Still Not Falling Behind

Earlier this week the Center for Public Integrity (CPI) published what purports to be an investigative look at broadband competition, speed, and prices, concluding that “U.S. Internet users pay more and have fewer choices than Europeans.” This is a common myth that we at ITIF have long worked to dispel, but apparently some myths die hard.

CPI duplicates the usual argument made in this space, attempting to compare a handful of French and U.S. cities, claiming that because French cities offer more broadband “choices,” they have lower price, and higher adoption. CPI suggests that the supposed success demonstrated in their “snapshot” should push us to emulate France’s open access policies. There are a number of problems with this picture.

CPI compares a set of advertised broadband rates in a handful of French cities to American cities. Let’s start by pulling back a bit and looking at country-wide data. If you are going to opine on national policy direction, it’s worthwhile to at least take a passing glance on the statistics at a national level, instead of comparing five French cities with five American ones. Compiled below are some data from Akamai’s latest State of the Internet report comparing U.S. and France:

 

France		Global Rank
Average Speed	7.1 Mbps	44th
Peak Speed	31.5 Mbps	58th
% over 4 Mbps	70%	43rd
% over 10 Mbps	15%	42nd
% over 15	5.5%	38th

U.S.		Global Rank
Average Speed	10.7 Mbps	16th
Peak Speed	49.4 Mbps	22nd
% over 4 Mbps	74%	40th
% over 10 Mbps	39%	17th
% over 15 Mbps	18%	17th

 

These high-level stats, which show the U.S. handily beating out France on coverage and speed, should be sufficient to caution against general pronouncements that current U.S. broadband policy is flawed. To be fair, CPI is less concerned with coverage and speed and more with “choice” and price.

But first, CPI makes some hand-wavy gestures towards “methodology” in an attempt to appease “critics,” and the disappointing follow-through warrants note. A quick example is the discrepancy between advertised and actual speeds. CPI, like OTI’s “Cost of Connectivity” study, relies on advertised, instead of actual, speeds. CPI acknowledges this is problematic, noting that “European Internet providers deliver just 74 percent of their advertised speeds, according to the European Commission” (U.S. operators provide 101 percent of advertised speed on average).

It is good of CPI to at acknowledge this issue, but if they had dug just a bit deeper, they would see that the study for the European Commission showed France to be, in fact, the worst offender in all of Europe, offering only 44.3% of its advertised DSL speeds.

 

Source: European Commission, Quality of Broadband Services in the EU (2013), at 86.

So it seems that French citizens—save those lucky 3 percent on fiber networks—pay more per megabit per second than Americans do, and get considerably slower speeds for the money. So much for falling behind…

Note that data on advertised vs actual speeds is only available for DSL networks in France. Why? Because in France, cable broadband and fiber only have 6 and 3 percent of the market share respectively. Intermodal competition is so low in France that the contractor that performed the study could not find a statistically significant number of users of fiber or cable to measure.

One more quick note on methodology. More troubling than the oversight on advertised vs actual speeds is CPI’s attempt to account for differences in population density. CPI says of its comparison cities:

The Center chose France because it has the same proportion of its population in urban areas as the U.S. In addition, the population densities of the cities are similar. Critics have said comparing U.S. broadband prices to cities abroad is unwise because the U.S. population is spread over a larger area. Greater population density leads to greater efficiencies and lower prices. By using France for comparison, the Center controlled for those differences.

They’re right—population density makes a big difference; it’s much more economical to serve packed cities than sprawling suburbs. Good of CPI to recognize this fact, but odd to then turn around and compare, for example, Columbus, OH (3,600/sq mile) to Nice (12,000/sq mile). Surprisingly enough, France has some remarkably dense cities. Check out this Wikipedia chart—the top 26 densest cities are dominated by India, the Philippines, and France.

When it comes to population density, CPI’s decision to compare Tulsa, OK to Bordeaux (13,000/sq mile) is like saying, with a straight face, “in order to control for differences in population density, we choose to compare Tulsa, OK (2,000/sq mile) to Chicago, IL (12,000/sq mile).”

The CPI report does its best to maintain the trappings of investigative journalism while forcing the facts into the well-worn “Falling-Behind” mold. This criticism is aimed more at the analysis of this genre as a whole than any explanation of French “success.”

Indeed, population density isn’t everything, as evidenced by France’s poor performance in the speed metrics above. Much of Europe struggles to upgrade for a wide variety of reasons. Protected, historic buildings and winding roads controlled by arcane zoning regulations are considerable impediments. Christopher Yoo notes that mistakes made by France’s top-down telecom planning, such as regulators’ fiber-fanaticism resulting in refusal to approve use of advanced VDSL technology, is to blame for France’s poor coverage of next generation networks (France has deployed networks capable of over 25 Mbps to only 24 percent of its population compared to 82 percent in the U.S.).

But all this is sideshow to the main argument CPI and others put forward: that open-access to unbundled local loops drives more choice and lower prices. This form of service-based competition can indeed offer more “choice.” But I submit this “choice” does not rest on meaningful competition, but instead is based by-and-large on the access rates set by the regulator. This is why in CPI’s charts virtually all the French providers are clustered in one corner—the price and speed are determined by regulation and the legacy network, not meaningful competition. CPI claims broadband providers here “deliberately carved up [territory] to minimize competition,” but ignores that fact there is wildly less overlap of any networks in France. Virtually all of the “choice” in France is on the same old France Telecom network.

CPI argues that we should prefer open-access as it drives down price, which will in turn drive adoption. The data shows that prices here are comparable to the rest of the world, and are unusually progressive, with lower speed tiers being some of the least expensive in the world. There is also all sorts of information indicating that getting everyone online is a much more difficult problem than price alone. Ok, set all that aside for a moment and acknowledge that price is a real impediment to some percentage of non-adopters. But why opt for a reversal in policy that would sap any incentive to invest and grow capacity when targeted subsidies could do the job? If folks spent half the ink they do on the tired “Falling Behind” genre on, say, transitioning Lifeline to broadband, we would all be better off.

The real international broadband success stories are those countries with high-levels of intermodal competition. We chose the route of the type of competition that drives investment and preserves incentives to upgrade networks, and have seen continued growth, improving speeds, and new entrants as a result. The U.S. is fortunate to have made smart policy choices during cable buildout that ensured two competing pipes into virtually every home. The successes of intermodal competition in the U.S. are impressive considering the hurdles we face with sprawling suburbs, rural states, low levels of computer ownership, and high rates of poverty.

I’m sorry, but finding advertisements for the fiber network that 3 percent of the French population is able to subscribe to does not do much to convince me we should reverse national broadband policy.