Archive for September, 2014
I’ve been searching for a good explanation of how net neutrality proponents think that Title II could be used to banish priority-delivery arrangements (“priority”) from the earth. Harold Feld of Public Knowledge was kind enough to alert me to this legal theory, which he filed with the FCC in May.
Before reading Mr. Feld’s submission, I had been laboring under the belief that netizens wanted Title II so they could regulate the rates of priority to zero. As pointed out to me by a telecom lawyer, this zero-price theory can’t overcome the legal requirement that common carriers receive a positive “fee” for services rendered, which must afford the carrier a reasonable chance for a return on investment.
Before assessing Mr. Feld’s alternative legal theory, it bears noting that economists don’t like to prohibit conduct that can be motivated by either efficiency or nefarious reasons. This “don’t-throw-out-the-baby-with-the-bathwater” concept is best captured in the rule-of-reason treatment of certain restraints in antitrust law. We don’t banish joint ventures from the face of the earth because some JVs could be used to coordinate prices. Ditto for vertical restraints. In contrast, price fixing is treated as a per se violation of the law—there is no plausible efficiency justification.
Because an ISP and its customers could reasonably want to prioritize the packets of a life-saving telemedicine application over those of a cat video (no offense to cat lovers), a rule that barred all priority would toss the baby overboard. And we love babies! While priority could be abused in theory—for example, an ISP could give preference to its own online gaming website or threaten to degrade the connection speeds of apps that declined priority—priory is not evil in all manifestations.
The netizens can’t see any good in priority. According to Mr. Feld’s legal theory, Title II has been used to banish conduct that the FCC found was “inherently unjust, unreasonable, or subject to abuse.” Because all priority is evil, goes the theory, it must be banished. Yet the odds that the D.C. Circuit would find that any discrimination across applications is “inherently unjust” are slim to none.
Setting aside this minor roadblock, let’s see if Mr. Feld can claim precedent. He cites two cases in which the FCC barred conduct under its “inherently unjust” power. The first is Carterfone, in which “the FCC found that it was inherently unjust and unreasonable to permit a carrier to interfere with the ability of a subscriber to attach a device to a network.” The analogue here would be barring ISP conduct that interfered with the ability of a broadband customer from accessing a website or from experiencing the website in all its glory.
I am in favor a rule that prohibits blocking (complete interference) and bars ISPs from degrading the treatment of packets from edge providers that decline priority (partial interference or extortion). But leaving one application provider whole (who declines priority) while giving priority to another (who pays for it) does not constitute interference in any way, shape or form.
Mr. Feld uses the ambiguous language “discriminating between devices” to bolster his Carterfone analogy. But recall that in that context, Ma Bell was trying to extend her monopoly to equipment. Offering priority to a third party online gamer—and extending the terms to all comers—is light years away from trying to be the exclusive provider of online gaming.
The Carterfone example is further distinguished from today’s debate because, according to a recent FCC report, only seven percent of U.S. homes are beholden to a single provider of wireline broadband at speeds of 10 Mbps (See Figure 5a). Mr. Wheeler offered alternative stats showing that 30 percent were subject to a single wireline provider at speeds of 10 Mbps, but even this is a far cry from the percentage of homes beholden to a single provider of local telephone service in the 1960s.
Mr. Feld’s second case is equally inapposite. In Computer II, the FCC determined that common carriers could offer “enhanced services” only through a standalone entity—that is, common carriers were forced to structurally separate voice from enhanced services. Offering priority to third parties is not equivalent to vertical integration. Certainly ownership could be effectuated via an exclusive priority arrangement. But if that’s the concern, let’s focus our attention on exclusive priority, as opposed to all priority. Don’t toss that cute baby.
In his preamble, Mr. Feld makes an important disclaimer: “[Public Knowledge] notes that while the Chairman is correct that Title II would permit reasonable discrimination, it does not, as some seem to insist, require reasonable discrimination and therefore Title II would require paid prioritization.” If Title II permits reasonable discrimination, and could bar priority only if the D.C. Circuit found that priority was “inherently unjust,” then Title II is likely being peddled to serve a different regulatory agenda.
Quants have been studying the million-plus comments submitted to the FCC during the Open Internet proceeding, and unsurprisingly, the vast majority favor net neutrality. But what does that mean?
Those pressing for heavy-handed regulations would like it to mean “support for Title II,” but the myriad comments that mentioned Title II were most likely form letters generated by advocacy groups: It is doubtful that ordinary citizens understand the legal nuances that distinguish the FCC’s authority to regulate Internet service providers (ISPs) under section 706 and Title II.
To understand which regulatory path to take, we need to clearly define what sort of conduct cannot be tolerated on the Internet. Consider the following offer (“Offer A”) by an ISP to a content provider: “If you don’t take my priority-delivery offering, I will degrade your connection speeds on my network.” Such repugnant conduct would diminish the absolute performance of any content provider who declined the offer.
Now consider a slightly different offer by an ISP (“Offer B”): “If you don’t take my priority-delivery offering, you will continue to receive the same connection speeds that you previously enjoyed. If you take it, however, your connection will be even faster.” In contrast to Offer A, this offer would not threaten the absolute performance of the content provider; the only impact for those who decline it would be a diminution in their performance relative to those who elected priority delivery.
A broad consensus has formed around the need for regulation to prevent the type of conduct associated with Offer A. There is also wide acceptance of rules that would bar ISPs from favoring affiliated websites over independents by, for example, slowing or blocking access to the competing content. Importantly, none of these regulations would require the FCC to engage in rate regulation. All would be achievable under the “light-touch” approach of section 706.
What section 706 cannot prevent, however, is the type of conduct associated with Offer B. The D.C. Circuit has said as much, ruling that any attempt to prevent ISPs and content providers from negotiating for priority delivery smacks of common-carriage regulation. In other words, if rates for priority delivery were set by regulatory fiat, then there would be no need for ISPs and content providers to negotiate over the rate.
Will FCC Chairman Tom Wheeler give a “thumbs up” to Title II?
Will FCC Chairman Tom Wheeler give a “thumbs up” to Title II?
Title II would not bar priority-delivery offerings out of the gate: Even under Title II, ISPs would be free to offer such services, so long as they did so in a non-discriminatory way—that is, each package would have to be available to all similarly situated websites. But Title II could empower the FCC to begin a rate proceeding for priority delivery, at which point interested parties could petition the agency for zero rates, which would effectively eliminate priority delivery from the marketplace.
Would it be a good thing to unleash rate regulation on ISPs to prevent the formation of priority delivery? Not if investment is the metric. In a new study released by the Progressive Policy Institute (PPI), Bob Litan and I analyzed the impact of rate regulation pursuant to Title II on the investment of incumbent telcos, entrants, and cable providers in the 1990s and early 2000s. The results should give regulators pause before dabbling in rate regulation again.
Telco entrants: The 1996 Telecom Act required the incumbent Regional Bell Operating Companies—the localized telephone monopolies that were part of the integrated AT&T before it was broken up by court order in 1984—to share or “unbundle” the pieces of their local exchange networks to telco entrants at regulated rates to allow the latter to begin breaking down the local monopolies. With two co-authors (including your fearless blogger), Bob Crandall of Brookings used cross-state variation in the price of constructing local phone lines relative to leasing unbundled loops at regulated rates to identify the sensitivity of the entrants’ investment in local lines to these regulated rates. The researchers found that facilities investment by telco entrants was actually greater in states with higher unbundling rates; in other words, the more generous the subsidy, the less facilities-based investment occurred by telco entrants.
Cable companies: Cable television providers were best positioned to challenge the telcos’ hegemony in voice and Internet services in the mid-1990s. But to enter, cable operators first had to upgrade their networks to support IP-based transmissions. Yet cable companies were reluctant to make such investments so long as regulators were providing a less expensive entry path to their competitors (the telco entrants). High margins in local telephony and Internet access were the signal for cable entry, but the FCC’s unbundling experiment was injecting unnecessary noise. It took a series of court orders that unwound the unbundling regime by 1999 for the cable operators to see the market signal through the noise. Using data from NCTA, we found that the average annual capital expenditure for cable operators during the three years following the 1996 Act was $6 billion. In comparison, the cable industry’s average annual capital expenditure during the three-year period after the unbundling rules were unwound was $15.1 billion.
Incumbent telcos: Perhaps the most pivotal regulatory decision concerning the fate of broadband occurred in 2003. In its Triennial Review Order, which became effective in October 2003, the FCC determined that there would be no unbundling requirement for fiber-to-the-home loops. Once the telcos understood that they were free of the obligation to lease their fiber-based networks to competitors at regulated rates, they entered into a race with their cable counterparts to begin building the broadband networks that are now transforming the telecom landscape. In the span of just five years, from the FCC’s adoption of a policy of regulatory forbearance for fiber and IP networks in 2003, the miles of optical fiber doubled from five to ten million. Annual wireline broadband investment by the telcos jumped to $15.5 billion by 2008.
Why should the FCC focus on investment when promulgating new Internet regulations? First, Congress instructed the agency to do so in section 706 of the Act. Second, and perhaps even more important, investment in the communications sector continues to play a pivotal roll in driving the U.S. economy.
This week, PPI released its third annual report on “U.S. Investment Heroes,” authored by Diana Carew and Michael Mandel, which analyzes publicly available information to rank non-financial companies by their capital spending in the United States. Once again, AT&T and Verizon ranked first and second, respectively, with $21 and $15 billion in domestic investment in 2013. Comcast, Google, and Time Warner also made PPI’s top 25 list, each investing over $3 billion. The authors credit investment in the core of the network with sparking the rise of the “data-driven economy.”
In light of the results from prior experiments in rate regulation, the FCC should eschew calls to regulate ISPs under Title II. The incremental benefits (potentially barring fast lanes) are dubious, but the incremental costs (less investment at the core of the network) would be economically significant. Given its size and contribution to the U.S. economy in terms of jobs and productivity, even a small decline in core investment in response to rate regulation would impose social costs beyond the immediate harm to broadband consumers from an atrophying network.
Let’s not repeat the mistakes of the past. If we focus on what’s important—preventing an absolute decline in the welfare of content providers and preserving incentives to invest—we can nurture our precious Internet ecosystem at both the edge and the core.
Readers of this blog who follow the net neutrality debate will recognize an important case called Cellco, cited repeatedly in the D.C. Circuit’s January decision to gut key provisions of the FCC ’s Open Internet Order that smacked of heavy-handed rate regulation.
In Cellco, the D.C. Circuit blessed the FCC’s 2011 Data Roaming Order, which established the terms by which wireless broadband providers should contract for data roaming. Importantly, the Court approved a light-touch approach to adjudicating roaming disputes: Rather than set a roaming rate by fiat, the Data Roaming Order granted the parties the freedom to negotiate towards a “commercially reasonable” roaming rate.
To ensure that his agency’s revised Open Internet Order survives scrutiny by the same court, the Chairman of the FCC, Tom Wheeler, has imported the “commercially reasonable” language from the Data Roaming Order to establish the terms under which content providers and Internet service providers (ISPs) should contract for priority delivery. I’ve written about his proposal glowingly here.
Not everyone agrees with my assessment. Proponents of strong net neutrality such as Netflix and Kickstarter claim that the current proposal would permit the development of “fast lanes” for content companies that could afford priority delivery. They seek to steer the Chairman toward heavy-handed rate regulation, which would bar any contracting for priority delivery by effectively setting its rate at $0.
While net neutrality garners press attention, similar efforts by competitors are underway to steer the FCC toward heavy-handed rate regulation for roaming disputes. In particular, they are trying to convert the “commercially reasonable” standard embraced in the Data Roaming Order into de facto rate regulation. And they have brought out the big guns.
Writing on behalf of T-Mobile, Dr. Joseph Farrell, former chief economist of the FCC and currently professor of economics at UC Berkeley, has proposed a new standard for determining whether a roaming rate is “commercially reasonable.” According to Professor Farrell, any roaming rate, expressed on a per-megabyte (MB) basis, that substantially exceeds the access provider’s retail rate for the incremental MBs used while roaming should be looked at with suspicion.
To make his proposal concrete, consider Sprint’s current wireless offering at $50 per month. Assuming the average wireless broadband user consumes 1,700 MB of data per month, Sprint’s roaming rate should not exceed three cents per MB (equal to $50 divided by 1,700 MB).
Not only would Professor Farrell’s proposal constitute heavy-handed rate regulation of the kind rejected by the Court, it would also impose the wrong rate. To see why, consider the following hypothetical:
Sprint competes with T-Mobile (as well as AT&T T -1% and Verizon) for business customers in Whatever, USA. To commute downtown, suburbanites ride a bus across a bridge that is covered by Sprint, AT&T and Verizon, but is not covered by T-Mobile. Although T-Mobile’s license covers the bridge, T-Mobile chooses not to build out its network there. Assume further that the typical commuter consumes five percent of her daily consumption of MB on the bridge. And most important, no commuter would ever subscribe to a wireless carrier that did not cover the bridge.
Does your wireless carrier cover this bridge?
Does your wireless provider cover this bridge?
By providing bridge coverage to T-Mobile via roaming, Sprint creates a new option for commuters that did not previously exist. It is now possible for a commuter who previously would have opted for Sprint, to now opt instead for T-Mobile. Is it any wonder why Sprint is reluctant to cut a roaming deal in this circumstance?
Sprint’s margins that are put in play via the roaming agreement are not just the margins associated with the commuter’s MB usage over the bridge. Instead, the entirety of Sprint’s retail margin is put in play! Accordingly, it would be perfectly commercially reasonable for Sprint to demand to be compensated for its forgone retail margins when setting its roaming rate.
At margins of roughly 40 percent, that would mean a $20 roaming fee per subscriber per month (equal to 0.4 x $50 per month). But if Sprint asked for anything more than $2.50 per subscriber per month (equal to 0.05 x 1,700 MB x 3 cents per MB), Sprint’s offer would violate Professor Farrell’s proposed standard. In other words, the roaming rate that would make Sprint indifferent between serving the customer indirectly (via a roaming agreement) and directly (as a retailer) is eight times T-Mobile’s asking price!
And herein lies the harm from rate regulation: Wireless providers made massive investments in broadband networks under the belief that those retail margins would provide a sufficient return on investment; dilute those margins too aggressively and the investments disappear. The access seeker sits back and then wants to cherry pick that investment at regulated rates rather than build the network itself. Yet an explicit goal of the Data Roaming Order was to provide incentives to “those providers to invest and deploy advanced data networks, and avoid potential disincentives for those providers to invest.”
Investment incentives are particularly important because wireless carriers are continually upgrading their networks. 4G is just the current flavor, but it followed 3G, and it will soon be followed by 5G, which might just serve as a viable alternative to wireline access technologies such as cable modem. If only wireless investment were complete, it might be possible to construct an economic model showing that Professor Farrell’s proposed solution maximized short-term consumer welfare. But when an industry is as dynamic as wireless, investment is never complete, and attempts to appropriate “sunk” investment will surely backfire.
To be fair, allowing access providers to capture the forgone retail margin may not be wise policy when retail markets are monopolized. But that is not the case here: The effective price per MB on U.S. wireless networks declined from $0.25 in 2008 to less than $0.05 by 2012. Retail competition among four national providers ensures that the voluntary access rates do not reflect monopoly rents.
So what is the lesson for net neutrality? Competitors are lobbying the FCC to set a regulated price for roaming, with little concern for investment effects. With the same recklessness, certain content providers are lobbying the FCC to set the regulated price for priority delivery at zero. In both instances, the FCC should refrain from getting embroiled in rate regulation.
The Chairman should stick with his original net neutrality proposal. ISPs and content providers should be free to contract for priority delivery pursuant to a commercially reasonable standard. Under such a standard, special arrangements for affiliated websites could be barred. And an ISP that tried to reduce the speeds of its standard offering—with the introduction of a “slow lane”—would be presumptively in violation of the standard. But there would be no regulated price for priority delivery.
If the FCC is dragged into rate regulation in these important cases, the incentives for private-sector investment will be undermined and, when the current flavor of technology is locked in place, broadband consumers will suffer.
Follow me on twitter @halsinger