vertically integrated operators may have an incentive to discriminate competitors on the logical and on the applications layers of Internet service provision. If in some geographic markets no infrastructureplatform competition exists or is likely to develop, then the FCC should use the instruments provided by the Telecommunications Act of 1996 to regulate unbundled access to remaining monopolistic bottleneck network elements, that competitors need to offer competitive high-speed Internet access services. Further, the assessment of the European regulation of telecommunications services lead to the conclusion that the possibilities, which the framework provides for limiting current overregulation, are not being used. A consistent application of the three-criteria test and a 164 The Commission here refers to the four “Net Freedoms” identified by the FCC, namely “the right for users to access and distribute (lawful) content, to run applications and connect devices of their choice” (EC, 2006: 32).