Canadas Usage Based Billing Controversy: How to - Michael Geist

Canada’s Usage Based Billing Controversy: How to Address the Wholesale and
Retail Issues
Michael Geist
•
March 2011
•
Canada Research Chair in Internet and E-commerce Law, University of Ottawa, Faculty of Law. My
thanks to Keith Rose for his remarkable research and analysis on the Bell deferral account and the
network costing exercise found at Appendix B, to Peter Waldkirch for his global UBB comparison
research that comprise Appendix A, to Tyler Nechiporenko and Rachel Gold for their timely research
assistance, and to several reviewers for their helpful comments. Thanks to Netflix Inc. for its support of
this research, which was offered after I published several long blog postings on usage based billing in
February 2011. Any errors or omissions remain the sole responsibility of the author.
Data caps on Internet service – frequently referred to as usage based billing (UBB) – took
the political world by storm in Canada earlier this year when over 450,000 Canadians
signed a petition calling for an end to the common practice.1 After the federal government
indicated that it would order the Canadian Radio-television and Telecommunications
Commission to reconsider its decision to allow large Internet providers such as Bell to
implement UBB when it sells wholesale access to independent ISPs, the CRTC
announced that it would delay implementing the decision for at least 60 days so it could
review it on the merits.2 That delay will now likely run until late 2011, as the
Commission has announced plans for an oral hearing in July 20113 and Industry Minister
Tony Clement has intimated he plans to address the broader policy issues as part of the
forthcoming national digital economy strategy.4
While the CRTC reconsideration holds the promise of an amended policy – the
government has made it abundantly clear it will reject any attempt to bring back the same
approach5 – the narrowness of the current regulatory issue, which is focused on wholesale
UBB, is likely to leave many dissatisfied. Indeed, the UBB issue is at least three issues
rolled into one – wholesale UBB, retail UBB (the imposition of UBB by the large,
incumbent providers to their retail customers), and the broader concerns about the
competitiveness of Internet access services in Canada.
This paper proposes policy solutions to all three issues, concluding that a one-size-fits-all
approach is both unwise and unworkable. Rather, the policy and regulatory measures
needed to address wholesale UBB are different from retail UBB, though both are closely
linked to the state of competition in Canada.
On the issue of wholesale UBB, the paper proposes several steps including rescinding the
earlier series of CRTC decisions, ensuring that speed matching is implemented and
vigorously enforced,6 reversing the CRTC ADSL-CO decision to promote more facilitiesbased competition by encouraging independent ISPs to locate closer to residential
customers,7 examining alternative approaches that allow independent ISPs to differentiate
their services, and prioritizing cable TPIA service as a viable alternative for independent
ISPs.
1
2
3
4
5
6
7
StopTheMeter.ca, Half a Million Canadians Against Telecom Price Gouging, www.stopthemeter.ca.
See e.g., Laura Payton, “Use-based billing court hurt economy: Clement”, CBC News,
http://www.cbc.ca/news/politics/story/2011/03/01/clement-talks-ubb.html.
Canadian Radio-Television and Telecommunications Commission, Telecom Notice of Consultation
CRTC 2011-77-1, “Review of billing practices for wholesale residential high-speed access services”,
http://www.crtc.gc.ca/eng/archive/2011/2011-77-1.htm.
“Where Canada Now Stands on Usage-Based Billing (UBB) for Internet Services”,
http://epistoli.blogspot.com/2011/03/where-canada-now-stands-on-usage-based.html.
See e.g., “CRTC must reverse internet usage ruling: Clement”, CBC News,
http://www.cbc.ca/news/canada/story/2011/02/03/crtc-internet-clement.html.
Canadian Radio-Television and Telecommunications Commission, Order 2010-632, “Wholesale highspeed access services proceeding”, http://www.crtc.gc.ca/eng/archive/2010/2010-632.htm.
Id.
2
Given the CRTC’s longstanding forbearance of retail Internet services, addressing retail
UBB is more difficult. While some have advocated for the elimination of UBB at the
retail level, there is little reason to foreclose all forms of UBB. The Canadian Internet
access market is best served by a competitive environment in which those providers who
wish to offer reasonable UBB plans are entitled to do so. The problem with the current
market, however, is that many Canadian consumers have few options that do not include
UBB. A truly competitive market, like those found in other countries, would consist of
UBB as an option, but not the only option.
The optimal method for addressing retail UBB is a robust competitive environment in
which it is consumer ability to switch providers in the event they are dissatisfied with
UBB that serves as the chief safeguard against unreasonable UBB practices. Given that
the Canadian market does not have that state of competition at the present time, I argue
that new guidelines are needed for Internet Billing Usage Management Practices
(IBUMPs). Modeled as an extension on the CRTC’s Internet Traffic Management
Practice guidelines, which received grudging support from across the spectrum as a
reasonable method of addressing the concerns associated with net neutrality and Internet
provider traffic management, I argue that additional guidelines are needed for retail UBB.
The guidelines would include transparency requirements and a reasonableness standard to
safeguard against the use of UBB for anti-competitive purposes and against exorbitant
pricing schemes that leave consumers on the hook for bills far disproportionate to actual
provider costs (including a reasonable profit).
The need for IBUMPs – much like the need for ITMPs – is closely linked to the state of
competition in the Internet access marketplace. I point to several steps that should be
taken to encourage a more competitive environment. In addition to facilitating
competition by revisiting the manner in which independent ISPs compete in the
marketplace, I raise additional measures that include opening the door to foreign
competitors, working with provincial and municipal governments to foster new
innovative broadband networks, expanding the current federal broadband programs to
address both access and competition, and enlisting CANARIE, Canada’s research and
education network, to identify local community networks that would benefit from
connecting to its broadband backbone.
The discussion and analysis is divided into two parts. Part one addresses several of the
most hotly debated UBB issues including claims that UBB is a reasonable response to
network congestion and heavy usage by a small number of subscribers, that UBB is a
common Internet provider practice and that the Canadian situation is consistent with
practices in other jurisdictions, and that UBB pricing reflects the costs associated with
data transmission.
The second part expands on the recommendations for wholesale UBB, retail UBB, and
Internet provider competition noted above.
3
Part One – Setting the UBB Record Straight
1.
a.
UBB and Network Congestion
A Decade of CRTC Congestion Concerns
UBB supporters have long pointed to concerns about network congestion as a key
rationale for imposing usage based billing measures on their subscribers. The argument
is easily understood – network capacity is limited and congestion on the network
negatively affects all subscribers. By imposing measures designed to limit network
congestion, the small number of “heavy” users who contribute most to the network
congestion do not adversely affect the large number of light or average users, whose
Internet use has a more limited impact on network congestion.
The link between UBB and network congestion dates back over ten years, when the
CRTC first approved volume usage rate restrictions for the wholesale service offered by
cable Internet providers known as Third Party Internet Access (TPIA).8 Shaw and
Videotron sought approval for volume usage rate restrictions and the Commission
approved the request based on the need to “ensure fair and proportionate use of the
service by all end-users.”9
At the time, cable Internet was a shared service and the potential for one end user to
affect the service of a neighbouring user was widely recognized. Accordingly, the CRTC
established the following policy, conditioned on the cable providers establishing the same
volume usage rate restrictions on their own residential customers:
Shaw and VidГ©otron propose to apply monthly volume usage rate restrictions to endusers. For example, Shaw proposed an $0.80 charge per 100 megabytes per end-user
whose monthly downstream traffic exceeds five Gb and a $2 per 100 megabytes charge
for end-users with monthly upstream traffic exceeding two Gb. The carriers generally
submitted that their cable network is a shared network where the quality of service
available to one end-user is a function of other end-users' usage. Accordingly, they
propose traffic thresholds to ensure fair and proportionate use of the service by all endusers. Shaw and VidГ©otron indicated they would apply the same usage restrictions and
charges to their Internet service end-users as they propose to apply in respect of ISPs'
end-users.
The Commission approves on a final basis the volume usage rate restrictions proposed
by Shaw and VidГ©otron. To the extent that these carriers do not apply these volume usage
rate restrictions and associated volume thresholds to end-users of their own Internet
services, the Commission is of the preliminary view that such action would be contrary to
s. 27(2) of the Act.
8
9
Canadian Radio-television and Telecommunications Commission, Order CRTC 2000-789, “Terms and
rats approved for large cable carriers' high speed access service”,
http://www.crtc.gc.ca/eng/archive/2000/O2000-789.htm
Id. at para 103.
4
Rogers and Cogeco have not proposed volume usage rate restrictions. However, the
Commission considers it would be appropriate for each of these carriers to adopt, at its
option, volume usage rate restrictions and associate volume usage thresholds similar to
those proposed by Shaw and VidГ©otron. In this event, to the extent Rogers or Cogeco do
not apply the same restrictions to end-users of its carrier's own Internet service, the
Commission is also of the preliminary view that such action would be contrary to s. 27(2)
of the Act.10
While cable providers received regulatory support for UBB, there has since been very
limited uptake of TPIA, with some cable providers rendering it difficult for independent
ISPs to adopt the service.11
The path to UBB for telecom providers such as Bell, which offer Digital Subscriber Line
(DSL) services, was somewhat different. With retail Internet services forborne in 1999,12
Bell was free to implement UBB for its retail customers, which it did in 2006.13
Shifting UBB to its wholesale customers was more challenging, however, since
independent ISPs purchase a regulated service known as Gateway Access Service. The
regulated GAS, much like the TPIA service for cable, is not an Internet service but rather
a connection between end users and the independent ISP. It is important to note that the
actual provision of Internet services comes from the independent ISP, not from carriers
such as Bell. Independent ISPs need the GAS in order to reach the end users themselves,
since only telecommunications and cable companies have the "last mile" connection to
the customer (i.e. the copper telephone line or cable connection to the home). Many
countries require some form of open access to this last mile (particularly for telecom) in
order to enhance competition among Internet providers.14
The Commission’s UBB support as a mechanism to address network congestion was
reinforced in the 2009 Internet traffic management practice guidelines as it emphasized
its support for economic measures as preferable to “technical measures” that restrict users
access through technical means such as “throttling.”15 The Commission started from the
position that “parties generally acknowledged that some traffic management is required
10
11
12
13
14
15
Id. at paras. 103 – 105.
Catherine Middleton & Annemijin van Gorp, Paper presented to the 37th Research Conference on
Communication, Information and Internet Policy: How Competitive is the Canadian Residential
Broadband Market? A Study of Canadian Internet Service Providers and Their Regulatory Environment
(Sept. 25-27, 2009), http://www.broadbandresearch.ca/ourresearch/middleton_vangorp_TPRC2009.pdf.
Canadian Radio-Television and Telecommunications Commission, Telecom Order CRTC 99-592,
“Forebearance from Retail Internet Services”, http://www.crtc.gc.ca/eng/archive/1999/O99-592.htm.
Karl Bode, “'New' Sympatico Caps For new customers starting December 3”, DSL Reports, 30 Nov.
2006, http://www.dslreports.com/shownews/79963.
See generally Yochai Benkler, Next Generation Connectivity: A review of broadband Internet
transitions and policy from around the world,
http://www.fcc.gov/stage/pdf/Berkman_Center_Broadband_Study_13Oct09.pdf.
Canadian Radio-television and Telecommunications Commission, Telecom Regulatory Policy CRTC
2009-657, “Review of the Internet traffic management practices of Internet service providers”,
http://www.crtc.gc.ca/eng/archive/2009/2009-657.htm. [ITMP]
5
to address congestion in order to ensure that all end-users receive acceptable Internet
service.” Given the need to address network congestion, the Commission noted that:
economic ITMPs would generally not be considered unjustly discriminatory, as they link
rates for Internet service to end-user consumption. Economic ITMPs also provide greater
transparency to users than technical ITMPs, as they are reflected in monthly bills.
Furthermore, these practices match consumer usage with willingness to pay, thus putting
users in control and allowing market forces to work.16
Indeed, the support for economic traffic management measures was firmly ensconced in
the test the Commission established for reasonable traffic management practices. The
test includes a requirement on ISPs to “explain why, in the case of a technical ITMP that
results in any degree of discrimination or preference, network investment or economic
approaches alone would not reasonably address the need and effectively achieve the same
purpose as the ITMP.”17
b.
The Carriers on Congestion
Although the CRTC has proceeded on the assumption that UBB is a response to network
congestion concerns, there is considerable evidence to the contrary. Indeed, both cable
and telecom operators have acknowledged – in both statements and practices – that UBB
is at best only loosely related to network congestion. For example, in November 2010,
Rogers, Shaw, Videotron, and Cogeco wrote to the CRTC in support of UBB and
acknowledged that it could not set pricing on the basis of costs:
UBB charges are first and foremost a mechanism for managing Internet traffic, that is to
say an economic ITMP. UBB rates and caps are set at levels where they are expected to
have an influence on the traffic patterns of end-users, particularly high consumption end
users, otherwise they do not achieve their purpose.
Any attempt to establish rates for wholesale UBB charges based on a narrow assessment
of incremental costs is destined to failure, as it would not take into account the broader
behavioral impacts that permit the attainment of the underlying carrier’s network
management objectives.18
Bell has been even more transparent about the profitability of UBB. In a February 2010
conference call with company analysts, CEO George Cope noted:
And, of course, we’re seeing the ever—the growth—the usage of the Internet with the
Internet sticks. And I think the really, really important thing for investors, which we are
obviously following and executing on, is making sure that it’s a usage-driven model—and
16
17
18
Id.
Id. at para. 43.
Cogeco Cable Inc., Quebecor Media Inc., & Shaw Communications, Submissions to CRTC re: TNC
2010-803, at paras. 23 – 24,
http://www.crtc.gc.ca/public/partvii/2010/8661/c12_201015975/1467242.PDF
6
that’s the model that we’re pursuing, and at this point, the model that the Canadian
market is pursuing. And that’s important to make sure we monetize this significant
opportunity for our investors, and at the same time, for our customers from a great
service perspective.19
In an August 2010 quarterly call, Cope stated:
our data revenue growth was 3.8% for our Residential Services business, particularly
driven through an increase in Internet ARPU of 3.3%. And interesting, almost all that
increase now coming from usage based billing as the demand for Internet use explodes
through the use of video services, and we’re continuing to see an increase in the revenue
per customer.20
Three months later in November 2010, Cope noted:
our residential services had an excellent revenue quarter from a data perspective, as
well, with data revenue growth of 5%, driven principally by the bandwidth usage revenue
being up 83% year-over-year.21
Later in the same call, Cope linked increased Internet video demands with growing
revenues – “as we see a growth in video usage on the internet, making sure we’re
monetizing that for our shareholders through the bandwidth usage charges.”22
In other words, while the company has claimed that UBB only affects a small percentage
of subscribers, it is the primary driver of increased revenues from its Internet access
services division.
Shaw has also told analysts that UBB represents an important monetization opportunity
closely linked to customer video usage. In a November 2010 presentation, executives
stated:
In the future, we believe our usage based billing plan will enable the further monetization
of our Internet business as data usage becomes more prevalent and common amongst our
customer base (ie. streaming of video)23
Competitors have made much the same point. For example, in reviewing Bell’s pricing,
Primus stated "it's an economic disincentive for internet use. It's not meant to recover
costs. In fact these charges that Bell has levied are many, many, many times what it costs
19
20
21
22
23
BCE Q4, 2009 Conference Call Transcript,
http://www.bce.ca/data/documents/reports/en/2009/q4/BCE_TRANSCRIPT_q4_09.pdf
BCE Q2 2010 Results Conference Call Transcript, at 6,
http://www.bce.ca/data/documents/reports/en/2010/q2/BCE_TRANSCRIPT_q2_10.pdf.
BCE Q3 2010 Results Conference Call Transcript, at 3,
http://www.bce.ca/data/documents/reports/en/2010/q3/BCE_TRANSCRIPT_q3_10.pdf
Id. at 11.
Shaw Communications Inc., Investor Update, at 5,
http://www.shaw.ca/uploadedFiles/Corporate/Investors/Presentations_And_Meetings/InvestorUpdateN
ovember2010Final.pdf.
7
to actually deliver it."24
In fact, Bell’s proposed UBB pricing for the GAS service similarly demonstrates that it
bears little relation to actual costs or network congestion. First, its plan is different in
Quebec, where there is a 60 GB cap, and Ontario, where the cap is set at 25 GB. The
difference is plainly a function of the competitive environment, where Videotron’s 60 GB
cap forced Bell to offer a similar cap in order to remain competitive.25 Moreover, Bell’s
plan features a 60 GB cap with an overage charge for the next 20 GB. After 80 GB, there
is no further cap until the user hits 300 GB.26 In other words, using 80 GB and 300 GB
costs the same thing. This suggests that the plan has nothing to do with pay-what-youuse but is rather designed to compete with similar cable ISP data caps.
c.
A Congested Network?
Network congestion is often treated as a catch-all for a network where demand exceeds
capacity and therefore cannot provide all users with maximum speeds. The CRTC has
defined it as “a situation whereby the amount of traffic transiting the network may lead to
a deterioration in service for some end-users.”27 A closer examination of incumbent
networks reveals that Internet traffic traverses at least three stages, two of which typically
do not raise significant congestion concerns.
The first stage is the so-called “last mile”, the link between the end-user and the “Central
Office” (CO), where ISPs begin to route user traffic to its intended destination. The last
mile is the most coveted part of the network since the ability to serve an end user with
Internet services depends on last mile access. While the last mile is a critical component
of the network, congestion is limited, particularly for DSL services provided by telecom
companies such as Bell. This should come as no surprise since the ISP is able to limit the
amount of data carried on the last mile by establishing a bandwidth speed consistent with
its network capabilities. Moreover, since DSL services involve a direct connection
between the end-user and the CO, there are no other users to congest the network.
The incumbent ISPs have acknowledged the absence of network congestion in the last
mile stage. At a recent Standing Committee on Industry, Science and Technology hearing
on UBB, Bell's Chief Regulatory Officer Mirko Bibic acknowledged:
There is a copper loop that goes from our central office to the home and all data travels
24
25
26
27
“Internet usage costs to rise”, CBC News, http://ca.news.yahoo.com/internet-downloading-costs-rise20110107-162621-824.html.
“Internet download limit slashed for many”, CBC News,
http://www.cbc.ca/news/business/story/2011/01/31/technology-internet-usage-based-billing.html.
Canadian Radio-television and Telecommunications Commission, Telecom Decision CRTC 2010-255,
“Bell Aliant Regional Communications, Limited Partnership and Bell Canada – Applications to
introduce usage-based billing and other changes to Gateway Access Services”,
http://www.crtc.gc.ca/eng/archive/2010/2010-255.htm.
Canadian Radio-television and Telecommunications Commission, Telecom Public Notice CRTC 200819, “Review of the Internet traffic management practices of Internet service providers”, at n. 6,
http://www.crtc.gc.ca/eng/archive/2008/pt2008-19.htm.
8
on that pipe: Internet traffic, television traffic, voice traffic, long-distance traffic. But
there are no congestion issues there. The real issue is when you get to the central office
and go behind it into the Internet. Fibe TV is completely different.28
The second stage runs from the CO until the ISP hands off the user to another provider.
This may occur at several different points. In some instances, an independent ISP may
co-locate with the incumbent ISP by installing their own equipment at the CO, so that the
user only uses the incumbent’s last mile, not their internal network.
In other instances, the incumbent ISP may aggregate the traffic of many users together,
delivering the collective traffic to the independent ISP (if the users are customers of an
independent ISP) at a later point in the network. Since the ability to deliver faster speeds
to users depends in part on shortening the distance between the user and the CO, the
investment in the network (which in many urban areas has been built for decades) is
focused on a closer connection to the end user. With a close connection to the end user,
the traffic is routed along the ISP’s internal network until it reaches the third stage – the
Internet.
Once the user traffic reaches the Internet stage, it moves between other network providers
to its end destination (e.g. a website or an email in-box). For traffic carried by
independent ISPs, this traffic transits separately from Canadian ISPs such as Bell since
independent ISPs acquire sufficient connectivity to handle their customers’ traffic
demands. For the incumbent ISP traffic that originates from its own residential
customers, it too is managed through peering and transit arrangements (described further
below). The Internet stage naturally involves a global network and there are no
significant congestion concerns at this stage either (or at least no congestion concerns
specific to Canadian ISPs).
The issue of network congestion in Canada is therefore largely limited to the second stage
when the user traffic enters the incumbent ISP’s own internal network (beyond the last
mile) until it is handed off to another provider such as an independent ISP, a peering
provider, a Content Distribution Network (CDN) provider, or a provider by way of a
transit arrangement for Internet services. While congestion can arise within this internal
network if the simultaneous aggregated traffic demands exceeds network capacity, there
are no additional costs to the ISP – the congested traffic all runs within its own network.
Moreover, there are many ways to address this congestion. In the case of independent
ISPs, the most obvious method is to hand off their user traffic closer to the CO so that it
does not contribute to the congestion.
For the larger incumbent ISPs, “technical measures” – also known as traffic shaping –
have become a common practice for addressing congestion issues. Virtually all of the
28
Canada, Standing Committee on Industry, Science and Technology, Evidence, (Feb. 20, 2011) at 1735
(Mr. Mirko Bibic),
http://www2.parl.gc.ca/HousePublications/Publication.aspx?DocId=4953635&Language=E&Mode=1&
Parl=40&Ses=3#Int-3739006.
9
larger Canadian incumbent ISPs use some form of traffic shaping to limit the bandwidth
allocated toward high-bandwidth applications during peak periods. The incumbent ISPs
apply these measures to all traffic regardless of whether it is their own retail traffic or
traffic that originates from independent ISP subscribers.
There is also reason to believe that incumbent ISP networks are large enough to handle
the Internet traffic without concern for congestion since the same networks
simultaneously carry other high bandwidth traffic such as IPTv. The issue may well be
one of bandwidth allocation. If Internet traffic demands continue to grow faster than the
other bandwidth demands running on the same connection, carriers could shift some of
the “space” reserved for services such as IPTv (or the chicken roasting channel in the
case of cable)29 to the Internet and thereby relieve some of the congestion pressures.
None of this suggests that consumer broadband demand is not growing rapidly. Driven by
increasing use of the network for streaming and downloading video as well as data
intensive games, it clearly is.30 However, the demands on the network are not outside
historical norms31 nor do they necessarily mean that the network is now “congested.”
Rather, it suggests that certain parts of the network may face greater congestion strain
during certain periods in the day, which can be addressed through several mechanisms,
including increased investment, technical measures such as traffic shaping, and a reexamination of bandwidth allocation on ISP networks.
2.
The Canadian Internet Access Market: Comparing UBB Around the World
Canadian telecommunications networks were once the envy of the world. No longer.
While there is some debate on the methodology and validity of the many studies that
have compared wireless and broadband networks worldwide, there is no disputing that
none rank Canada as a leader in either area. Indeed, while Canadian broadband networks
were once viewed as a global leader – Canada ranked as high as second only a decade
ago on some metrics – today it is at best a middle of the pack player with mounting
frustration among both consumers and businesses reliant on a world-class digital
infrastructure.32
The Canadian debate over the competitiveness of its Internet access services has been
shaped by the experiences in other countries. Many international studies have concluded
29
30
31
32
“All-chicken roasting TV set for Canada”, UPI News, http://www.upi.com/Odd_News/2011/02/24/Allchicken-roasting-TV-set-for-Canada/UPI-84491298567988/.
“Online Video Continues Steady Growth Streak”, Mashable.com, http://mashable.com/2010/02/05/deccomscore-video-data/.
In fact, the CRTC recently found that network demands are growing at roughly 40 to 50 percent
annually. See CRTC, supra note 10 at para 5. That is rapid growth, though historically growth rates
have been even faster at 100% per year: K.G. Coffman & Andrew Odlyzko, “Internet Growth: Is There
a 'Moore's Law' for Data Traffic?”, http://papers.ssrn.com/sol3/papers.cfm?abstract_id=236108.
Organisation for Economic Co-operation and Development, OECD Broadband Portal,
http://www.oecd.org/document/54/0,3343,en_2649_34225_38690102_1_1_1_1,00.html; Michael Geist,
“OECD Report Finds Canadian Broadband Slow, Expensive”,
http://www.michaelgeist.ca/content/view/4019/135/.
10
that Canada is no longer a leader, with consumers facing higher prices and slower speeds
than those found in other countries. For example, consumers in South Korea, the world
Internet leader, have access to basic plans from SK Broadband that feature guaranteed
download speeds of 100 MB per second with no data caps and no long-term contracts for
33,000 won or C$28.62 per month.33 There is no comparable plan available with any
provider in Canada.
The OECD has noted that Canada is one of the only jurisdictions in the world where
virtually all providers utilize some form of UBB.34 For most providers, data caps have
been in place for years: Rogers established a monthly cap to some of its plans in 200535
and Bell implemented data caps in 2006.36
While much of the focus has been on the widespread use of UBB in Canada, more
instructive is the near-uniform implementation of UBB. In Canada, UBB invariably
means a cap on monthly data use combined with overage charges for all data used that
exceeds the cap. The plans do not distinguish between peak and off-peak use, the source
of the data, or contemplate other solutions beyond overage charges.
The near-uniform Canadian approach is notable because other, more competitive
countries feature a wider variation in approach, as highlighted in the comparative chart at
Appendix A. Countries such as Germany give consumers the ability to select from
among unlimited and capped services. Deutsche Telekom, the largest providers, offers
unlimited plans, while providers such as 1und1 offer lower cost plans with a 100 GB data
cap. The U.S. market similarly features a mix of capped and unlimited plans. Comcast
offers a 250 GB cap and AT&T recently announced plans to implement a 150 GB cap.37
Other leading ISPs such as Cablevision and Verizon offer unlimited plans.
In addition to choice from among unlimited and data capped plans, other countries
feature far more variation in approach to UBB. In Australia, data caps are very common,
yet exceeding the cap does not result in overage fees. Instead, Australian ISP “rate limit”
subscribers that exceed their cap by reducing download speeds for balance of the
month.38 In other words, all plans are effectively unlimited plans but with different
speeds once a data threshold is met. The use of rate limiting is a relatively common
approach in countries with data caps as it is also found in New Zealand, Germany, and
India.39 Moreover, in response to competitive pressures, British Telecom recently
33
34
35
36
37
38
39
http://www.skbroadband.com/eng/iproduct/evalley.asp
Organisation for Economic Co-operation and Development Directorate for Science, Technology and
Industry, OECD Broadband Portal, Prevalence of explicit bit/data caps among surveyed offers, by
country (Oct. 2009), http://www.oecd.org/dataoecd/11/38/39575261.xls.
Karl Bode, “Rogers Applies 'Generous' Caps 'Keeping pace' with 'evolving Internet needs'”, DSL
Reports, http://www.dslreports.com/shownews/60361.
Supra, note 13.
Nate Anderson, “Is AT&T's new 150GB DSL data cap justified?”, Ars Technica,
http://arstechnica.com/tech-policy/news/2011/03/is-atts-new-150gb-dsl-data-cap-justified.ars
“Fairly comparing broadband ADSL plans in Australia can be difficult”, Internode Blog,
http://blog.internode.on.net/2010/09/13/comparing-isp-plans-in-australia/.
See Appendix A.
11
announced plans to eliminate data caps but retain rate limiting for heavy Internet users.40
The Australian market also features another notable variation on UBB. Many ISPs
feature two separate data caps each month – one for peak time usage and a second for
off-peak usage. For example, Optus offers several broadband plans which each feature
separate data caps for peak and off-peak usage. The peak period runs from noon until
midnight, while the off-peak period runs from midnight until noon. If a subscriber
exceeds the peak or off-peak cap, Optus reverts to a rate-limited service.41 IiNet offers a
similar peak/off-peak data capped service with consumers able to purchase as much as 1
TB of data per month.42
Another alternative approach is to distinguish between domestic and international data.
For example, Vodaphone Iceland offers a 10 GB capped plan but “all data downloaded
domestically is unlimited without any extra cost.”43 The effort to distinguish between the
origin or destination of data reflects different costs – regional data transfers may remain
within a single ISP’s network or is subject to a peering arrangement that reduces transfer
costs. By contrast, international data will typically require a transit arrangement that
increases data transfer costs.
Yet another approach is to distinguish between upstream and downstream traffic. Data
caps are still relatively rare in Japan, but those ISPs with data caps limit their application
to upstream traffic. Moreover, the caps are so high – 900 GB of upstream traffic per
month is standard – as not constitute caps in comparison with the policies in Canada.44
What makes these UBB approaches noteworthy when viewed from a Canadian
perspective is that consumers can typically choose between unlimited and capped plans.
Moreover, the capped plans elsewhere bear a more direct relationship to cost and
congestion concerns. In contrast, Canadian UBB bears little relation to actual cost, but is
instead largely a function of market dynamics and the lack of competitiveness.
3.
UBB’s Black Box: How Much Does it Cost To Transport a Gigabyte?
Costs related to Internet access pricing structures sit at the heart of the UBB debate, yet
the most important data point remains shrouded in secrecy. The incumbent ISPs have
long been reluctant to disclose their actual costs in maintaining their networks, arguing
that the information is sensitive, confidential commercial data. In recent months, owing
to the fact that data caps and overage charges are typically based on gigabytes of data, the
40
41
42
43
44
Matthew Lasar, “BT to UK Infinity subscribers: no more usage caps!”, Ars Technica,
http://arstechnica.com/telecom/news/2011/03/bt-to-uk-fiber-subscribers-no-more-usage-caps.ars.
Optus, Naked (standalone) Broadband,
http://personal.optus.com.au/web/ocaportal.portal?_nfpb=true&_pageLabel=Template_woRHS&FP=/p
ersonal/internet/nakedbroadband&site=personal.
iinet, broadband plans, http://www.iinet.net.au/broadband/plans.html.
http://www.vodafone.is/en/internet
Chiehyu Li & James Losey, “Bandwidth Caps for Residential High-Speed Internet in the U.S. and
Japan”, http://www.newamerica.net/files/Bandwidth%20Caps%20for%20HighSpeed%20Internet%20in%20the%20U.S.%20and%20Japan.pdf. [Losey]
12
cost issue has crystallized around the question of the cost to transfer 1 GB of data.
Reliable cost information would be extremely helpful in order to respond to at least two
issues. First, the claims regarding light users subsidizing heavy users would be a more
informed discussion, since it would allow for a realistic assessment of the actual costs of
servicing both light and heavy Internet users. Second, reliable cost information would
allow for analysis of the reasonableness of current overage charges. While retail Internet
access pricing is unregulated, efforts to analogize Internet access to regulated utilities
raises the specter of assessing the reasonableness of the markup for Internet access
services.
i.
Internet facing data costs
Calculating the cost for an ISP to transfer 1 GB of data involves at least two cost
components. The first component is the Internet side of the equation. As discussed above,
the user’s traffic travels through three stages – the last mile, the ISPs internal network,
and the public Internet. The public Internet cost is relatively easy to calculate since all
ISPs depend upon peering and transit arrangements to carry data between networks.
Independent ISPs advise that the cost of a 1 Mbps (megabit per second) connection is
roughly $5 per month (the cost is likely considerably less for large incumbent ISPs given
their economies of scale as some estimates put the price at $3 per Mbps).45
With a 1 Mbps connection, an ISP is permitted to transfer 1 Mbps of data per second on a
continuous basis. That is the equivalent of 7.5 MB (megabytes) per minute (60 megabits
is equal to 7.5 megabytes), 450 MB per hour, and 10,800 MB per day (or approximately
10.5 GB). Over a 30 day month, it adds up to 316.4 GB or roughly 1.58 cents per GB.
For larger ISPs able to negotiate lower costs, the $3 per month fee allows them to transfer
a GB of data for under one cent. In fact, when off-peak traffic or demand shifting is
factored into the equation, the costs may be further reduced.
Estimates of transit arrangement pricing typically point to between $3 and $5 per megabit
per second per month. This figure is declining rapidly, with some estimates suggesting
that it may drop to under $1 per megabit by 2014.46 Transit arrangements are used when
two ISPs do not have a direct connection and therefore rely on an intermediary network
provider to complete the Internet connection. Some ISPs may directly connect with one
another (known as a peering arrangement) that would further reduce costs since the data
transfer travels directly from one ISPs network to another ISPs network with the ISPs
agreeing to offset each other’s traffic. Yet another alternative for web-based traffic and
video streaming is the use of Content Distribution Networks, such as Akamai. These
companies are often used by content distributors to manage traffic more efficiently and
provide ISPs with the advantage that the traffic may not run on their networks, thereby
further reducing costs.
45
46
DrPeering International, “Internet Transit Prices – Historical and Projected”, http://drpeering.net/whitepapers/Internet-Transit-Pricing-Historical-And-Projected.php.
Id.
13
These numbers are consistent with other public numbers for large-scale data transfers.
For example, the Amazon EC2 service, a leading cloud computing provider, charges one
cent per GB for all regional data transfers.47 Regions include the U.S. East Coast, U.S.
West Coast, and Europe. For data transfers outside a single region, the Amazon EC2
service costs ten cents per GB for all in-bound data transfers in North America and
Europe and as little as 8 cents per GB for out-bound data transfers.
ii.
Internal Network Costs
The far more difficult cost calculation involves the internal ISP network, namely the last
mile and the internal network. ISPs zealously guard this information, demanding that
virtually any potential disclosure remain confidential. While the ISPs often disclose this
information to the CRTC, most of the public record of CRTC proceedings has been
redacted of information that would be useful to estimate the costs of increased network
traffic on confidentiality grounds.
There is one notable exception, however. As discussed at Appendix B, unredacted
information about total costs and total subscriber numbers for Bell's proposals for
funding broadband expansion to underserved communities from the deferral account is
available. Three separate sets of data were disclosed in 2010, representing different
designs for the access network. Bell's preferred option is based on High Speed Packet
Access (HSPA) technology. Other options investigated included a pure DSL technology
option and a hybrid option, combining both technologies.
It should be noted that Bell’s data describes a “worst case scenario” in the sense that Bell
was making the case that such a network would not be economically viable and would
therefore merit additional support through the deferral account process. The Bell data
disclosed the total cost of the project (over 15 years), the number of communities and
premises where service would be available, the projected revenue, and the peak number
of subscribers.
The detailed analysis of the Bell data, which admittedly involves a number of
assumptions, suggests that the costs of building a new network would lead to estimates of
roughly seven cents per GB. The actual costs vary considerably between networks –
capital expenditures, network usage, subscribers, and properties of the backhaul links all
influence actual costs. While the actual numbers may be higher or lower, the ten
cents/GB estimate is fairly consistent in several network models.
When combined with the Internet costs of roughly one cent per GB for larger ISPs, a high
end estimate of the per gigabyte costs for large Canadian ISPs is approximately 8 cents
per GB.48 This assumes the creation of a new network and accounts for all aspects of the
data transfer – the last mile, internal ISP network, public Internet transfers, and other
47
48
Amazon Web Services, “Amazon EC2 Pricing”, http://aws.amazon.com/ec2/pricing/.
Others have even lower estimates, particularly for large U.S. ISPs. For example, Dave Burstein
estimates that large U.S. ISPs spend 2 to 5 cents per GB: FastNewsNet, “Wireline Costs and Caps: A
Few Facts”, http://dslprime.com/dslprime/42-d/4148-costs-and-caps.
14
associated expenditures. While this is higher than the 3 cents per GB that has been
invoked in some discussions, it is far lower than overage costs imposed by some ISPs,
which run as high as $10 per GB in Canada.49
Part Two – What Should Come Next
With both the CRTC and the federal government now seized with the UBB issue, all
parties are looking for market-based solutions that will foster increased competition and
fair pricing. As noted above, a one-size-fits-all approach will not work. Instead,
politicians and regulators would be well advised to identify targeted solutions for
wholesale UBB, retail UBB, and the broader competition concerns for Internet access
services in Canada.
i.
Wholesale UBB
The immediate issue before the CRTC is limited to wholesale UBB - the effort by Bell to
impose UBB on independent ISPs who pay Bell for the GAS service. The CRTC
continues to paint the issue as being about "subsidizing" heavy users,50 but rather than
focusing on the heavy vs. light usage, the CRTC should be driven by a single priority fostering a competitive market by establishing rules that allow independent ISPs to
compete. In the context of wholesale UBB, there are five steps that should be taken to
address the competition concerns.
First, the UBB decisions to date should be rescinded. The federal government has made
it clear it will accept nothing less, so the Commission appears to have little alternative but
to come forward with an alternative approach. The government has been criticized for
meddling into a regulatory issue, but in this instance the CRTC’s approach desperately
requires reform.
Second, the CRTC should ensure that the speed matching decision is implemented and
vigorously enforced. Independent ISPs cannot compete with incumbents if they are
offering inferior speeds. The CRTC has acknowledged that effective competition requires
the ability to offer equivalent speeds to consumers and it should prioritize efforts to
ensure that any remaining barriers are eliminated and that incumbent ISPs are fully
compliant with their obligations.
Third, the CRTC should reverse its decision on ADSL-CO, a proposal that would have
allowed independent ISPs to locate closer to the end customer. As noted above, policies
that encourage independent ISPs to affordably locate closer to end users promotes more
facilities-based competition as independent ISPs would expand their networks, reduces
concerns with network congestion by lessening their use of the incumbent ISP networks,
and positions independent ISPs to offer greater customer choice.
49
50
Josh Kerr, “Internet penalties challenged”, Yukon News, http://www.yukon-news.com/news/22089/.
Statement from the Chairman of the CRTC on usage-based billing, February 3, 2011,
http://www.crtc.gc.ca/eng/com100/2011/i110203.htm.
15
Fourth, for those independent ISPs that do not use ADSL-CO and cannot currently afford
to invest in co-location facilities, the CRTC should ensure that those independent ISPs
have access to a regulated bulk wholesale service that would allow them to allocate the
data usage as they see fit. This would provide for the same overall network usage but
without the UBB that limits service differentiation.
Fifth, the CRTC should turn its attention to cable providers with respect to wholesale
Internet services. While wholesale access is available using TPIA, it has not proven
popular. Some independent ISPs advise that the situation is improving and that they are
contemplating offering cable-based Internet access services. The shift to cable has taken
far too long since a more competitive environment necessitates wholesale access to both
cable and DSL. A marketplace that steadily reduces DSL market share but leaves cable
untouched runs the risk of creating de facto cable monopolies for Internet services. The
CRTC should prioritize making the cable TPIA service a viable alternative for
independent ISPs.
ii. Retail UBB
Given the CRTC’s longstanding forbearance of retail Internet services, addressing retail
UBB represents a more challenging regulatory and policy issue. While some have
advocated for the elimination of UBB at the retail level, there is little political support for
directly intervening in the pricing choices of retail ISP services. The optimal method for
addressing retail UBB is a robust competitive environment in which it is consumer ability
to switch providers in the event they are dissatisfied with UBB that serves as the primary
safeguard against unreasonable UBB practices.
In many respects, the current debate resembles the early battle over net neutrality in
Canada.51 Indeed, substitute the words video streaming, Netflix, and heavy users for peerto-peer, BitTorrent, and file sharers and the discussion points are virtually identical.
Net neutrality advocates maintained that fears about traffic shaping practices and
preferential treatment of incumbent ISP content were the product of an uncompetitive
marketplace that required regulatory intervention. Incumbent ISPs argued that there was
no need for CRTC involvement since a market-oriented approach was sufficient to
address any public concerns.
The CRTC ultimately adopted guidelines that ensured that Internet traffic management is
not a free-for-all.52 It rejected arguments that the market would ensure ISPs provide
adequate disclosure on how they manage their networks. Instead, it mandated full
disclosure of traffic management practices, including information on when they occur,
which applications are affected, and their impact on Internet speeds. Moreover, it adopted
a new test to determine reasonable traffic management practices. Where a consumer
complains, ISPs are now required to describe their practices, demonstrate their necessity,
and establish that they discriminate as little as possible.
51
“Battle over 'net neutrality' arrives in Canada”, CBC News,
http://www.cbc.ca/news/technology/story/2006/11/02/tech-neutrality.html.
52
ITMP, supra, note 15.
16
a.
IBUMPs – Transparency and Disclosure
Given the current competition concerns and the fears that the UBB could be used for
uncompetitive purposes, the CRTC should consider new guidelines for retail UBB Internet Billing Usage Management Practices (IBUMPs) - expanding on the CRTC’s
Internet Traffic Management Practice guidelines.
Much like the ITMPs, the IBUMPs would include a comprehensive transparency
requirement so that consumers can better understand the limits of their service and
reliably track their monthly Internet usage. While ISPs already provide some tracking
tools, they have proven unreliable with admissions that there have been errors.53 The
CRTC’s ITMP decision appears to adopt the position that monthly usage bills provide
sufficient transparency to Canadian consumers.54 Yet the incumbent ISPs have also
sowed some marketplace confusion with marketing that breeds consumer uncertainty
about what their data plans actually provide in practice.
Marketing materials often overstate the benefits of data plans by citing unusually
compressed and shorter MP3 music or video files to suggest that consumers can
download far more than is likely to be the case.55 Similarly, discussions often conflate
streaming video with downloaded video, leaving consumers with the mistaken
impression that they can access hundreds of hours of video each month.56 This claim may
only be true for low-definition streaming, not for downloaded video from various
commercial services.
If UBB is to remain part of the retail Internet access landscape, the transparency and
public disclosures must improve. The CRTC should adopt similar requirements as those
found with ITMPs to ensure that consumers are better informed about the benefits and
limits of their capped services.
b.
IBUMPs – Reasonableness
Much like the ITMPs, greater transparency and disclosures are a necessary but not
sufficient step to address current concerns. The ITMP process also led to guidelines
designed to ensure that ISP traffic management practices are reasonable. The CRTC’s
basic test for acceptable ITMPs was as follows:
When an ISP is responding to a complaint regarding an ITMP it has implemented, it will
use the ITMP framework. In doing so, the ISP shall:
53
54
55
56
See e.g., “Bell admits internet metering problem”, CBC News,
http://www.cbc.ca/news/technology/story/2011/02/09/technology-bell-internet-usage.html.
ITMP, supra, note 15.
For example, Rogers currently warns consumers that estimates on video and song downloads are
subject to “The times specified are approximations and will vary depending on size and quality of
content” without providing specific information for the basis of the advertised claims.
Mirko Bibic, “Pay for Usage”, Financial Post, http://opinion.financialpost.com/2011/02/03/pay-forusage/.
17
Describe the ITMP being employed, as well as the need for it and its purpose and effect,
and identify whether or not the ITMP results in discrimination or preference.
In the case of an ITMP that results in any degree of discrimination or preference:
• demonstrate that the ITMP is designed to address the need and achieve the purpose
and effect in question, and nothing else;
• establish that the ITMP results in discrimination or preference as little as
reasonably possible;
• demonstrate that any harm to a secondary ISP, end-user, or any other person is as
little as reasonably possible; and
• explain why, in the case of a technical ITMP, network investment or economic
approaches alone would not reasonably address the need and effectively achieve
the same purpose as the ITMP.57
The ITMP second-level analysis is limited to instances of discrimination or preference.
Although UBB may raise discrimination or preference issues, particularly for competing
video products, many UBB plans do not. Applied to the issue of retail UBB, the
establishment of an IBUMP policy would would ensure that retail UBB is reasonable in
light of network congestion concerns and marketplace conditions. It could include the
following four conditions.
First, CRTC approval for UBB has long been focused on addressing network congestion.
Indeed, incumbent ISPs now argue that the use of UBB is related to congestion concerns
and the need for fairness among all subscribers. In the case of a complaint about UBB
retail practices, ISPs should be required to demonstrate that the UBB approach is
designed to address specific network congestion concerns.
Second, the UBB model should be the least restrictive possible to achieve its intended
goals. Much like ITMPs, this may mean giving consumers the option for accounts that
are rate-limited after a certain monthly cap is reached instead of imposing overage
charges (as noted above, many ISPs around the world have adopted this approach).
Third, if the ISP has implemented traffic shaping or other technical measures, it should be
required to demonstrate why those approaches alone would not reasonably address the
same network congestion concerns. This requirement is a mirror image of the ITMP
requirement that envisions ISPs relying on network investment or economic ITMPs (ie.
UBB) rather than technical ITMPs such as traffic shaping. If traffic shaping is in place,
ISPs should be required to explain why that has not adequately addressed the network
congestion concerns.
Fourth, while the CRTC should not engage in reviews of the retail pricing or size of
monthly data caps, it should, under certain circumstances, be entitled to examine overage
57
ITMP, supra, note 15, at para 43.
18
charges should ISPs use this economic model (as noted above, many ISPs around the
world rate limit rather than impose overage charges). The review of overage charges
would only occur in local markets where both the cable and DSL provider employ UBB,
thus leaving consumers with limited non-UBB alternatives. In such instances, the CRTC
should review overage charges to allow for a reasonable profit but establish safeguards
against price gouging in light of actual ISP costs.
The IBUMP approach would provide additional guidance the CRTC neglected to provide
in the ITMP decision. That decision recognized that guidance on reasonable traffic
management practices could co-exist with a largely unregulated retail Internet services
market. The same decision left economic ITMPs largely unregulated, effectively giving
the Commission's blessing to UBB. Given recent marketplace developments, it is
increasingly clear that was a mistake. Much like the ITMP decision, the Commission
should not dictate pricing models, but it should ensure that pricing models are transparent
and that those premised on addressing network congestion do so in the least restrictive
manner possible.
iii.
Fostering Greater Competition
In recent months, there has been considerable public discussion about the benefits of
relaxing telecom foreign ownership restrictions as a solution to the competition
concerns.58 The days of retaining Canadian control over physical telecommunications
infrastructure connected to millions of homes are over. The notion that Canadian control
guarantees Canadian jobs is part of a by-gone era. Canadian carriers regularly outsource
some of their customer service jobs out of the country. Meanwhile, other parts of the
organization - retail and business sales as well as network building - involve jobs that will
remain in Canada regardless of a company's country-of-origin. While some head office
jobs may be at risk, new companies operating in Canada could potentially create more
jobs, not fewer.
Beyond the jobs issue, concerns that a foreign competitor will not abide by Canadian law
are unfounded. Foreign companies obviously operate in Canada all the time and
compliance with local laws - whether privacy requirements, environmental regulations, or
reporting obligations - are simply a cost of doing business. There is no reason to believe
that a local telecom company is any more likely to comply with Canadian law than a
foreign competitor.
Foreign ownership restrictions should be dropped, but there are several reasons why no
one should be under the illusion that opening the doors to foreign competitors will solve
the competition problems with Canadian broadband. First, the "last mile" will remain a
major problem. There may be some willingness by new entrants to lay fibre in urban
areas, but it is unlikely to extend much beyond that. Second, the foreign competitors may
be just as likely to purchase Canadian companies, rather than investing in new facilities,
58
Bea Vongdouangchanh, “Clement top priority: making $40-billion telecommunications industry more
competitive”, Hill Times, http://www.hilltimes.com/page/printpage/clement-03-14-2011.
19
yielding deep pocketed competitors but less in the way of new facilities. Third, the
foreign ownership issue is most relevant in the wireless context, where the forthcoming
spectrum auction59 may provide the best hope for new competitors as consumers "cut the
cord" and opt for anywhere connectivity that offers reasonable (though not fibre-like)
speeds.
The prospect of wireless broadband competition is hugely important - the forthcoming
spectrum auction may represent the best case for new competition - and the government
policies on that auction (open access requirements for device interoperability, set asides
for new competitors, open to foreign bidders, white spaces) are absolutely crucial.
Opening the door to foreign competitors is a necessary but not sufficient step to address
the competition problem. At the federal level, the current Canadian broadband programs
should be expanded to address both access and competition. To date, the programs have
been geared primarily toward providing access to the communities that still do not have
broadband. Ensuring all Canadians have access to broadband networks remains essential,
but Canada must also begin to transition toward a second policy goal - encouraging new
business models, particularly community-based fibre initiatives that hold the promise of
new local competition. Providing federal support for innovative approaches might help
the business case for new ventures and lessen the risk associated with establishing new
facilities. The government's condition on financial support for new local fibre initiatives
should be open access requirements - the fibre gets laid, but it is openly available to all
competitors.
The federal government should ask CANARIE60 to identify local community networks
that would benefit from connecting to its broadband backbone as well as support
experimental broadband deployments. The U.S. has embarked on something similar,
allocating $60 million in stimulus funds toward USUCAN. That project is described as
follows:
The United States Unified Community Anchor Network (U.S. UCAN) is a new national
project dedicated to connecting community anchor institutions, including public libraries,
schools, community colleges, research parks, public safety and health care institutions
with advanced broadband capabilities. Utilizing the Internet2 Network and in
collaboration with regional research and education networks across the country, U.S.
UCAN will enable these anchor institutions to serve their communities with telemedicine,
distance learning and other life-changing applications not possible with commercially
available Internet services.61
The federal government can also play a role in promoting new community fibre network
initiatives by provincial and municipal governments. Indeed, some of the most
competitive Internet access environments come from cities that seized the issue and
59
60
61
Industry Canada, Press Release, Minister Clement Updates Canadians on Canada's Digital Economy
Strategy (Nov. 22, 2010), http://www.ic.gc.ca/eic/site/ic1.nsf/eng/06096.html.
See generally CANARIE, http://www.canarie.ca/.
United States Unified Community Anchor Network, http://www.usucan.org/.
20
developed a strategy for fostering high speed open access networks.
For example, the City of Stockholm founded the Stokab system in 1994 after the federal
government passed a law with the objective of developing a "sustainable information
society for all.”62 The Stokab network now has thousands of km of cable that is openly
available to all competitors. That has led to 90 operators and service providers along with
hundreds of additional customers. It is working with building developers to link new
developments to the network - it will connect each building to the network and the
developer establishes a network within the building. Tenants are then able to choose
from several providers. The city has announced plans to provide fibre connections to 90
percent of all households by 2012.63
Measures to spur greater competition is essential, but it will realistically take several
years before new competitors can make their mark on the market. In the meantime, it is
also crucial to addressing the potential for abuse. The Competition Bureau has not been
active on this issue, despite the potential for serious anti-competitive behaviour as the
incumbent ISPs could use their position to favour their own content or create economic
incentives that favour services such as their own video-on-demand over Internet based
alternatives. The Bureau should aggressively investigate abusive behaviour as well as
questionable marketing tactics.
The lack of competitiveness of Canadian Internet access is the product of ten years of
policy neglect. While it may be true that there are hundreds of independent ISPs, the
practical reality for most Canadian consumers is that they see little alternative beyond a
single cable provider and a single DSL provider. Moreover, there is little product
differentiation between the two dominant providers with similar pricing, data plans, and
data caps. The widespread use of data caps in Canada is a function of a highly
concentrated market where a handful of ISPs control so much of the market that they can
impose wildly unpopular measures without much fear of losing customers. Addressing
public concerns over ubiquitous UBB plans therefore requires more than just a quick
regulatory fix. It demands a broader public strategy designed to foster a more robust
competitive environment for Internet access services in Canada.
62
63
Prop. 2004/05:175 Från IT-politik för samhället till politik för IT-samhället (Swed.).
ITU News, A Look at Digital Cities, http://www.itu.int/net/itunews/issues/2010/04/36.aspx.
21
Appendix A – Comparative Global Internet Access Services, March 2010
Market
ISP
Offers unlimited
data plans?
Details if cap:
Australia
BigPond (Telstra) No64
Australia
Optus
No65
No additional usage charges. All plans include
rate limiting that slows speeds once usage
allowance is reached. Plans include peak and offpeak caps.
Australia
iiNet
No66
No additional usage charges. All plans include
rate limiting that slows speeds once usage
allowance is reached. Plans include peak and offpeak caps.
Chile
VTR
Yes67
n/a
68
n/a
No additional usage charges. All plans include
rate limiting that slows speeds once usage
allowance is reached. No peak or off-peak
restrictions. Offers BigPond Unmetered which
offers entertainment content that does not count
toward monthly usage limit.
Chile
Movistar
Yes
France
Orange
Yes69
n/a
France
Free
Yes70
n/a
Yes
71
n/a
72
n/a
France
Teleconnect
Germany
T-Home
(Deutsche
Telekom)
Yes
Germany
1und1
Yes (both
Rate limited: after exceeding the data cap on the
limited and
limited plans, download speed reduces from
unlimited plans 16,000 kbps to 1,000 kbps.
available)73
64
65
66
67
68
69
70
71
72
http://go.bigpond.com/broadband/
http://personal.optus.com.au/web/ocaportal.portal?_nfpb=true&_pageLabel=Template_woRHS&F
P=/personal/internet/nakedbroadband&site=personal
http://www.iinet.net.au/broadband/plans.html
http://vtr.com/productos/principal/internet/index.php
http://www.movistar.cl/PortalMovistarWeb/appmanager/PortalMovistar/portal?_nfpb=true&_page
Label=P10001010511291900031885&itemSeleccionado=4&
http://abonnez-vous.orange.fr/residentiel/forfaits/decouverte.aspx
https://adsl.free.fr/cgv/brochure_tarifaire_28122010.pdf?0.07150634494610131
http://www.teleconnect.fr/Pages/Internet/Internet.aspx
http://www.telekom.de/is-bin/INTERSHOP.enfinity/WFS/EKI-PK-Site/de_DE//EUR/ViewProductDetailsStart;sid=AmoUS9KsuB4US5_Nk1Eb_9mpvdxtycttcFz9DEKNlXPCgZ2cmf8lithclXPCgQ==?Catalog
CategoryID=jqYFC7IS4T0AAAEdU7cMVfLJ&ProductRefID=0110001000058%40EKIPK&StageProductRefID=0802101000948_0007%40EKI-PK
22
limited and
limited plans, download speed reduces from
unlimited plans 16,000 kbps to 1,000 kbps.
available)73
Germany
Kabeldeutschland Yes74
n/a
75
n/a
Hong Kong
Hong Kong
Broadband
Network
Yes
Iceland
Siminn
No76
Users are notified by SMS or email when they
exhaust their cap. Extra bandwidth is available
for 1600 kr for 10 GB or 5040 kr for 50 Gb.
Iceland
Vodafone
No77
Only data that originates from outside Iceland
counts against the cap; “all data downloaded
domestically is unlimited without any extra
cost.”
Users are notified as they approach the download
limit. When the limit is reached, Vodafone
“cap[s] the usage” so the customer is not billed.
The user can also purchase 5 GB extra for 990 kr
[$5.33 CAD].
India
Bharat Sanchar
Nigam Limited
Yes (both
limited and
unlimited plans
are available)78
BSNL offers a wide variety of plans, both with
and without data limits. Limited plans are
available that are either rate-limited (ie download
speed slows after exceeding the cap) or that incur
per MB charges for exceeding the cap.
Japan
NTT
No79
900 GB cap on upload only
80
Japan
KCN
Yes
n/a
Malaysia
TM Streamyx
Yes (though
n/a
with an
ambiguous Fair
Use Policy)81
Malaysia
PersiaSys
Yes82
n/a
73
74
75
76
77
78
79
80
81
82
http://dsl.1und1.de/xml/order/DslLeistungsuebersicht;jsessionid=1E146AAECAEF54E3318F7A8
5B73BEF44.TCpfix168b?__frame=_top&__lf=Einstieg&linkOrigin=DslEinstieg&linkId=hd.tab.tarifve
rgleichkomplett&linkType=txt
http://www.kabeldeutschland.de/static/media/Preisliste_Internet_und_Telefon.pdf
https://apply.hkbn.net/orp/eng/orp_nc_step2_plans.jsp
http://www.siminn.is/einstaklingar/netid/verd/nanar/store466/item106478/
http://www.vodafone.is/en/internet
http://210.212.169.201/erp/bbplan.php?action=N
Losey, supra, note 44.
Id.
http://www.tm.com.my/ap/personal/internet/broadband/Pages/standardpackages.aspx
http://www.persiasys.com/
23
Malaysia
Maxis
Yes (both
For the limited plans, exceeding the allowed
limited and
bandwidth incurs a charge of 5 sen / MB
unlimited plans
available)
Mexico
Telmex
Yes (both
limited and
unlimited plans
are available)83
Mexico
Telecable
Unknown –
?
Telecable does
not advertise its
data usage
policy.85
Nigeria
JuniSat
Yes (both
Only incoming traffic counts against the cap for
limited and
the limited plans. Consequences of exceeding the
unlimited plans limit are unclear.
are available)86
Nigeria
CobraNet
No87
Cobranet uses McWiLL wireless technology.
Most plans have data limits; some, however,
have unlimited data but limited time.
New Zealand Actrix
No88
When the bandwidth cap is exceeded, the
customer has a choice: “rate limited”, in which
there is no extra charge, but the speed is reduced
to that of dial-up; or the user can pay $2 (NZD)
per gigabyte.
New Zealand Telecom
No89
When the bandwidth cap is exceeded, the
customer has a choice: “rate limited”, in which
there is no extra charge, but the speed is reduced
to that of dial-up; or the user can pay $2 (NZD)
per gigabyte.
83
84
85
86
87
88
89
Some plans are advertised as having data caps.
However, even the unlimited data plan appears
to be covered by the following Fair Use Policy:
“Furthermore, in order to protect the quality of
the Service, you acknowledge and agree that it
may: (i) terminate any SESSION
INTERNET when failure to detect the use
thereof by the CUSTOMER, (ii) limit the
volume of data transferred and (iii) limit
time periods, according to the modality of the
service.”84
http://www.maxis.com.my/personal/broadband/packages.asp
http://www.telmex.com/mx/hogar/pdf/pt_descarga.jsp?a=Terminosycondiciones-Infinitum.pdf
http://www.telecable.net.mx/internetPrecios.aspx
http://www.junisat.com/index.php?option=com_content&task=view&id=46&Itemid=48
http://www.cobranet.org/ugoz.php?id=3
http://www.actrix.co.nz/page.php?id=144
http://www.telecom.co.nz/totalhomebroadband/details?pid=int091
24
per gigabyte.
90
n/a
n/a
Philippines
PLDT
Yes
Philippines
Digitel
Yes91
Russia
Koptevo.net
Yes (both
For the limited plans, users are charged a fee for
limited and
each MB of excess usage.
unlimited plans
are available)92
Russia
__________
(CentroSet)
Yes (both
For the limited plans, users are charged a fee for
limited and
each MB of excess usage.
unlimited plans
are available)93
Russia
MegaBistro
Yes (both
For the limited plans, users are charged a fee for
limited and
each MB of excess usage.
unlimited plans
are available)94
South Korea
SK Broadband
Yes95
n/a
96
n/a
Singapore
SingNet
Yes
United
Kingdom
BT
Yes (both
limited and
unlimited plans
available97)
Users receive a usage notice after reaching 80%
of their cap. The first month one exceeds the cap,
nothing happens; the second month, the user is
charged ВЈ5 per 5GB.
United
Kingdom
Virgin Media
Yes98
n/a
United
Kingdom
CCS Leeds
No.99
Data cap applies only to “peak” period usage
(peak period is generally 8 am – midnight
weekdays).
Users may choose between “top-up” packages or
paying per GB. Top-up packages are: 10Gig for
ВЈ10.38, 25Gig for ВЈ24.98, 50Gig for ВЈ49.00,
75Gig for ВЈ73.00 and 100Gig for ВЈ95.95.
Without a top-up the price is ВЈ1.31 per Gigabyte.
90
91
92
93
94
95
96
97
98
99
http://myworldmydsl.com/plan.aspx
http://lec.digitel.ph/content/view/60/88/
http://www.koptevo.net/index.php?alias=uslugi
http://centroset.ru/texts.php?way=159&id=159
http://megabistro.ru/view/services/9/docs/14.html
http://www.skbroadband.com/eng/iproduct/evalley.asp
http://www.singnet.com.sg/plans_and_services/broadband/snbb.asp?snbb=15Matl
http://www.productsandservices.bt.com/consumerProducts/displayCategory.do?categoryId=CONTOTAL-BB-R1&s_cid=con_FURL_broadband
http://shop.virginmedia.com/broadband/compare-broadband-packages.html
http://www.ccsleeds.co.uk/residential-dsl.html
25
Yes100
n/a
United States Comcast
No101
250 GB data cap
United States Verizon
Yes102
n/a
United
Kingdom
Breathe
Country
Are unlimited plans available at all?
Australia
No
Chile
Yes
France
Yes
Germany
Yes
Hong Kong
Yes
Iceland
No
India
Yes
Japan
Yes
Malaysia
Yes (although with ambiguous Fair Use Policy)
Mexico
Yes (although with ambiguous Fair Use Policy)
Nigeria
No
New Zealand
No
Philippines
Yes
Russia
Yes
South Korea
Yes
Singapore
Yes
United Kingdom
Yes
United States
Yes
100
101
102
http://www.breatheinternet.com/
Losey, supra, note 44.
Id.
26
Appendix B – Estimating the Cost to Transfer 1 GB of Data Based on Bell’s
Network Costs
Unredacted information about total costs and total subscriber numbers for Bell's
proposals for funding broadband expansion to underserved communities from the deferral
account is available. Three separate sets of data were disclosed in 2010, representing
different designs for the access network. Bell's preferred option is based on High Speed
Packet Access (HSPA) technology. Other options investigated included a pure DSL
technology option and a hybrid option, combining both technologies.
i. Raw Data
The raw data for this analysis are taken from Table 2, which is reproduced from the
public record of proceedings related to the disposition of funds in the deferral account.103
The table was part of an economic study filed by Bell in support of its proposal to use
deferral account funds to cover the shortfall between the cost of this proposed network
expansion and the revenues that would be derived from it. It represents an essentially
new network, including both access network and backbone segments. Since Bell is
attempting to justify its claim that, without deferral account funds, this project would be
uneconomical, the study adopts a comprehensive approach to costs.
However, some of the cost inputs have been redacted. In particular, the entry for Costs
Causal to the Program, which covers expenses such as advertising and billing costs, is not
directly itemized. Note 4 in that Table indicates these costs have been included in the
uneconomic costs. Specifically, the sum of the uneconomic costs plus the revenues must
equal the total costs, including the capital and costs causal to the program. Bell has also
included an amount to cover the loss of revenue from dial-up customers switching to the
new broadband service. From Table 2, the total cost of the program, including the dialup cross impact is $498,345,813.82.
Table 1 presents a higher level summary, comparing three slightly different proposals. It
has been organized in a different way. Here, the total cost is reported as $494 M, and the
dial-up cross impact is not included. From this, we can conclude that $494 M figure
includes the costs causal to the program. This figure therefore represents the cost that is
of interest for this analysis.
103
Bell Aliant/Bell Canada, “Response to Interrogatory The Companies(CRTC)9Mar10-506 DAAttachment- Appendix 1” (30 March 2010), CRTC File No 8638-C12-200817505 at para. 18 online:
CRTC <http://www.crtc.gc.ca/public/partvii/2008/8638/c12_200817505/1376452.zip>.
27
Proposed HSPA
Solution
($M)
DSL-Only
Scenario
($M)
Hybrid Scenario
($M)
Capital (including replacement)
#
#
#
Costs Causal to the Program (Capital & Expense)
#
#
#
Expenses
#
#
#
Total Cost
Less: Broadband Revenues (Note 2)
$ 554
$ 494
$ 508
$ 94
$ 92
$ 93
Dial Internet Service Cross Impact
#
Uneconomic Cost before Contingency (Note 3)
Plus: Contingency (10%)
Uneconomic Cost
#
#
$ 463.6
$ 407
$ 419
(Note 4)
$ 41
$ 42
$ 463.6
$ 447
$ 461
Table 1: Summary of Results Estimated Present Worth of the Companies' Broadband
Expansion Program Over the 15-Year Study Period (2010$)
Number of approved communities (Note 1)
112
Number of premises passed (Note 2)
61,369
Number of subscribers (end of 2024) (Note 3)
41,048
PWAC Broadband Capital and Expenses (2010 $)
$489,587,262
PWAC of Broadband Backbone Capital (2010 $)
#
PWAC of Broadband Access Capital (2010 $)
#
PWAC of Broadband IP Core Capital (2010 $)
#
PWAC of Broadband Expenses (2010 $)
#
PWAC Costs Causal to the Program (2010 $)
#
PW of Broadband Revenues (2010 $)
$91,823,281
PW of Dial Internet Service Cross Impact (2010 $)
#
PW of Uneconomic Cost (2010 $) (Note 4)
$406,522,533
Total Upfront Capital Expenditures (CapEx) (Note 5)
$277,050,827
Upfront CapEx per premise passed
$4,515
Upfront CapEx per subscriber (based on number of subscribers at end of 2024)
$6,749
Notes:
1. The number of communities approved for the program by the Commission.
2. Eligible lines (premises passed) located in the approved communities.
28
3. Estimated number of broadband subscribers located in the approved communities at
the end of the 15-year study period (in 2024).
4. PW of Uneconomic Cost is equal to PWAC Broadband plus PWAC Costs Causal to
the Program minus PW of Broadband Revenues minus PW of Dial Internet Service Cross
Impact.
5. Total Upfront Capital Expenditures include Upfront Capital Expenditures per
Community plus Upfront Capital Expenditures Causal to the Program.
# Filed in confidence with the CRTC.
Table 2: Bell Canada and Bell Aliant's Proposed Broadband Expansion Program
Deploying DSL Technology - Summary of Results
ii.
Service Roll-out
The economic study assumed a progressive service roll-out over a finite period of time.
The precise details about number of customers have been redacted.
It was assumed that # of the potential residential subscribers would
actually subscribe to the retail broadband service in Year 1, growing to
by Year 5 and levelling out at a constant
# in Years 10 to 15.104
#
Bell's model is based on a rising subscription base over ten years, followed by a constant
level for the remaining five. It is not clear from the redaction whether the growth is
linear, or exponential; nor whether the reference to the Year 5 level represents a change
in the pattern at that point. Some assumptions are therefore necessary.
A simple working assumption is that the growth is linear, and that there is only one
slope (m) over the ten year period.
N ?1?= N 1
(1)
t
N ?t ?= N 1? ∑ m ,t = 2..10
(2)
N ?t ?= N max , t≥ 10
(3)
2
Nmax is a known quantity. N1 and m can therefore be related.
m=
N max- N 1
9
(4)
The model assumes a constant average revenue per subscriber:
For the same communities, the Average Revenue Per Unit (ARPU) for each
residential and business subscriber over the 15-year study period was
derived as follows: the monthly rate of $31.95 plus $2.50 per Gigabyte (GB)
of additional usage, assuming # GB of additional usage per month, less the
104
Bell Aliant/Bell Canada, “Response to Interrogatory The Companies(CRTC)9Mar10-506 DAAttachment” (30 March 2010), CRTC File No 8638-C12-200817505 at para. 18 online: CRTC
<http://www.crtc.gc.ca/public/partvii/2008/8638/c12_200817505/1376452.zip> [footnotes omitted,
redactions per original].
29
$5.00 Bell Bundle discount per subscriber per month. The average monthly
usage of # GB per customer is based on the average customer usage metrics
associated with the Companies' Internet Essential Plus service in Ontario for
the January to December 2009 period. The revenues generated from the
application of the $35.00 one-time activation fee are included in the
calculation of the total revenues.105
The average overage charges are an unknown parameter. However for a given
annualized revenue per subscriber R, and an annual interest rate of i, the present value of
revenues can be computed.
15
PV R = ∑
t= 1
R∑N ?t ?
?1? i ?t
(4)
Although it is not entirely clear what interest rate should be applied, the study does
specify an average cost of capital of 8.47%.106 It seems reasonable to adopt this value for
all present value calculations.
PVR is a known quantity.107 Using equation (3), (4) and (1) can be solved for N1, using
the goal seek function in a spreadsheet. This allows the number of subscribers to be
computed for each year of the study.
iii.
Network Usage
Knowing the number of subscribers allows for a model prediction of network usage.
Assuming a monthly cap of C, and an average monthly overage O, predicted total usage
U is:
15
U = ∑ 12 ?C ? O?N ?t ?
t= 1
(5)
However, it must be noted that this quantity may not be of much interest, since it depends
on cap thresholds that are not obviously related to network properties. In the present
study, C is known to be 2 GB/month. O is unknown, but is likely to be less than 1
GB/month, based on the following redacted discussion.
As noted in the response to The Companies(CRTC)28Jan10-2 Decision 20081, based on the usage levels of the Companies' Essential Plus-Ontario
customers who can only realize a maximum of 2 Mbps of download speed,
the usage of the majority of customers ( #) does not, on average, exceed
the 2 GB monthly usage. Further, of the
# of Essential Plus-Ontario
customers whose usage exceeds the 2 GB monthly usage allowance, about
one third (i.e., only
# of customers) only used one extra GB per month.
105
106
107
Ibid at para. 22 [redactions per original].
Ibid, Table 2.
The revenue figure must be adjusted to remove the activation fees of $35/subscriber that have been
included in the model. However, on the assumption that the model includes no subscriber churn, this is
just a fixed offset of $1,436,685.53.
30
The Companies anticipate that only a small proportion of customers in the
approved communities will have high monthly usage levels. As a result, the 2
GB monthly usage allowance will satisfy the vast majority of customers.108
Usage of 3 GB per month, on a 2 Mbps download/800 Kbps upload line, represents only
0.336% of the theoretical line capacity.
The communities in question are presently underserved specifically because the revenue
that would be obtained in serving them is not enough to fund the cost of the network. It
is therefore of little surprise, and of little use, to discover that the cost per GB obtained by
dividing the total cost by that small usage figure is quite high: in the range of $10$40/GB.
Figure 1 shows the result of solving the subscriber numbers in equation (4) for revenues
based on 0 to 5 GB of usage above the 2 GB cap. Total traffic is then computed by
summing the average traffic per user over the subscribers, over the study period. Cost
per GB of usage is computed by dividing the total cost by the total usage
This merely confirms that Bell has modelled an unprofitable network. However this is
plainly not a reflection of the costs of network operations, generally. If it was, Bell
would not be in the business of providing network services at all.
Given a fixed total cost, there is an inverse relationship between traffic levels and cost per
GB. Figure 2 shows the cost per GB if total traffic is computed as a fraction (labeled as
Utilization) of the advertised link capacity of 2 Mbps download/800 Kbps upload, with
the number of subscribers determined as per Figure 1. However this is only meaningful
to the extent that the network is capable of sustaining the traffic. Although, in proposing
to fund this project primarily from the deferral account, Bell is not really spending its
own money, it nonetheless seems unlikely that it would provision backhaul links that can
carry 300 times the predicted average traffic levels unless it needs to. It is therefore
necessary to estimate the aggregated carrying capacity of the network links included in
the study.
108
Response to Interrogatory The
Companies(CRTC)9Mar10-106 DA at 2 [redactions per original].
31
Figure 1: Cost/GB in low-usage
scenario
40
30
20
10
Cost/GB
($)
0
0
1
2
3
4
5
Average Overage (GB)
Cost/GB ($)
Figure 2: Cost/GB in high-usage
scenario
100
0.01
0.01%
1.00%
100.00%
Utilization
iv.
Network Capacity and Usage
Bell's proposed network architecture for the project is complex, but relatively typical of a
DSL-based network. It involves a multi-tied backbone architecture linking serving
Central Offices (COs) to the existing Core network, and in some cases extension feed
lines (which may be copper or fibre-optic) from serving COs to remote Digital Subscriber
Line Access Multiplexers (DSLAMs), and local copper loops from the DSLAMs to the
subscriber premises.109 The proposal contemplated using existing facilities where
possible, or building new where necessary. Consequently, the precise path linking any
given subscriber to the core would vary from community to community.
In order to estimate the total capacity of the proposed network, two simplifying
assumptions are necessary. First, it will be assumed that subscribers are uniformly
109
See e.g. Bell Aliant/Bell Canada, “Broadband Expansion Proposal” (1 September 2006), CRTC File No
8638-C12-200602708 at 22-26, online: CRTC
<http://www.crtc.gc.ca/public/partvii/2006/8638/c12_200602708/662133.zip>.
32
distributed among the 112 wire centres involved, such that each represents an equal
traffic load. Table 3 summarizes the average properties of each wire centre. Second, it
will be assumed that each wire centre is served by a single CO, with a fibre-optic link to
the Core network.
subscribers
41048
wire centres
112
subscribers/ wire centre
366.5
DL bandwidth/subscriber (Mbps)
2
UL bandwidth/subscriber (Kbps)
800
Total wire centre DL demand (Mbps)
733
Total Wire centre UL demand (Mbps)
286.33
Table 3: Wire Centre summary
Under these assumptions, the properties of the backhaul links will determine the overall
capacity of the network. Three hypotheses will be considered: OC3 (155 Mbps), OC12
(622 Mbps), or OC48 (2500 Mbps), each of which will be assumed to be operating in full
duplex mode. Each will be restricted to 80% of theoretical capacity, to reflect Bell's
typical working fill factor.110 Table 4 summarizes the traffic characteristics of these
links, if each subscriber link was 100% saturated both in uplink and downlink, 100% of
the time. The contention ratio indicates the extent to which the capacity of the access
network exceeds the capacity of the backhaul link. Where the ratio is 1:1, the traffic on
the backhaul is determined entirely by the load of the access network. Otherwise, the
backhaul is capped at the target 80% of maximum.
As can be seen in the table, an OC3 would be oversubscribed both in uplink and
downlink capacity, while an OC48 would be overprovisioned in both. The intermediate
case of an OC12 is oversubscribed in downlink, by a contention ratio of 1.47:1, but is
overprovisioned in uplink.
Total usage over the 15 year period must therefore reflect this upper limit on network
capacity. This has been included by computing annual usage based on the N(t). The
model assumption is that annual usage rises with the number of subscribers until the
backhaul saturates. Any subsequent increase in subscribers does not result in any
increase in potential traffic.
110
See e.g. Bell Aliant/Bell Canada, “Response to Interrogatory The Companies(CRTC)20Aug09-7 TN
242 & 7181”, CRTC File No 8740-B2-200904989 at 3, online: CRTC
<http://www.crtc.gc.ca/public/8740/2009/b2/1281758.zip>. However, note that Bell’s thresholds run as
high as 90% in some instances.
http://www.crtc.gc.ca/public/partvii/2008/8622/c51_200805153_1/926702.zip
para. 71.
33
OC3
OC12
OC48
155
622
2500
80% fill (Mbps)
124
497.6
2000
DL contention
5.91:1
1.47:1
1:1
UL contention
2.31:1
1:1
1:1
DL/month (GB)
39806.54
159739.8
235308.94
UL/month (GB)
39806.54
91917.55
91917.55
Traffic/month (TB)
77.75
245.76
319.56
Half-duplex
(Mbps)
Table 4: Backhaul traffic
v.
Per GB Network Costs
Table 5 summarizes the effective cost per GB of traffic within networks (not for Internet
transit) under a variety of scenarios. Cases B through F are based on the total cost of
$498M, representing the pure DSL scenario, with differing total usages. Case B
computes the number of subscribers from the solution to Equation (4) with 1 GB/month
of average overage. Case C uses the same method, but assuming 0 GB average overage.
Case D assumes identical usage to Case C over the first 10 years, but then assumes that
the number of subscribers continues to grow until, at year 15, all premises that could
receive the service are subscribers. Case E assumes that all premises are subscribers for
the entire study period, with no ramp-up in numbers. Case F representing maximum
possible usage (and therefore lowest possible costs per GB) assumes the backhaul is fully
saturated over the entire study period. Finally, Case A uses the higher costs ($554 M)
associated with the HSPA deployment, with the same usage as in Case B.
case
OC3
OC12
OC48
Notes
A
$0.3452
$0.1192
$0.1007
HPSA costs; usage per Case B
B
$0.3078
$0.1063
$0.0898
subscribers assuming 1 GB overage
C
$0.3078
$0.1016
$0.0844
subscribers assuming 0 GB overage
D
$0.3078
$0.0980
$0.0759
more subscriber growth
E
$0.3078
$0.0825
$0.0501
max subscribers, no roll-out
F
$0.3078
$0.0767
$0.0191
saturated backhaul
34
Table 5: Cost per GB Capacity
It is difficult to know with precision which, if any, of these scenarios best reflects average
network costs, across the country. First, it must be recognized that the costs are not
actually independent of the choice of backhaul links. Consequently, the variation of up to
an order of magnitude reflects a real uncertainty in these estimates, rather than
alternatives that can be selected. At best, the assumptions about the choices in the
network design are educated guesses. Second, the 80% saturation may vary, as Bell has
indicated that its thresholds for existing networks range as low as 61% to as high as 90%.
Significant variation on either end would naturally increase or decrease per GB costs.111
Indeed. the scenario of a completely new construction project involving remote
communities, may significantly inflate the costs, when compared to on-going expansion
of capacity in existing networks. Without more information, it is impossible to know
how well this would offset the inflated estimate of network capacity. However, if the
growth in subscriber numbers is viewed as a proxy for growth in traffic demand from a
stable subscriber base, this model may nonetheless provide a conceptually reasonable
estimate. In any case, it would seem to be the best estimate—or range of estimates—that
can be obtained from this data.
vi.
Alternative Capital Model – Cost Estimates With More Typical Capital
Expenditures
As noted in Table 2, Bell's model for this project included an upfront capital expenditure
(CapEx) of $4,515 per premise passed. This value is unusually high, reflecting the
increased cost of construction in reaching low density populations. Table 6 canvases the
range of CapEx per premise passed reported for a variety of Next Generation Access
Networks, worldwide. Note that in some cases costs have been reported as per
subscriber, rather than as per premise passed. For the sake of comparison, a conversion
111
“Link utilization measurements are taken every 15 minutes, however these figures hide the
instantaneous traffic peaks that can be higher than actual measured utilization. This is highlighted by
the fact that latency and dropped packets occur well below 100% link utilization. Bell Canada has
developed thresholds for each type of link using the same methodology:
DS-3: 61%;
OC-3: 84%; and
OC-12 and OC-48: 90%.
[…]
The Company is able to actively monitor congested links in order to take appropriate action to resolve
the problem. For a link to be considered congested today, the threshold must have been exceeded at
least once on 5 or more different days of a 14 consecutive day period. Bell Canada believes that this
approach to measuring and identifying congestion in its network is consistent with the methodology
used by many other service providers throughout North America and the UK.”
http://www.crtc.gc.ca/public/partvii/2008/8622/c51_200805153_1/926702.zip
para. 71.
35
factor of 2/3 was applied, which corresponds to the approximate take rate in Bell's model.
(That is for 60k premises passed, Bell forecasts 40k subscribers.)
While reported costs clearly vary significantly based on factors such as choice of
technology, network architecture, labour cost and population density, none of the
reported costs are remotely close to Bell's value. Canada will likely be on the high-end of
the cost spectrum based on both labour costs and population density. However the worst
case scenario based on Table 6 is the New England network modelled by the Telemarc
Group, which suggests a plausible upper limit of around $2000 per premise.112 In reality,
since Canada is highly urbanized and network builds will concentrate on the high-density
areas, true deployment costs may well be similar to the ~$1000-$1500 per premise range
that seems to be typical of other industrialized countries. It must also be noted that these
figures are mostly for FTTx networks. Lower capacity VDSL, DOCSIS 3.0 and 4G
networks should be somewhat cheaper to build, while still providing a significant
upgrade relative to the 5-10 Mbps access networks that are widely deployed in Canada,
currently.
Network Type and Location Source and Year
CapEx per Premise Passed
FTTx deployment, Central
and Eastern Europe
Analysys Mason, 2010113
$100-$800
DOCSIS 3.0 upgrade of
existing cable network, US
Columbia Institute for
Tele-Information, 2009114
$70-$100
Verizon FiOS (FTTP), US
Tellabs Inc., 2006115
$825-$1000
FTTP, Vermont & New
Hampshire (low density)
The Telemarc Group,
2006116
$1500-$2300 ($2000-$3500 (USD)
per subscriber)
112
113
114
115
116
By comparison, Australia’s NBN is $2,700 per premise passed, which includes the satellite and fixed
wireless build. See p. 134 NBN Co Limited (2010). Corporate Plan 2011 – 2013.
http://www.nbnco.com.au/wps/wcm/connect/main/site-base/main-areas/publications-andannouncements/latest-announcements/nbn-co-corporate-plan-released.
“Central and Eastern Europe” (2010), online: Analysys Mason
<http://www.analysysmason.com/Research/Content/Reports/RDTW0_FTTx_rollout_forecast_Aug2010/Sample-pages/>.
Robert C Atkinson & Ivy E Shultz, Broadband in America: Where It Is and Where It Is Going
(According to Broadband Service Providers) (New York: Columbia Institute for Tele-Information,
2009) at 22, online: Columbia University
<http://www4.gsb.columbia.edu/null/download?&exclusive=filemgr.download&file_id=7212786>.
Bruce Ross, “Challenges Operators Face in Deploying FTTX” (2006) at 12,online: Commsday
<http://www.commsday.com.au/Archive/2006/presentations/Bruce_Ross_Tellabs.ppt>
Terrence P McGarty, “Fiber to the Home; Capital Costs and the Viability of Verizon’s FIOS” (2006) at
21, online: The Telemarc Group
<http://www.telmarc.com/List_By_Date/2006%2009%2001%20FTTH%20Capital%2001.pdf>
[McGarty].
36
Network Type and Location Source and Year
CapEx per Premise Passed
CSMG, 2007
117
~ $700 (ВЈ670 per subscriber)
FTTP, UK
CSMG, 2007
118
~ $1075 (ВЈ1030 per subscriber)
FTTB, Athens
Athens Information
Technology, 2007119
$1250
Generic WiMAX
deployment
Huawei, 2009120
$400 ($600 USD per subscriber)
FTTP, Japan
Institute for Information
and Communications
Policy, 2006121
~ $1200 (ВҐ158,000 per subscriber)
Generic Next Generation
Access, France
Caisse des De_po_ts,
2011122
Dense urban: $270-$540 (200€-400€)
Non-Urban: $1350-$1625 (1000€1200€)
WiMAX, India
Beyond 4G, 2010123
$160-$550 ($240-$800 USD per
subscriber)
DSL, India
Beyond 4G, 2010124
$350-$800 ($500-1200 USD per
subscriber)
FTTH, Netherlands
Reggefiber, 2011125
~$1350 (~1000€)
VDSL, UK
Table 6: Estimates of CapEx per Premise Passed for various networks
Bell's model predicts $277 M in total CapEx, which represents approximately 56% of the
total costs of the project. It is a simple matter to calculate an adjusted cost by scaling this
CapEx value. Figure 3 presents the variation of cost per GB with CapEx, based on the
same assumptions as in Case B of Table 5. However, as noted above, 80% utilization of
117
118
119
120
121
122
123
124
125
CSMG, “Broadband Comes of Age: Bandwidth-intensive Video” (2007), online: TMNG Global
<http://www.tmng.com/print/236>.
Ibid.
Siamak Azodolmolky & Ioannis Tomkos, “Techno-Economic Study of a Modeled Active Ethernet
FTTB Deployment” (2007) at 4, online: The Universitat Politècnica de Catalunya
<http://personals.ac.upc.edu/siamak/articles/CSNDSP08_v0.8.pdf>.
Li Zheng, “Fast track to the broadband market”, (2009) 48 Huwei Communicate 19 at 20, online:
Huawei <http://www.huawei.com/publications/view.do?id=5853&cid=10809&pid=61>.
Keisuke Takachi & Hitoshi Mitomo, “Estimating the Cost of Nationwide Optical Fiber Network
Development in Japan: A Step toward a Cost-Benefit Analysis of the Fiber-to-the-Home Initiative”
(2006) at 11, online: Institute for Information and Communications Policy
<http://www.soumu.go.jp/iicp/chousakenkyu/seika/pdf/2006-01.pdf>.
He_lide_o Costa Elias, “When and why PPPs are an option for NGA ?” (Presentation delivered at
Rome, 15 February 2011) at 2, online: European Investment Bank
<http://www.bei.org/epec/resources/presentations/nga-roundtable-costa-elias.pdf>.
BP Tiwari, “State and Opportunity of Broadband in India” (2010) at 12, online Beyond 4G
<http://www.beyond4g.org/whitepaper-state-and-opportunity-of-broadband-in-india>.
Ibid.
Mark Meijer, “Financing FTTH Networks in Netherlands” (Presentation delivered at the FTTH
Conference, Milan, 8 February 2011) [unpublished], online: Slideshare
<http://www.slideshare.net/ceobroadband/ftth-conference-2011-workshop-mark-meijer>.
37
the backhaul may not be a supportable assumption. Figure 4 therefore repeats the
calculation with the backhaul traffic limited to only 15% of theoretical capacity,
reflecting reported utilization levels of backbone segments. Table 7 summarizes the
results.
With the exception of the case of the OC3 link limited to 15% capacity, which is highly
oversubscribed, the results indicate a range of $0.05-$0.30. This is consistent with Table
5, which would tend to confirm that this is the best estimate available.
Given that the installation labour is substantially more expensive than the material, it
seems rather unlikely that Bell would lay new fibre with less than 48 strands. In fact, 96
strand cable is slightly cheaper, per strand-mile. The Telemarc model of the New
England FTTH network assumes 48 strand fibre throughout, except for single strand
drops to the house.126 Construction costs per mile are approximately $17.6k, vs. $2.6k
material cost per mile for the fibre.127 This tends to support a result near the low end of
the range.
$0.35
$0.30
$0.25
$0.20
OC4 (80%)
OC12 (80%)
OC48 (80%)
$0.15
Cost/GB ($)
$0.10
$0.05
$0.00
$0.00
$1,000.00
$2,000.00
$3,000.00
$4,000.00
CapEx per Premise Passed ($)
Figure 3: Cost/GB, max 80% utilization of backhaul links
126
127
McGarty, supra note 116 at 18-19.
Ibid.
38
$5,000.00
$1.80
$1.60
$1.40
$1.20
$1.00
OC4 (15%)
OC12 (15%)
OC48 (15%)
$0.80
Cost/GB ($)
$0.60
$0.40
$0.20
$0.00
$0.00
$1,000.00
$2,000.00
$3,000.00
$4,000.00
$5,000.00
CapEx per Premise Passed ($)
Figure 4: Cost/GB, max 15% utilization of backhaul links
CapEx Per
Premise Passed
$500.00
$1,000.00
$1,500.00
$2,000.00
OC3 (15%)
$0.8373
$0.9393
$1.0412
$1.1432
OC12 (15%)
$0.2087
$0.2341
$0.2595
$0.2849
OC48 (15%)
$0.0635
$0.0712
$0.0789
$0.0867
OC3 (80%)
$0.1570
$0.1761
$0.1952
$0.2144
OC12 (80%)
$0.0542
$0.0608
$0.0674
$0.0740
OC48 (80%)
$0.0458
$0.0514
$0.0569
$0.0625
Table 7: Cost per GB capacity, variable CapEx
Bell may well prefer to leave some of the installed fibre dark for future use. So it is not
possible to rule out the low-capacity alternative. And it is noteworthy that the resulting
estimate of up to $1.15/GB is comparable to Ericsson's published estimate of the end-to-
39
end costs of HSPA networks.128 Nonetheless, an installed-base of dark fibre would
simply provide a low cost path to scale up the backhaul capacity, as needed. Moreover,
there is no reason to presume that an underprovisioned link would be limited to 15% of
its actual capacity. On the contrary, the low-capacity link is much more likely to be
saturated.
So, in the long run, the low end of the range would again seem to be the better estimate as
demand for peak bandwidth increases. Consequently, while there is a range of plausible
values, a comparison of Tables 5 and 7 suggests a figure of $0.07/GB may be the best
single estimate for the costs of the internal network.
128
Ericsson claims 0.92€/GB, all-in. See “Don’t worry – Mobile broadband is profitable” (2009) 2009:2
Ericson Business Review 54 at 56, online Ericsson
<http://www.ericsson.com/ericsson/corpinfo/publications/ericsson_business_review/pdf/209/209_BUSI
NESS_CASE_mobile_broadband.pdf>.
40