COMPETITION ISSUES AT THE INFRASTRUCTURE LEVEL

At the infrastructure level two competition issues have received most attention: first, price- or margin-squeeze cases, where incumbent network providers charge retail prices for Internet access that make it unsustainable for competitors to operate in the market, given the incumbent’s retail prices, and second the debate surrounding net neutrality and the risk that network operators or Internet service providers engage in price and/or quality discrimination with respect to different content providers or types of content.

9.5.1 Price and Margin Squeezes in Internet Access Markets

One of the most common allegations in Internet access markets is that vertically inte­grated incumbent network operators abuse their dominant position by engaging in so- called price or margin squeezing, that is by strategically lowering retail prices or raising access prices at the wholesale level in order to constrain reasonably or even equally efficient downstream competitors. In the academic literature one of the debates has been about the incentives for regulated companies to engage in price or margin squeezing at all (see, e.g., Bouckaert and Verboven, 2004). This debate is quite similar to well-known discussions about the rationality of predatory pricing in unregulated markets (see Hovenkamp and Hovenkamp, 2009). In principle, the regulated access price at the whole­sale level reflects the incumbent’s opportunity cost of serving a particular customer. While the economic logic of a margin squeeze largely resembles that of a predatory pricing strategy or, alternatively, refusal to deal, there is a lively legal debate as to whether price and margin squeezes should be treated as a separate competition policy concern or whether they should be subsumed as a particular case of predatory pricing or refusal to deal (e.g., Sidak, 2008).

Further policy debates concern the proper efficiency standard that an incumbent has to adhere to when setting its retails prices: Which competitors need to be able to survive in the retail market, reasonably efficient operators or only equally efficient operators? While most economists argue for the equally efficient operator test (e.g., Bouckaert and Verboven, 2004; Gaudin and Saavedra, 2014), Clerckx and De Muyter (2009) defend the reasonably efficient competitor standard and emphasize that incumbent network operators tend to have inherited their positions from a government enterprise or other forms of protection.

The legal analysis of margin squeeze, furthermore, diverges into two different philoso­phies: some experts consider margin squeezes as a form of the classical refusal-to-deal abuse or predatory pricing, whereas others consider margin squeezes as a peculiar form of abuse warranting separate analysis and remedies.

In Europe, the four most prominent of the numerous price-squeeze cases have been Deutsche Telekom (2003/2010), Wanadoo (2003), Telefonica (2007) and TeliaSonera (2011), but there are several other cases in almost all EU member states. For detailed discussions, see Crocioni and Veljanovski (2003), Fernandez Alvarez-Labrador (2006), Motta and de Streel (2006), Bravo and Siciliani (2007), Polo (2007), Clerckx and De Muyter (2009), Heimler (2010), Hay and McMahon (2012) as well as Gaudin and Saavedra (2014).

The general approach in the EU, following the Deutsche Telekom case, is that the European Commission as well as national competition or regulatory authorities tend to consider a price squeeze to be abusive per se and liable to prosecution under Article 102 TFEU, regardless of the economic effects on competition and consumers. When the European Union’s Court of First Instance (CFI) endorsed the Commission’s decision in the Deutsche Telekom case, it also clarified that an abusive margin squeeze can be dis­covered through the so-called imputation test.

The formalistic approach of the imputation test contrasts heavily with the more eco­nomic or effects-based approach to which the European Commission has moved in other areas. Interestingly enough, all four cases were decided on grounds of predatory (retail) pricing, not on grounds of excessive (wholesale) pricing. The alleged margin squeeze did not result from an excessive wholesale/access price for an essential input (access to the fixed local loop) but from a retail access price that was considered too low.

The USA follows a rather different approach since Trinko and linkLine. The linkLine decision concerned four California Internet service providers (ISPs) supplying retail digital subscriber line (DSL) services. These ISPs purchased wholesale transmission services from the vertically integrated Pacific Bell (doing business as AT&T), which itself supplied DSL Internet access to the retail market. In July 2003, the ISPs brought a private antitrust suit against AT&T alleging that it had monopolized and attempted to monopolize the regional DSL market in violation of Section 2 of the Sherman Act in several ways, including the creation of a price squeeze. While the four ISPs prevailed in the District Court and the Circuit Court of Appeals, the US Supreme Court saw no need to view price squeeze as a distinct exclusionary strategy for antitrust purposes.¹³ It decomposed vertical price squeeze into two parts: first, the high wholesale price is an exercise of monopoly power, and the exercise of lawfully obtained market power does not violate the Sherman Act Section 2 prohibition of monopolization.

9.5.2 Net Neutrality

Traditionally, the Internet has developed within a non-discriminatory architecture. All data packages are treated equally (with so-called ‘best effort’), independent of their content and origin. Recent technologies, however, enable network providers to distin­guish and to differentiate and discriminate between different packages. Following the advent of new traffic management technologies a debate has emerged in policy circles and in academia - originally only in the USA (see Lessig, 1999; Wu, 2003), but subse­quently spreading to Europe - on whether the traditional principle of ‘net neutrality’ would need to be secured by means of regulation.

The term net(work) neutrality is not clearly defined and has several meanings. As Kramer et al. (2013, p. 797) point out, the meaning ‘is often ambiguous and can mean anything from blocking certain types of undesired or unaffiliated traffic (Wu, 2007), to termination fees (Lee and Wu, 2009), to offering differentiated services and taking meas­ures of network management (Hahn and Wallsten, 2006)’.

One of the core elements of net neutrality in any case is the best-effort principle. As long as network operators do not identify the origins of the various packages, all packages are obviously treated equally (best-effort rule). In addition, network operators cannot charge the data sender as long as they do not identify them. This in turn implies zero prices for sending traffic (zero-pricing rule).

From an antitrust policy perspective, the concept of net neutrality as a non­discrimination rule is of particular interest. Strict net neutrality then prevents ISPs from prioritizing any traffic from any origin. Advocates of strict net neutrality fear that any departure from this rule would induce vertically integrated ISPs to behave in anticompeti­tive fashion, for example by blocking or discriminating against rivals’ content in terms of prices and quality of service. ISPs should thus be subject to a prophylactic regulation to prevent any such behavior right from the start.

We would like to shed light on the incentives of ISPs to engage in such behavior and on to the consequences for competition and welfare.

9.5.2.1 Blocking and degradation

Vertically integrated ISPs that own the network infrastructure and act, at the same time, as content providers may have an incentive to degrade the quality of rivals’ content on their network or to entirely block services in order to reduce competition in the content market and enhance the demand for their own content. The concern is that ISPs will use their control over the last mile to favor their own proprietary content over content provided by competitors (see Kramer et al., 2013).

These are common examples of so-called vertical foreclosure practices. According to Rey and Tirole (2007) foreclosure is a dominant firm’s denial of access to an essential facility with the intent of extending market power from one segment of the market (the bottleneck segment) to an adjacent segment (the competitive segment). By treating some groups of customers preferentially and offering less attractive terms to others, companies can achieve the same results as a vertically integrated company even without vertically integrating.

The theoretical findings on vertically integrated firms’ incentives to foreclose rivals downstream are mixed. Bowman (1957) famously made the argument that there is only one monopoly rent in any vertical chain of production and, thus, a monopolist in the upstream market would have no incentive to monopolize the downstream market (and vice versa). According to the Chicago School’s ‘single-monopoly-rent hypothesis’ a verti­cally integrated company can only earn a monopoly profit in one of the markets, either upstream or downstream, but not two separate monopoly rents in both markets. As a result, a monopolist either has no incentive to vertically integrate in order to leverage its dominant position from the upstream to its downstream market or, in the case of imper­fect downstream competition, vertical integration would actually benefit consumers and increase welfare as it removes the inefficiencies from double marginalization.¹⁶ Post­Chicago economists, however, have shown that the validity of the hypothesis depends, among other things, on the assumption that market participants have perfect informa- tion.¹⁷ The modern economic literature identifies various circumstances where vertical foreclosure can be profitable (for an overview see Rey and Tirole, 2007).

In telecommunications markets, there is at least some evidence that ISPs may some­times foreclose rival services in practice. In the USA, the net neutrality debate evolved in several steps. First, in 2005, the FCC took action against the Madison River Telephone Company.¹⁸ Madison River Communications, a regional company offering both tel­ephone and Internet services, blocked ports used for Voice-over-IP (VoIP) services, thus preventing its subscribers from using third party VoIP services. The FCC regarded this action as an infringement of principles of an open Internet, first expressed by then Chairman Michael Powell in 2004, which should generally enable customers to access any legal content.¹⁹ Following the Madison River case, the FCC adopted the Open Internet Principles in 2005, establishing four consumer rights. However, these principles were not legally binding but were only a declaratory policy statement by the agency. When, in 2008, the cable network operator Comcast slowed BitTorrent peer-to-peer traffic in response to heavy usage by private customers,²⁰ the FCC required Comcast to disclose the details of its discriminatory network management practices to the Commission within 30 days. In addressing the BitTorrent case, the 2005 Open Internet Principles were applied but an appeals court overturned the decision. In response, the FCC adopted its 2010 Open Internet Order. This order was subsequently challenged by Verizon and referred back to the Commission by an appeals court. That court reiterated that the FCC had the authority to classify and reclassify broadband access services as information or telecom­munication services (the two major US legal categories), that the agency had the author­ity to promulgate rules assuring non-discrimination in the Internet, and that there was concern that broadband access providers might abuse their market power. However, the court found that the 2010 order had applied common carrier principles to information service providers. After a lengthy political and legal process and many changes in direc­tion, the FCC adopted a new order in February 2015, designed around the following core principles (see FCC, 2015):

• Bright line standards (no blocking, no traffic degradation, no paid prioritization).

• Broadband access services, both fixed and mobile, were reclassified as common carrier services according to Title II of the Communications Act.

• Additional safeguards for edge providers and customers (along the lines of the 2005 declaratory order) were adopted.

In Europe, broadband markets are less concentrated than in the USA (see Kramer et al., 2013). This may explain the European Commission’s cautious view on any ex ante regulation of ISPs. In a less concentrated market there may be less potential for unlaw­ful behavior of ISPs as long as consumers are able to figure out that certain services are blocked or degraded and can switch in a reasonable time. Moreover, European competi­tion and telecommunications law already provides tools sufficient to deal with many of the problems of net neutrality. The risk of discrimination through (potentially) vertically integrated content and network providers can be addressed by means of sector-specific regulation, as an ISP with significant market power can be obliged to provide access to its facilities under current law. In this case, a regulation of access fees already pre­vents discrimination by (vertically integrated) ISPs.²¹ Moreover, discrimination can be addressed by means of competition law. Article 102 of the TFEU prohibits the abuse by one or more undertakings of a dominant position within the internal market or in a substantial part of it. Hence, the European Commission’s Universal Service Directive²² acknowledges the positive effect of prioritization traffic and product differentiation, as long as consumers have a free choice of services and the conditions of these services are transparent to consumers.

9.5.2.2 Quality of service and price discrimination

Strict net neutrality prevents ISPs from prioritizing and discriminating against certain traffic. However, one of the main arguments against this best-effort principle is that serv­ices differ in their sensitivity to delay. Streaming services and VoIP are more sensitive to delay than, for example, web browsing and emails. Moreover, services like e-health may be highly sensitive to delay and require guarantees of prioritization. Hence, proponents in favor of price and quality differentiation and against net neutrality state that it makes sense to allow for traffic management by (1) offering different categories of quality of services and (2) price differentiate according to the sensitivity of delay. In contrast, advocates of strict net neutrality fear that this kind of traffic management may result in competition and welfare-distorting behavior by ISPs.

In the theoretical literature, there are arguments both in favor of and against discrimi­nation. Hermalin and Katz (2007) and Litan and Singer (2007) widely acknowledge the positive effects of differentiation in service quality. In their model, ISPs only offer a single medium quality under net neutrality, whereas differentiation makes it possible to offer efficient high-quality to high-valuation providers. Moreover, due to the structure of the Internet one has to consider a further effect of strict net neutrality. All services use the network as a common resource. As stated above, services differ in their sensitivity to delay. Too much traffic of less sensitive services, such as file sharing, can cause capacity overload and delay or loss of data packages. This capacity overload, however, mainly affects the high-sensitive services like IP television. Finally, if this happens reasonably often, the high-sensitive services may be crowded out by low-sensitive services, which is a well-known phenomenon of the ‘tragedy of the commons’. To avoid crowding out, ISPs have to manage traffic according to the sensitivity of delay and provide different quality class contracts (quality of service).

On the other hand, Economides (2008) points out that in such cases an ISP may abuse its market power and also force low valuation providers to accept priority pricing. Choi and Kim (2010) find ambiguous effects of prioritization on welfare. They state that for a large set of parameters a discriminatory regime may lead to lower short-run welfare. Bauer (2007) furthermore points to potential dynamic inefficiencies. Summarizing the theoretical literature, Schuett (2010) concludes that while welfare effects are not entirely clear, in many scenarios there are likely to be positive effects of non-neutrality.

Clearly on the other hand, service and price discrimination may carry the potential for ISPs to behave in an unlawful manner and distort competition and welfare. ISPs with significant market power, however, are subject to control of dominant behavior in Europe as well as in the USA. Article 102 of the TFEU prohibits discriminatory behav­ior by companies enjoying a dominant position. In the USA, Section 2 of the Clayton Act prohibits price discrimination if such discrimination substantially lessens competi­tion or tends to create a monopoly. In a competitive environment, ISPs are free, as is any other company, to offer differentiated services. The parties will monitor whether the respective qualities promised are really maintained, and otherwise they are free to switch. Hence, the European Commission puts a special emphasis on transparency of providers’ terms and condition. The EU Commission’s Universal Service Directive forces national regulatory authorities to put transparency obligations into national law. According to the directive, providers with a significant market share have to provide the regulatory authority with the terms and conditions for access to and usage within their network. Moreover, all providers are to disclose information to consumers about their net neutrality policy. This should enable consumers to choose between provid­ers. Practically, it may be questioned whether consumers are able to evaluate and compare the net neutrality policy or whether this requires great expertise. Finally, the EU Commission’s Universal Service Directive entitles national regulators to secure a minimum quality level if necessary.

A case for strict net neutrality regulation is not compelling, as many violations are already recognized as cases subject to antitrust and competition law (see Yoo, 2005, 2007; Sidak, 2007). Strict net neutrality, where all services are treated equally, is economically inefficient, since services differ in their sensitivity to delay and users differ in their will­ingness to pay for these services. Although network management can provide incentives for discriminating, competition policy already provides sufficient tools to deal with many of the concerns and further ex ante regulation of net neutrality is not urgent. Finally, a departure from strict net neutrality may allow ISPs to deviate from the zero-pricing rule for content providers and split the charges between content providers and users. Economically, this seems to be more efficient than the current ‘receiving-party-pays’ principle since both parties share a benefit from the content.

9.6