Facts about US Internet access

If you ever want to understand just how powerful the cable distribution giants in America are, just listen to former TCI CEO John Malone. In 2011, he said “cable is basically a monopoly now.”  He said he was sorry that he had sold his TCI systems to AT&T. And he also said that the threat of wireless competition had been way overblown.

We learned this summer that John Malone is re-entering the cable distribution marketplace in America. He’s bought a large stake in Charter, and Charter has tax losses that Malone would like to use to buy Time Warner Cable. The goal: have the same bargaining strength as Comcast, by rolling up a bunch of cable distributors, without the pain of being a vertically integrated company – Comcast has to manage NBCU, and Malone doesn’t. Malone can then let the returns roll in. John Malone’s company is already the largest cable distributor in the world (larger than Comcast), and Malone himself is very smart and tough as nails.

John Malone knows an unregulated monopoly when he sees one. And the key fact he’s glommed onto is that high-capacity, low-latency data connections are addictive. Americans can’t get enough of them. An American’s only choice when he or she wants to buy a high-capacity connection to the Internet, in most parts of the country, is his or her local cable monopolist.

The large cable distributors in America – who never compete directly with one another – have clearly become the nation’s monopoly suppliers of terrestrial wired connections, each in its own footprint. Their market power is unrestrained. They can charge whatever they want for whatever services they choose to provide. They have little incentive stemming from either market pressure or oversight to upgrade to symmetrical fiber, which is the world standard, or to charge reasonable prices for world-class access.

Meanwhile, our former telephone companies, Verizon and AT&T, have retreated almost entirely to wireless services – complementary, non-threatening to the cable guys, and highly lucrative. Nothing shows the existence of this market division reality more clearly than the co-marketing, co-development agreement among Verizon Wireless and cable giants Comcast and Time Warner Cable (among others) that was approved by the FCC in 2012.

Recently, the incumbent communication companies in America arranged for the publication in The New York Times of two op-eds (in a single week) claiming that America was doing just fine when it comes to high-speed Internet access. But they are trying to confuse you. They are hoping that Americans don’t notice that they’re focusing on the wrong definition of high-speed Internet access and blurring two separate markets – mobile wireless and fixed connections.

First, the relevant market for everyone is (or should be) high-capacity, low-latency, symmetrical fiber connections to homes and businesses of at least 100Mbps. That’s what they have in South Korea, Japan, Sweden, and (soon) Australia and China. Right now, the vast majority of Americans are stuck with the cable guys’ product, which is very expensive (three or four times as expensive for the same download services as in other countries http://newamerica.net/publications/policy/the_cost_of_connectivity) and second-best (because it doesn’t provide symmetrical, or equal, upload capacity). It’s not fiber, and it’s under the complete price/service control of individual companies that, again, are subject to neither oversight nor competition and have no incentive to make the upgrade to fiber.

Second, mobile wireless access sold (at healthy margins) by Verizon and AT&T is a complementary product that isn’t substitutable for a high-capacity, low-latency wire at home. People who can afford it have both a mobile and a fixed connection. At least 83% of people who have a smartphone also have a wired connection at home. The steep capacity limitations of mobile wireless and, because of those capacity limitations, its enormous expense relative to wired or fixed access dictate that it isn’t a competitive substitute for that fixed connection.

Now, many Americans are not rich enough to afford the total cost of satisfying their needs for voice and Internet connectivity as well as access to video (or other high-capacity applications) and mobility. And so they choose wireless as the option that satisfies some of their needs at a cost they can afford. But this situation, in which some Americans rely on smartphone access sold by wireless carriers because that’s all they can afford, is merely amplifying and cementing existing inequalities in American society. It’s a bug, not a feature. You can’t cost-effectively enter a classroom virtually using a smartphone, or have a visit with your doctor online, or do many other things that require first-class, very-high-capacity, reasonably-priced access. This problem is about to get much worse as gigabit applications come online.

Because the existing incumbents (i.e., Comcast, Time Warner Cable, Verizon, AT&T) are doing very well in their separate worlds, they have no incentive to allow for any loose talk about changing the status quo. Hence, op-eds in mainstream media claiming that the U.S. is doing much better than everybody thinks.

Interestingly, the telephone industry mounted exactly the same campaign back in 1916 when they were worried about municipal telephone networks disrupting their local private monopolies: The industry claimed that America was succeeding with its private telephone networks in ways that Europe wasn’t.

The incumbents are missing the point. Northern European and Asian countries are upgrading to standard fiber Internet access connections to homes and businesses. The availability of very-high-capacity, inexpensive communications facilities in those countries will trigger direct and indirect economic and social benefits for their citizens, in the form of increased productivity and new ways of making a living. Here in America, we have no national plans for such an upgrade.

If we had a stagnant market in the U.S. we would be seeing high margins for leading companies that are not losing market share to competitors—and we are. We would be seeing that we are paying more for the same product than people in other countries are—and we are seeing that.

When it comes to connections that allow 10 Mbps or more in downloads and 200 gigabytes of data – a measure of capacity, or volume – per month, we are heading with increasing speed toward a series of regional cable monopolies in wireline high-speed Internet access.

Americans know that our communications picture is dire. Mr. McAdam’s and Mr. Bennett’s comments should be viewed with skepticism; the status quo is serving our existing giant telecommunications companies well.

Here are the facts.

Claimed “Intermodal” Competition in the U.S. is Marginal

McAdam: Almost everyone in the country has several competitive choices for high-speed broadband service (with wireline, satellite and wireless options).

Bennett: Much of the recent growth has come thanks to our system of facilities-based competition — that is, each service provider is responsible not only for broadband service but for the underlying infrastructure, which encourages them to improve network quality to win customers.

McAdam and Bennett claim that competition between differing connection technologies (fiber, cable, DSL, satellite, mobile wireless, and a handful of nascent fixed-location wireless technologies) is driving growth and protecting consumers. But these different modes of connection – we call the whole issue intermodal competition – don’t actually compete with one another.

Satellite connections suffer from latency, are expensive, emphasize downloads over uploads, and are generally considered unsuitable for 21st century online uses (tele-education, telemedicine, telepresence).  Satellite is the connection of last resort: Where DSL or cable is available, people don’t subscribe to satellite Internet access services.

DSL, which runs over copper phone lines, suffers from the limitations of physics. Its copper lines can carry only a narrow range of frequencies. In order to achieve higher speeds (or throughput) of data, engineers transmit data at the highest frequencies DSL can handle. But the higher the frequency, the steeper the falloff of a signal over distance. So very often the distance between the processing electronics and a customer’s home is too great to deliver a satisfying digital experience. The FCC recently reported that only 0.3% of DSL connections in America provide speeds of 25 Mbps or higher.

Moving forward, most people think DSL will not be able to compete with cable or fiber, knocking telcos without fiber networks out of the marketplace and leaving us with just cable or fiber as our primary fixed-location competitors. (Foreshadowing: For the vast majority of Americans, fiber to the home is not currently available. We’re left with just cable.)

Both Verizon and AT&T have recently raised DSL prices by $5 per month, showing that they are milking this asset rather than investing in fiber in DSL neighborhoods. And both Verizon and AT&T are backing away from their wired assets in general; they are dumping their wires because their overall direction is towards wireless. Although DSL took roughly half the new high-speed Internet access subscriptions in 2002, in 2012, less than one out of every eight new high-speed Internet access subscriptions went to the nation’s telcos–and that includes not only DSL but also their fiber deployments. With a choice between DSL and cable, it’s clear that Americans overwhelmingly choose cable. Why? Far faster speeds.


Fiber to the node won’t get us to the speeds we’ll need. AT&T’s U-Verse system is fiber to the node, meaning that AT&T brings fiber into neighborhood nodes and then runs signals over a pair of copper wires into homes. Speeds of 25 Mbps can be achieved by U-Verse, or even 50 Mbps downloads if two pairs of copper wire are bonded together (although AT&T is apparently offering only 24 Mbps speeds at the moment). But, as with DSL, actual U-Verse speeds are distance sensitive and so are often much lower than those cited in AT&T advertising. And U-Verse uploads remain very cramped. In general, takeup of U-Verse has been slow; it has taken U-Verse six years to reach the penetration that FiOS reached right off the bat.


Cable wins. Cable uses fiber to neighborhood nodes but links customers to those nodes using coaxial cable. While the latter has a lot more capacity than telco twisted pair copper wires, its capacity remains far below what an all-fiber network can deliver. The DOCSIS 3.0 standard adopted by cable operators fuses together four of 125 (or more) cable channels for the purpose of providing Internet access. DOCSIS 3.0 can make possible 100 Mbps download services. But even with DOCSIS 3.0, cable upload speeds remain far below their download speeds. Craig Moffett, a leading cable industry analyst, predicts that, in the third of the country that will be served by DSL services, the high-speed access business will shift almost entirely to cable by 2020. http://moffettresearch.com/letters

Fiber to the home is future-proof and symmetrical. Fiber optic connections to homes and businesses will allow for virtually unlimited capacity – a single strand of fiber, even now, can support 90,000 TV channels. Once the fiber is in the ground, its capacity can be upgraded by just swapping out the electronics in the network.

Verizon’s FiOS product is FTTH, fiber to the home. It’s objectively a better communications product than cable because it allows for symmetrical, equal uploads as well as downloads. The good news is that Verizon’s FiOS product is is a relatively strong competitive alternative to cable’s Internet access offering. The bad news is that Verizon halted expansion of its FiOS buildout in March 2010, which means that just 14% of the country will have access to FiOS. And, like cable’s high-speed offering, FiOS is very expensive. Currently, less than 5% of Americans (14.5M people, according to Verizon CEO Lowell McAdam) have access to FiOS; 5.6 million subscribe.

Comcast faces competition from Verizon’s FiOS in less than a fifth of its territory. Cablevision, by contrast, is competing with Comcast in almost two-thirds of its territory. Some cable companies are bigger and more important to one another than others; Comcast and Time Warner (with just 11% FiOS overlap) are strategically aligned in a way that sometimes leaves out Cablevision. The numbers show the difference: Comcast’s average revenue per video customer went up by 21% between the fourth quarter of 2010 and the fourth quarter of 2012; Cablevision’s average revenue went up just 2% over the same period. Malone is interested in Charter because only 4% of its territory overlaps with FiOS.

Verizon stopped expanding FiOS for a simple reason: Its existing phone lines are made of twisted copper wire. To build FiOS, it has to install a complete second network–roll in the trucks, rip up the streets, and put in fiber–essentially cannibalizing the existing network on which it still sells DSL service. That’s an expensive procedure, and Wall Street hates steep, long-term, up-front capital expenditures. Wall Street wants to see high free cash flow, ample dividends, and frequent buybacks.

Comcast, meanwhile, only has to swap out some electronics to shift its existing cable network to DOCSIS 3.0 services. Much, much cheaper. And a death knell to potential competition. While FiOS is objectively better because uploads and downloads across its fiber optics are evenly fast, these services will only be available on a very limited scale in comparison to the country as a whole.

When it comes to world-class fixed Internet access services, most Americans will be stuck with their local cable company. Those services are addictive and increasingly essential; John Malone has said, according to The New York Times, that people “would give up food before they would give up the Internet.”

Mobile Wireless is Not Substitutable. As John Malone said in 2011, “the threat of wireless has been way overblown.”

Although mobile speeds have gotten fast enough to meet the extremely low 3 Mbps threshold that the FCC has set as a definition of high-speed Internet access, mobile connections are not going to provide the speeds that we will need for next generation applications – and particularly not in a cost-effective way.

Let’s start with physics. The cable network currently offers hundreds of megahertz of usable spectrum to homes. But mobile wireless networks offer only tens of megahertz of spectrum to every cell, which is then shared with every person connected to that cell. Because the amount of data you can send depends on how much bandwidth – or range of frequencies – you can muster to carry that data, the laws of physics dictate that a coaxial connection has far greater capacity (thousands of times more capacity) than a wireless connection. In addition,wireless transmissions have to travel through the open air, which makes them subject to all kinds of interference from weather and objects. Frequencies inside coaxial connections travel free of interference. And fiber connections have even greater capacity than cable. So mobile wireless will never provide the capacity of coaxial cable or fiber.

Although mobile technology will continue to advance, and engineers will figure out ways to cram more and more data into the available mobile wireless spectrum, it won’t be enough to substitute for what’s possible over a high-capacity wire. The move from 3G to 4G roughly triples network capacity, but throughput – amount of data being jammed into networks by users – is growing much more quickly. The OECD says that mobile data traffic grew 15-fold in the five years between 2006 and 2011.  There is a new protocol coming for data transmission, but it is probably 5-10 years away from commercial deployment. Capacity constraints on mobile wireless will therefore be even greater in the near future than they are today. That’s why Cisco estimates that wireless data accounts for only 2% or so of all data traffic, and that wireless will still represent just 9% of traffic in four years.

Americans love mobility, and mobility is a plus for mobile wireless, but a fixed connection will always trump wireless when it comes to data capacity. And a connection shared with the members of a single household will always trump a connection shared with an entire neighborhood. This isn’t a question of legislation. It’s because of physics.

Because of these capacity constraints, Verizon Wireless and AT&T have tiered data plans, and most people sign up for about two or three gigabytes of usage per month.  (According to Verizon, “An account with 1 Smartphone and a monthly allowance of Unlimited Talk, Unlimited Text and 2GB of data would cost $100 per month, not including taxes and surcharges; An account with 2 Smartphones and 1 Basic phone sharing a monthly allowance of unlimited talk, unlimited text and 4GB of data would cost $180 per month, not including taxes and surcharges.”)  If you’re using an iPad 2 on a 3G connection, you’ll far exceed a 2GB cap – or even a 3GB cap – by simply downloading ten apps, watching five hours of YouTube video, and watching 10 hours of Netflix content in a month. Higher-quality HD watching will increase data usage greatly: For a user with an iPad HD downloading Retina-quality apps and streaming video over a 4G connection, Gizmodo estimates the same media habits will lead to more than 5GB of data usage a month: 14MB(10 apps) + 350MB(5 hrs YouTube) + 350MB(10 hrs NetFlix) = 5390MB or 5.26GB for the month. Not counting Web surfing or sending emails. This has worked out well for the telcos; they can charge steep overage fees (right now, $15 per gigabyte of data, according to Verizon) for data usage that exceeds caps chosen by customers.

As a result of both these capacity constraints and the expense of mobile connections, mobile wireless isn’t cost effective for HD video or next-generation gigabit applications. According to Craig Moffett, the median wired data user already consumes twenty times as much data as a mobile wireless user.  Mobile wireless (non Wi-Fi) use of Netflix, the most popular online application, is in the low single digits.

It’s important to note that the capacity limitations of cellular wireless are so great that in most areas of the world, including the U.S., the wireless carriers offload (or cause users to offload) half or more of their traffic onto Wi-Fi for delivery. You can think of Wi-Fi as a shared wire within a home or business; it’s not the same as mobile wireless services sold by AT&T and Verizon. Wi-Fi does carry a lot of data, but only across very short distances. According to Diffraction Analysis, “only a quarter to a third of data traffic consumed by mobile devices is actually delivered over mobile networks (except in India and Japan where it’s half).” People and devices use cellular wireless only when their free Wi-Fi (shared wire) connections aren’t available.

The bottom line is that a mobile connection doesn’t compete with a wired one. These are complementary products, not competitive markets. At least 83% of Americans with smartphones also have a wire at home.

So don’t be fooled by claims that there is wild competition between mobile wireless and other forms of Internet connectivity. Although mobile wireless subscriptions have increased, and now represent about 43% of the connections over 3 Mbps in America, mobile connections represent less than 11% of connections over 10Mbps in America. And these wireless connections come with usage caps that restrict their value for popular high-bandwidth applications like video streaming.

To sum up: Intermodal competition in America is sharply limited. The cable distributors have won the wired race and the telcos have ceded that territory to them. Mobile wireless, dominated by AT&T and Verizon Wireless, does not substitute for a 21st century wired connection.

It is true that some small percentage of Americans relies on mobile access for Internet connections, but that’s often because those subscribers can’t afford the total cost of having both wired and mobile connections and satisfying their need for voice and data connectivity (as well as entertainment) and mobility – so they choose mobile access as the option that satisfies more of their needs at a cost they can afford.

Some European countries – particularly in Northern Europe – are doing much better than the U.S. is when it comes to rolling out reasonably-priced fiber to the home connections to consumers.

Bennett: Over the last three years America’s broadband systems have doubled in speed, while Europe’s have remained stagnant. And that will continue, because broadband companies here are installing advanced fiber-optic technology faster than Europe, and most of the world’s users of the fastest mobile broadband technology, 4G/LTE, live in America.

Bennett is deliberately avoiding discussing fiber to the home connections. He is apparently focused on the very lowest speeds of Internet connectivity (“America’s broadband systems have doubled in speed”). He is right that abysmal connections in America have slowly been upgraded to very very slow connections. Those very very slow connections (in comparison to what’s possible with fiber) include mobile wireless.

Bennett is probably right that most users of 4G mobile wireless live in America. He is also probably right that our communications companies are installing fiber to some cell towers and in parts of the network providing dedicated access at great expense to companies.

But all of that – doubling in abysmal consumer speeds, focus on cellular wireless, installation of enterprise fiber – is beside the point. Consumers want fast wireline connections when they can get them, but in the U.S. wired connections remain both expensive and second-rate.

The truth is that in terms of fiber to the home availability (and fiber to the node availability), the U.S. is behind South Korea, the UAE, Hong Kong, Japan, Taiwan, Latvia, Lithuania, Norway, Sweden, Slovakia, Bulgaria, Portugal, Iceland, Denmark, Estonia, Finland, and Norway. Very few Americans have the option to buy fiber to the home connections at reasonable prices.

The truth is that in countries where fiber to the home is the standard, it’s available at far lower prices than it is here.

The truth is that mobile wireless is complementary to wired connections – unless it’s all you can afford.

The truth is that adoption of high-speed Internet access has plateaued in America.

We are subject to several great digital divides: There is a divide between Americans who have a wire at home and those that don’t, a divide between affluent Americans who will be able to pay for cable’s high-capacity (yet still second-class) high-speed connections and those who won’t, and a divide between America as a nation and those countries that have prioritized fiber access as a widely-available, reasonably-priced utility. All of these divides cast shadows over our nation’s future.

American growth in high-speed Internet access has plateaued.

Bennett: Much of the disparity between perception and reality has to do with timing. Before the recession, American Internet service was on a very different path, not keeping pace with large sections of Western Europe and East Asia.

But that began to change as the economy turned around. Private investment and advances in technology, brought about by a competition policy that encouraged cable and phone companies to improve their networks, have propelled America’s networks forward.

The FCC has set a low standard for what counts as high-speed Internet access.  They’re keeping track of anything over 3 Mbps download and 768 kbps upload speed – speeds that mobile wireless can provide. Recently, the FCC released a report that relies on industry-provided data about those connections. That report says that only 45% of U.S. households subscribe to a residential fixed-location connection with speeds that meet or exceed the (very low) FCC threshold of 3Mbps. (About 10% of fixed connections in the US are slower than 1.5Mbps. This makes us worse than Europe: In the EU as a whole, only 3.9% of lines are below 2 Mbps, and you’d have to go to Estonia, Poland, and Slovenia to be in a European country where more than 10% of lines are slower than 2Mbps.)

When it comes to ALL fixed-location connections – not just the ones that exceed the FCC’s very low threshold, but also the connections that are even slower than that – Bennett is correct that during the 10 year period between June 2002 and June 2012, residential fixed-location connections in the U.S. grew from 14 million connections to 82 million connections – at a annual growth rate of 19% per year. During that time, household adoption increased from 13 to 67 connections per 100 households (13% to 67%).

But growth in adoption slowed around 2007 and hasn’t picked back up. In 2008, over 100 million Americans, about one-third of the population, did not have an Internet connection at home. As of June 2012, 31% still do not have this utility service at home. We were at about two-thirds adoption five years ago, and that’s about where we are now.

We know the following about high speed subscriptions in Europe:

According to the EU, 14.8% of European fixed connections provide speeds of at least 30 Mbps (up from 9% a year ago), mainly thanks to the expansion of cable DOCSIS 3.0 lines. Penetration has gone from 2.5% to 4.2%.

The FCC sets a benchmark at 25Mbps in the U.S., and tells us that only 9-14% of fixed connections meet it.

We also know that about 2% of European homes are subscribing to at least 100Mbps, and that 3.4% of the fixed connections in the EU are above 100Mbps as of January 2013, a rate which doubled from 1.6% a year earlier.

In the US, the FCC numbers from June 2012 tell us that less than half a percent of fixed connections in the US meet the 100Mbps mark.  If we include mobile connections in the denominator, that number shrinks to less than 0.2%. http://transition.fcc.gov/Daily_Releases/Daily_Business/2013/db0520/DOC-321076A1.pdf

American investment in upgraded communications networks for consumers has slowed to a crawl.

The large communications companies in America have made their investments in their networks. They are done; they are now in harvesting mode, reaping enormous revenues while slowing capital investment. Instead of investing in better connectivity for all Americans – and upgrades to fiber to the home – they are instead sending large dividends to their shareholders.

Comcast and Time Warner Cable:  In 2012, Comcast invested just 12% of its revenue in capital expenditures, down from 37% in 2001; Time Warner Cable in 2012 invested just 14% of its revenue in capital, down from 30% in 2001.

For Comcast, total shareholder returns (dividends and buybacks) as a percentage of capital expenditure have climbed steadily, from 30% in 2009 to 81% in 2012.

Similarly, Time Warner Cable’s shareholder returns as a percentage of capital expenditure increased from 37% in 2010 to 82% in 2012.

AT&T and Verizon: Verizon’s wired capital spending as a percentage of revenue declined from 28% in 2000 to less than 16% in 2012; wireless, from more than 30% in 2000 to less than 12% in 2012. AT&T’s capital spending as a percentage of revenue was at 15% (wires) and 16% (wireless) in 2012.

Between 2002 and 2012, AT&T’s dividend increased by 64%, while Verizon’s expanded by 47%.  And for the past two years, both telcos have been paying out substantially more dividends per share than they’ve earned in profits. In 2012, AT&T’s shareholder returns-to-capital investment ratio actually exceeded 100%. This means that the nation’s largest telco returned 17% more cash to shareholders than it invested in both its wired and wireless networks combined.

It is not true that only 2% of households in the EU have access to 100Mbps service.

McAdam: Contrast this with the European Union, where innovation and investment in advanced networks have stagnated under an onerous regulatory regime that limits investment and innovation, and where today only about 2 percent of households have access to broadband networks with 100-megabit-plus speeds.


McAdam’s statement concerning access by ‘about 2 percent of households’ is flatly untrue. He is mixing apples and oranges: In the U.S., he is counting available connections – even if no one subscribes. In Europe, he is counting actual subscriptions, which are very low. Availability figures in several European countries are far higher than they are here.


For example, 12.2% of the households in Europe are reached by a FTTH connection that is certainly capable of 100 Mbps downloads – and equal uploads. If such a speed is not offered, it’s a deliberate choice by the operator.


Similarly, 39.4% of the households in Europe are reached by a cable DOCSIS 3.0 connection that is certainly capable of 100 Mbps downloads. If such a speed is not offered, it is a deliberate choice by the cable operator.


McAdam’s math is wrong.


There were 202.8 million households in Europe as of 2010.

2% of that number = 4.1 million.


In the Netherlands and Belgium, more than 95% of households are reached by a cable DOCSIS 3.0 connection that is capable of 100 Mbps.  . The Netherlands has 7.3 million households and Belgium has 4.5 million. That means that in just two European countries over 11 million households have access to DOCSIS 3.0. That accounts for over 5% of total European households, which debunks McAdam’s statement.


Now, just 2% of all households actually subscribe in Europe to a connection of more than 100 Mbps. .(Similarly, 2.6% of Americans actually subscribe to FTTH, according to the FCC.)  But McAdam said “have access to” – and that is misleading.


To compare apples to apples, the US is in fact doing worse than Europe when it comes to actual subscriptions to high-capacity connections. FCC numbers from June 2012 tell us that 0.3% of fixed connections in the US meet the 100Mbps mark. And, demonstrating once again that mobile is not a substitute, if we include mobile connections in the denominator that number shrinks to 0.2%.


Cost is an important reason that not all Americans have a reliable connection to the Internet at home.


Bennett: Our broadband subscription… could easily surpass 90 percent if computer ownership and digital literacy were widespread.


Bennett’s argument that computer ownership and “digital literacy” are the answer to low adoption in America –  implying that cost is irrelevant – is misleading. Most people have multiple reasons why they don’t adopt high speed Internet access. A recent study from NTIA asked non-adopters a single question – “what is the main reason you don’t have broadband?” and allowed only one answer.  In the NTIA study, 48% of non-users reported that they did not use the Internet at home because they didn’t need it and were not interested; 28% reported that it was too expensive; 13% reported that they did not have a computer, or that their computer was inadequate.
In other studies, where respondents are allowed to choose more than one reason, we see a different picture. In a 2009 FCC survey, 47% of non-Internet users in the survey said monthly cost was a reason they didn’t subscribe, while 42% said that the activation and installation fees were too high and formed a barrier, and 40% said they could not afford a computer. Of those who said monthly cost was a barrier, about 40% said they’d pay an average of $26 per month for Internet access at home. A 2012 study on high-speed Internet access adoption in Illinois found that 52% of non-Internet users point to monthly subscription cost as a reason, 51% point to an inability to afford activation and installation fees, and 43% point to their inability to afford a computer.


There’s a chicken-egg problem that may be at work here. Many people who can’t afford an Internet access connection don’t bother to acquire a computer, or can’t afford one. So the distinctions among these reasons are blurry and the reasons themselves are interrelated. But one thing is clear: cost is a major factor.


This isn’t to say that digital literacy isn’t a real issue. It is. But it is also true that those who can’t afford a connection have a more difficult time becoming digitally-literate, and that “I’m not interested in the Internet” is sometimes a response that masks an underlying affordability problem.


One recent 2012 study showed that even after going through digital literacy training 22% of participants still did not have a connection. Of those unconnected, a total of 54% said that they could not afford commercial broadband services. Of those citing cost as the barrier, $12.00 per month was the average price point that would make the service affordable.


A separate follow-up study, focusing on parents of school-age children, showed that 48% of non-connected respondents claim cost to be the biggest barrier to their ability to establish a home broadband connection. When these respondents were asked what price would make broadband affordable, the average cost cited was $9.75 per month. Although 42% feel that a broadband connection in their home is not relevant, 71% of these feel this way because they have been unable to afford a computer.


A study on variations between Chicago neighborhoods tells us that those who are not interested are generally older and that age correlates strongly with interest. In Chicago, income is the major factor explaining concerns about cost, and residents who say they can’t afford Internet access are statistically more likely to be low‐income. Latinos and African‐Americans are among those more likely to cite cost concerns.


Similar patterns exist in Europe, although usage patterns vary sharply among different EU countries. The three main factors in the EU for non-use are cost, lack of skills, and lack of interest. Cost is particularly important as a reason for non-use in Belgium, Bulgaria, Germany, Estonia, France, Hungary, Portugal, Romania, and Slovenia. In countries with fewer non-users, lack of interest is more important. Cost is definitely a major issue, though, and the importance of cost as a barrier to adoption has remained strong since 2006. .


High-capacity, reasonably-priced, low-latency connections are too rare in America; second-rate cable connections are provided here at monopoly prices. In other countries, fiber to the home connections are or are becoming standard.


Bennett: There is a popular story going around about the state of America’s broadband networks: service is pitifully slow, hugely overpriced and limited to the richest neighborhoods — whereas in Europe, service is cheap, fast and widespread because regulators force big companies to make room for smaller service providers


When it comes to the future of American industry and society, our focus should be on high-capacity, low-latency, reasonably-priced connections. That’s where much of Asia and Nordic countries are heading – and many of them are already there. America, by contrast, doesn’t have a plan for a fiber upgrade, and we’re stuck with second-rate cable connections that are not reasonably priced, don’t provide symmetric (equal) upload capacity, and don’t allow for retail competition across their lines. This is an enormous social policy problem for the country.


The fiber story in the U.S. is meager. Verizon FiOS is the largest provider of FTTH access in North America, with most of its installations on the east coast. Verizon serves about 6 million customers and will pass (make fiber available to) about 18 million lines when its buildout is complete. According to the FCC, as of June 2012, just 2.6% of the American population subscribed to fiber. According to the Fiber to the Home Council Europe, if current trends continue, the US is predicted to reach 20% fiber to the home penetration in 2019.


Other countries are moving ahead. All of Russia, Denmark, Portugal, China, Australia, France, and the Netherlands are slated to reach the 20% household adoption rate of fiber to the home one to four years ahead of the U.S. In Japan, 42% of all fixed high-speed Internet access lines are already FTTH. In South Korea, 58% of all lines are FTTH.


Asia is marching ahead. Japan, Korea, and Taiwan reached the 20% adoption mark for fiber to the home between 2007 and 2009. In China, South Korea, and Japan, fiber passes by (is available to, even if people don’t subscribe) an average of 32% of households. At the end of 2011, China had over 37 million FTTP subscribers.


In 2012, China experienced the biggest growth of any country with respect to FTTP being made available.


In Hong Kong, FTTP services have been available for longer than in China, and, while much smaller scale in absolute terms than mainland China, deployments and adoption as a proportion of households are significant. In Hong Kong, 39% of households subscribe to FTTP, and in Taiwan, FTTH adoption has already overtaken DSL; incumbent Chunghwa Telecom has been installing FTTH and already has over 2.3 million subscribers. (Diffraction Analysis.)


Other European countries are marching ahead. Many Nordic and Eastern European countries have already passed a high percentage of their homes with FTTH.

In Denmark, Norway, and Sweden, fiber deployment has been driven mostly by local and regional utilities. Sweden (less densely populated than the US) reached the 20% adoption mark for FTTH in 2012; according to Diffraction Analysis, Sweden was at the forefront of the fiber access revolution, in part due to public policies designed to help municipalities connect public buildings. There are over 250 municipal networks in Sweden, most of them providing wholesale access at a fixed price as an input into retail competition. In Norway, there are FTTH networks in many tier-2 cities. FTTH passes 39% of Bulgarian households. Even Moldava has hit 20% penetration for FTTH. The Baltics are going quickly: In Lithuania, 47% of homes have FTTH, and Latvia is seeing high take-up rates of FTTH – 67.7%. Turkey is in 8th place among the G20 in terms of FTTH adoption. (Diffraction Analysis.)


Other countries have policies we seem incapable of adopting. As the OECD recently pointed out, other countries are imposing genuine regulatory policies that ensure true universal coverage and competitive prices. In Israel, the Israel Electric Corporation (IEC) utility plans to wire up over 66% of the country with FTTH over the next seven years, with some contributions from public funding contributions. Australia is continuing to follow a plan for National Broadband Network (NBN) that will bring FTTH to 93% of households in the country by 2021. http://www.diffractionanalysis.com/blog/2012/03/22/free-report-world-of-fiber-2012.html Several countries (S.Korea, Hong Kong, China) require wholesale providers to allow retail competition over their lines. http://www.oecd-ilibrary.org/science-and-technology/oecd-communications-outlook-2013_comms_outlook-2013-en  at 40-43.

In Korean cities, and in Hong Kong and China, direct competition by retail providers allows multiple, competing fiber networks to be run right into the basements of many apartment buildings, intensifying competition. http://www.oecd-ilibrary.org/science-and-technology/oecd-communications-outlook-2013_comms_outlook-2013-en

The U.S. is, at best, in the middle of the pack of developed nations when it comes to speed.


Observed speeds: Although ranking numbers are certainly contestable, the U.S. is not doing well in terms of speed of online connections. Bennett and other “all is well” commentators have focused on Akamai numbers; even according to Akamai, we’re 8th. Here are some other numbers.

Pando Networks, another content delivery network (CDN), puts US high-speed Internet access speed at 26th worldwide, at about a quarter of the speed of world leader South Korea. According to PandoNetworks, Eastern European nations dominate the top of the list (Romania, Bulgaria, and Ukraine), with speeds that are about double or triple those in the US.

Ookla, which pulls its figures from Speedtest.net, a popular self-indexing site, and reports them on NetIndex.com, puts the US at 33rd, behind the EU average, and well behind the UK, the Nordic countries, most of Eastern Europe, and Japan and South Korea. http://www.netindex.com/download/allcountries/  Unlike Akamai, Ookla employs a method that aims at “filling the pipe” of a user initiating a test. (Akamai’s tests don’t measure the unused capacity of fast connections, and tend to collapse the differences between fast and super-fast connections.)

M-Lab also puts the US somewhere in the middle, slower than Belgium, Denmark, Finland, Germany, Hungary, Japan, Luxembourg, Netherlands, Norway, Sweden, and Switzerland.

If we look at just Netflix subscribers, and focus on just streaming video speed, the US sits just ahead of the UK and Ireland, well ahead of Mexico, but behind all of the Nordic countries. (Netflix reports on only these countries).

Tussling over contestable rankings is not a good use of our time. The undeniable truth is that the U.S. is not building the last-mile access networks — the FTTH connections — we will need to compete in the future. Other countries are. We should be looking in our rearview window at the rest of the world. Instead, we’re behind the leaders. We’re also handing our local cable monopolists extraordinary market power that affects consumers and businesses that would like to reach those consumers.

Offered Speeds: Even without focusing on fiber, we’re behind Europe. When it comes to offerings, the average advertised download speed for OECD nations is 44.44 Mbps. The average advertised download speed for the US is 44.69, behind Canada, Denmark, Finland, France, Japan, Korea, Luxembourg, Netherlands, Norway, Portugal, Span, Slovenia, Sweden, and the United Kingdom.

Speeds of Actual Subscriptions: When it comes to actual subscriptions, we are behind.

We’re behind for speeds 10 Mbps and higher. In the EU, 59% of all fixed connections now provide speeds of 10 Mbps and above. In Bulgaria and France, already roughly 90% of lines are at least 10 Mbps.


In the US, that number is only 48.3% as of 2012.


            We’re behind for speeds 25 Mbps and higher. At least 14.8% of European fixed connections provide speeds of at least 30 Mbps (up from 9% a year ago), mainly thanks to the expansion of cable DOCSIS 3.0 lines.

The FCC sets a benchmark at 25Mbps, and tells us that only 9-14% of fixed connections meet it as of June 2012.  This is an interesting fact: DOCSIS 3.0 and fiber lines are certainly capable of providing 25 Mbps speeds, but the companies that operate these networks generally choose not to implement this capability. Only 14.2% of cable connections and 21% of fiber connections reach the FCC’s 25Mbps benchmark.

In the U.S., cable high-capacity service is expensive. Now, the cable industry and its brethren claim that 82% of Americans have access to wired high-speed Internet access of speeds exceeding 100 Mbps. But these services, to the extent they’ve actually been implemented, are provided in at least 80% of the country by only the local cable monopoly providers. They’re not routinely available to Americans, and if they are they’re very expensive.

McAdam: More than 80 percent of American households live in areas that offer access to broadband networks capable of delivering data with speeds in excess of 100 megabits per second.

Let’s unpack the cable industry’s assertions. Verizon CEO Lowell McAdam tells us that “more than 80 percent of American households live in areas that offer access to broadband networks capable of delivering data with speeds in excess of 100 megabits per second.” He cites the the FCC’s Eighth Broadband Progress Report (August 2012).  The FCC report actually says, “According to industry reports, DOCSIS 3.0, which is capable of 100 Mbps speeds and even higher speeds, has been deployed to 82% of U.S. households,” and cites to the The National Cable & Telecommunications Association (NCTA), the principal trade association for the U.S. cable TV industry, and a Comcast press release on DOCIS 3.0 rollout.

Not only are the sources suspect, but DOCIS 3.0 deployment doesn’t mean that the cable company is actually offering a 100Mbps plan or that it plans to do so soon. It means only that DOCIS 3.0 technology has been deployed in the cable modem termination system (CMTS) located at the CATV headend in a particular neighborhood.

Because of the uncertainty of using industry-provided data on DOCIS 3.0 rollout as a proxy for available network speed, the FCC’s Eighth Broadband Progress Report also pointed to government collected data (collected semi-annually through state-led efforts and maintained by NTIA for the National Broadband Map), which tells us that these speeds were only available to 26.9% of the population as of June 2011 – much less than the 80% cited by McAdam.

A better, updated source would be the NTIA numbers as of June 2012, published May 2013, which tell us that 100Mbps speeds are available to just 47.09% of the population, still nowhere near the 80% cited by McAdam. In fact, according to the FCC, as of June 2012 only about 14% of cable connections subscribed to by Americans actually reached 25 Mbps.

These low subscription numbers may be explained by the fact that the cable industry’s high-speed Internet access services are extraordinarily expensive: Comcast charges $114.95 per month for 105 Mbps download services.

As I have written in the past, people in places like Seoul, Paris, and Amsterdam pay only $35-$45 for high speed connections that are faster than those for which we pay $90 or more per month. 2013 data from the OECD shows that people in the U.S. buying high-speed connections pay at least $1.75 per Mbps, where people in South Korea pay $.16 per Mbps.  And, unlike U.S. offerings, these connections are also symmetrical, delivering these very high speeds in both directions. You would have to move to Turkey or Mexico to pay more per Mbps than we do.

The OECD has developed a methodology for reporting market prices of fixed high-speed Internet access subscriptions. The method involves dividing ISP offerings into 10 baskets– five “lower use” baskets of various speeds with low data caps, and five “higher use” baskets of various speeds with higher data caps.

The OECD’s most recent 2013 report states that for the slowest two baskets, the United States ranks 20th and 18th of 24 OECD nations for price. For the remaining eight baskets, the United States ranks from 26th to 31st of 34 nations. We pay a lot for our fixed line internet connection subscriptions.

The separate, complementary service of mobile wireless is cheaper in Europe than in the U.S.

Bennett is probably right that the U.S. has more LTE subscribers than any other country. We’re a big country. European coverage of 4th generation LTE networks tripled in 2012.

And the prices of mobile data in the US are among the highest in the world.  For those who fear that ever-increasing consolidation and “family budget-eating” cellular service bills are unavoidable, it’s useful to compare the U.S. and European markets for mobile communication services.

In a recent report, Bernstein Research analysts Robin Bienenstock and Craig Moffett undertook such a comparison.  Noting that Verizon and AT&T enjoyed much higher market valuations than European telco incumbents, the Bernstein analysts pointed to the pricing power of the two dominant U.S. wireless providers as a key reason for these very different valuations. As they explained:

Pricing power! It’s a concept unheard of in Europe where ARPU per capita has fallen by 15% over the past 5 years despite smartphone penetration of over 40%. In Europe today 1GB of data and unlimited minutes and texts can now go for as low as EUR 9.00 (or $12) a month. A comparable service in the US goes for $50 with a BYO T-Mobile plan to $90 for Verizon (almost an  order of magnitude higher…!)…

The nature of European wireless is a market structure that is far more conducive to competition than that of their American cousins. In Europe this structure has created room for upstarts whose secret weapon is a new lower cost model, and through years of intense competition nurtured a generation of managers whose default competitive tool is price-cutting.

Among the industry dynamics impacted by the relative lack of competition in the U.S. wireless market is ARPU – far higher per capita in the US than in Europe. Similarly, churn is significantly lower in the U.S. market.

The Bernstein analysts went on to review key elements they believe distinguish the U.S. wireless market from its European cousins.

1.      “The USA is big and empty.”

2.      In the U.S., “the two largest operators have historically enjoyed a near monopoly of the best spectrum giving them unparalleled network differentiation.”

3.      “The absence of Chinese equipment manufacturers has kept telecom equipment prices high (versus Europe…), making it harder for new entrants to build a big network cheaply and quickly.”

4.      “The use of different technologies and the nature of CDMA have created huge device-related barriers to exit for consumers.”

5.      “The disproportionate success of Apple’s iPhone, and its continuing semi-exclusivity, has wildly skewed the attractiveness of operators.”


“In this environment,” says the Bernstein report, [Verizon and AT&T] “managers have learnt that if they invest in their networks they can give consumers more, and charge more for new products and meaningful network differentiation.”

The resulting industry dynamic, according to the Bernstein analysts, has unfolded something like this:

“The weakest players haven’t the margins to support price cuts and are happy to follow VZW and T. And boy, are T and VZW not afraid to use it. Verizon and AT&T have gradually killed off unlimited usage and raised fees for new devices. To extend their advantage further the US operators have introduced family and data sharing plans.”

In addition to the fundamentally anti-competitive deal Verizon has struck with cable operators (that has been approved by regulators), the fact remains that Verizon and AT&T control most of the lower-band spectrum that has especially strong propagation characteristics, including the recently licensed 700 MHz spectrum and the original 800 MHz cellular band.

And both wireless giants appear to be pursuing a strategy in which their LTE devices will operate only on their own 700 MHz bands, not on those used by their competitors, which include rural and regional carriers that were successful bidders for more localized 700 MHz licenses.  Unless the FCC takes some action on this front, as proposed by the Competitive Carriers Association, these smaller competitors may never achieve the critical mass needed to support high-volume, low-cost LTE devices and genuinely competitive LTE offerings.

If this restrictive device strategy and other actions by the Big Two weaken (or even destroy) competition in the key 700 MHz band, the prospects may be greatly diminished for the U.S. wireless market to evolve into a healthier competitive scenario, in which “good spectrum” is more evenly distributed,  mobile technologies are more interoperable, and the cost for consumers to switch carriers and support their wireless communication needs is significantly reduced.

I’ll keep talking about these issues. Don’t be misled by our existing giant providers. Even if their words are published by The New York Times.

Leave a Comment