The United States and Canada occupy places somewhere in regulatory purgatory for having - respectively - scrapped the requirement that phone companies share their lines with broadband competitors, or supposedly pursuing such policies only half-heartedly. Conventional wisdom has it that they have consequently fallen behind the rest of the developed world, and this belief seems to be backed up by an abundance of comparative data, combined with calls for re-regulation. Yet many of the voguish comparisons between North America and Europe (which is often held up as a pacesetter) are really comparisons between apples and oranges. Economists with extensive practical experience of telecommunications regulation have already rebutted the Berkman Center report that harshly assessed Canadian broadband performance, but it is also worth pointing out how much room for interpretation there is in broadband comparisons.
The standard take is that Canada is 10th and the United States is 15th among 30 OECD countries in broadband penetration. The OECD measures broadband penetration as the number of broadband lines per 100 persons, and mixes business and residential connections. Residential broadband subscriptions, however, are taken at the household level, not at the individual level. And big businesses often connect several hundred employees with one "line." The United States and Canada have 2.6 individuals per household, compared with 2.2 in Germany and some other European countries. Thus, if North American household sizes fell to German levels, and all households subscribed to broadband, the United Statse and Canada would have an additional seven lines per 100 persons.
Moreover, large businesses account for a larger share of employment in North America than in Europe. In Canada, firms with 500 or more employees account for 43 per cent of total employment, while in Europe, firms with more than 250 workers employ just 33 per cent. Thus there could well be more employees "connected" in North America, although there might be fewer connections. If all large businesses were magically split into two, the number of business broadband lines could be doubled in a fell swoop. The "lines per 100 persons" measure is misleading and less conceptually solid than measures based on household and business adoption rates. Indeed, both business and household adoption rates should be measured separately and in different ways.
TRUE PENETRATION RATE
Canada has a true broadband penetration rate of close to 70 per cent of households. And North Americans use the Internet somewhat more intensively than do Europeans, according to Cisco Systems data on Internet traffic. Further, business Internet traffic in North America appears to be at levels substantially higher than elsewhere in the world. Sadly, there is little systematic effort by international agencies to measure the intensity of Internet usage.
Instead, we see comparisons of advertised speeds and "price per advertised megabit," which are especially misleading. Advertised broadband speeds vary from actual speeds. In North America, this is largely a result of "network overhead," and is quite modest. In Europe, however, the variation is often dramatic.
Real-world speed testing efforts, while not perfect, tell a dramatically different story from comparisons of advertised speeds. Using real-world data on the amount of time taken to deliver files to end users from its global network of servers, Akamai Technologies reports that the average download speed for Canada was 4.2 megabits a second, against 3.2 Mbps for France, whereas the OECD finds that the average advertised speed from French ISPs was a staggering 51 Mbps. Akamai data ranked the United States and Canada ahead of all the larger European countries, though they trailed an elite group of smaller northern European and East Asian countries. Similarly, Cisco Systems, Oxford University and the University of Oviedo in Spain, put the United States and Canada ahead of France, Germany, Britain, Spain and Italy, and behind only Japan and Korea among the larger countries, in a study measuring broadband quality and leadership.
NOT ALWAYS AS ADVERTISED
Fifty-Mbps speeds (and their prices) are representative of user experience only where advanced fibre and cable networks are widely on offer. Although parts of France have developed impressively in this regard, such networks are accessible to at most 25 per cent of households, and the take-up of high-speed services is very low. Canada is likely soon to have a proportion substantially higher than France's of homes served by advanced fibre and cable networks that can deliver such speeds, thanks in part to the ubiquity of cable networks that are less costly to upgrade. Advertised speeds and price comparisons are apt to be remarkably inconsistent with broadband realities, even when they seem fairly straightforward. In any case, household and business adoption rates and usage levels are good evidence of affordability and availability, and are at least somewhat easier to compare.