Catching the Wave: Should Canada
Follow the Global Trend Toward
Spectrum Auctions?
Martin C. Glass and David M. Rhodes’
This article examines different methods of spectrum allo-
cation to determine which one best accomplishes Canada’s
spectrum allocation goals. The June 1996 amendments to Can-
ada’s Radioconmunicadons Act permit licenses, previously al-
located by administrative selection, to be sold by auction to the
highest bidder. The experiences of other countries have shown
that spectrum auctions only accomplish public policy and effi-
ciency goals when they are carefully designed.
After examining the advantages and disadvantages of dif-
ferent methods of spectrum allocation, the authors conclude that
Canada would be best served by a hybrid system which uses a
screening process to qualify bidders to enter into an auction for
licenses. The screening stage would determine
spectrum
whether bidders meet certain socially desirable criteria, thus
being eligible for bidding credits. The auction stage would in-
volve a simultaneous multiple round auction in which licenses
could be purchased either individually or in groups. Licenses
acquired pursuant to the auction process would be transferable
and divisible. They would have a fixed duration of fifteen years
and would be re-auctioned at the end of that period.
A spectrum allocation system designed in this fashion
would provide the proper balance between market forces and
social policy considerations for the Canadian context.
Cet article examine les diffdrentes mdthodes d’attribution
de spectre afin de dterniner laquelle mdthode permet de
mieux accomplir les objectifs du Canada dans cc domaine. En
juin 1996, des amendements h la Loi sur les radiocommunica-
dons canadienne ant pennis d’attribuer des licences par vente
aux enchies, alors que pr6cdemment elles daient attribudes
par salection administrative. L’expdrience des autres pays a
ddmontr6 que l’attribution de spectre par vente aux ench;res
n’accomplit les objectifs d’ordre public et d’efficaeit6 que lots-
qu’elle est organisae soigneusement.
Apris avoir examin6 les avantages et ineonv6dnients des
difflrentes rnthodes d’attribution de spectre, les auteurs con-
cluent que le Canada serait mieux servi par un systame hybride
qui utiliserait une procddure d’6valuation pour qualifier les en-
ch disseurs A participer aux ventes aux ench~res et A obtenir des
licences de spectre. L’dtape d’6valuation ddterminerait si les
ench6risseurs rencontrent certains critares considdeds ddsirables
dans notre societ6 et s’ils sount par consdquent dligibles pour des
erdlits d’ench re. An m~me temps, l’tape de vente aux encha-
res impliquerait une vente i multiples tourades par laquelle des
licences pourraient atre achetes soit individuellement, soit en
groupe. Les licences aequises en vertu de la procddure de vente
aux enchres serait transfdrable et divisible. Elles araient une
durde fixe de quinze ans et seraient revendues par vente aex en-
chies a la fm de cette pdriode.
Un syst~me d’attribution de spectre organis6 de cette fa-
gn pourrait founir pour le contexte canadien l’&luilibre cher-
ch6 entre les forces du march6 et les considerations de politique
sociale.
. Martin C. Glass, B.C.L., LL.B. (McGill, 1998), is an associate at White & Case, New York, and
David M. Rhodes, B.C.L., LL.B. (McGill, 1998), is a law clerk to the Honourable Madame Justice
Desjardins of the Federal Court of Appeal, Ottawa, and future associate at Sullivan & Cromwell, New
York. We are very grateful for the assistance and commentary of many people in the production of
this article. We would like to thank Professor David Johnston, C.C. for his guidance and encourage-
ment throughout the writing process; Ian Munro and Michael Connolly-both of Industry Canada-
for their insight and commentary; and Assistant Deputy Minister Michael Binder for his support. We
would also like to thank Professor Richard Janda, Dr. Sunny Handa, and Dr. Michael & Mrs. Brenda
Rhodes for having read previous versions of this article. We take full responsibility, however, for any
errors or omissions.
McGill Law Journal 1999
Revue de droit de McGill 1999
To be cited as: (1999) 44 McGill LJ. 141
Mode de rdf6rence: (1999) 44 R.D. McGill 141
MCGILL LAW JOURNAL I REvUE DE DROIT DE MCGILL
[Vol. 44
Table of Figures
Summary of Recommendations
Introduction
1. Nature of the Spectrum
I1. Spectrum Allocation
A. History of Spectrum Allocation in Canada
B. History of Spectrum Allocation in the United States
C.
1. Comparative hearings
2. Lotteries
3. Auctions
International Spectrum Allocation Processes
1. New Zealand
2. Australia
3. Columbia
Ill. Comparative Advantages of Spectrum Allocation Methods
A. What Goals should Canada Seek to Achieve Through Spectrum
Allocation?
1. Promotion of Canadian culture
2. Diversity of service providers
3. Availability to businesses of all sizes
4. Enhanced competition and customer choice
5. Speed of allocation
6. Rapid network deployment
7. Efficient use of the spectrum
8. Access to services
9. Ensuring sufficient spectrum for essential public services
10. Transparency of the allocation process
11. Revenue for the government
B. Relative Advantages of Different Methods of Allocation
1. First come/first served
2. Lotteries
3. Comparative analysis
4. User fees
5. Auctions
C. Specific Objections to Pure Market Auctions
D. Meaning of these Results for Canada
1999]
M.C. GLASS AND D.M. RHODES – SPECTRUM AUCTIONS
IV. Tailoring an Auction System for Canada
Introduction
A.
B. Hybrid Models-Blending Auctions with Other Methods of Spectrum
Allocation
1. Reservation of spectrum for specific purposes
2. Bidding credits
3. Screening auction applicants
4.
Imposing conditions on spectrum licenses
C. Selecting an Auction Method
1. Criteria to be used in comparing different auction methods
2. Oral ascending (English) auction
3. Oral descending (Dutch) auction
4. First-price sealed bid auction
5. Second-price sealed bid auction
6. Sequential English auction with sealed-bid package bidding
7. Simultaneous multiple round auction
D. Conclusion
V. Proposed Method of Spectrum Allocation in Canada
A. Stage 1: Expression of Interest
B. Stage 2: Screening Stage
C. Stage 3: Auction
D. Post-Allocation Stage
E. Transfer and Change in Use of a License
F Divisibility of Spectrum Rights
G. Duration of License and the Possibility of Renewal
Conclusion
144
MCGILL LAW JOURNAL / REVUE DE DROIT DE MCGILL
[Vol. 44
Table of Figures
Figure 1: The inherent features of different pure methods of allocation. Note that first
come/first served, user fees and lotteries are significantly less desirable than auctions
or comparative analysis.
Figure 2: The strengths of the simultaneous multiple round auction system as com-
pared to other auction methodologies is clear when the features of all these systems
are compared.
Figure 3: Formula for calculating the reimbursement of a bidding discount when a li-
cense is divided.
1999]
M.C. GLASS AND D.M. RHODES – SPECTRUM AUCTIONS
Summary of Recommendations
Canada should adopt a simultaneous multiple round auction system which is tai-
lored to complement the tendency of pure market auctions to satisfy only our effi-
ciency goals. This would involve a three stage process to allocate commercial spec-
trum bands.’ At the first stage, potential participants would express their interest in
bidding for a spectrum license. At the second stage, a two-tier screening process
would ensue. The first tier would involve an evaluation of the business/service plans
of all potential bidders to ensure that they satisfy certain criteria. The second tier
would screen for compliance with desirable social criteria which an applicant might
choose to fulfill in order to earn bidding credits for the auction stage. The third stage
would involve an auction between all applicants who pass the screening stage, with
bids starting at a designated reserve price. Bidders would be able to withdraw any
bids they wished to change, but would be subject to a penalty equivalent to the differ-
ence between the price eventually paid and the amount of the withdrawn bid. Winning
bidders would pay 25% of their bids at the conclusion of the auction and would pay
the remaining 75% over the first half of the license term.
Violations of the first tier criteria would result in the forfeiture of the license. Li-
censees which qualified for bidding credits under the second tier criteria would have
to maintain their qualifying status or refund the amount of the bidding discount plus
interest and a 50% penalty.
Licensees would be able to transfer their licenses or change their use of imple-
mentation. Upon the transfer of a license, the transferee would have to convince the
Spectrum Management Branch (SMB) that its proposed business/service plan satisfies
all criteria of the first tier. If the transferor of the license benefited from bidding cred-
its, the transferor would have to refund the amount of the bidding discount on transfer
plus a 50% penalty, unless it demonstrated to the SMB’s satisfaction that the trans-
feree fulfilled the criteria upon which the bidding credits were awarded. Where the
original licensee desires to change the use of its license, it would have to satisfy the
SMB that its new business/service plan would continue to fulfill all first tier criteria,
and if it wished to avoid refunding any bidding discounts and penalties, it would also
have to show continued compliance with the second tier criteria.
Licensees could divide their spectrum rights either by geography or by frequency.
All holders of spectrum rights would have to meet all first tier criteria. Failure to sat-
isfy any applicable second tier criteria would be treated in the same way as a transfer
or change in use.
The duration of a license would be fifteen years to allow sufficient time for licen-
sees to recoup their investments. Two years before the end of the term, a license
would be re-auctioned. In this re-auction, bidding credits would be awarded to in-
‘ Spectrum bands for essential public services would be allocated by other means.
146
MCGILL LAW JOURNAL / REVUE DE DROIT DE MCGILL
[Vol. 44
cumbents to alleviate the potential for successful predatory bidding by their competi-
tors. The amount of this bidding credit would depend on the revenues of the licensee.
Our belief is that this system would provide the proper balance between market
forces and social policy considerations for Canada at present and in the future.
1999]
M.C. GLASS AND D.M. RHODES- SPECTRUM AUCTIONS
Introduction
The radio spectrum is a public resource which is used as a medium of communi-
cation by a diverse group of entities ranging from cellular telephone companies and
paging services to fire departments and air traffic controllers. Canadians benefit from
spectrum-related services every day. In a knowledge-based economy, much of the
wealth in society rests in ideas. Dependence on these services will only increase as
they become the primary means by which ideas are shared. Therefore, the way in
which spectrum use is allocated between service providers has a direct impact on the
lives of all Canadians-and the method of allocating this scarce resource deserves
careful examination.
Spectrum allocation in Canada is presently at an important crossroad which will
determine its evolution for decades to come. Since the inception of radiocommunica-
tion in this country, spectrum licenses have been allocated through an administrative
process which compared the proposed uses of each applicant according to public in-
terest criteria. The applicant who came closest to meeting those criteria was selected
as the winner. Now things are poised to change. In June 1996, the Radiocommunica-
tions Act,’ was amended to permit the allocation of spectrum licenses through auc-
tions. This represents an increased faith in market forces as mechanisms to guide the
development of the economy and a belief that the existing administrative process
would be unable to adapt to the pressures generated by increasing spectrum scarcity.
Canada’s interest in this method reflects the growing use of spectrum auctions in the
international community.
The experiences of other countries have shown that the success or failure of an
auction is dependent upon its design. Correct design is important from the outset be-
cause the rights being auctioned can last for decades, and if they are allocated badly,
the entire Canadian telecommunications industry will suffer the consequences.
In this article, different methods of spectrum allocation are examined with a view
to assessing their relative strengths and weaknesses, and determining which method
would be most desirable in the Canadian context. First, the nature of the spectrum and
how it is used for communications will be examined. Second, the history of spectrum
allocation in Canada, the United States and other countries will be reviewed. Third,
the discussion will shift to elaborate a series of goals which Canada’s allocation
method should achieve. Various proposed methods of allocation will be assessed ac-
cording to their proficiency in satisfying these goals. Fourth, an examination as to
how these methods could be tailored to improve their effectiveness will be under-
taken. Finally, a method of spectrum allocation will be proposed which is believed to
be best suited to meet the spectrum allocation goals of this country.
‘R.S.C. 1985, c. R-2.
148
MCGILL LAW JOURNAL / REVUE DE DROIT DE MCGILL
[Vol. 44
I. Nature of the Spectrum
The electromagnetic spectrum is a continuum which includes all possible forms
of radiation arranged according to the wavelength and frequency properties of waves.
Wavelength is the distance between any two wave crests, measured in metric units of
distance ranging from millimeters to kilometers. Frequency is calculated by measur-
ing the number of wave crests that pass a stationary point in one second. It is repre-
sented in cycles per second, or hertz (Hz). Since all electromagnetic waves travel at
the speed of light, a wave with a longer wavelength must have a lower frequency.
For purposes of identifying waves which share similar wavelength and frequency
characteristics, the electromagnetic spectrum is conceptually grouped into broad sec-
tions. The section we are most familiar with is visible light. Other familiar groupings
include X-rays, ultraviolet rays and radio waves.
The radio wave spectrum is simply a grouping of waves in the electromagnetic
spectrum which range from 3 kHz to 3,000 GHz. With present technology, we are
only able to use the portion of the radio spectrum between 3 kHz and 100 GHz. How-
ever, as technology advances, it steadily increases the useable portion of the radio
spectrum. The portion of the radio spectrum that can be used at any given time to
transmit information is known as the “spectrum”.
Radio waves can be made to carry information through a process known as
“modulation”. This is a way of coding the information to be transmitted and causing
the radio wave to vibrate according to the coded pattern. The coded radio wave energy
is then radiated in all directions by a transmitting antenna. Some of these coded waves
can be collected by a receiving antenna and the information is extracted by the proc-
ess of demodulation, which simply involves separating the coded information from its
carrier waves and decoding it. One important prerequisite for successful transmission
is the absence of interference from other radio waves. Each transmitter must have the
exclusive use of a portion of the radio spectrum between itself and each receiving an-
tenna. If two transmitters try to use the same portion of the spectrum in the same geo-
graphic area, interference between the two waves will garble the information which
they are carrying and prevent accurate demodulation.
Because of their physical properties and the limitations of current technology to
code and decode information accurately, certain frequencies of radio waves are more
suited to carrying particular types of information. In practice, there is a balance be-
tween the information-carrying capacity of a wave and its distance of travel. All parts
of the spectrum can carry the same amount of information, but because of the inter-
ference problem, there is less low frequency spectrum available to carry information.
Lower frequency radio waves-those between 150 and 1,500 kHz-travel great dis-
tances because they are not easily stopped by obstacles. This means that low fre-
quency transmitters must be spaced further apart to avoid interference, and there is
less opportunity to reuse the same frequency in different areas of the country. With
many people wanting to use a portion of these low frequency waves locally, there is
less bandwidth available at lower frequencies. The limited availability of bandwidth
restricts the use of low frequency waves to those purposes which do not require large
amounts of information to be transmitted. Low frequency waves are suitable for the
1999]
M.C. GLASS AND D.M. RHODES – SPECTRUM AUCTIONS
long distance transmission of audio information such as voice and music used in
commercial radio broadcasts.
Higher frequency waves, those between 30 and 3,000 MHz, travel only short dis-
tances because they are easily blocked or absorbed by people, buildings and the ter-
rain itself. Because higher frequency waves have more limited range, transmitters can
be placed close together providing greater bandwidth across Canada for transmitting
information. The availability of bandwidth makes these waves suitable for transmis-
sion of video information, including ordinary television and high definition television
(HDTV) signals.
Portions of the spectrum which share similar information-carrying capacity are
conceptually grouped into units known as “bands” or “blocks”. With the need to use a
band having the proper balance between information-carrying capacity and distance
of travel, there are a limited number of spectrum bands suitable for any given type of
communications service. However, as technology advances, the amount of spectrum
suitable for any given service generally increases. Improvements in technology make
it possible to compress more information into lower frequency waves and tailor higher
frequency transmission systems to carry information across greater distances without
absorption. The upper limit of the useable portion of the spectrum, moreover, is con-
stantly being pushed higher by new transmission technologies. Despite these ad-
vances, new technologies and the growth of new services tend to increase the demand
for spectrum faster than it becomes available.’ For this reason, the spectrum is consid-
ered a scarce resource whose allocation must be managed with care.
II. Spectrum Allocation
Due to interference, transmission of information via the spectrum requires each
radio transmitter to have exclusive use of a spectrum band. This necessitates a method
of allocating spectrum bands whenever more than one person wants to use a given
band. Indeed, every country has a national agency to allocate the right to use spectrum
bands within a given area of the country. In Canada, this function is performed by In-
dustry Canada’s SMB. The same role is performed in the United States by the Federal
Communications Commission (FCC) and in Australia by the Spectrum Management
Agency. Each of these bodies has established elaborate methods of allocating the
spectrum among different services and competing service providers. In order to ap-
preciate the controversy surrounding spectrum auctions, it is important to examine the
history of spectrum allocation.
3 Examples demonstrating how improvements in technology change the way we use the same spec-
trum bands for new services include the shift from ordinary radio broadcasting to digital radio, and
the shift from analogue cellular telephones to personal communications services (PCS) digital cellular
telephones. Innovations in spectrum band use include the broadcast of internet information, satellite
telephony and wireless local area networks of personal computers.
MCGILL LAW JOURNAL / REVUE DE DROIT DE MCGILL
[Vol. 44
A. History of Spectrum Allocation in Canada
Statutory authority to allocate the spectrum is given to the Federal Minister of In-
dustry by sections 5(1)(a)(i.1) and 5(1)(e) of the Radiocommunication Act. These
sections empower the Minister to plan the allocation and use of the spectrum by
granting licenses to use a specified radio frequency area for a certain purpose within a
defined geographic area. The statute does not specify any method of allocation or any
clear criteria according to which one is to be selected.” It only states that the Minister
“may have regard to” the objectives contained in section 7 of the Telecommunications
Act in exercising his or her powers of allocation.’ Recent amendments stipulate that
the Minister may use competitive bidding to award an applicant with a spectrum li-
cense However, there is no requirement that this process of allocation be used. The
choice of allocation method is ultimately left to the Minister and-by derogation-to
the SMB. The method which this agency has traditionally used is known as a “com-
parative selection process”.
This comparative process involves three major stages. The first stage is the ex-
pression of interest. Here the SMB is involved in determining what type of use is most
appropriate for a given spectrum band.’ For example, the SMB has selected the bands
between 1,850 and 1,990 MHz for PCS. Once a service is selected for the designated
spectrum bands, the SMB invites anyone who wants to provide the service to make its
interest known to the SMB before a specific closing date, at which point the names of
all interested parties are published. This usually results in mergers and joint ventures
between parties intending to provide similar services, since applicants with more re-
sources are better able to provide the service. After these mergers have occurred, if
there is sufficient spectrum available for all parties who express an interest, the spec-
trum licenses are allocated among them on a first come/first served basis.
‘S. 5(1) indicates that the Minister is to “takre] into account all matters which he considers relevant
for ensuring the orderly establishment or modification of radio stations and the orderly development
and efficient operation of radiocommunication in Canada”
S.C. 1993, c. 38.
‘ Radiocommunication Act, s. 5(1.1).
‘Ibid. at ss. 5(1.2)-(1.4).
‘ In making this determination, the SMB will frequently request public submissions as to the best
match between services and available spectrum bands. See Notice, C. Gaz. 1994. DGTP-006-94, for an
example of a request for public comment on the proper spectrum bands for PCS. In making these deci-
sions, the SMB is influenced by the choices of the FCC since significant differences between Canada
and the United States in the designation of spectrum bands for the same services requires modifications
of American communications equipment for the Canadian market. This has the effect of delaying the
deployment of new technologies and increasing equipment costs. See Industry Canada, “Wireless
Personal Communications Services
in
Canada” (15 June 1995) at 6, s. 5.2, online: Industry Canada
the 2 GHz Range:
in
Implementing PCS
‘PCS telephones are the next generation of mobile communication. They rely on digital as opposed
to analogue technology to code information onto radio waves, permitting more callers to use the
spectrum at the same time and ensuring the privacy of communications.
1999]
M.C. GLASS AND D.M. RHODES – SPECTRUM AUCTIONS
If there is insufficient spectrum for all of the interested parties (a situation known
as “mutual exclusivity”), then the process of allocation moves to the second stage
where the parties submit detailed applications concerning the way in which they in-
tend to provide their communications services. These submissions are used by the
SMB to select the applicant, or applicants, best qualified to provide the services in
question. The SMB requires a great deal of information to make this decision-con-
sequently, applications are long and detailed. They typically include engineering de-
signs of communications equipment, construction plans for facilities, long term busi-
ness and marketing plans, proof of sufficient financing to successfully operate the
services in question, and proof of Canadian control and/or ownership if the service in
question is subject to such restrictions. These applications can easily run up to a thou-
sand pages in length and cost $100,000 or more to prepare.
These applications are reviewed by two levels of SMB experts, and their recom-
mendations go to the Minister who makes the final decision about which applicants
should be granted spectrum licenses. There is no right of appeal or review of the
Minister’s allocation decisions. The goal of this selection process is to determine
which applicant’s proposed services would produce the greatest public benefit. While
no fixed criteria exist to determine what constitutes a benefit to the public, certain
factors are commonly examined. These include: (i) the efficiency of proposed use
(number of users per portion of spectrum), (ii) the suitability of the proposal to the
spectrum band available, (iii) the need for spectrum in order to provide the service as
proposed; (iv) the financial capacity of applicants to perform their proposals, (v) the
potential for interference between the proposed service and those of other spectrum
users, (vi) the resulting increase in competition in the communications sector, (vii) the
use of new technologies, (viii) the demonstrated competence of the applicant to run a
communications undertaking, (ix) the resulting economic benefits to society, and (x)
the extent to which the social objectives of section 7 of the Telecommunications Act
would be fulfilled. In this context, the relevant social objectives could be summarized
as:
(1) strengthening Canadian culture;
(2) providing high quality service to remote areas of the country; and
(3) increasing competitiveness both nationally and internationally.”
Normally, the issue is whether the same service could be adequately provided over wires.
The full text of Telecommunications Act, s. 7 reads as follows:
It is hereby affirmed that telecommunications performs an essential role in the mainte-
nance of Canada’s identity and sovereignty and that the Canadian telecommunications
policy has as its objectives
(a) to facilitate the orderly development throughout Canada of a telecommunications
system that serves to safeguard, enrich and strengthen the social and economic
fabric of Canada and its regions;
MCGILL LAW JOURNAL / REVUE DE DROIT DE MCGILL
[Vol. 44
All applications are considered in light of these factors in order to best determine
which applicant’s proposed use of the available spectrum will generate the greatest
public benefit. The successful applicant, or applicants, then proceed to the third stage
of Canada’s spectrum allocation process which involves equipment certification.
Once the successful applicant is awarded a spectrum license, it can begin con-
structing its communications facilities, networks and transmitters. Before this equip-
ment can begin to transmit, however, it must be certified by the SMB as meeting the
technical standards established for equipment of that type. These standards help en-
sure that interference is not created by substandard equipment. Certifications are not
difficult or costly, and they are mostly a formality if the licensee has used high quality
communications equipment.” Once the certification process is complete, the licensee
can begin to provide communications services. The licensee, however, is not free to
do anything it wishes. Each spectrum license comes with several important conditions
attached.
The most important condition requires the licensee to provide the services it
promised in its application. This includes both the type of service and the geographic
area in which those services are provided. There is also a requirement that the licensee
pay an annual license fee for its use of the spectrum band. These fees are determined
by the SMB according to the anticipated revenues generated by the services provided.
As such, they vary widely from service to service. The SMB presently receives over
$150 million each year in license fees, and this amount is expected to increase as
more PCS providers begin to provide service. Failure to pay license fees and failure to
honour the terms of an application can lead to a license being revoked. Conversely, if
an applicant has provided its service as promised and paid its fees on time, it can gen-
(b) to render reliable and affordable telecommunications services of high quality ac-
cessible to Canadians in both urban and rural areas in all regions of Canada;
(c) to enhance the efficiency and competitiveness, at the national and international
levels, of Canadian telecommunications;
(d) to promote the ownership and control of Canadian carriers by Canadians;
(e) to promote the use of Canadian transmission facilities for telecommunications
within Canada and between Canada and points outside Canada;
(f) to foster increased reliance on market forces for the provision of telecommunica-
tions services and to ensure that regulation, where required, is efficient and effec-
tive;
(g) to stimulate research and development in Canada in the field of telecommunica-
to encourage innovation in the provision of telecommunications
tions and
services;
(h) to respond to the economic and social requirements of users of telecommunica-
tions services; and
(i) to contribute to the protection of the privacy of persons.
12 If the equipment or network design is found to be insufficient, the SMB will require the licensee
to make changes before providing services.
1999]
M.C. GLASS AND D.M. RHODES – SPECTRUM AUCTIONS
erally expect to have its license renewed by the SMB when the term expires.” The fi-
nal important condition attached to a spectrum license is that it cannot be transferred.
The costs and delays involved in this allocation process are considerable. On av-
erage, six to eighteen months elapse between the time the SMB requests expressions
of interest and the time the spectrum license is allocated to the successful applicant.
The total cost of applying for a license and having equipment certified ranges from
$200,000 to $700,000.
Despite the cost, delay, complexity and uncertainty of the comparative selection
process, it has received very wide support from the Canadian wireless telecommuni-
cations industry. In recent public hearings, 95% of industry submissions strongly fa-
voured the comparative selection process over spectrum auctions.” Despite the oppo-
sition of the industry, the SMB decided to conduct auctions of the spectrum bands
designated for local multipoint communications services starting in the fall of 1998.”
This was decided by the SMB because with the increased demand for the spectrum
and the need for government fiscal restraint,” the traditional comparative process will
become a great deal slower if its use is continued in the future.
B. History of Spectrum Allocation In the United States
In the early days of broadcasting in the United States, chaos and interference
reigned because the law was interpreted to give the United States Secretary of Com-
reject applicants on the
merce the power to grant spectrum licenses, but not to
grounds that interference would result.” The resulting spectrum congestion threatened
radio communications and prompted the United States Congress to resolve the prob-
lem with the Radio Act of 1927″ and the Communications Act of 1934.” These stat-
“The duration of a spectrum license varies considerably depending on the type of service it pro-
vides. Generally, licenses last for at least five years and for no more than twenty years. However, since
renewal is relatively assured, the only important consequence of the expiration of a license is that re-
newal can involve an alteration in license fees.
” Notice, C. Gaz. 1996. DGRB-001-96: Industry Canada, “Review of the Comparative Selection
and Radio Licensing Process-Findings” (20 February 1996) at 1, online: Industry Canada
parative Selection Review”].
“Notice, C. Gaz. 1997. DGRB-003-97: Industry Canada, “Consultation on Issues Related to Spec-
trum Auctioning” (1 August 1997) at 1, online:
accessed: 24 January 1999) [hereinafter “Auction Consultation Review”].
“Ibid. at 4.
“United States v. Zenith Radio Corporation, 12 F2d 614 (N.D.III. 1926), online: WL (DCT-OLD);
Hoover v. Intercity Radio Co., 286 F 1003, 52 App. D.C. 339 (D.C.C.A. 1923), leave to appeal to
U.S.S.C. refused 266 U.S. 636, 45 S.Ct. 10 (1924). Note also that the Secretary of Commerce did not
have the authority to limit the power of transmission used by licensees, their frequency or hours of
operation.
“47 U.S.C. 81ff.
“47 U.S.C. 151ff.
MCGILL LAW JOURNAL/REVUEDEDROITDE MCGILL
[Vol. 44
utes created the FCC-giving it broad authority to allocate spectrum licenses and
control the use of the spectrum in accordance with public interest, convenience and
necessity criteria.”0
1. Comparative hearings
Until 1981, the FCC used a comparative approach to allocate spectrum licenses,
similar to the process adopted by the SMB. The FCC determined the services best
suited to each spectrum band and invited interested parties to apply for spectrum li-
censes to provide these services. Where mutual exclusivity occurred, detailed appli-
cations” were submitted and a quasi-judicial comparative analysis to select appli-
cants-based on the criteria of public interest, convenience and necessity-ensued.22
One important difference between the FCC process and the SMB process was that the
selection criteria were applied by FCC administrative judges in a quasi-judicial forum
in which competing applicants could argue why they should be awarded the license
over competitors. Interested third parties were allowed to argue for or against an ap-
plicant. The final decision of these FCC judges could be appealed to an internal FCC
appeal panel, and ultimately to the Commissioners themselves. In addition, the final
allocation decision of the Commissioners could be appealed through the ordinary
court system all the way up to the Supreme Court. ‘
Because of intense competition between many qualified parties for very limited
amounts of spectrum, the FCC regularly faced great difficulties in deciding which of
two equally qualified applicants would better use the spectrum license in the public
interest. In such situations, the FCC frequently decided between applicants based on
rather tenuous grounds. In one case, an applicant was selected because it pledged to
build a twenty-four cell cellular network instead of the twenty-three cell cellular net-
works promised by its competitors.”4 In another instance, an applicant was selected
20 G.L. Rosston & J.S. Steinberg, “Using Market-Based Spectrum Policy to Promote the Public
Interest” (1997) at Part II, online: Federal Communications Commission
2 These applications would generally contain the same information as the SMB would require.
However, they tended to be longer and even more expensive to produce, with the cost ranging from
$US 150,000 to $US 400,000 dollars.
22 It is important to note that the FCC selection criteria lack any reference to cultural or political
goals. The only social goal included is universality of service. Another important difference is the
somewhat greater emphasis on market forces and free competition that have existed in the FCC crite-
ria from the beginning. Conversely, the references in the Canadian law to reliance on market forces
are a relatively recent addition.
“The delay caused by these levels of appellate review could protract the granting of a spectrum li-
cense by up to eight years.
2 J. Blake, “FCC Licensing Methods: From Comparative Hearings to Auctions” (1994) 47 Federal
Comm. L.J., online: Federal Communications Law Journal
1999]
M.C. GLASS AND D.M. RHODES – SPECTRUM AUCTIONS
because it promised to provide more washroom facilities at its broadcasting station. ‘
These types of decisions had three notable results. First, they led to a great deal of
litigation and delay as disappointed applicants sought to overturn the FCC’s decisions.
Second, they made the comparative selection process appear arbitrary, effectively dis-
crediting it in the eyes of the communications industry and the general public. Third,
they undermined the requirement that an applicant’s license was conditional upon
meeting the terms of its application. The apparent arbitrary nature of the selection
process pushed many qualified applicants to make promises they could not keep.
When these applicants failed to honour the terms of their inflated applications, the
FCC rarely revoked their licenses. Instead, it recognized that all applicants were
promising more than they could provide, estimating that the applicant selected was
probably as close to fulfilling its promises as any other applicant was likely to be. The
effect of this reasoning was to turn the comparative selection process into a breeding
ground for laudable-but unfulfilled-promises.
Not only was the process’ transparency and effectiveness seriously questioned, it
was also derided as expensive and time consuming both for applicants and for the
FCC. For example, over two hundred applicants applied for the initial thirty licenses
to provide cellular service, and the second and third set of licenses attracted nearly
one hundred applications. Each of these rounds took over two years to complete-not
including litigation time-and strained the FCC’s resources to the limit. Factoring in
litigation, the delay in receiving a spectrum license could be nearly eight years. The
costs involved were equally large, ranging from $500,000 up to several million.” Be-
cause of these deficiencies, the FCC’s comparative selection process failed to win the
support of the American wireless communication industry.”
2. Lotteries
In 198 1, because of the difficulties encountered in the comparative selection proc-
ess, the FCC was granted the authority to allocate spectrum licenses by lottery. Lotteries
were initially proposed as a way to assign a license when two or more equally qualified
applicants had been chosen by the comparative selection process. They were seen as a
method to avoid the costs and delays of litigation and to remove the apparently unfair
elements from the FCC’s spectrum allocation process. However, because of the costs
and delays of the comparative process, the FCC used lotteries as its general method of
deciding between all applicants, not just between those most qualified to provide service.
2Ibid.
26 See Federal Communications Commission, “In the Matter of Amendment of the Commission’s
Rules to Allow the Selection from Among Mutually exclusive Cellular Applications using Random
Selection or Lotteries Instead of Comparative Hearings” CC Docket No. 83-1096, Report and Order,
(1984) 98 F.C.C. (2d) 175, online: WL (FCOM-FCC).
27 Note that the FCC selection process involves an equipment certification stage similar to the one
used by the SMB. This stage applies to all broadcasting equipment regardless of the method used to
allocate the spectrum license.
MCGILL LAW JOURNAL / REVUE DE DROITDE MCGILL
[Vol. 44
With spectrum licenses now equally available to anyone at little cost,” a specula-
tive market in the licenses began to emerge. Licenses were auctioned by the winners
of the lotteries to the serious service providers waiting in the wings. In some cases, li-
,censes fetched $US 30 million or more for the lucky winners. The FCC was soon
flooded with applications; for some lotteries, the agency received over 65,000 in a two
day period. Since screening would have taken over two years to complete, the high
volume of applicants forced the FCC to remove even the minimal screening proce-
dures it had proposed for weeding out applicants clearly unable to provide service.
Moreover, lotteries did not avoid the costs and delays of litigation. Schemes to rig the
lotteries prompted legal challenges on the grounds that they had been improperly
conducted and thus were invalid. In short, FCC spectrum lotteries appeared to have
many more flaws than comparative hearings ever did.’
With a few modest changes, however, the lottery system might have worked ef-
fectively. The first important change would have been to prohibit multiple applica-
tions from the same person or group of persons. Charging significant application
fees-and then much larger license fees to the eventual winner-would have limited
the applicants to only those who were serious service providers. Another step in the
same direction would have been to screen applications for proof of technical capacity
and financing. Finally, imposing and enforcing transfer restrictions would have pre-
vented the winner of a lottery from transferring its spectrum license either directly or
indirectly. With these changes, the lottery process might have been an effective
method of allocation rather than a disaster.
3. Auctions
In the Omnibus Budget Reconciliation Act of 1993,” the United States Congress
amended the Communications Act of 1934
requiring the FCC to design and conduct
auctions to allocate spectrum licenses. At the outset, auctioning was not enacted for
the purpose of raising revenue. Rather, auctioning was conceived as a way to quickly
assign licenses to those who would use them, to avoid the problems associated with
speculation in the spectrum license market, and to reduce administrative costs. The
” Lottery applicants only needed to pay the $US 35 filing fee for each application which had to be
submitted along with an engineering plan for the proposed service. Most applicants would file be-
tween one and several dozen applications. This rate of filing was facilitated by lottery application
mills which provided applicants with everything necessary to apply en masse, including numerous
pre-designed engineering plans.
‘ One undesirable side-effect of lotteries that was never present with the comparative selection pro-
cess was the emergence of investment fraud schemes. The highly visible revenues earned by the win-
ners of the FCC’s lotteries were used as a lure by con-artists to induce people to invest in a chance to
win a spectrum lottery. Generally, these con-artists would overestimate the chance of winning a li-
cense and charge their victims application fees much larger than necessary to enter the lottery. These
practices were investigated by various United States law enforcement agencies.
o Pub. L. No. 103-66, 6002, 107 Stat. 312 at 387-392 [hereinafter Budget Act].
5, 47 U.S.C. 309G) was added.
1999]
M.C. GLASS AND D.M. RHODES – SPECTRUM AUCTIONS
philosophy behind auctions was that the high costs of winning a license would dis-
suade all but the most serious service providers from applying, and the winners of the
auction would be those service providers who valued the spectrum most and would be
able to implement services the quickest. 2
In both the design and the implementation of auctions, the Budget Act stipulated
that the FCC’s primary goal was to protect the public interest. The fee was also to
promote certain specific objectives. These objectives included: (i) quick deployment
of new technologies and services to the public, including rural areas, (ii) economic
development and competition through broad distribution of licenses to a diverse group
of license holders, (iii) recovery for the public of some of the value of the spectrum,
and (iv) efficient spectrum usage.” The FCC was also directed to ensure the participa-
tion of small business, rural telephone companies and businesses owned by women
and minorities.” Raising revenue was a secondary concern, and the FCC was prohib-
ited from making decisions based solely on the amount of revenue that would be gen-
erated. There is some evidence, however, that Congress’s clarity of purpose wavered
once it became obvious that the auctions would raise much more revenue than previ-
ously anticipated. This shift occurred after the fourth and fifth auctions (PCS blocks
A, B and C) which raised ten times the anticipated revenue. One indicator of this shift
in focus is that Congress ordered the FCC to auction the bands designated as “cellular
unserved” in time to balance the United States budget despite the FCC’s protestations
that the market was uncertain about how to use those bands, making it inefficient to
auction them at that time. The FCC’s concerns appear to have materialized because
the cellular unserved auction raised only a terith of the money that Congress had ex-
pected.
As of September 30, 1997, the FCC has conducted fourteen auctions which have
allocated over 4,300 licenses and raised over $US 23 billion. According to the FCC,
in cases where demand exceeds supply, auctions are a more efficient way to assign
spectrum licenses than any previously employed mechanism. Compared with lotteries
and the comparative selection process, auctions rapidly award licenses to productive
users,” encourage the emergence of innovative firms and technologies, generate mar-
ket information about the value of a spectrum license, encourage the participation of
small businesses, and compensate the public for the use of public resources. They are
“47 U.S.C. 309(j)(3)(A)-(D).
m,/bid., 309()(4)(D).
32 Federal Communications’Commission, “In the Matter of FCC Report to Congress on Spectrum
Auctions” WT Docket No. 97-150, [1997] WL 629251 at Part II, online: WL (FCOM-FCC) [herein-
after “FCC Reporf’].
3 Ibid. 3090).
36 Note, however, that spectrum auctions can take a long time when many licenses and bidders are
involved. For example, the bidding phase of the United States auctions for blocks A and B of the PCS
spectrum lasted three and a quarter months, block C lasted for five and a half months, and blocks D, E
and F lasted for four and a half months. This represents only the duration of the bidding phases. See
“FCC Report”, supra note 32.
158
MCGILL LAW JOURNAL / REVUE DE DROIT DE MCGILL
[Vol. 44
supposed to do all these things at a lower cost than other methods of allocation.”
However, it is too soon to reach any definitive conclusions because the long term ef-
fects of auctions on the United States wireless communications industry are unknown.
Auctions, moreover, may generate some unforeseen countervailing effects which un-
dermine their supposed virtues.
C. International Spectrum Allocation Processes
The spectacular revenues generated by the FCC’s early spectrum auctions and the
increasing importance of economic theory have induced many countries to consider
auctions as a method of spectrum allocation. Seven years ago, no country had con-
ducted spectrum auctions. Today, over a dozen countries have done so,’ and many
others have announced their intention to conduct auctions in the future or are in the
process of making that decision. Auctions have received particular attention as a
method of allocating the most valuable private commercial spectrum bands. While the
auctions conducted to date have been met with varying degrees of success, most of
the difficulties can be traced to poor auction design. 9
1. New Zealand
New Zealand was the first country to conduct spectrum auctions. Its experience in
1990 was somewhat of a failure. New Zealand used single round, sealed bid auctions
which prevented each bidder from knowing how much its rivals had bid, and the win-
ner was only required to pay the amount of the next highest bid.’ The results were
dramatic and some of the licenses were awarded for winning bids as low as $NZ 1.
There was also great deviation in the estimated value of the licenses. In one case, the
“‘ Auctions can, however, be extremely costly for participants. In preparation for the FCC PCS auc-
tions, Pacific Bell Mobile Service Inc. lobbied the FCC for a year and a half to ensure the auction’s
rules would not jeopardize its business strategy. It also employed a core team of over a dozen experts
to design fail-proof bidding strategies and supported them with many economists and attorneys. To
ensure that the electronic bidding process would not be interrupted by power failures, third parties or
natural disasters, it built special secure facilities with encrypted communications networks and estab-
lished duplicate facilities forty miles apart. Pacific Bell Mobile won two PCS licenses for which it
paid $US 696 million. See Nordicity Group Ltd., The Stakes are High and the Meter is 7icking…
(Report prepared for WIC Ltd.’s submission to Industry Canada Re: Notice C. Gaz. 1996. DGRB-
001-96: Industry Canada, “Radio Authorization Fees for Local Multipoint Communication Systems
in the 28 GHZ Range (LMCS)” (16 May 1996) [hereinafter Nordicity Report]).
38 Countries which have already conducted auctions include New Zealand, Australia, the United
Kingdom, Egypt, India, Colombia, Argentina, the United States, and Spain. See Nordicity Report,
ibid.
39 ibid.
,0 See the discussion of second-price sealed bid auctions in Part IVC.5, below. The winner only
pays the amount of the second highest bid because, inter alia, certain auction theorists believe that the
highest bid may represent an over-valuation of the license and that the second highest bid is a closer
approximation of true value.
1999]
M.C. GLASS AND D.M. RHODES – SPECTRUM AUCTIONS
highest bid was $NZ 100,000 and the second highest bid was $NZ 6. In an even more
extreme case, the highest bid was $NZ 7 million and the next highest bid was
$NZ 5,000.” In both of these cases, the highest bidder only had to pay the lower
amount. This was politically embarrassing since it failed to raise the anticipated
amount of revenue for the government. As a result, New Zealand has radically redes-
igned its auction methodology. 2
2. Australia
Australia also suffered embarrassing difficulties with its first auctions because of
their faulty design. While the Australian auctions for two satellite television service li-
censes required the winners to pay the amount of their bids, they did not make sure
that all bidders had adequate financing or had made advance deposits. The two groups
which won the auction used this deficiency to pay much less than their winning bids.
They accomplished this by setting up a series of shell corporations which together
submitted twenty bids staggered by $Aus 5 million-ranging from very low to very
high amounts (i.e., much more than anyone would bid). After they knew the amount
of the highest bid by a third party, they allowed all the corporations to default except
the one which had bid just above the third party bid. When a winner defaults, the li-
cense passes to the next highest bidder. Thus, the license passed from one defaulting
corporation down to the next. Only the corporation which had bid immediately above
the highest third party bidder actually paid its bid for the licenses. The difference be-
tween the highest bids and the amounts actually paid was large. For one license the
highest bid was $Aus 212 million, but only $Aus 117 million was actually paid. The
other license was won by a high bid of $Aus 177 million, but the eventual licensee
only paid $Aus 77 million. These embarrassing events caused a one year delay in the
introduction of satellite television services in Australia. Furthermore, they almogt led
to the dismissal of the Minister of Communications.”3 Perhaps this debacle could have
been avoided if Australian authorities had required up front deposits from all bidders.
This would have covered any default penalties that might have been incurred.
3. Columbia
The auction system used by Colombia in 1994 to allocate three regional cellular
licenses was much more successful. The process involved an auction conducted be-
tween those applicants who had passed a technical and financial screening process de-
signed to exclude bidders who were unqualified to provide wireless service.’ The
government raised over $US 1 billion, and services were quickly deployed to the
41 See Nordicity Report, supra note 37.
2 New Zealand now uses a first-price sealed bid method. See the discussion of this method in Part
IV.C.4, below.
43 J. McMilan, “Selling Spectrum Rights” (1994) 8:3 J. Econ. Persp. 145 at 150.
” The auction methodology was a simultaneous single round auction. See the discussion of this
method in Part IVC.6, below.
MCGILL LAW JOURNAL / REVUE DE DROIT DE MCGILL
[Vol. 44
public. The success of this method should be attributed to the use of screening proc-
esses.
‘s
Ill. Comparative Advantages of Spectrum Allocation Methods
A. What Goals should Canada Seek to Achieve Through Spectrum
Allocation?
A successful Canadian spectrum allocation system is one that reflects Canada’s
spectrum allocation goals and serves the public good. The purpose of this section is to
define these goals. It will be seen that some of the criteria are germane to any spec-
trum allocation system, while others seek to reflect the distinctiveness of Canada it-
self. The following are the goals which should be promoted by a system of spectrum
allocation, and they are the criteria by which the various methods of allocation will be
subsequently judged. In establishing our criteria, it should be noted that we are mind-
ful of the letter and spirit of the Telecommunications Act.
1. Promotion of Canadian culture
Canadians want to see more of their culture reflected in the mass media. They
overwhelmingly support the government’s protection of their cultural industries and
its attempt to protect these industries in free trade agreements.” However, as the final
report of the Information Highway Advisory Council (IHAC) states, “[m]arket forces
have never provided fully for Canada’s unique needs-not because Canadians are
uninterested in their own stories, debates and heritage-but simply because there are
not enough Canadians to form a viable market, particularly for higher-cost forms of
content”‘ Canadian content helps Canadians understand their country and each other.
The view is repeated in section 7(1) of the Telecommunications Act which states that
“telecommunications performs an essential role in the maintenance of Canada’s iden-
tity and sovereignty,” and in section 3(l)(b) of the Broadcasting Act” which states that
“the Canadian broadcasting system … makes use of radio frequencies that are public
property and provides … a public service essential to the maintenance and enhance-
ment of national identity and cultural sovereignty?’ The promotion of Canadian cul-
ture and content, therefore, is a consideration that must figure prominently in the se-
“See Part V.B., below.
40 “[A] December 1996 Angus Reid survey found that 78 percent of Canadians supported govern-
ment protection of Canada’s cultural industries, while 84 percent (93 percent in Quebec) believed the
federal government should protect Canada’s cultural industries when negotiating trade agreements”
(Industry Canada, Final Report of the Information Highway Advisory Council, Preparing Canada for
a Digital World (9 September 1997) at 60, online: Industry Canada
“Ibid. at 57.
R.S.C. 1985, c. B-9.01.
1999]
M.C. GLASS AND D.M. RHODES – SPECTRUM AUCTIONS
lection of the appropriate spectrum allocation process. As telecommunications and
broadcasting converge, this concern will become even more important.” To identify
what qualifies as “Canadian” content and culture, one must look to the Broadcasting
Act as well as the content-related recommendations of the IMAC.’
2. Diversity of service providers
Diversity, tolerance and multiculturalism are hallmarks of Canadian society-so
much so that the promotion of multiculturalism has been elevated to a constitutional
imperative in this country.” In the chapter on access to the information highway, the
final report of the ilAC states that Canadians have always worked hard within their
society to overcome social factors creating barriers to equality!2 Canadian spectrum
policy should continue to ensure that these barriers are reduced. Hence, any system of
spectrum allocation should ensure that aboriginal groups, French language minority
groups, students, and other groups that are traditionally under-represented in their
spectrum use or consumption, have access to the spectrum. The United Kingdom
White Paper on spectrum allocation recommends that a spectrum allocation system
should “place due weight on … social and cultural applications of radio and protect
the continuing diversity of radio use within the community'” This means that regu-
lators should ensure: (i) that wireless service providers offer a wide range of services
including those catering to special groups, and (ii) that these special groups have suf-
ficient access to spectrum licenses to enable them to provide services.
3. Availability to businesses of all sizes
Just as under-represented social groups should have access to the spectrum, small
businesses-which are also less represented on the spectrum-should enjoy the bene-
fits of radio communication. Examples of small businesses that use spectrum are taxi
co-operatives, messenger services, and bus companies. In its submission to Industry
‘ At present, spectrum allocation is primarily relevant for telecommunications services since a
separate regulatory regime run by the Canadian Radio-television & Telecommunications Commission
(CRTC) applies to the broadcast media. However, we believe that the system outlined in this article
could be useful in the broadcast context.
0 Industry Canada, Final Report of the Information Highway Advisory Council, Connection,
Community, Content: the Challenge of the Infornation Highway (1 August 1995) at c. 3, online: In-
dustry Canada
more detailed definition of “Canadian culture and content” is beyond the scope of this paper.
“‘ Canadian Charter of Rights and Freedoms, s. 27, Part I of the Constitution Act, 1982, being
Schedule B to the Canada Act 1982 (U.K.), 1982, c. 11 provides that the “Charter shall be interpreted
in a manner consistent with the preservation and enhancement of the multicultural heritage of Cana-
dians.”
52Preparing Canada, supra note 46 at 41.
U.K., Department of Trade and Industry, Spectrum Management: Into the 21st Century (17 June
1996), online: Radiocommunications Agency
accessed: 14 February 1999) [hereinafter UK Report].
MCGILL LAW JOURNAL / REVUE DE DROIT DE MCGILL
[Vol. 44
Canada, Quick Messenger Service Ltd. (a courier company in Ottawa) expressed con-
cern about the impact of changes in allocation and licensing procedures on the cost of
doing business.’ In the technology sector, the small businesses of today could very
well be the corporate leaders of tomorrow. Their progress should not be stifled by lack
of access to the spectrum. Spectrum allocation policy should ensure that frequency
continues to be used by businesses of all sizes in order to generate wealth and jobs,
and to improve competitiveness.”
4. Enhanced competition and customer choice
Since Canada is a diverse nation, the services offered to Canadians should be
equally diverse. Customer choice is enhanced when wireless service providers offer a
wide range of programming and services. An allocative process should strive to en-
sure enhanced competition and customer choice. As broadcasting and telecommuni-
cations converge, these objectives will become increasingly important.
5. Speed of allocation
In the United Kingdom, an economic impact study calculated that a two year de-
lay in introducing PCS as competition to existing cellular services would have re-
sulted in economic losses of over 2 billion per year by the end of the century, and the
loss of over 7,000 jobs.” A prominent member of the Progress and Freedom founda-
tion has stated that a three year delay in allocating a license can easily wipe out one-
fourth of the present value of the license. 7 Furthermore, spectrum that has not been
allocated is spectrum that is not being utilized-thereby depriving customers of serv-
ices to which they might otherwise have access. The speed in which the computer in-
dustry changes is universally acknowledged. If companies such as Apple, Lotus and
Intel had required spectrum to get to market, it is arguable that the delays they might
have encountered could have resulted in a postponement of the personal computer
revolution.” Similarly, the increasing convergence of broadcasting and telecommuni-
cations could be significantly slowed if unnecessary delays occur in the allocation of
” Submission of Quick Messenger Service Ltd. to Industry Canada Re: Notice, C. Gaz. 1996.
DGRB-002-96: Industry Canada, “Consultation on Radio License Fees-Phase 1” (22 April 1996).
Submission of U-need-a-cab limited to Industry Canada Re: Notice, C. Gaz. 1996. DGRB-002-96:
ibid. (13 May 1996).
UK Report, supra note 53 at para. 4.5.
“Ibid. at para. 3.7. Similarly, it has been estimated that the cost of regulatory delay in the introduc-
tion of cellular telephone service in the United States is close to $100 billion with more than $25 bil-
lion lost in a single year (J.H. Hausman, “Valuing the Effect of Regulation on New Services in Tele-
communications” (1997) Brookings Papers on Economic Activity, Microeconomics, online: The
Brookings Institution
” Testimony of Peter K. Pitsch before the FCC, online: En Banc Hearing on Spectrum Allocation
” Ibid.
1999]
M.C. GLASS AND D.M. RHODES – SPECTRUM AUCTIONS
spectrum licenses. It should also be noted that delays in the introduction of new serv-
ices can effectively subsidize existing services which would otherwise face increased
competition from the new services. The speed of spectrum allocation, therefore,
should be a priority in any new regime of license distribution.
6. Rapid network deployment
Quick spectrum allocation is important, but by itself it does not ensure the rapid
proliferation of new services. Individuals and corporations with capital, established
use plans, and existing infrastructure are more likely to rapidly put spectrum to use.
The longer spectrum is not actually being used, the longer consumer choice and the
overall economy suffer. Hence, spectrum policy should select the market participants
who would most rapidly develop the communications infrastructure. Participants who
have paid for their spectrum will have the greatest incentive to deploy their networks
quickly in order to recover their investment.
7. Efficient use of the spectrum
Once spectrum has been allocated and is being used, Canadians should maximize
their enjoyment of the resource. As such, licensees should maximize the amount of
information that they transmit by way of the frequency allocated to them. For exam-
ple, Canadians would be better served if spectrum services where offered twenty-four
hours a day. Similarly, they would be better served if they could receive voice, video
and digital information over the spectrum, rather than simply one or two of those
forms of information. Increased spectrum efficiency might require that users relin-
quish unused or under-used spectrum, or that they engage in greater sharing of chan-
nels.” For example, if the military no longer needed a certain frequency, an ideal
spectrum allocation model would provide for the re-allocation of that unused fre-
quency, or provide for shared use with another user until the military again required
the band. It is clearly in the interest of all Canadians that the method of spectrum allo-
cation encourage efficient use.
8. Access to services
Spectrum users should be encouraged to provide services to as many people as
possible within the area covered by their licenses. Canadian consumers should be able
to choose from a wide range of commercial and non-commercial spectrum-based
services at a reasonable cost, regardless of location. Access to spectrum-based serv-
ices is limited for people living in rural and remote areas-particularly the far
North-because the low population density makes it inefficient to deploy networks. ‘
Spectrum allocation policy should address these concerns.
‘9 UKReport, supra note 53 at para. 4.4.
‘0 Preparing Canada, supra note 46 at 45.
MCGILL LAW JOURNAL / REVUE DE DROITDE MCGILL
[Vol. 44
9. Ensuring sufficient spectrum for essential public services
The armed forces, air traffic control, emergency services, meteorologists and cer-
tain charities all require spectrum for their everyday operations. Because of their pub-
lic service missions, these organizations frequently lack the financial resources to ac-
quire their spectrum on the open market. The major difference between public service
providers and commercial users is their inability to charge for services and thereby
generate revenue. For example, in Barrie, Ontario, the police service expressed con-
cem that if they had to compete on the same financial terms with private sector com-
panies for their right to use the spectrum, they would be hindered in their ability to
provide policing services.” Similarly, in Etobicoke, Ontario, the fire department ex-
pressed concern that if its access to spectrum was impeded, it would place the public
in grave danger.” Canadian spectrum policy should therefore ensure that these vital
organizations continue to have access to the spectrum.63
10. Transparency of the allocation process
Ideally, the spectrum allocation process should be as objective as possible so that
applicants can see that it is fair and non-discriminatory. It is important that all partici-
pants see that spectrum allocation decisions are based solely on the stated criteria.”
The process, in particular, must ensure that there is no appearance of favoritism to-
ward established spectrum service users over new entrants. The SMB is sensitive to
the requirement for procedural fairness in the spectrum allocation process and has
strived to instill the comparative process with a greater degree of transparency.’
11. Revenue for the government
As illustrated in the discussion of the FCC auction process,” fiscal austerity has
led to a temptation on the part of government to maximize the revenues it obtains
from spectrum allocation. This temptation should be resisted. Instead, as one com-
mentator has argued, “spectrum should be released as rapidly and efficiently as possi-
ble, even if that reduces the government’s auction proceeds, because the goal is to free
up spectrum for the benefit of the … economy.”‘ Not requiring spectrum users to pay
for their use, however, will encourage inefficient and excessive use of scarce spectrum
resources. Hence, the amount of revenue that the government raises should be a crite-
6, Submission of Barrie Police Service to Industry Canada Re: Notice, C. Gaz. 1996. DGRB-002-
96: supra note 54 (29 March 1996).
62 Submission of Etobicoke Fire Department to Industry Canada Re: Notice, C. Gaz. 1996 DGRB-
002-96: supra note 54 (29 March 1996).
“This fact is recognized in UKReport, supra note 53 at paras. 4.11-4.13.
“Ibid. at c. 4.
“”Comparative Selection Review”, supra note 14 at 13.
See discussion at Part ll.B.3, above.
6’ Pitsch, supra note 57 at 4-5.
1999]
M.C. GL4SS AND D.M. RHODES – SPECTRUM AUCTIONS
non in the spectrum allocation process, but only to the extent that charges for use are
the best way to ensure efficient spectrum use.
B. Relative Advantages of Different Methods of Allocation
Having identified the goals and values which a Canadian spectrum allocation
scheme should seek to achieve, the capability of various models to achieve these goals
will now be assessed. Keep in mind that the following discussion assumes the “pure”
form of each model, i.e., a system that is not tailored to circumvent some of the inher-
ent defects in the selection process. A tailored process usually exhibits traits of other
allocation methods and is therefore known as a hybrid. Despite the benefits of hybrid
processes, an examination of each method of allocation in its pure form better reveals
its inherent advantages and disadvantages.
1. First come/first served
According to this model, the first person to request a license obtains it. Since the
first claimant may have no interest in promoting social policy goals, this model does
not ensure that content reflecting the distinctiveness and diversity of Canada will find
its way onto the airwaves. Similarly, given its random nature, this allocation method
does not ensure that spectrum will be available to businesses of all sizes or that con-
sumer choice will be enhanced. While the first come/first served method is certainly a
speedy method of allocation, the rapidity of network deployment and efficient use of
spectrum will depend entirely upon the intentions, financial resources and technical
know-how of the first claimant. Furthermore, there is no guarantee that public service
providers will be among the first claimants for a license. Sufficient spectrum is there-
fore not ensured for essential public services. Additionally, giving away spectrum to
the first person who claims it does not generate any revenue for government. Aside
from being speedy, the only redeeming feature of this allocation method is its trans-
parency. Giving the right to utilize a frequency to the first person in line certainly en-
tails an objective determination. In sum, the adoption of a first come/first served
model amounts to leaving the fulfillment of policy goals to chance. This is clearly not
a good way to ensure that Canada’s spectrum allocation objectives are met.
2. Lotteries
In a lottery, applicants participate in a random draw for the available spectrum.
Lotteries suffer from the same drawbacks as the first come/first serve model. Spurred
by low capital requirements, unqualified, undercapitalized and inexperienced parties
are able to obtain spectrum licenses.” Since anyone can obtain a license, this assign-
W. Kummel, “Spectrum Bids, Bets, and Budgets: Seeking an Optimal Allocation and Assignment
Process for Domestic Commercial Electromagnetic Spectrum Product, Services, and Technology”
(1996) 48 Federal Comm. L.J. 511 at 526, online: Federal Communications Law Journal
MCGILL LAW JOURNAL / REVUE DE DROIT DE MCGILL
[Vol. 44
ment method delays development of spectrum services, limits diversity of service
providers and limits universality of customer access to services-and more often than
not leads to delays in network deployment and overall inefficient use of spectrum.
The lottery process primarily results in private gains without generating direct public
revenue. The American experience shows that the lottery process fuels speculation,
and in many cases allows license winners to realize windfalls by selling their licenses
in secondary market transactions.’
In a typical case, a little-known group called
RACDG Partnership was chosen by lottery in 1989 to operate cellular telephone
services in Cape Cod. The partners sold their license to Southwestern Bell for $41
million.”0 This delayed the assignment of the license to the eventual service provider
while allowing lucky private citizens to pocket money that could have gone into gov-
ernment coffers. Lotteries do not quickly and efficiently get spectrum licenses into the
hands of those who can actually use them. The need for after-market transactions re-
sults in significant losses because-as illustrated-time is of the essence in the spec-
trum allocation process.” Furthermore, the lottery process does not ensure that Cana-
dian culture will be provided via the radio spectrum, or that small businesses and es-
sential public services will have access to spectrum. The only positive feature of this
method of allocation is its speed of license allocation and its transparency. Having a
machine choose a number is quick and objective. These are the same features that
prompt millions of Canadians to play various lotteries each year. While this method is
fair and equitable, it is does not ensure that Canada’s spectrum allocation goals are
met.
3. Comparative analysis
In a comparative process of allocating spectrum licenses, proposals are judged on
their merits with licenses allocated to applicants who best meet certain specified crite-
ria.” The criteria can be tailored to meet Canada’s spectrum policy goals, and can be
made to reflect the desire to protect and foster Canadian culture and a diversity of
service providers. Of course, any applicant can promise to satisfy certain social policy
goals; it is up to the government to ensure that the licensee actually fulfills them.’
Making the size of the applicant’s business and the nature of the organization relevant
criteria can ensure that small businesses and essential public services get their fair
share of spectrum. A comparative analysis guarantees that only applicants with a
credible business plan, sufficient capital, and the requisite level of expertise are allo-
cated spectrum licenses-leading to a greater probability that the spectrum services
” M. Spicer, “International Survey of Spectrum Assignment for Cellular and PCS” (1996) at
Part 111.4, online: FCC Wireless Telecommunications Bureau
Survey”].
‘0 “Selling Spectrum Rights”, supra note 43 at 146.
7, See discussion at Part IIIA.5, above.
S”Comparative Selection Review”, supra note 14 at 3.
,”International PCS Survey”, supra note 69 at 6.
1999]
M.C. GLASS AND D.M. RHODES – SPECTRUM AUCTIONS
and networks will be rapidly deployed. The costs of preparing an application in a
comparative process are considerably high because of the large amount of detailed in-
formation that has to be presented. A comparative process can also help to ensure that
service is provided to rural areas, by making the granting of a license covering major
urban centers contingent upon servicing outlying areas.
It has been argued, however, that determining efficient use of spectrum is a com-
plex task requiring vast amounts of continually updated information beyond the ca-
pacities of the administrators charged with performing the analysis.” Furthermore,
under the comparative analysis process, licensees who no longer need some or all of
the spectrum covered by their licenses have no real incentive to return the unused fre-
quencies for reassignment or to migrate to uncongested frequencies.’ Since the allo-
cating agency must process information on a wide range of factors, comparative
analysis suffers from delays brought about by the need to wade through voluminous
submissions and deliberate at length on their contents. The FCC has reported that the
average number of days from the filing of an application to the granting of a license is
720.” Another major disadvantage of the comparative analysis process is its lack of
transparency. The final decision of an allocating agency is discretionary, and this may
cause some applicants to feel that they have been treated unfairly. Such suspicions are
hard to dispel and often lead to judicial review of the allocation decision where avail-
able. Discretionary decision-making by government officials can also invite political
interference into the allocation process. In France, the fact that senior politicians be-
came involved in the selection process for a national PCS license just weeks before
the license was allocated made it appear that the final decision was influenced by po-
litical considerations.’ In Ireland, although several major international companies ap-
plied for spectrum licenses, the government awarded a Global System for Telecom-
munications (GSM) license to a little-known consortium. The unhappiness of the un-
successful participants was compounded by the government’s refusal to divulge its
reasoning in awarding the license in that manner. One disappointed participant ap-
pealed the government’s decision to the European Commission.’ While the integrity
of the Canadian comparative analysis process has never been questioned, the subjec-
tive nature of any such process inevitably leads to a lack of transparency.
4. User fees
Under a user fee system, a license is granted to the first person who is willing to
pay a set fee. Policy goals-like the promotion of Canadian content guidelines-are
not assured in the user fee model, as the first person willing to pay for the license will
74 Pitsch, supra note 57 at 2.
75 UK Report, supra note 53 at para. 3.5.
76 This can be contrasted with the much shorter delay of 233 days for auctions. See “FCC Report”,
supra note 32 at s. 5(a).
“International PCS Survey”, supra note 69 at 13.
,’ Ibid. at 12.
MCGILL LAW JOURNAL / REVUE DE DROJTDE MCGILL
[Vol. 44
not necessarily be interested in advocating such objectives. Access by small busi-
nesses and diversity of service providers is also not assured. In fact, such a system fa-
vors better capitalized organizations who are able to devote significant time and fi-
nancial resources to monitoring the allocation of spectrum licenses in order to claim
them as they become available. Enhanced competition and customer choice are not
assured by this variation on the first come/first served method. Moreover, this method
of allocation fails to ensure that sufficient spectrum is available for essential public
services. Public service organizations are not well funded in this era of budget cuts,
and they could be hard-pressed to pay the required fees.’ The absence of administra-
tive analysis, however, makes the user fee system a speedy and transparent method of
allocating spectrum licenses. Rapid network deployment would be encouraged. Eager
to recoup the money they paid for the spectrum license, licensees would have some
incentive to use the spectrum in the most efficient way possible, in order to maximize
their return on investment. This method would also raise revenue for the government.
However, it would be difficult for the government to set an appropriate price for the
use of spectrum because licenses are difficult to price without the possibility of mar-
ket feedback. The government would be forced to estimate the price that users would
be willing to pay if spectrum were available on the free market. The haphazard nature
of the pricing mechanism can send the wrong signals to the market about the amount
that should be invested to exploit the spectrum. A high price would result in over-.
investment and a low price would lead to under-investment. These pricing inefficien-
cies would result in higher costs and risks, forcing spectrum users to seek higher re-
turns on their investments.’
Under a user fee system, universality of service-particularly in rural areas with
lower density and lower incomes-would improve as licenses in these areas would
cost less than those in major metropolitan areas with higher density and lower in-
come, thereby reducing development costs.
5. Auctions
Under an auction system, spectrum licenses are allocated to the highest bidder.
Auctions would not by themselves ensure the promotion of Canadian culture on the
radio spectrum, as the highest bidder may not have any interest in promoting it. Di-
versity of service providers is impeded because the communities whose interests
should ideally be promoted rarely have the financial resources to compete effectively
on the open market. Essential public services and small business are among the
As such, the Emergency Measures Organization of Nova Scotia feels that these organizations
should receive special consideration for radio frequency license purposes. See Submission of Emer-
gency Measures Organization of Nova Scotia to Industry Canada Re: Notice, C. Gaz. 1996. DGRB-
002-96: supra note 54 (19 May 1996).
‘0 The foregoing discussion on setting user fee prices borrows heavily from Kummel, supra note 68
at 526; and Ti. Shroepfer, “Pee-based Incentives and Efficient Use of Spectrum” (1992) 44 Federal
Comm. L.J. 411 at 419.
1999]
M.C. GLASS AND D.M RHODES – SPECTRUM AUCTIONS
groups vulnerable to a market-oriented allocation process.” Since large enterprises
generally have more capital than small enterprises, small business access to the spec-
trum is also not assured. For example, if a taxi company or emergency dispatcher had
to compete directly with a cellular service provider for the same spectrum band, it is
unlikely that the former would ever win. Enhanced competition and consumer choice
is also not assured. While specialty channels might generate some interest among
consumers, that interest might not be sufficiently widespread to justify a bid that is
competitive vis a vis more mainstream services. The speed of allocation is much
higher under the auction system than under the comparative analysis process, as auc-
tions do not require administrative officials to confront a morass of papers or to delib-
erate before deciding. 2 Auctions, however, are slower than either the first come/first
served or the user fee methods of allocation. Speed of network deployment is in-
creased because the cost of the license induces successful bidders to launch services
as soon as possible in order to recoup their initial investment.’3 For the same reason,
efficient use of the spectrum is highly likely. Proponents of the auction process also
argue that markets can “parallel-process” information gathered by millions of con-
sumers and businesses to continually provide signals as to the relative usefulness and
efficiency of a service-thereby rendering markets a more effective method to deter-
mine which use is or is not “efficient”.” Service to less populated areas would in-
variably improve. With licenses in remote areas costing comparatively less than those
in major urban centers, startup costs would diminish, providing incentive for wireless
service providers to expand into outlying markets.” The auction process itself is very
transparent and objective. Participants know the rules in advance, and it is clear why
the bidders finish as they do.” Auctions, moreover, also raise significant revenue for
the government.
To alleviate this problem, the UK Report recommended that safeguards be put into place to ensure
that essential public services have continued access to spectrum licenses: see supra note 53 at para.
4.11.
“See supra note 76 and accompanying text.
“Kummel, supra note 68 at 527; and UK Report, supra note 53 at para. 6.2. Some industry appli-
cants to Industry Canada have suggested that the converse would actually result if auctions were to be
used. Opponents worry that auctions would attract speculators whose intentions would be to hoard the
spectrum and then resell it at a profit to firms who wish to provide services (see “Comparative Selec-
tion Review”, supra note 14 at 11). This fear can be effectively addressed through the adoption of
“use it or lose if’ provisions. Opponents also fear that the large up-front sum that would have to be
paid by an auction winner would constitute a barrier to entry and rapid service development. While
this is a legitimate concern, it can be alleviated by allowing the winner to pay for the spectrum over
the duration of the license (see “Comparative Selection Review” at 10).
‘, Questions and Answers: Framework for Spectrum Auctions in Canada at question 7, online: In-
(date accessed: 24 January
dustry Canada
1999).
‘4″Comparative Selection Review”, supra note 14 at 10.
Kummel, supra note 68 at 526.
MCGILL LAW JOURNAL / REVUE DE DROITDE MCGILL
[Vol. 44
The chart below gives a general overview of the pros and cons of various methods
of spectrum allocation. Comparative analysis and auctions appear to be the best ap-
proaches. It is beyond the scope of this article to discuss the current comparative
analysis model. Rather, the focus is the advisability of adopting an auction model in
Canada. Objections that have been raised to the auction process should be confronted.
Only then may one examine the extent to which auctions can be tailored to overcome
its inherent problems.
COMPARATIVE ADVANTAGES OF SPECmUM ALLOCATION METHODS
FirstCome/First
Served
Lotteries
Comparative
Analysis
User fees
Low (random)
Low (random)
Very High
Low (random)
Pure
Market
Auctions
Low
(random)
Low (random)
Low (random)
High
Low (random)
Very Low
Low (random)
Low (random)
Very High
Low (random)
Low
Low (random)
Low (random)
High
Low (random)
Low
Very High
High
Low
Low (random)
Low (random)
High
Low (random)
Low (random)
High
High
High
High
Low (random)
Low (random)
High
High
High
Very
High
Very
High
High
Low (random)
Low (random)
Very High
Very Low
Very Low
(1) Promote Cana-
dian Culture
(2) Diversity of
Service Providers
(3) Availability to
Businesses of all
Sizes
(4) Enhanced Com-
petition and Cus-
tomer Choice
(5) Speed of Alloca-
tion
(6) Rapid Network
Deployment
(7) Efficient and In-
tensive Spectrum
Use
(8) Universality of
Service (rural and
urban)
(9) Ensuring Suffi-
cient Spectrum for
Essential Public
Services
(10) Transparency of Very High
Decision Making
Process
Very High
Very Low
Very High
(11) Revenue for
Government
Very Low
Very Low
Very Low
High
Very
High
Very
High
Figure 1: The inherent features of different pure methods of allocation. Note that first come/first
served, user fees and lotteries are significantly less desirable than auctions or comparative analysis.
1999]
M.C. GLASS AND D.M. RHODES – SPECTRUM AUCTIONS
171,
C. Specific Objections to Pure Market Auctions
In addition to the defects inherent in the auction process described above, several
specific objections have been made against auctions as a method of spectrum alloca-
tion. These objections will now be explored along with possible answers to these le-
gitimate concerns.
It has been argued that while auctions may be workable in the United States, they
are not suitable for a country like Canada with a smaller population and lower levels
of spectrum congestion.” By the late 1980s, spectrum scarcity was evident for certain
bands in Canada’s major urban centers.” However, countries such as Panama ‘ and Is-
rael-whose populations are much smaller than Canada’s-have successfully con-
ducted spectrum auctions.
Opponents argue that the development of spread spectrum technology-which
allows devices to scan the spectrum and utilize any frequency that is free at that given
moment-will render the current paradigm of exclusive usage of a single frequency
obsolete. Hence, allocation methods like auctions that perpetuate the current paradigm
retard the development of technology. According to Industry Canada, however, the
extent to which new technology may render spectrum scarcity a moot point in the fu-
ture cannot be known with certainty.” In any case, should proponents of spread spec-
trum technology turn out to be right, it would be much easier for the government to
expropriate the licenses in Canada than in the United States, where license holders
enjoy a constitutional guarantee to property rights. Alternatively, when the duration of
a license ends, the SMB could refuse to renew it and instead designate the associated
portion of the spectrum for spread spectrum use. It would therefore appear that
adopting an auction method of license allocation would not hamper Canada’s ability
to adapt to the emergence of spread spectrum technology in the future.
Critics fear that auctions may give rise to cases where one party would out-bid its
rivals for spectrum rights simply to prevent competition, thus creating monopolies.’
The report of the Radiocommunications Agency in the United Kingdom acknowl-
edged this potential problem. It thereby recommended that caps be placed on spec-
trum holdings to prevent a single bidder from acquiring an undue concentration of li-
censes so that it could generally affect competition.’ Another possible counter-
measure that could be adopted to solve this problem is the extension of bidding pref-
erences to smaller companies by way of bidding credits. 3 This measure would be de-
sirable in Canada. Competition law would presumably apply to prevent monopoliza-
“Comparative Selection Review”, supra note 14 at 11.
“Auction Consultation Review”, supra note 15 at 2.
“”International PCS Survey”, supra note 69 at 4.
9″Comparative Selection Review”, supra note 14 at 11.
” Ibid.
9, UKReport, supra note 53 at paras. 6.12-6.13.
‘3 See discussion at Part IV.B.2, below.
MCGILL LAW JOURNAL / REVUE DE DROIT DE MCGILL
[Vol. 44
tion of the spectrum because it represents an essential facility for all competitors in
the wireless industry.
Opponents of auctions also argue that any amount paid in an auction for a license
would be passed on to consumers in the form of higher prices. However, it is more
likely that licensees will charge what the market can bear regardless of whether or not
they have to pay to get spectrum. Even if costs are passed onto consumers, it could be
argued that it is more Pareto efficient4 to make consumers who actually use the serv-
ice pay for the cost of spectrum, rather than to make all members of society pay by
foregoing the public revenue that could be raised.
A major objection to spectrum auctions is that they require winning firms to make
large payments at the outset, thereby depriving them of the capital required to develop
their wireless networks and services. This leads to slower network deployment and a
higher risk of firm failure. This objection stems from the United States experience.
After the auctions for Block C PCS licenses, many of the winning bidders asked for
extensions of their payment deadlines because they had insufficient funds to simulta-
neously pay their winning bids and construct their PCS networks. This demonstrates
that an auction system must be careful to select market participants who have suffi-
cient financing to simultaneously absorb both costs. In many of these situations, the
firms defaulted on their payments and forfeited their licenses to the FCC. This re-
quired new auctions which delayed the introduction of PCS services in some parts of
the United States by six months.
An auction can be structured to accommodate the payment of a winning bid over
a given time frame. The financial burdens imposed on winning bidders thus can be
spread evenly across the entire term of the license, or deferred until the network is de-
ployed. Significant up front payments, however, offer one critical benefit: they reduce
the probability of speculation in spectrum licenses.
Speculation occurs when a firm acquires a spectrum license in the hope that eco-
nomic and technological changes in the near future will increase its value and allow
the speculator to sell the license at a profit. Speculation can prevent the efficient as-
signment of licenses which might otherwise be acquired by firms intending to use
them to provide services immediately. Speculation is especially likely in auctions
where winning bidders are allowed to pay the value of their bids over the entire dura-
tion of the license. Indeed, the possibility that a license will ultimately be worth less
than anticipated is not a large risk because the speculator can always stop paying and
permit the license to be reclaimed by the allocating agency. The simplest way to avoid
speculation is to require the winning bidder to pay the entire amount of its bid near
the beginning of the license termY. As previously discussed, however, large up-front
9, Pareto efficiencies arise when market participants maximize the internalization of the benefits and
detriments of their conduct. In essence, an allocation is pareto efficient if there is no other allocation
which would make some people better off and no one worse off.
9 This was done in FCC auctions: see “FCC Report”, supra note 32.
1999]
M.C. GLASS AND D.M. RHODES – SPECTRUM AUCTIONS
payments could be burdensome to winning bidders, since they will occur at the time
when they are investing heavily in the wireless networks necessary to exploit their
spectrum licenses. It is therefore proposed that a middle course be adopted. The allo-
cating agency should require a significant up-front payment (20-35%) and then spread
the remaining payments over the first half of the license period. Thus, a winning bid-
der who bid $1 million for a ten year license might pay $250,000 after the auction and
then make annual payments of $150,000 for the next five years. The balance of pay-
ments should be adjusted to take into account special factors arising from different
types of licenses. This recommendation is a simple, yet effective compromise.
Another effective method of accomplishing the same result is to make it a condi-
tion of a spectrum license that the holder provide services within a stipulated period or
risk forfeiture. The difficulty with this approach is that service providers may only
provide the minimum service necessary to avoid losing their licenses, thus the public
will not receive the benefit of efficient spectrum use.
D. Meaning of these Results for Canada
Despite the criticisms leveled against the auction process, it remains a viable op-
tion. According to Figure 1-outlining the comparative advantages of different spec-
trum allocation methods-the comparative analysis process is the best way to allocate
spectrum licenses. However, as previously mentioned,” Figure 1 only takes into ac-
count the “pure forms” of the various methods of allocation. As such, it does not con-
sider the fact that auctions can be tailored to the Canadian context. Comparative
analysis, as set out here, is already customized for the Canadian context because of
the special selection criteria it uses when judging applicants. Therefore, before draw-
ing any conclusions about which system is best for Canada, the extent to which an
auction system can be tailored to the Canadian context must be analyzed.
IV. Tailoring an Auction System for Canada
A. Introduction
As previously illustrated, the major drawback of the auction method lies in its
failure to promote goals unrelated to efficiency. This difficulty would make auctions
as a method of spectrum allocation unattractive to many countries were it not possible
to develop hybrid auction models to incorporate some of the beneficial elements of
other methods of allocation. There are as many hybrid models as there are ways of in-
corporating non-auction elements into a market-driven method of spectrum allocation.
Some of the more popular modifications to the strict auction are the reservation of
spectrum for specific purposes, the use of bidding credits, the screening of auction
96See discussion at Part 11.B., above.
MCGILL LAW JOURNAL / REVUE DE DROIT DE MCGILL
[Vol. 44
applicants and the imposition of license conditions. The following section will con-
sider each of these variations.
B. Hybrid Models-Blending Auctions with Other Methods of
Spectrum Allocation
1. Reservation of spectrum for specific purposes
Reserving spectrum for particular purposes is not really a hybrid auction process
since the allocating agency simply determines that certain spectrum bands will be al-
located in some manner other than by auction. Despite this complete exclusion of the
market mechanism, every country which has conducted spectrum auctions has re-
frained from auctioning those portions of the spectrum which are used for public
services. This makes sense because most groups providing public services have insuf-
ficient revenues to outbid private commercial interests.’ A wide variety of spectrum
users fall into this category: ambulance services, air-traffic controllers, the military
and any government body providing essential services. On account of the benefit de-
rived from their providing free television to the public, another group that could be
exempted from the auction process are public broadcasters.”
The advantages of reserving spectrum are clear. It is an excellent way to ensure
sufficient spectrum for essential public services. It is also a good way to promote Ca-
nadian culture, diversity of service providers and availability of spectrum to busi-
nesses of all sizes. Because of the high degree of administrative control over spectrum
reservation, it is very likely that these goals can be met. The disadvantages, however,
are equally apparent. Reservation of spectrum can easily lead to inefficient spectrum
use by the groups exempted from competition or payment for spectrum licenses. It
also encourages these groups to use spectrum technologies where non-spectrum sub-
stitutes are available. It may be difficult to determine whether the benefits derived
from reserving spectrum licenses for these groups justifies the loss of revenue because
without an auction, the value of the reserved spectrum licenses can only be estimated.
It is also possible to combine spectrum reservation with an auction system by re-
serving certain spectrum bands for auction among designated entities. For example, in
the FCC auction process, only small businesses could bid on certain spectrum li-
97See discussion at Part liA. 11, above.
9’The spectrum bands assigned to television broadcasters were excluded from the FCC auction pro-
cess. However, there was some debate about whether this should be the case. In Canada, the SMB has
indicated that the spectrum assigned to broadcasters licensed by the CRTC are not subject to auction-
ing (see “Auction Consultation Review”, supra note 15 at 8). Canadians would benefit if broadcast
spectrum were treated in the same fashion. As the country moves toward convergence in communica-
tions technologies, maintaining a distinction between broadcasting and telecommunications spectra is
tenuous and unnecessary. One allocation system could adequately serve both fields.
1999]
M.C. GLASS AND D.M. RHODES – SPECTRUM AUCTIONS
censes. This compromise may represent the best way to reserve spectrum-at least for
those groups which face some competition in providing their public services.
2. Bidding credits
A system of bidding credits promotes particular groups or goals by al’ -wing cer-
tain auction participants to pay only a specified percentage of their actual bids. The
FCC relied heavily on this method to promote the participation of small businesses in
the auction process. The FCC awarded a 15% bidding credit to those businesses
which had gross annual revenues of less than $3 million, and a bidding credit of 10%
to those businesses which had gross annual revenues of less than $15 million.’ Ac-
cordingly, a business with a 15% bidding credit would only have to pay 85% of its
winning bid–enabling it to make higher bids and compete more effectively with
better capitalized entities. According to the FCC, this system was a success because a
high proportion of licenses were won by small businesses. However, the success of
the FCC’s use of bidding credits has been called into question.'” For example, of the
1,020 SMR licenses that were auctioned, 250 were awarded to small businesses. ‘
Bidding credits are especially effective where the goal to be promoted is specific and
can be associated with one or more auction participants. Thus, they could effectively
promote diversity of service providers, availability of spectrum licenses to businesses
of all sizes, and availability of spectrum for public services. Conversely, bidding cred-
its are not useful for promoting goals like Canadian culture and content, speed of allo-
cation and efficient spectrum use. The major disadvantage of bidding credits is that
they require a price preference to be assigned to each designated entity which pro-
motes the chosen goal without over-compensating that entity. A more mundane prob-
lem with bidding credits is the fact that a targeted group’s capital-while clearly infe-
rior to that of other auction participants-must still be significant for a price prefer-
ence to be useful.
Economic studies have shown that a proper bidding credit system does not result
in decreased government revenue. A price preference for the targeted firms stimulates
the bidding competition, forcing other auction participants to bid higher. If the price
preference is set at the correct level, its revenue raising effect from higher bids by
non-targeted firms outweighs its revenue lowering effects from chance wins by tar-
geted firms paying lower prices. As a rule of thumb, a 10% preference would result in
about 10% more of the licenses being won by targeted firms than would have hap-
pened without the preference.'”
” “FCC Report”, supra note 32 at s. 5(e).
’00 See P Cranton, “The FCC Spectrum Auctions: An Early Assessment” (1997) 6 J. Econ. & Mgt.
Strategy 431.
101 Ibid.
” “Selling Spectrum Rights”, supra note 43 at 145.
MCGILL LAW JOURNAL / REVUE DE DROIT DE MCGILL
[Vol. 44
3. Screening auction applicants
When the allocating agency conducts a screening of auction applicants it requires
all potential bidders to submit relevant information and show that they meet certain
policy driven criteria for admission to the auction process. Any criteria can be chosen,
but the process must be carefully designed to ensure it does not unwittingly import the
shortcomings of a full blown comparative analysis. The key is to avoid trying to select
the best bidders from the pool of applicants. Instead, the goal should be to exclude
those potential bidders who are clearly unsuitable. This means that applicants should
only be required to meet those criteria which are truly essential. If the screening stage
is too demanding, there is a risk that it will prejudge the auction by excluding appli-
cants who might have been able to use the spectrum efficiently. In particular, there is a
risk that start-up firms might not be admitted, thus depriving Canadians of the benefits
and innovations of competition. Screening criteria, therefore, should not include those
elements which are already accounted for in the auction process. These include effi-
cient use of the spectrum, suitability of the applicant’s proposed services to one or
more spectrum bands, and the relative competitiveness of the applicant.
The more objective the criteria used to screen applicants, the quicker and more
transparent the system will be. This does not mean, however, that less objective crite-
ria cannot or should not be used. It simply means that less objective criteria should not
be the overriding concern, and when used, it should be as clearly defined as possible.
If administered correctly, a screening process can ensure the strengthening of Cana-
dian culture, the expertise and financial capacity of applicants, and the basic feasibil-
ity of the services applicants propose to provide.
Eliminating competition among applicants will markedly speed up the screening
process since applicants will not feel obliged to submit large amounts of detailed in-
formation to offset the risk of being denied spectrum licenses. Most of the delays in-
volved in a comparative analysis stem from the need for applicants to prepare volumi-
nous submissions and the need for spectrum managers to read them. Two other ways
to speed up the process are to prohibit the use of hearings and to permit applicants ex-
cluded by the allocating agency to submit additional information before the close of
the screening phase.
The great benefit of a screening process is that it can take into account non-
objective criteria which are not easily factored into a pure auction system. For an auc-
tion to function properly, however, the addition of a screening mechanism should re-
main true to the efficiency advantages of the auction process.
4.
Imposing conditions on spectrum licenses
The imposition of conditions on spectrum licenses is attractive because it pro-
vides a way to ensure that all licensees do things which the allocating agency consid-
ers to be in the public interest. However, conditions are only compatible with auctions
where they are sufficiently limited and well defined for bidders to accurately estimate
the costs associated with compliance. If too many vague conditions are attached to a
spectrum license, bidders will be unable to accurately predict its value-and the auc-
1999]
M.C. GLASS AND D.M. RHODES – SPECTRUM AUCTIONS
tion may not allocate the license to a bidder who is actually able to exploit the spec-
trum most efficiently. License conditions, therefore, are only suitable for promoting
those policy goals which can be clearly articulated in an objective fashion-such as
stipulated network construction requirements and price caps on wireless services.
Promoting Canadian culture or ensuring sufficient spectrum for essential public serv-
ices is simply not specific enough to be a good license condition. Attaching major
conditions to licenses can also significantly reduce revenues for the government be-
cause bidders will likely over-estimate the cost of compliance with these conditions.
C. Selecting an Auction Method
If Industry Canada were to decide that auctions are the mechanism of choice for
spectrum allocation in Canada, it would then have to determine the actual design of
the auction. The importance of auction design should not be underestimated. In Aus-
tralia, the failure of the auction design to ensure that bidders actually paid for the
spectrum they won almost resulted in the replacement of the Minister of Communica-
tions and delayed the introduction of satellite-television services by one year. ” Any
oversight in auction design can have harmful repercussions, as bidders can be counted
on to find ways of outsmarting the allocation method.
1. Criteria to be used in comparing different auction methods
Since one of the goals of auctions is to raise money for the government, the level
of revenue generated by a particular method of auctioning is a relevant consideration.
Spectrum licenses are interdependent, since firms trying to build a regional or na-
tionwide presence may have to purchase several licenses. In essence, they can be seen
as bricks in a larger wall, rather than as separate and discrete bundles of rights. Since
licenses can complement each other, the value of an aggregation of licenses may also
greatly exceed their independent value. Aggregation is also economically efficient.
Firms that have several licenses can spread the fixed costs of developing and acquir-
ing technology as well as building a customer base.” Consumers of spectrum services
also benefit from aggregation. For example, consumers value the ability to use the
same cellular phone anywhere in Canada. This can only be achieved if the cellular
phone provider can aggregate enough licenses to cover the whole country.”‘ An ideal
auction design, therefore, should be flexible enough to allow the bidders to construct
their preferred license aggregations.
An ideal auction system would also minimize the so-called “winner’s curse’. This
concept characterizes the highest bidder as the one who most over-estimates the value
“o See discussion at Part IT.C.2, above. See also “Selling Spectrum Rights”, ibid at 150.
,04 Ibid. at 151.
“‘Ibid.
MCGILL LAW JOURNAL / REVUE DE DROITDE MCGILL
[Vol. 44
of the item for sale, and finds that it has paid more for the item than it is worth.'” Bid-
ders who experience winner’s curse are more likely to have second thoughts about
their winning bids and to default on them. Winner’s curse has a close causal connec-
tion to bidder’s default-and the two should therefore be analyzed together.
Auction designers must also ask whether their design minimizes the potential for
collusion among the bidders. Collusion occurs when several bidders unite to manipu-
late the results of the auction. While such collusion is prohibited under Canada’s
Competition Act,’7 this does not mean that it will not occur.
In sum, the level of revenue generated by the auction, the possibility of aggregat-
ing auctioned items, the likelihood of winner’s curse, the probability of default on
winning bids and the likelihood of collusion among the bidders should all be taken
into consideration when designing an auction system for Canada. The next section in-
volves an analysis of the extent to which. various methods promote these goals.
2. Oral ascending (English) auction
In an oral ascending auction, bidding is open and escalates until one party re-
mains. In this manner, individual licenses are sold off one by one in a series of succes-
sive auctions. This is called a “sequential auction”. Bidders get feedback from their
competitors’ bids about how their estimates of the license’s true value compares with
that of their rivals. Participants, therefore, are more comfortable submitting higher
bids because their fear of experiencing winner’s curse is eased by the continuing in-
terest of other market participants. However, this market feedback can be a false
safety blanket, sometimes causing the winner to bid too high in the exhilaration of the
moment. When the FCC used this method in the July 1994 IVDS auction, several
winners experienced winner’s curse and later defaulted on their bids.”
High prices and subsequent government revenue are the usual results of such a
system. The sequential system impedes aggregation by requiring a bidder to win each
license independently. Increased costs and delays can result as participants seek to
construct aggregations through secondary market transactions.'” License aggregation
can also be thwarted by predatory bidding. This occurs when auction participants’
bids are not meant to reflect thd true market value of the spectrum, but rather to hinder
the progress of their competitors. A bidder may try to drive the prices up to excessive
levels for the early licenses in order to prevent the winners from competing for the
later licenses-thereby further decreasing the chances of aggregation.
” See generally J. McMillan, “Why Auction the Spectrum?” (1995) 19 Telecomm. Pol’y 191 at
199. See also “Selling Spectrum Rights”, ibid. at 153.
o7 R.S.C. 1985 (2nd Supp.), c. 19.
“00 Kummel, supra note 68 at’530.
‘0’ Federal Communications Commission, “In the Matter of Implementation of Section 309G) of the
Communications Act-Competitive Bidding” PP Docket No. 93-253, (1993) 8 F.C.C. Rcd. 7635 at
para. 35, online: WL (FCOM-FCC).
1999]
M.C. GLASS AND D.M. RHODES – SPECTRUM AUCTIONS
The risk of collusion under the oral ascending auction system is high. The identity
of each bidder is known to all others, making it easier for participants to communicate
with one another and reach collusion agreements. Through their bidding patterns,
participants can signal to others what they consider to be “their territory”, and what
they consider to be “other’s territory”. In sum, revenue under this model is high and
the likelihood of winner’s curse is increased. This model, however, does not provide
for the efficient aggregation of licenses, and participants find it relatively easy to col-
lude with one another.
3. Oral descending (Dutch) auctions
Under the oral descending auction system, the auctioneer-by open outcry-
starts the price high and descends until one party bids. The bidder automatically wins
the license.’ Since spectrum licenses are auctioned off sequentially under this model,
efficient aggregation is not provided for. The winning bidder does not have the benefit
of feedback from the market about the value of the auctioned license. As such, firms
tend to bid low in an effort to avoid winner’s curse. Revenue for government also
tends to be lower under this method than under the oral ascending system. Collusion
among the bidders, moreover, is made very difficult since the first price expression by
a bidder ultimately ends the auction.
4. First-price sealed bid auction
In a sealed bid auction, bidding occurs secretly and escalates through multiple
rounds until one party-the highest bidder-remains. Anonymous bids prevent bid-
ders from knowing the amount and origin of the bid. Since participants are ignorant of
the bid amounts pledged by their opponents, they tend to bid low to avoid winner’s
curse. This conforms to the general theory that government can increase its revenue
by publicizing any available information that affects the licenses’ assessed value.”‘
Due to the bidders’ fear of winner’s curse, bids and revenue tend to be relatively low.
The likelihood of winner’s curse, however, is also rather low. Aggregation is not pro-
vided for because the licenses are auctioned off one by one. Only through good luck
can a firm win all the licenses it needs for its business plan; bad luck can either mean
that a firm wins more licenses than it needs, or that it wins a collection it cannot use
with any plausible business plan.”2 The sealed bid system does, however, reduce the
likelihood of collusion between bidders because, given the anonymous nature of the
auction, it is very easy for a party to renege on a collusion agreement.
“o This system is habitually used in Canada to auction tobacco.
“Selling Spectrum Rights”, supra note 43 at 153.
“Auction Consultation Review”, supra note 15 at 27.
A2
MCGILL LAW JOURNAL / REVUE DE DROIT DE MCGILL
[Vol. 44
5. Second-price sealed bid auction
Under this method, bidding is secret and escalates through multiple rounds until
one party is left, but the winning party only pays the price of the second-highest bid.
Government revenue clearly decreases under this model. This was dramatically illus-
trated in New Zealand after they used this method to auction spectrum licenses in
1990. In one case, a firm that had bid $NZ 100,000 only paid the second highest bid
of $NZ 6.” In another auction, the high bid was $NZ 7 million and the second highest
bid $NZ 5,000. An Otago University student bid $NZ 1 for a television frequency that
serviced a small city. As the only bidder, he got the license for free. Revenues fell far
short of expectations-having only received $NZ 36 million as opposed to the pre-
dicted $NZ 240 million. These developments led to harsh criticism of the govern-
ment’s policy, and New Zealand promptly switched to first-priced sealed bid auctions.
Second-priced sealed bids, however, do alleviate the problem of winner’s curse.
Since winning bidders only pay the second highest bid, they are less likely to feel that
they are paying too much for the frequency. The risk of default is reduced commensu-
rably. Aggregation remains difficult because the frequencies are still being auctioned
individually. Due to the anonymous nature of the bidding, the likelihood of collusion
between bidders also remains rather low.
6. Sequential English auction with sealed-bid package bidding
This model is a variation on the sealed-bid methods. Parties are allowed to submit
a sealed bid for any combination of licenses they desire. Oral ascending bids are then
taken for the individual licenses one after the other. When the oral bidding is com-
pleted, the auctioneer compares the winning oral bids to any package bids and awards
the license to a package bid where it proves greater than the sum of the individual oral
bids.”‘ At the conclusion of the auction, all bids are made public–enhancing the
transparency of the process. This model is expressly designed to provide for aggrega-
tion. Because an aggregation of licenses across various geographic locations is often
worth more than the value of the individual licenses on their own, bidders are willing
to pay a premium to obtain the aggregation. Bidders are ble to purchase the aggrega-
tion they, desire rather than having to purchase a block pre-selected by the govern-
ment. This flexibility leads to increased government revenue. Two factors greatly re-
duce the likelihood of collusion: (i) the sealed bid nature of the auction, and (ii) the
number of bidders involved in the auctioning of multiple licenses probably would be
greater. The likelihood of winner’s curse and consequent default on the bid is quite
low because, lacking full price expression from other parties at the moment the bid is
made, firms are likely to bid conservatively to avoid an overbid.
While this system increases government revenue by allowing for aggregation, the
premium received could be greater if the package bidders had the benefit of full price
“Selling Spectrum Rights”, supra note 43 at 149.
” Kummel, supra note 68 at 531.
1999]
M.C. GLASS AND D.M. RHODES – SPECTRUM AUCTIONS
expression from other market participants, and were thus reassured that they would
not fall victim to winner’s curse. To allow for this greater premium, the simultaneous
multiple round auction was designed.
7. Simultaneous multiple round auction
In the simultaneous multiple round auction, multiple licenses are open for bidding
at the same time, and bidders have the opportunity to bid on as many licenses as they
desire in successive discrete bidding rounds. Bidding remains open as long as there is
some bidding on any of the licenses. Bidding occurs over rounds with the results of
each round announced to the bidders before the start of the next round. This auction is
best run by a computer capable of processing on-line bids.”‘ The two hallmarks of this
system-the simultaneous bids and the multiple round (ascending) bids-aid effi-
ciency. Multiple round bids allow participants to witness the full price expression of
their competitors and adjust their bids accordingly. This information increases the
likelihood that licenses will be assigned to bidders that value them most. Since bid-
ders are reassured by the market, their fear of winner’s curse diminishes, and they are
-willing to bid more for the license. Since winner’s curse diminishes, so too does the
probability that winning bidders default on their bids. As a result, revenue to the gov-
ernment increases.
A simultaneous auction with multiple rounds of bidding gives bidders full flexi-
bility in building license aggregations, as well as the ability to switch to their back-up
aggregations should their first choice aggregations prove to be too expensive. Since
the auction induces the bidders to express their ideas about desirable license aggrega-
tions, the market process determines the outcome. Therefore, the allocating agency
need not know how the available licenses complement or substitute for each other.”‘
The symmetry of the system prevents the occurrence of certain kinds of predatory
bidding. In a sequential system, a bidder may try to drive up prices to excessive levels
for the early licenses so that winners will be unable to compete for later licenses. This
strategy would fail in a simultaneous system because bidders can vary their bids on
any license at any time. Anonymous bidding decreases the likelihood of collusion be-
cause any bidder can secretly alter its agreement to collude. The potential for aggre-
gation and the level of revenue for the government increase, while the potential for
winner’s curse diminishes.
Based on the table below, one can clearly see that simultaneous multiple round
auctions are clearly the best way to auction spectrum licenses. They provide for li-
cense aggregation, maximize revenue for the government, and greatly decrease the
probability of winner’s curse and collusion. If Canada decides to adopt the auction
“5 For a discussion of this auction method, see generally J. McMillan & P. McAfee, “Analyzing the
Airwaves Auction” (1996) 10 J. Econ. Persp. 159. See also “Auction Consultation Review”, supra
note 15 at 26.
“6 McMillan & McAfee, ibid at 173.
MCGILL LAW JOURNAL / REVUE DE DROIT DE MCGILL
[Vol. 44
method of spectrum allocation, a simultaneous multiple round system is recom-
mended.
D. Conclusion
COMPARATIVE ADVANTAGES OF DIFFERENT TYPES OF AUCTIONS
Oral
Ascending
Oral
Descending
First-price
Second-
price
Sequential
Sealed-Bid
Simultaneous
Multiple
Round
(I) Revenue
High
Medium
Medium
Low
Very High
Highest
(2) Aggregation of
Auctioned Items
Not Provided
for
Not Provided
for
No
No
Yes
High
High
Low
Low
Low
Yes
Low
(3) Likelihood of
Winner’s Curse
(and probability of
default)
(4) Likelihood of
Collusion Among
Bidders
High
High
Low
Low
Low
Low
Figure 2: The strengths of the simultaneous multiple round auction system as compared to other auction
methodologies is clear when the features of all these systems are compared.
Auctions can be tailored to meet certain public policy goals not otherwise pro-
vided for by a pure auction method. Spectrum can be allocated directly to essential
public services, thereby ensuring their continued access to the spectrum. Bidding
credits can also be granted to certain groups to help them win spectrum licenses.
Furthermore, auction participants can be screened to ensure that they all have de
ininimus technical and financial capability to operate a wireless service. Finally, li-
censes can be granted on the condition that licensees meet certain conditions. These
innovations alleviate the defects inherent in the pure auction system, and should be
adopted to compose the appropriate hybrid auction method for allocating spectrum.
With the above assertions in mind, our final objective is to propose a comprehensive
method of spectrum allocation for Canada.
V. Proposed Method of Spectrum Allocation in Canada
As demonstrated, comparative analysis and auctions appear to be the most desir-
able methods of spectrum allocation. The comparative process allows for the promo-
tion of social goals. However, it is ill-suited in promoting efficiency goals because the
allocating agency cannot absorb and process enough information to make licensing
decisions accurately. As for the pure auction model, it cannot effectively promote so-
cial goals which cannot be expressed in efficiency terms. On the other hand, auctions
are ideal for promoting efficiency goals because they can “parallel process” informa-
tion gathered by millions of consumers and businesses about the relative desirability
and efficiency of a spectrum-related service. Because of the different strengths and
1999]
M.C. GLASS AND D.M. RHODES – SPECTRUM AUCTIONS
weaknesses of these two methods, synergies can be achieved if the two are merged
correctly. This section outlines a proposal for a three-stage hybrid auction method of
spectrum allocation. In harnessing the advantages of the existing comparative process
as well as the benefits of an auction system, Canada’s spectrum allocation goals
would be achieved.
The first stage of this hybrid process would duplicate the existing expression of
interest procedure used by the SMB. At the second stage, all applicants would be
screened through an assessment process designed to select only those applicants who
meet the criteria which administrators deem to be essential. Those applicants who also
satisfy certain secondary criteria would be assigned bidding credits for use in the third
stage of the process. The third stage would involve a simultaneous multiple round
auction conducted among those applicants who pass the screening stage.
A. Stage 1: Expression of Interest
At the beginning of the process, the SMB must determine what spectrum is avail-
able for allocation and divide it into suitable licensing units. A suitable licensing unit
is the smallest band of contiguous frequencies that can be put to practical use given
the current state of technology (i.e., the smallest band that can be used to transmit in-
telligible information). To ensure continued spectrum access by public service provid-
ers, certain spectrum bands would be reserved for those groups and allocated by non-
market methods. For those bands not reserved for public services, the SMB would
publish a notice defining the available blocks-and invite all parties interested in par-
ticipating in an auction of these bands to express their interest. After all expressions of
interest have been received, the names of these parties would be published-thereby
facilitating mergers and joint ventures among the parties. If there was sufficient spec-
trum available for all parties after these groupings have been formed, then the avail-
able spectrum licenses would be allocated among them on a first-come/first-served
basis. If there was insufficient spectrum for all interested parties, then the SMB would
request that all parties file a business/service plan describing their proposed use of the
spectrum.”‘ These plans are needed to demonstrate to the SMB that the applicants
meet the conditions of the screening stage.
17 Note that in the FCC spectrum auctions, bidders are not free to provide any services they choose
with their licenses. Instead, they must provide the service which the FCC has determined to be best
suited to the spectrum in question. The authors are of the view that the market should play a greater
role in deciding which services are provided over which spectrum bands. Thus, it is proposed that ap-
plicants be allowed to choose the services they wish to provide. The SMB’s division of available
spectrum into licensing units would not determine the services a bidder would have to provide be-
cause the units would be small enough that a bidder would never be forced to purchase more than it
needed. Further, as explained below, it would be open to a purchaser to acquire licensing units from
other owners or to sell any units it did not need.
MCGILL LAW JOURNAL / REVUE DE DROIT DE MCGILL
[Vol. 44
B. Stage 2: Screening Stage
The purpose of the screening stage is to ensure that all participants who advance
to the third stage have de minimus ability to: (i) operate a feasible spectrum-based
service, and (ii) fulfill the essential public interest goals of Canadian telecommunica-
tion policy. This means that the screening process would not be comparative in the
sense of seeking to select the best firm, but rather would focus on ensuring that the
business/service plan proposed by each applicant would-at a minimum-fulfill these
two requirements.
The SMB would use a two-tier screening approach that would distinguish those
criteria which are essential in a candidate from those which are merely desirable. At
the first tier, each applicant would have to demonstrate, to the satisfaction of the SMB,
that it is capable of meeting all of the following criteria:
(1) Its proposed service will not interfere with existing spectrum services. When
two spectrum users try to use portions of the spectrum which are too close to
each other, the resulting interference will prevent either one from communi-
cating over their respective frequencies. Interference can also be generated if
broadcasting equipment is improperly located or operated.
(2) It has never violated any SMB auction rules.
(3) It has the basic technical competence to operate the proposed service.
(4) Its business/service plan is feasible.
(5) It has (or will have) sufficient capital to provide the proposed service. This
helps prevent speculation in spectrum licenses.
While it is recognized that other goals could have been included as essential require-
ments for acquiring a spectrum license, those other criteria are best handled through
the provision of bidding credits at the auction stage. Too many criteria in the first tier
of screening would make the process slower and more discretionary, perpetuating the
very weaknesses of the comparative process. Therefore, once the applicant has dem-
onstrated that it meets all of the first tier criteria, it should advance to the auction stage
of the allocation process.”‘
An applicant, however, could choose to demonstrate to the SMB that its busi-
ness/service plan would promote other goals of Canadian telecommunications policy
which would entitle it to receive bidding credits. At this second tier of the screening
process, the applicant would have to demonstrate any one or more of the following:
“‘ The SMB would provide written reasons for any refusals. These would enable the unsuccessful
applicant to modify its proposed business/service plan and resubmit it before the deadline for admis-
sion to the auction stage.
1999]
M.C. GLASS AND D.M. RHODES – SPECTRUM AUCTIONS
(1) It is an entity whose participation in the provision of spectrum services ought
to be encouraged (it increases diversity of service providers, i.e., native
groups, students, francophones, etc.).
(2) It is a small business (less than the stipulated level of annual gross revenues).
(3) It would provide universality of service (offering services to a stipulated pro-
portion of the persons within the area covered by its licenses).
(4) It would provide service to remote areas of the country.
(5) It will strengthen Canadian content and culture.
Satisfaction of one or more of these criteria would give the applicant the right to exer-
cise a bidding credit in the auction stage of the allocation process. The calculation of
bidding credits should be weighted according to the following recommendations:
(1) If the applicant is an entity whose participation in the provision of spectrum
services should, according to the SMB, be encouraged: 20%.
(2) If the applicant is a small business, based on a sliding scale depending on the
firm’s revenues: 5% to 15%.
(3) If the applicant would provide universality of service: 10%. Universality of
service is generally a licence condition for broadcasters and local service
telephone companies. Universality should not be a condition of a spectrum
license because it is not an achievable goal for many spectrum services.
(4) If the applicant would provide service to remote areas of the country: 10%.
This should not be imposed as a license condition because many licenses
would cover both remote and non-remote areas of the country. Service pro-
viders holding those licenses might find that there is simply no demand for
their services in the remote areas covered by the license.
(5) If the applicant would strengthen Canadian content and culture: 5% to 20%.
The maximum cumulative bidding credit to which any applicant would be entitled
should be limited to 50%. This would be the best way to incorporate legitimate social
policy concerns into the process of spectrum allocation while maintaining the integ-
rity of the free market auction process. Many of the selection criteria used by the
SMB in the existing comparative analysis process do not need to be taken into ac-
count at the second stage because an auction is already designed to reflect these goals.
These goals include: (i) the speed of network deployment, (ii) efficient use of the
spectrum, (iii) suitability of spectrum bands to the proposed services, (iv) revenue for
government, (v) enhanced competition and customer choice, and (vi) secondary bene-
fit to the economy.
186
MCGILL LAW JOURNAL / REVUE DE DROIT DE MCGILL
[Vol. 44
C. Stage 3: Auction
The auction methodology proposed here would be a standard simultaneous mul-
tiple round auction, similar in principle to the system used by the FCC.”‘ This is the
best form of auction because it allows for full price expression by all parties, thereby
reducing the probability of winner’s curse. If bidders’ fears of winner’s curse are alle-
viated, higher bids are made, thus leading to greater revenue for the government. The
proposed auction provides for license aggregation by allowing bidders to bid on many
licenses at the same time. Due to the larger number of participants and licenses being
auctioned at the same time, collusion is difficult.’20
Parties who earn bidding credits would make bids like any other bidder. If they
win, the amount of their winning bids would be discounted by the amount of the bid-
ding credits. For example, if a firm has qualified for a 20% bidding credit because it is
operated by natives, and has also qualified for a 10% bidding credit because it pro-
poses to provide services to remote areas of the country, then it would only have to
pay the SMB 70% of its winning bid. Thus, if it had bid $100,000, it would defeat a
non-qualifying firm which had bid $90,000, and it only would be required to pay
$70,000 to the SMB.
Bids would start at a minimum amount set by the SMB for each license. This
amount would represent a reserve price below which the SMB would not allocate the
license. In setting the level of the reserve price, there are two competing influences.
High reserve prices would prevent the spectrum from being allocated to a bidder who
is willing to provide services, but who places less value on the license than anticipated
by the SMB. This is an inefficient outcome because it means that valuable spectrum is
not being used. Low reserve prices contribute to long, drawn-out auctions since bid-
ding can start at levels far below the actual value of a license. These delays in allocat-
ing a license lead to lost revenue and reduced economic growth. Reserve prices would
not be set at very high levels because efficient allocation is more important than gov-
emiment revenue.'”‘ The current practice is to employ a declining reserve price. This
reserve price is initially set at a high level and if there are no bids, the auction author-
ity can reduce it until bids are made. This overcomes the difficulty of establishing a
fixed reserve price.
Bidders would be able to withdraw any bids which they later wanted to change. In
order to ensure that only serious bids were tendered, a bid withdrawal penalty would
be imposed. When registering for the auction, bidders would be required to tender an
up-front security deposit. This deposit would be used to pay any withdrawal penalties.
This would prevent people from making bids which did not represent their estimated
value of the spectrum. It would also prevent a repeat of the events in Australia where
“” See discussions at Parts 1.B.3 and 1V.C.7, above.
0 See discussion at Part IV.C.7, above.
‘,
,’ “Auction Consultation Review”, supra note 15 at 29.
1999]
M.C. GLASS AND D.M. RHODES – SPECTRUM AUCTIONS
bidders made excessively high bids with no intention of ever honoring them.'” The
penalty would correspond to the potential loss in revenue caused by the withdrawal of
the bid. If the license for which a bid has been withdrawn ends up selling for more
than the withdrawn bid, no penalty would be charged to the bidder. If the license ulti-
mately sells for less than the withdrawn bid, the penalty would be the difference be-
tween the withdrawn bid and the eventual final selling price. The withdrawal penalty
would be due as a lump sum payment shortly after the close of the auction.'”
Unlike the FCC auctions, bidders would not be required to pay their winning bids
as lump sums at the conclusion of the auction. Instead, they would have to submit an
up front payment of approximately 25% of the amount of their bids in order to receive
their licenses. If the winning bidder fails to make the up front payment, then the li-
cense would go to the next highest bidder. If the winning bidder makes the up front
payment, then the remaining 75% of the bid would have to be paid in biannual in-
stallments over the first half of the license term. Interest on these payments would be
computed at the market rate. This payment scheme represents a compromise between
the need to avoid speculation and the desire to avoid draining firms of much needed
capital. It also ensures that winning bidders will have enough capital left over after
acquiring their licenses to effectively develop their wireless service networks while
still requiring a large enough initial investment to deter speculators. To illustrate, a
winning bidder who bid $1 million for a ten year license would pay $250,000 after
the auction, and then make biannual payments of $75,000 (plus interest) for the next
five years. This balance of payments would be modified to account for special factors
arising from different types of licenses or different categories of winning bidders.”
As illustrated above, the auction process ensures rapid network deployment and
efficient use of the spectrum because bidders will be eager to recoup the cost of their
bids by quickly maximizing the return on their investment. This means that they will
have a strong incentive to provide the services which customers demand the most.
.22 See discussion at Part II.C.2, above.
‘ “Auction Consultation Review”, supra note 15 at 31.
2 For example, the FCC set up special installment plans for small businesses which varied the du-
ration of payments and the rate of interest according to the annual gross revenues of the business: see
“FCC Report”, supra note 32 at s. 5(e). The proceeds of spectrum auctions could be designated for
some special purpose, or they could simply go to the general treasury. The danger of allocating pro-
ceeds to some specific task, such as balancing the budget, is that it can induce spectrum managers to
design the auction process in order to meet earnings targets rather than to achieve efficient spectrum
allocation. However, if the auction process is kept separate from the administration of the resulting
proceeds, then it would be possible to use auction proceeds for a specific goal. For the purposes of
determining how the spectrum should be allocated, it does not matter which goals the auction pro-
ceeds are used to promote. However, it is important that the proceeds are not used to undermine effi-
cient allocation by subsidizing auction participants or spectrum-based services. Either of these things
will result in distortions in demand which could undenine the auction process. Those goals should
be promoted exclusively through the bidding credit system in the manner described. If greater incen-
tives are desired than the credit amounts proposed here, they should take the form of larger bidding
credits.
MCGILL LAW JOURNAL / REVUE DE DROIT DE MCGILL
[Vol. 44
The result will be greater customer choice. The revenue generated from a given spec-
trum band will be greatest if it is used for the service which is most suitable for its in-
formation-carrying properties. Therefore, the market will select those services most
suitable to each spectrum band. Since the auction mechanism ensures rapid network
deployment, it will result in more jobs and quicker economic growth-benefiting all
Canadians indirectly.
Despite the advantages of the auction process, it is still subject to market failures
that may lead to undue concentrations of spectrum licenses in the hands of a few
firms. To avoid this problem, it is proposed that caps be instituted on the total amount
of spectrum which any person or group of persons can hold. Every country which has
experimented with market-based methods of spectrum allocation has instituted such
measures. A system of spectrum caps would have to be established before any auc-
tions were conducted. Competition law authorities, however, should be encouraged to
provide their views about the implications of this system before it is implemented.
D. Post-Allocation Stage
After winning a spectrum license, a bidder would have to pay the 25% up front
payment in order to have the license allocated. Failure to pay within the stipulated
time-usually thirty days after the auction-would lead to forfeiture and a re-
auctioning of the license. Once the winning bidder has actually received its license, it
would have to honor the business/service plan submitted to the SMB at the screening
stage of the process. This means that it would meet the first tier criteria by providing
the promised services pursuant to the conditions stated in its business/service plan. It
would also comply with the second tier criteria by continuing to fulfill the social pol-
icy goals it had promised to promote. A licensee which qualified for bidding credits
would also have to maintain its qualifying status. For example, a licensee which quali-
fied for a 20% bidding credit because it is native-operated must continue to be oper-
ated by natives. Small businesses would not be required to maintain their annual gross
revenues below the stipulated level, but they could not be acquired by a non-
qualifying business.
If a licensee fails to fulfill any first tier criteria, it would forfeit its license to the
SMB. If a qualifying applicant fails to fulfill any applicable second tier criteria, the
penalty would be less severe. The violator would have to pay the SMB an amount
equivalent to the bidding discount to which it would be no longer entitled (plus inter-
est). A violating licensee, moreover, would be subject to a further penalty equal to
50% of this amount. This would dissuade firms from fraudulently claiming to be
qualifying entities. If this penalty were not implemented, firms would seek to qualify
for bidding credits in order to unfairly defer a portion of the license costs during the
capital intensive period when they were constructing their wireless networks.
E. Transfer and Change in Use of a License
After having received a license, a licensee might desire to provide services other
than those promised in its business/service plan. This might occur because improve-
1999]
M.C. GLASS AND D.M. RHODES – SPECTRUM AUCTIONS
ments in technology or changes in the demand for wireless services would make other
uses more efficient than those the licensee originally wished to provide. It is very im-
portant that these changes be allowed in order to permit licensees to respond to market
forces and shifts in consumer demand. However, the policy goals set out in the tiers of
selection criteria must still be met.
It is proposed that if a licensee wants to make significant modifications to its
business/service plan, it should have to submit a new plan to satisfy the SMB that all
the first tier criteria would still be met. If the new plan did not fulfill the first tier crite-
ria, the licensee would not be able to make these changes. However, the licensee
would always have the option of submitting another business/service plan for evalua-
tion. If the licensee qualified for bidding credits in the auction, then it should satisfy
the SMB that the new business/service plan also meets the same criteria as the origi-
nal one submitted. Should the new plan not meet one or more second tier criteria, the
licensee would have the option of withdrawing the new plan or paying an amount
equal to the bidding discount (plus interest), as well as a 50% penalty. If a licensee-
which did not qualify for one or more bidding credits in the auction-were to submit
a new business/service plan which satisfied one or more second tier criteria, it would
be entitled to a discount equal to the amount of the associated bidding credit. The
same system would apply to the transfer of a license by the winning bidder. In that
case, the transferee would have to submit a business/service plan in the same manner
as a licensee who wanted to modify its original plan. The transferee would have to
meet all of the first tier criteria and would have to satisfy any applicable second tier
criteria if it wished to avoid reimbursing the SMEB for the bidding discount plus a 50%
penalty.
In order to account for future contingencies and avoid imposing undue restric-
tions on licensees, the 50% penalty should apply to violations of second tier criteria
which occur only within the first two-thirds of the license’s duration. After that point,
a licensee in violation of those criteria would have to reimburse only the bidding dis-
count.
F Divisibility of Spectrum Rights
For the same reasons that licensees should be permitted to transfer their licenses
or change the services they provide, they also should be able to divide their licenses
and transfer portions to others. This means that licenses should be divisible both ac-
cording to bandwidth and according to area covered. There are several advantages to
doing this. First, it would remove the barriers faced by many small businesses in get-
ting access to spectrum licenses, and thereby encourage competition. Second, it would
allow winning bidders to keep only those portions of their spectrum licenses which
they actually need in order to provide their desired services. This would enhance
customer choice by allowing more services to be provided with the same spectrum li-
censes. The same requirements for SMB screening would also apply to the division
and transfer of a fraction of a license as it would to the transfer of an entire license.
With respect to the services it proposed, the receiver of a fraction of a license would
still have to comply with all first tier criteria. If a fractional transferee did not comply
MCGILL LAW JOURNAL / REVUE DE DROIT DE MCGILL
[Vol. 44
with the second tier criteria under which the license was granted, then a portion of the
bidding discount (and interest) equivalent to the portion of the license used by the
transferee would have to be refunded. A 50% penalty would also be assessed on the
amount of the refund. The refund should be paid by the winning bidder since it would
be in the best position to insist that the fractional transferee either comply with the
second tier conditions or pay the cost of not complying. The amount of the refund
could be calculated according to the following formula:
A = Bandwith of the transferred fraction.
B = Bandwith of the license as a whole.
A XC XE
B
D
C = Population of the geographic area covered by the transferred fraction.
D = Population of the geographic area covered by the license as a whole.
With respect to refunds involving provision of service to remote areas, the
reference to population in variables C and D would have to be replaced with a
reference to the areas actually served.
E = Price paid for the whole license at auction.
Figure 3: Formula for calculating the reimbursement of a bidding discount when a license is divided.
Any instalments of the winning bid due after transferring a fraction of a license would
be divided between the original licensee and the fractional transferee in proportion to
the respective population/area and bandwidth they have allocated between them-
selves. The duration of the license would not be affected by the transfer of a fraction
to another person. The fractional transferee’s portion of the license would have the
same duration as that remaining on the original license at the time of division and
transfer.'”
G. Duration of License and the Possibility of Renewal
Two important considerations must be balanced in determining the appropriate
duration of a license. While a longer duration provides the licensee with increased
certainty when determining how much to invest in the construction of its wireless
network, it makes it more difficult to reallocate spectrum to new uses as wireless
technology evolves. A shorter license duration coupled with uncertain renewal dis-
courages the licensee from making large investments which may be needed to exploit
the spectrum license. Given the speed of technological change, however, shorter du-
ration does make spectrum reallocation much easier.
Different countries have come to different conclusions about the best balance
between these concerns. Australia has proposed an extension to the maximum dura-
“‘ In the event of a division of a license, the holders of the fractions would have to provide the SMB
with detailed information specifying the bandwidth and area held by each party.
1999]
M.C. GLASS AND D.M. RHODES – SPECTRUM AUCTIONS
tion of a spectrum license from the current ten years to fifteen years. At the end of this
period, the license would be re-auctioned.’26 In New Zealand, legislation currently
permits license duration of up to twenty years, but the government is considering in-
creasing this to a quasi-perpetual duration.’27 The United Kingdom government has
also noted the need for some security of tenure when licenses are auctioned.’2’ In the
United States, the FCC has auctioned licenses with relatively short ten year terms, but
these licenses also feature a “significant expectancy of renewal” which for most pur-
poses amounts to perpetual licenses. Non-renewal is only likely if the licensee
breaches fundamental license conditions or deceives the FCC.'”‘
Instead of providing shorter terms with a high probability of renewal, licenses
should have longer durations-perhaps fifteen years-and be re-auctioned two years
prior to expiration. This duration provides the proper balance between the legitimate
concerns of incumbent licensees and the need to allow other market participants to
gain access to spectrum licenses. Such an approach would also result in greater un-
certainty for incumbent licensees than the “significant expectancy of renewal” pro-
vided by the FCC. However, advance re-auctioning would provide incumbent licen-
sees with foreknowledge of whether or not they would receive another license term.
Due to their established wireless networks and client base, incumbent licensees would
also presumably have a significant advantage in the re-auction. Therefore, if they have
been using the spectrum in an efficient manner, they should be able to consistently
outbid the competition and retain their licenses. Alternatively, if improvements in
technology or changes in demand for wireless services render the incumbent’s tech-
nology obsolete or its services redundant, the market would ensure that the license is
transferred to more efficient uses. As a general rule, a license which has been divided
into two or more pieces would be re-auctioned as separate pieces. The SMB would
have the discretion to recombine the separate pieces of a license at the re-auction
stage. This has the benefit of allowing market forces to determine how spectrum
rights should be divided.
The downside of this approach is the potential for predatory bidding. This occurs
when auction participants’ bids are not meant to reflect the true market value of the
spectrum, but rather to impede the progress of their competitors.”‘ On re-auction, an
initial licensee may be targeted by an investor who hopes to win the spectrum license
… Australian Communications Authority, “Spectrum Licence Allocation: 800 MHz and 1.8 GHz
Bands-Applicant Information Package” (6 February 1998) at 19, online: Australian Communica-
tions Authority
127 W. Wedderspoon, “Radio Spectrum Management: The New Zealand Approach” (Speech, Na-
tional Radio Spectrum Policy, Communications Division, New Zealand Ministry of Commerce, Oc-
tober 1995), online: Radio Spectrum Management
(date accessed: 14 February 1999).
.. UK Report, supra note 53 at paras. 8.7-8.9.
“‘”uction Consultation Review”, supra note 15 at 17.
30 See discussion at Part IVC.2, above.
MCGILL LAW JOURNAL / REVUE DE DROITDE MCGILL
[Vol. 44
and then acquire the incumbent licensee’s network at a low price.’3′ Alternatively, par-
ticipants may bid solely for the purpose of breaking-up an incumbent’s license aggre-
gation.’ To alleviate the problem of predatory bidding in license re-auctions, it is
proposed that incumbent licensees be granted a bidding credit of 5 to 20%, depending
on their annual gross revenues. In this manner, a proper balance between the incum-
bent’s need for certainty in renewal and society’s need for maximum allocative effi-
ciency would be ensured.
Conclusion
Canada’s movement toward a knowledge-based economy means that a major
portion of its future wealth will rest in ideas. Spectrum technologies, in effect, will
become the main method by which we exchange our ideas. It follows that the quality
of our spectrum-based communications services is important to the evolution of the
whole economy. The transition to a knowledge-based economy, moreover, will be im-
peded if spectrum-based services are implemented inefficiently. Since the develop-
ment of spectrum services takes many years, ensuring high quality services in the
coming decades requires optiumum allocation of spectrum licenses today.
Spectrum allocation in Canada is now at an important crossroad. It has been dem-
onstrated that auctions are an attractive course to take, but they must be carefully de-
signed in order to ensure their success. Only when an auction takes into account all of
Canada’s policy goals will it be a suitable method of allocation for this country. A
pure auction system that focuses on efficiency concerns fails to account for policy
goals. Canada must therefore look for ways of incorporating these policy goals into its
auction system.
A hybrid auction model is proposed which draws upon the experience of other
countries. It represents a compromise between the comparative analysis method
which Canada now uses and the FCC auction system operational in the United States.
It is designed to address the two key considerations for a successful auction system:
market efficiency and social policy goals. To ensure efficient allocation of the spec-
trum resource, the proposed hybrid auction model would employ a simultaneous
multiple round bidding system providing for license aggregation, reduction in the
probability of winner’s curse, maximization of government revenue and limitation of
the potential for collusion. To maintain Canada’s cultural identity, it is recommended
that a screening process be adopted to incorporate essential policy goals. In addition, a
system of bidding credits should be implemented reflecting secondary policy goals
that, while important, are not considered to be fundamental.
“‘ “Auction Consultation Review”, supra note 15 at 18.
1’ For example, an economically efficient assignment of spectrum licenses might involve one licen-
see acquiring a block of similar frequencies covering Vancouver Island. A competitor may seek to
damage this licensee by bidding very highly in a re-auction for just one key license, such as Victoria,
and thus significantly disrupt the business of the incumbent licensee: see ibid. at 18.
1999]
M.C. GLASS AND D.M. RHODES – SPECTRUM AUCTIONS
193
This system is designed to ensure a fair and efficient allocation of spectrum li-
censes. It balances efficiency concerns and cultural objectives in a way that allows
Canada to advance into the knowledge-based economy with confidence.
