A number of incumbent European telecom operators are lobbying, via their trade association ETNO, to get more money for carrying Internet traffic. Faced with large investments in fiber, last-mile operators are seeking new sources of revenue. Competing with content delivery networks may be a promising new business model. But regulatory safeguards may be necessary.

This summer, the French digital economy Minister, Fleur Pellerin, said that France should avoid an interpretation of net neutrality which would overly favour US Internet companies to the detriment of French operators. Implicitly, the Minister is supporting the idea of reasonable compensation for French operators in the context of their commercial dealings with large over-the-top service providers, such as Google. The Minister’s comments share a theme of realigning payment flows on the Internet recently proposed by ETNO, the association of European telecom operations. Its proposal would modify the International Telecommunications Union treaty to include the principle of reasonable compensation for last-mile network operators that carry Internet traffic. It would, roughly speaking, help network operators negotiate for added fees, using a “sending party pays” system similar to the model used for telephone calls. ETNO’s proposal is controversial. US operators and the US government are against any modification of the ITU treaty that would open the door to regulation of the Internet. Critics fear a telecoms-style regulation of Internet traffic, as well as content-based regulations that could lead to censorship and harm freedom of expression.

European authorities are studying the economics of Internet traffic exchange, to determine whether some form of regulation is necessary. The French regulator ARCEP recently issued an order requiring operators to provide detailed information regarding their IP transit and peering agreements. AT&T and Verizon both challenged ARCEP’s order, arguing that it exceeds ARCEP’s statutory authority and that IP transit and peering have not shown evidence of any market failure. Cogent filed a complaint with the French competition authority because of a problem negotiating a peering agreement with France Telecom. The French Competition Authority just released its analysis of the case, finding that France Telecom was guilty of no abuse. The competition authority found that the relevant market consists of access to France Telecom customers via either peering or transit. In other words, peering and transit are substitutable from a demand standpoint. If this were not the case, France Telecom would have a 100% market share on the market for access to France Telecom’s customers via peering with France Telecom. As it stands, the competition authority found that France Telecom held a market share of approximately 50% on the combined transit and peering market, and that given its market share, France Telecom “might” hold a dominant position. Nevertheless, Cogent was unable to show that France Telecom had abused its potential dominant position. For the competition authority, France Telecom could reasonably impose fees on Cogent to compensate for highly unbalanced traffic. To allay fears that it was guilty of margin squeeze, France Telecom volunteered to develop internal transfer price protocols between France Telecom’s network division and its in house transit operator “Open Transit.”

To our knowledge, this is the first time, outside the context of a merger case, that a competition authority has examined the IP peering and transit market. The Polish telecommunications regulator UKE attempted to impose obligations on the Polish incumbent operator in connection with its Internet peering agreements, but UKE’s initiative was vetoed by the European Commission, in part because of defects in the definition of the relevant market.

Would it be Possible to Regulate Internet Peering and Transit Agreements in Europe?

In theory, it would be possible for telecom regulatory authorities in Europe to regulate Internet peering and transit agreements. Unlike the US, Europe makes no distinction between voice interconnection and data interconnection. Regulators in theory have jurisdiction over data interconnection. In practice, however, regulators cannot regulate unless they show an enduring market failure, or that end-to-end connectivity is jeopardized.

The first theoretical route for regulating Internet peering or transit agreements would require that the regulator identify an operator with significant market power, i.e. an operator that has the equivalent of a dominant position under competition law. This can be challenging given the many different routes that Internet traffic can follow. Moreover, large content providers may exercise countervailing buying power. Finally, the operator’s own retail customers may complain and ultimately change operator if Internet traffic were disrupted. This puts a strong competitive constraint on any last mile operator that may negate any finding of dominance. The French competition authority’s decision shows that the market for peering and transit are highly dynamic, but that dominant positions may nevertheless emerge. Dominance, however, is not sufficient to justify ex ante regulation under the European Framework. Regulators must also show that the market is not evolving toward competition and that competition law is not sufficient to deal with market problems. It is unclear that these two conditions would be satisfied in the Internet peering and transit market.

A second route for regulating the exchange of Internet traffic under the European framework would exist if there were a problem of end-to-end connectivity. If a problem of connectivity existed, a regulatory authority would be able to intervene and impose a form of “symmetric” regulation on all operators, whether or not they held significant market power. In the context of the exchange of Internet traffic, end-to-end connectivity never seems to be in jeopardy - at least not so far - because of the Internet’s architecture. Internet traffic trends to end-run any possible obstacle that may arise, whether the obstacle is technical or results from unreasonable commercial conditions.

Because both routes of ex ante regulation are closed, the imposition of a regulated data termination tariff seems impossible without a revision of the European directives. Yet telecom operators are finding other ways to be remunerated. For major websites, ultrafast load time is critical to maximizing advertising revenues. Global content delivery networks (“CDNs”) such as Akamai, EdgeCast, or Highwinds help content providers reduce load time. Certain telecom operators are beginning to compete with CDNs by offering local caching of content at low levels of the network. ARCEP recently described this phenomenon in its report to the French Parliament. The European Commission also discussed CDNs in its July questionnaire on net neutrality.

Would Telco-Operated Cdns Raise a Regulatory Issue?

Content delivery networks help reduce load times for web pages. Advocates of strict net neutrality will argue that last-mile operators should never be able to offer CDN-type services to upstream content providers. If a telco offered this kind of CDN service on an exclusive basis to only certain content providers, regulatory authorities in Europe would no doubt interpret this as a violation of net neutrality, although even then, the service might qualify as a form of managed service. However, if the service is available to all content providers on a non-discriminatory basis, the situation would not be different from what exists today: content providers today can, and routinely do, pay independent CDNs to provide this same service.

If the telco’s CDNs make use of network resources that make the telco’s service better than what independent CDNs can offer, there would be a clear threat to competition that might require regulatory intervention. One could imagine requiring the operator to offer to its competitors the same network elements as it offers to its own downstream CDN service. The imposition of an unbundling obligation such as this would still require a finding that the operator holds significant market power. This in turn requires a relatively narrow definition of the relevant market for wholesale CDN inputs. If the telco’s in-house CDN elements are substitutable from a demand perspective with other more traditional CDN technologies, then the scope of the market would be broad, and the telco would not have significant market power.

Page loading time has a direct impact on advertising and e-commerce revenues for any web-based service, which is why there is a vibrant market for the CDN services. A French start-up, Cedexis, has built a business of routing in real-time content providers’ traffic over various competing CDNs depending on their respective performance levels at a given time and place. Large content providers may make use of several CDNs and balance traffic between them to obtain optimal page loading performance.

Last mile telecommunications operators are particularly well placed to enter the CDN market. They can store content on servers located at decentralized points in the network and potentially offer service that is superior to classic CDNs. For a regulatory standpoint, it would seem disproportionate to prohibit telcos from entering this vibrant market. However, it may prove necessary to impose non-discrimination and “equivalence of input” obligations on the relevant telco so that it provides to competing CDNs and operators the same network resources it provides to its own in-house CDN service.