Berkman broadband study got basic data wrong

Harvard’s Berkman Centre was commissioned last year by the FCC to produce a report on international broadband deployment and usage. Its report was widely reported as being damning to the broadband industry structure in Canada and criticisms of the report were ridiculed based on the interests of funders of contradicting studies.

We pointed out numerous times that the report re-hashed flawed OECD studies [for example: here, here, here, here, here].

It turns out that Harvard’s researchers got it wrong when looking at one of the fundamental inputs that led them to conclude that American and Canadian regulatory frameworks need to be overhauled to support for European-styled open access, rather than the facilities-based platform competition enjoyed in North America. 

Yesterday, Bell Canada was asked by the CRTC to place on the public record a letter it sent to the Berkman study author, Yochai Benkler. The Berkman study had been made part of the record of a CRTC proceeding looking at mandating wholesale access to high speed services.

Two of the central themes of the Berkman study have come under question during the CRTC proceeding. First is the assertion that Canadian speeds and prices are too high. The CRTC commissioned its own study that appears to contradict this point.

But a more fundamental assertion of the Berkman study used erroneous pricing data to support its conclusion that Canada’s entire regulatory framework was flawed. Berkman said:

During the first generation transition and to this day, Canada has had some of the highest regulated rates for unbundling anywhere in the OECD.

And later, Berkman said:

Indeed, Canada has the highest monthly charge for access to an unbundled local loop of any OECD country.

Bell Canada has been trying to replicate the data that led Berkman to its conclusions. And according to the letter sent to the author and filed by Bell, Berkman didn’t look at the underlying data properly.

Bell Canada’s territory is divided up into seven (7) rate bands, specifically Bands A through G. Although Rate Band D is the middle rate band it represents only two percent (2%) of Bell Canada’s ULL, and is mostly associated with small towns and rural areas. Ninety percent (90%) of Bell Canada’s ULLs fall into Rate Bands A and B, the two least expensive rate bands for ULL.

So the Harvard study appears to have used Band D as a proxy for the average Canadian unbundled loop price, not knowing that it was actually completely unrepresentative of the Canadian marketplace.

As can be seen from the tables above, Bell Canada’s weighted average monthly ULL rate is among the least expensive in the OECD data set, with only five countries having lower ULL rates.

This critique seems to echo problems identified by NTT last fall which said that Berkman misrepresented the Japanese situation, a problem that could have been cleared up with a simple phone call.

As NCTA said in its criticism of the Berkman study, the authors abused their academic standing:

The Berkman Report, in short, is an advocacy piece, not the work of dispassionate scholarship that the Commission requested.

As I indicated yesterday, Len Waverman, Dean of University of Calgary’s Haskayne School of Business and author of the Nokia Siemens Networks Connectivity Scorecard will be speaking at The 2010 Canadian Telecom Summit next week. On June 8, we will host a number of sessions helping to develop strategies for Canada’s digital economic future.