Study of Market-based Exclusive Spectrum Rights

5. Analysis

5.1 The Rationale for Market-based Exclusive Spectrum Rights

The traditional administrative system of spectrum management can be characterized as an island of administrative allocation of an input into production processes in the much larger ocean of the use of markets processes — at least in market economies. To that degree, it is an anomaly.

This observation does not by itself:

  • imply that administrative methods have led to poor results, or
  • mean that they should be replaced by market methods, or
  • suggest that market methods always work well.

It does, however, raise a number of questions which this chapter will seek to answer, such as:

  • can spectrum be managed by market methods?
  • what conditions are necessary for spectrum markets to operate effectively?
  • should they extend to:
    • select commercial bands?
    • all commercial bands?
    • public sector use, or more generally, use in the public interest, as well?
  • can interference be successfully avoided with a system of marketed spectrum?
  • while Canada has introduced some limited market methods, how should the process be managed in Canada?

The first three issues are addressed in 5.2; the third in 5.3, the fourth in 5.4.2, and the fifth is embarked upon in 5.3.2.

top of page

5.2 The Potential Benefits of Market Methods

5.2.1 Markets vs. Hierarchies

Markets permit the free exchange of labour and land, goods and services, and physical financial and intellectual assets among the households, firms and public organizations in the economy. Because transactions are based on free exchange, markets do not — as administrative methods of allocation do — entail hierarchies with 'deciders'and 'implementers' (although firms operating in a market economy do exhibit such hierarchies internally).Footnote 96 It is important to note against the background of some of spectrum reforms discussed above that a spectrum market entails free exchange and flexibility of use of spectrum — in other words, secondary trading without restrictions on use. Other changes, such as auctions or administered prices, do not qualify because they do not break out of the straitjacket of the regulator having to fundamentally determine the use of the spectrum.

A useful and standard way of evaluating market and other methods of allocating resources is to ask:

  • Do they promote economy in production? (In the case of spectrum, this would involve asking if spectrum were wasted by some users.);
  • Do they promote the production of the right set of goods and services? (For example, is spectrum available to produce the services which end-users want?);
  • Do they promote innovation? (In other words, is spectrum available for new services?).Footnote 97

Much economic analysis has, since the days of Adam Smith's invisible hand, been devoted to establishing the conditions in which markets will score well or badly on these criteria.

Particularly relevant is the result that a market system with a great deal of competition (a 'perfectly' competitive one) will maximize economic welfare by both driving out waste and promoting the production of the right set of goods and services.Footnote 98

Administrative methods can, in principle, score equally well or better than markets on the criteria noted above. However, they have to operate in a fundamentally different way with respect to the gathering and processing of information to make decisions. In an administrative process, information about alternatives has to be sent up the hierarchy for decisions to be made. (Thus the spectrum regulator receives and adjudicates new requests for spectrum from firms.) In a market, this is done in a decentralized manner, for example by two or more firms, each with its own business plan which generates an implicit value of spectrum to the firms, bidding against one another for the right to acquire access to spectrum.

There is a general consensus that the administrative system of allocating spectrum by national spectrum regulators, conducted under the auspices of global rules set by the ITU, worked well or at least adequately in the past. Spectrum was not heavily utilized and technical developments were slow and relatively predictable. As a result, conflicts among alternative users were not acute nor did cumbersome administrative processes have effects which were severely detrimental to end users.

These days have now gone. Since 1990, nearly three billion people have subscribed to mobile telephones, the spread of which has far outstripped the diffusion over more than a century of fixed networks. Spectrum-using data services can provide broadband, either in competition with cable and DSL or, in large parts of the world, in competition solely with each other. Broadcast signals can be received on the move. Special spectrum technologies can exploit access to bands ' below the noise floor' to provide new services. These are only the most conspicuous and developed of a much larger number of technologies.

Whatever view is taken of the past performance of administrative methods of spectrum management, technical and market developments require a re-evaluation. In particular:

  • competition for access to high-value spectrum below 3 GHz (or even 1 GHz) is far more acute; this places a premium on efficient use of spectrum, as an inefficient allocation can have a huge opportunity cost;
  • the time to market of new technologies, now typically reliant on firm-led de facto rather than government-led de jure standardization, is a great deal faster, increasing the cost of delays in introducing new technologies.Footnote 99

In our view, these fundamental shifts in technology and consumer demand tip the balance strongly in favour of decentralized market methods as the means for allocating and assigning spectrum. This conclusion is based partly on a priori reasoning relating, for example, to the growing inability of spectrum managers fully to apprehend current and future developments and the resource conflicts between them, and partly upon the observation of how market methods have applied in practice where they have been utilized. We do not claim that a preference for market methods has been proven beyond any possible doubt; such a claim will require more data from a variety of countries over a longer period. Nonetheless, we believe there is sufficient evidence for the increased adoption of market-based policies, subject to a proper risk analysis.Footnote 100

5.2.2 How Spectrum Markets Operate

Spectrum is a non-depleting resource access to which, in the first instance (via primary issue), has been and continues to be, even in countries with spectrum markets, controlled by the government. Spectrum management has traditionally been undertaken through a two-stage process of allocation, assignment to particular users for specified purposes usually for a definite period subject to periodic renewals, and, on cancellation of the licences, return to the regulator. This is to be contrasted with a fully flexible secondary market for spectrum in which during the period of a licence, any frequency can be traded and used to provide any service using any technologies (i.e., what is known as full service and technology neutrality). This approach makes no distinction between allocation and assignment. Intermediate positions might also be adopted in which a set of frequencies is restricted to a particular set of services (e.g., communication services); this would be a loose form of allocation.

Clarity about what is being traded is fundamental to the success of any market. This places a burden on the spectrum regulators to produce an initial definition of spectrum user (or licence holder) rights. This is necessary to initiate the process of spectrum trading. But as trading of rights occurs, they may be aggregated or disaggregated in various dimensions (just as holdings of land have historically been changed through aggregation and sub-division). We would also expect to see adjustments made on the boundaries between licensees via micro bargaining in relation to detailed issues.

Because such trading and the bargaining which underlies it may have a transaction cost, it is desirable for property rights initially to be set in ways which minimize the subsequent reshuffling of rights in the interests of greater efficiency.Footnote 101

We return in Section 5.4 below to a further discussion of spectrum usage rights and interference.

5.2.3 Do the Conditions Exist for an Efficient Spectrum Market?

Only certain types of markets score well on the three criteria noted in 5.2.1 above. The most obvious problems arise when services are provided by a monopolist (and often in a duopoly). Such a firm is likely to:

  • be wasteful in producing goods and services for its customers;
  • over-charge (i.e., make monopoly profits) and thus restrict consumption of the service to below what it would be if prices reflected costs; and
  • have little incentive to innovate.

If spectrum markets led to monopolization of the supply of downstream services (i.e., if a single firm could corner the entire spectrum capable of producing such a service), and there were no other technologies, then a spectrum market could easily produce worse results than an administrative system which led to competition among downstream suppliers of services.

Is it likely that spectrum markets will lead to monopolization?

In the first place, this depends upon the degree of flexibility the regulator allows the market to exhibit. If there were no prior allocation restrictions (limiting certain services to certain bands) and if the arena in which the market operated was extensive, a firm seeking to build a spectrum monopoly leading to dominance in downstream markets would probably not succeed; in the case of major services such as mobile voice or data, or mobile broadcasting, the required spectrum holding would be very large indeed. In practice, most of the spectrum below 3 GHz is subject to administrative allocation. However, the danger does increase if either there are allocation restrictions or if the scope of the market is small (and other barriers to entry are high).

Secondly, Canada, as other countries, has a competition law which can be deployed against abusive conduct, either the original monopolization of spectrum or in the downstream market.

Thirdly, if competition law is considered to be inadequate, special procedures can be put into place to limit the acquisition of spectrum licences. This might require prior approval of transfers or the application of merger-control procedures which vet a proposed concentration of spectrum for its impact on the relevant anti-trust market.

Fourthly, spectrum regulators can construct auction rules for the release of new spectrum in ways which promote competition. The current debate in the USA over conditions to apply to the forthcoming 700 MHz auction (including a possible requirement that some spectrum should be auctioned subject to an open access obligation) is a case in point. Other more straightforward methods include dividing spectrum for auction into lots and preventing any firm from buying more than one lot, closing auctions to operators above a certain size already active in the market, and so on.

Fifthly, spectrum caps, or limits on the amount of spectrum which any firm can have licences for, can be imposed.

These five points represent a list of possible anti-monopoly measures. Some of them, especially the last, run the risk of penalising legitimately successful firms and have undesirable side effects. In our view, provided the scope of the market is large, as we recommend it should be, competition law should be enough, although a merger control procedure might yield extra comforts in the initial stages of market operation.

A second key requirement for an effective market is that it have enough liquidity (i.e., volume of trades) to provide participants with a reliable method of transacting. Liquid financial markets notoriously exhibit high spreads or differences between the buy and sell price, to compensate the intermediary for the cost of holding stock.

These transactions costs are an obstacle to the efficient allocation of resources. Spectrum markets might exhibit an even more malignant form of illiquidity, namely the inability to trade because of an absence of a counterparty.

International experience of secondary trading was collated in Chapter 3 above. It exhibited the following similarities and differences:

  • there were no signs of intermediaries being active in the market;
  • there were no signs of speculators entering the market;
  • some countries exhibited significant levels of trade (Guatemala and El Salvador) or a number of major ($ hundred million) trades (the U.S.);
  • in Australia and New Zealand, levels of trade were fairly low (roughly equal to the turnover of commercial property);
  • in the U.K., trades in the limited bands available have been infrequent, but the number of traded bands has been small and the spectrum regulator is in the middle of a large programme of spectrum awards which may provide an alternative source of spectrum to those who want it.

Thus the liquidity of spectrum markets remains a real issue and the design of liberalization measures should be in the foreground.

Other possible adverse features of spectrum markets seem less important. In particular (subject to the discussion of the definition of spectrum usage rights in 5.3 below), information problems are unlikely to be a problem among technically knowledgeable purchasers of spectrum access.

5.2.4 Windfall Issues

One of the key problems in moving to a market-based system of spectrum management is the way in which the transition from administrative methods affects the value of licences currently held by firms. In Canada, as in most countries, these are subject to quasi-automatic renewal. It would therefore be difficult, against the background of licensees' reasonable and legitimate expectations of their security of tenure, to call the licences back and re-issue them through, for example, a competitive auction process. However, the situation is different when licensees have been provided with an alternative means to deliver their services. This happened in Canada when UHF TV channels 70–83 were taken away from broadcasters. A similar situation will arise with analogue switch-off.

Converting existing licences which are confined to particular uses to licences which are flexible and tradable appears to confer a considerable windfall gain on the holder; the firm can now convert the spectrum to more valuable use for itself or sell the licence for a higher price, thus raising shareholder value.

However, to some degree, the example described above is a fallacy of composition. If tradability and flexibility of use were conferred on just one licence, its price would go up. But if tradability and flexibility are conferred on all, the equilibrium price of previously non-traded spectrum which was in short supply will fall (because in the new situation, more spectrum will be converted to that use) while the equilibrium value of 'abundant' spectrum will rise.

This proposition has been verified in work undertaken by Thomas Hazlett.Footnote 102 He examined the effect of a liberalized spectrum markets on the proceeds of auctions for more than one thousand mobile wireless licences in 27 countries. After allowing for other factors, such as GNP per head, he found that the value of licences in regimes where more property rights, such as the right to change use of the spectrum, were conferred was appreciably lower (by 38%) than in countries with a more restricted regime. This would mean that windfalls form liberalization to existing licensees would be negative.

For those who find this evidence less than wholly convincing, there are ways of finding a compromise between discouraging efficient trading (which would move spectrum to an economically better use) and conferring large windfall gains on already affluent companies. These include some sharing of the gains from selling newly tradable spectrum when the event occurs. There is precedent for this in some countries' approaches to large capital gains which follow privatization.

5.2.5 Summary

  1. Administrative methods of spectrum management have performed adequately in the past, but many countries, including Canada, have moved along a continuum towards greater reliance on market mechanisms. Often, administrative methods and their informational requirements, slowness, inflexibility and immunity to the special interests of incumbents, tip the balance in favour of spectrum markets as a means of enhancing static and dynamic efficiency in the provision of services to end users.
  2. The key to a spectrum market is to allow purchase and sale of spectrum licences subject to flexibility in use, leading to complete service and technological neutrality; this requires a clear definition of licensee rights, initially set by the regulator but subject to change by subsequent transactions.
  3. Markets which are monopolized are bad for end users and it is therefore essential to consider whether spectrum markets can be 'cornered' in some way through trading. If the scope of the market is large, this is unlikely, but various regulatory measures can be put in place to prevent abuses; markets also need liquidity to allow trades to be implemented and here the evidence is mixed, some countries exhibiting many trades, others a relatively small number; the market's scope and rules should be designed with liquidity in mind.
  4. It is often suggested that converting existing inflexible licences to a flexible form will confer a windfall on licensees, although there are theoretical grounds for doubting the significance of these gains. Furthermore, if deemed necessary, it is quite possible to put procedures in place to mitigate windfall gains which might accrue to current licensees.
top of page

5.3 What Spectrum Bands Should Be Subject to Market Methods of Allocation?

5.3.1 Unlicensed Spectrum (Spectrum Commons)

The previous section has focussed on the alternatives of administrative and market methods of spectrum management. There is, of course, a third variant, namely the use of unlicensed spectrum or 'the commons,' which has always been a part of the traditional administrative method of spectrum management but which more recently has attracted much attention. Unlicensed spectrum is a hybrid in which administrative methods are used to choose the unlicensed frequencies and to set technical conditions for spectrum access (power limits, etc.), but thereafter access is available to all qualified users.Footnote 103

A more complex hybrid which again has been a feature of the traditional administrative method of spectrum management involves a primary user with pre-emptive access to any frequency (i.e., other uses must give way when the primary licensee requires access) and a secondary 'commons,' (i.e., a regime in which other users can jointly access the spectrum when it is unused or can use the spectrum on the condition of not disrupting the operations of the primary licensee).

Unlicensed spectrum and more complicated derivatives are discussed in section 5.4.13 below. This section focuses on the alternatives of administrative and market methods.

5.3.2 Use of Markets for Spectrum Deployed for Commercial Purposes

The impact of a market is measured by the transformation or substitution possibilities it enables. In relation to the former, making spectrum tradable enables licensees to exit the market for spectrum-run services and to convert their licences into cash. The key benefit here is to admit new uses and (possibly) new users of spectrum into the system. The increase in value of the new applications compared with those they replace broadly measures the benefits of the market and will differ from frequency to frequency depending on end-user valuations of the services produced and the degree of disequilibrium inherited from the administrative system.

In addition, some licensees will wish to substitute one frequency for another, while maintaining an existing service; fixed links provide a possible example since they can occupy a range of frequencies and users may wish to switch within them to reduce costs. Alternatively, it might be desirable for a firm running a paging system to switch from VHF to UHF. Such a switch is only possible if the market arrangements apply to both frequencies between which an exchange is sought. But if the market applies only to the lower or upper part of the feasible range, this will not happen.

These considerations suggest that, other things being equal, a spectrum market exhibits increasing returns to scale. In other words, a market covering twice the number of frequencies will confer more than twice the economic benefit of one covering a smaller area. This provides an argument for wide-scale use of markets and, subject to a risk-management analysis, a speedy transition.

It also provides an argument for the release of the maximum possible spectrum into the market place, rather than its retention under the control of the regulator. This goes against the grain of past administrative practice when spectrum assigned to a use or a licensee which failed was 'frozen.'This encouraged conservatism in issuing licences, because it was difficult to retrieve mistakes. Moreover, new demand could only be met by newly issued or refarmed spectrum as operators could not go into the market to acquire it.

In a system with secondary trading, the same does not apply. The more spectrum available in the market place, the easier it is, in principle, to acquire a suitable frequency. The argument for 'hoarding' by the regulator no longer applies.

There still may be an issue about the comparative speed of response of the market and the regulator. If the approval process for an ownership transfer is speedy and if the procedure for the regulator to make a primary issue of spectrum is slow (as is likely to be the case), then releasing spectrum into the market in advance will speed up the realization of potential users'plans. For this reason, we believe that spectrum for which there is a predictable use should be made available as soon as possible (i.e., offered to the market without a high reserve price).

5.3.3 Applying Market Methods to Public Sector Use Spectrum?

The first two enabling guidelines of the June 2007 Spectrum Policy Framework for Canada are:

  1. market forces should be relied upon to the maximum extent feasible; and
  2. notwithstanding (a), spectrum should be made available for a range of services which are in the public interest.

The document on spectrum reform issued by the European Commission in June 2006 makes a different and less persuasive assertion. It states the aim of the proposals is:

"to shift away from administrative decisions towards market-based valuations of spectrum, including in particular trading of exclusive spectrumFootnote 104 usage rights to optimize efficiency and flexibility of usage and reduce access costs. However, public administrative control over spectrum usage should be maintained where specific public interests objectives – such a safety of life – are at stake."

Thus, unlike the Canadian guidelines, the European Commission prescribes not an outcome (availability of services) but a process (public administrative control). The former thus, properly in our view, leaves open the door to the extension of market methods to spectrum used for (some) public interest purposes.

The enabling guidelines do not indicate which services are 'in the public interest.'For the purposes of this report, we assume that they include those directly supplied by government or public sector bodies, those which are financed by public funding, and those where legislation or administrative practice imposes exceptionally heavy constraints on the providers of services, causing them to behave in a manner significantly different from an unconstrained profit-maximising provider.Footnote 105 The delineation of such services is clearly and properly a political decision.

Services in the 'public interest'have traditionally been furnished with spectrum by administrative methods. This is a viable way of proceeding. But there are doubts about whether it is necessary and efficient. It would not be necessary if a market method of securing the availability of spectrum were feasible, and it would not be efficient if those administrative levels led to wasteful use of spectrum, as we have suggested above is likely to be the case in connection with non-government use spectrum.

It may seem counter-intuitive to consider the use of market methods to allocate spectrum to uses in the public interest but three contextual points can be made in its favour:

  • it is not the normal practice in market economies to allocate inputs by administrative methods to services which are considered essential in the public interest; as examples, public service broadcasters pay energy bills; the ambulance service buys vehicles; police forces recruit officers;
  • it is widely believed that poor incentives to economize on spectrum have left many public sector users (especially Ministries of Defence in some countries) with excessive holdingsFootnote 106 but this factor may not operate to the same degree in Canada;
  • it is a basic result of economic analysis that there are always better ways of promoting the production of a service than by furnishing it with a free or subsidized input simply because such an approach encourages extravagant use of the input in question. In other words: for the sake of economic efficiency, do not mess with (or subsidize) input prices.Footnote 107

Most of the arguments for spectrum markets in the commercial world read across to spectrum used to provide public interest services. Such a policy can be implemented by the following set of measures, which seem fully to satisfy the enabling guidelines of the new Canadian spectrum policy:

  • a significant part of spectrum used to provide services in the public interest should be made tradable: the key point here is that this does not require the licensees actually to sell (or lease) it, but it does permit the disposal of any permanent or temporary surpluses;
  • equally (and more controversially), organizations producing services in the public interest should be required to anticipate the need for more output and be responsible for acquiring (and given the means to acquire) the necessary spectrum by going to the market;
  • there should be a fall-back position for unexpected emergencies, under which additional spectrum would be made available by administrative methods (i.e., by non-competitive primary issue or by compulsory purchase from existing users).

This approach represents one way of continuing the use of market forces while ensuring the availability of services in the public interest. In our view, it is preferable in the long term to the conventional view that the provision of services in the public interest entails permanent use of administrative methods of spectrum management. It is also noteworthy that the United Kingdom adopted this policy in 2006 and that OFCOM has recently published a consultation document on its detailed implementation.Footnote 108

A more gradual approach to unifying all spectrum management is also available. A number of steps, culminating in the application of markets to services in the public interest are set out in Section 3.24 above.

top of page

5.4 Transitioning to Spectrum Usage Rights and Flexible Use Spectrum Rights

5.4.1 Contrasting the Three Approaches to Spectrum Management

As indicated earlier, there are three models of spectrum management and each have an important role to play in certain circumstances:

  1. the traditional administrative method;
  2. the flexible exclusive licensedFootnote 109 method; and
  3. the commons/shared access method.Footnote 110

The traditional administrative method assumes that the regulator/spectrum manager plays a direct role in assigning the spectrum to users and in choosing the technology and service model for how the spectrum is accessed and used. This is often used to characterize the traditional model of spectrum management in which direct regulatory oversight manages most features of how spectrum is used.

In contrast, the other two models decentralize to market forces significant responsibility for assigning those who use the spectrum, the choice of technology and how the spectrum is used but they accomplish this goal in different ways.

The exclusive licensed model focuses on making the access rights to the spectrum itself tradable, treating the licences virtually as private property and leaving control over who is allowed to share that spectrum to the licensee. In this model, the primary user (licensee) is granted strong interference protection to protect their exclusive right.

In contrast, commons/shared use spectrum assigns equal rights to use the spectrum to a limited number of multiple users or, in the most extreme case, to all users who comply with the access rules (protocol) for accessing the spectrum. In this model, there are multiple co-primary users with explicit rights for interference protection. The exclusive licensed model is often regarded as more suitable for wide-area wireless network deployment by traditional service providers such as over-the-air broadcasters or mobile service providers, while the latter may be more suitable for low-power/limited range end-user/equipment-maker driven deployment models such as associated with WiFi-style networking.

The purpose of the discussion here is not to suggest which model is better but to contrast the relative ways in which each model may differ from the legacy approach to spectrum management and help realize the goal of transitioning more management responsibility to market forces. Sequentially, we discuss for each model what is needed to facilitate the transition.

Operationally, the transition from the traditional administrative approach to spectrum management toward market flexibility may be thought of as being implemented in three related, but logically-distinct, directions:

  • Service neutrality: right to change use of spectrum (e.g., move from broadcasting to mobile services);
  • Technology neutrality: right to choose technology that accesses spectrum (e.g., CDMA or GSM);
  • Ownership transferability: right to transfer spectrum to another licensee who assumes rights to access, obligations, and protection associated with the licence. (In the first instance of assignment, the preferred market-based approach is to make the assignment via auction.

The discussion below focuses on what happens after the licence has been assigned, so subsequent transfers of rights ownership are a form of secondary trading. Obviously, the form and success of the secondary market will be influenced by the auction design that shapes the initial assignment).

In the extreme version of the traditional administrative method, the use, technology and assignment of rights are all under direct regulatory control. While this is a caricature of the actual situation in most cases, we use it as a straw man to characterize the status quo situation which may be reformed by allowing new forms of market flexibility.

In contrast, in the extreme version of a transition to market forces, the Invisible Hand of the decentralized market is relied upon to manage the choice of user, technology and service choices for accessing spectrum. Of course, just as the administrative extreme vision is a caricature, so too is this vision of the market process. In fact, as we will discuss further below, the market will be subject to regulation. One may consider by way of analogy the situation in real estate markets which mix different frameworks for ownership (government and commercial, public and private) and encumber titles granting rights to use with various obligations (responsibilities not to pollute) and limitations on use (zoning bylaws, rights of way easements, restrictive covenants, etc.).

Similarly, under the various market models for spectrum access, the need to manage legacy transition issues, protect against harmful interference and promote the public interest will require on-going regulation. However, relative to the traditional framework, the regulation will be less administrative and more market-based, relying on general competition law and adjudication processes to manage a well-defined framework of rights. As we shall see and as will be discussed further below, the definition of well-defined framework of usage rights is an important first step to enable market processes.

5.4.2 Flexible Exclusive Licensed Spectrum

Under current licensing regimes, these licences are heavily encumbered with administrative controls that limit their flexibility to a single licensee that has primary usage rights granting the licensee strong interference protection. Strong interference protection is required to protect the value of complementary investments in network infrastructure without which commercial investment will be deterred.

While this approach worked reasonably well in the past, especially when demand for spectrum access was less intense, it suffers from a number of important limitations in light of the rapid pace of innovation in wireless technology and services, growth in demand and the move toward increased reliance on market processes and competition across the communications value chain.

Our analysis of implementation issues related to spectrum management reform and implementation of an exclusive licensing regime includes a review of the following topics:

  • isolation of spectrum management from other policy considerations to allow focus to be on interference management
  • interference management
  • easements may be limited
  • tradable licences (primary and sub-leasing)
  • predictable/transparent licence terms
  • service and technical neutrality, as far as possible
  • long but fixed term limits
  • shared information (spectrum database)
  • Spectrum Usage Rights (SURs) technically defined
  • enforcement framework

5.4.3 Resetting Spectrum Management Focus

From the perspective of overall regulatory policy, it seems advisable to limit the use of spectrum management as a general policy tool. Spectrum is an essential and scarce resource, the management of which entails significant technical complexity.

Managing spectrum technical complexity appropriately is difficult enough without conflating the challenge with other policy issues such as content diversity, universal access, or industry structure. While these other issues remain valid policy concerns, they need to be addressed separately and independently from the question of how to manage shared access to the radio frequency spectrum while providing appropriate interference protection. The focus of spectrum management should be on providing adequate interference protection in a market framework. The goal of licensing reform should be to strip away, to the extent possible, all aspects of the regime that are not related to interference management in a market-allocation framework such as:

  • build out requirements for licences
  • service or technical restrictions that are not consistent and motivated by interference protectionFootnote 111

Since the focus of the above will be to move to a licensing regime focused on interference management as opposed to other possible goals for spectrum management, it is helpful to provide a summary discussion of the sources of harmful interference in radio systems. A more complete discussion can be referred to in Annex 4.

5.4.4 Interference Management

Harmful interference arises in radio systems when a transmitter's ability to communicate with its intended receiver(s) is limited because of the transmissions of other transmitters. The problem may be thought of as arising from the limitations of the receiver: better receivers are more able to extract the desired signal from a noisy environment of background radiation and other transmitters.Footnote 112

There are three categories of interference that are of principal concern:

  1. In-band interference from adjacent areas: because radio waves do not simply stop at geographic boundaries, transmissions from an alien Tx may interfere with reception for Rx that are distant from the desired Tx (i.e., when the signal at the target Rx is stronger from the alien Tx than from the desired Tx). The principal means for addressing this challenge is to adequately separate Tx and ensure that the power flux outside the licensed service area is below some maximum threshold. Limitations on Tx power and antenna height/directivity can effectively address this challenge.
  2. In-band interference from adjacent frequencies: because Rx frequency selectivity is imperfect, energy may leak in from adjacent frequencies, once again causing interference. Analogous to the preceding, the principal means for addressing this is to ensure adequate frequency separation (guard bands) to ensure that interference from adjacent bands is limited. Once again, limitations on Tx power (out-of-band emissions) and antenna height/directivity can address this.
  3. Out-of-band interference: once again, because of Rx imperfections, a strong signal in an adjacent band may interact with elements in the Rx front-end to produce new signals that are both hard to predict and hard to address since they can be highly non-linear. This is the hardest type of interference to address since presumably both the Tx and Rx are operating within their normal performance bands and it results when the victim Rx is near the boundary of the alien Tx's operating area.

Taken together, by properly defining the electrospace – and the size of the volumes – it is possible theoretically to specify Tx or Rx occupancy rights so that Tx/Rx must operate in different and distinct electrospace volumes to ensure non-interfering operation. Thus, to deploy a radio system the provider needs to first acquire the rights for the electrospace that is needed to ensure adequate interference protection given the existing allotment of usage rights. A full allocation of the electrospace volumes will provide a map of the maximum energy flux allowed from the system to be deployed (so as to preclude interference to others) and to be tolerated from alien systems (to anticipate potential interference from others so that the target system may be designed appropriately).

Before proceeding, it is worth noting that the definition and management of the electrospace framework is dependent on the physics of the frequency band and radio propagation, local conditions (terrain, weather, occupancy of adjacent channels) and radio system engineering constraints (limitations of Rx/Tx design, propagation modeling standards). These are technically complex issues which are generally beyond the understanding of almost everyone who is not a highly-trained radio engineer.

The exclusive use licence defines the rights to occupy the spectrum volume. The primary user has a presumptive right to exclude other users from occupying their electrospace. Secondary users may have the right to occupy the electrospace if they can do so without causing interference to primary users, although they have no interference protection rights of their own. The rationale and implications of allowing easements to facilitate secondary use are discussed below.

5.4.5 Easements

There are two types of easements that are commonly discussed: underlay and overlay easements. An underlay easement is the right to operate a low power device in primary licensed spectrum. The logic is that sufficiently low powered devices may exist below the level of background radiation or below the level of sensitivity of primary user receivers and, hence, would not result in interference for primary users. The unlicensed devices such as WiFi radios, remote controls and cordless phones that operate under Part 15 rules in the U.S. are an example of such a secondary use.

The newest version of such a use is provided by Ultra-wideband (UWB) radios that are used for such services as cable substitution to link wireless multimedia devices in the home, for ground-penetrating radar and potentially for a wide array of other radio services. UWB is able to offer very high-bandwidth services with very low power in any particular frequency band because it spreads the signal over a very wide bandwidth (e.g., in excess of 500 MHz or more). From a regulatory perspective, it seems likely that acquiring the sub-leases to allow such low-power operation across a swath of spectrum as wide as 500 MHz would likely represent an insurmountable transaction cost challenge that would preclude the commercialization of valuable UWB technologies. Consequently, for UWB to be commercially feasible, it needs to be granted an easement to operate in primary spectrum. Such radios have been approved in the United States, but not in the U.K. Opponents have argued that allowing such radios diminishes the value of primary users' systems and will pose a threat of future interference especially as such UWB devices aggregate and increase the level of background noise, resulting in an interference "pollution" externality. The risk of even small increases in background noise is especially problematic for wireless systems such as radio telemetry and astronomy that rely on very sensitive receivers.

The second type of easement is an overlay easement associated with the deployment of so-called cognitive or software radios that may opportunistically be able to access the "white space" where there is a pre-existing primary allocation. Cognitive radios incorporate intelligence and sensors that allow them to sense their environment and modify their behaviour in response to local conditions. For example, a radio equipped with GPS and able to access a database of spectrum users may be able to identify areas where there are no active Tx (e.g., over-the-air television broadcasters or radars) and could then access the spectrum selectively so as to avoid interfering with primary users. In other situations, a secondary user could "listen-before-talking" to detect gaps over time when primary Tx are silent and only transmit in those silent periods. Cognitive radios (most often implemented in software, and hence are also sometimes called software radios) are in the early stages of commercialization and the capabilities and uses of such systems are not well demonstrated yet. Consequently, there is much discussion as to whether such overlay uses could reliably and predictably be deployed without posing an undue risk of interference. The likelihood that such systems could be deployed successfully is much clearer if one considers overlaying a legacy system such as traditional over-the-air broadcasters using fixed Tx than if one contemplates overlaying, say, a mobile telephony service. Much of the argument against allowing such cognitive radio access to broadcast white space appears motivated by a desire to limit competition and to extend exclusive access rights beyond what may be socially optimal. It appears that the technical case for allowing such radios with minimal risk in at least a portion of the broadcasting spectrum is favourable although this remains a topic of hot debate in the ITU and elsewhere.

The rationale behind easements is to provide spectrum access options for technologies and business models, especially appropriate for end-user deployed or other new wireless business models that may not have the resources available to acquire long-term exclusive access rights or do not have the need for the strong predictability and interference protection afforded by (and priced into the access for) exclusive licensed spectrum. While it seems unlikely that technologies making use of such easements will be more than marginal additions to the wireless landscape, such marginal options nevertheless may be very important for the development and robustness of wireless innovation and competition.

With respect to UWB, there may be no other option for deploying this technology if an appropriate underlay easement is not granted and this looks as if it might be a very promising addition to the wireless landscape. With respect to overlay easements to enable cognitive radios, the case is less clear. It seems desirable to provide some scope for experimenting with such secondary use easements and the logical spectrum to do this in is the broadcasting spectrum. This will provide real-world experimental evidence and experience which will help in commercialization efforts and will position Canada for fuller expansion of such use should the appropriate technologies and market demand for them evolve as proponents anticipate.

If easements are imposed after spectrum is auctioned, the licensee may claim that a portion of the value of the license has been expropriated since the winning bidder could have elected subsequently to sub-lease the secondary usage rights (assuming that the license will be adequately flexible as described above). In contrast, if the licences are auctioned with easements specified then bidders can factor any risk and value impacts of such potential encumbrances into the auction. Assuming that the purpose of acquiring the licence is to ensure access to adequate spectrum and interference protection to ensure that the anticipated wireless use will work as planned, purchasing stronger-than-needed protection (i.e., easement-free spectrum) would increase spectrum access costs above what is needed. Thus, there does not appear to be a strong ex ante reason for favouring or opposing easements associated with newly auctioned spectrum.

5.4.6 Tradable Licences (Primary and Sub-leasing)

This is the core to the market-based approach where spectrum licences are fully transferable (primary users may replace each other), and sub-leasing/sub-division should be possible.

The core characteristic of enabling market mechanisms for exclusive licences is to allow such licences to be tradable and amenable to division and aggregation (although with respect to aggregation, there may need to be controls to protect against the creation of undue market power).Footnote 113 For tradability, it is best that the primary licensee be able to transfer its licence to a new primary licensee; whereas with respect to leasing, the primary licensee retains the principle liability and regulatory responsibility for ensuring compliance with licence rules although this may be attenuated under the terms of the lease agreement.

The goal of tradable licences has several beneficial implications:

  1. First, if licence ownership may be transferred, then it is continuously possible for market forces to provide incentives for spectrum to be allocated to its highest value use, and for whatever use it is currently being employed in to reflect its true opportunity cost. This induces high-powered incentives to use spectrum efficiently.
  2. Second, the more spectrum that is tradable in this way, the more liquid will be secondary markets and the lower the average opportunity cost or scarcity rents associated with spectrum access rights. Encouraging spectrum prices to be as low as possible, consistent with aggregate demand and supply factors, will enable low-cost access for new applications and services which is an important overall goal of spectrum reform.

As will be discussed further below, the transition to tradable licences may need to take place in a two-phase approach. However, the more spectrum that may be transformed into tradable spectrum relatively quickly and to a common licensing framework, the more liquid will be secondary markets and the lower will be transaction costs for selling and acquiring new spectrum.

A phased approach is in contrast with the approach described by Evan Kwerel and John Williams (2002)Footnote 114 in their "Big Bang" proposal, which seeks to move the maximum feasible amount of spectrum to a new tradable license regime in a single two-sided auction. This ensures that there is lots of spectrum available to allow incumbents and new entrants to flexibly re-optimize their current holdings and addresses the challenge of inducing incumbents and entrants to participate.Footnote 115 While the Kwerel & William' Big Bang proposal has much to recommend it, it may be too ambitious in light of existing legacy allocations and the desire to move quickly before all of the impediments to significant reform may be addressed.

As an alternative, a rolling "Big Bang" may be advisable, under which policy-makers in Canada commit to moving as much spectrum as possible to the new regime over time, starting with spectrum to be auctioned in the near term associated with the switch-off of analogue television.

Allowing licences to be divisible and subject to aggregation will also enhance flexibility. As will be discussed further below, establishing the licences with reference to appropriately defined electrospace volumes will facilitate the ability of market participants to divide and combine access rights to accommodate the needs of evolving wireless technologies, business models, and markets. We expect that when spectrum is sub-leased, the primary licensee retains primary responsibility for ensuring compliance with spectrum rules.

The current Industry Canada method for dealing with leasing of spectrumFootnote 116 represents a significant improvement relative to the overly restrictive previous situation but is only a small incremental change relative to the sort of divisibility and sub-leasing anticipated in the recommended reforms here. However, even the current framework leaves much to be desired. Industry Canada requires review and approval of every lease which seems designed to impose unnecessary bureaucratic delays and costs thereby increasing spectrum access transaction costs and hampering the sort of flexibility and secondary trading that the reform appears intended to enable.

5.4.7 Predictable/Transparent Licence Terms

For spectrum trading to evolve, it is necessary for there to be a clear definition of the rights that are being traded since wireless infrastructure and service participants need assurances that access to essential and scarce resources will not be subject to arbitrary and uncertain government action.

Clear, open, predictable and as-simple-as-possible spectrum management policies will help reduce regulatory uncertainty (which deters investment) and will assist in promoting the transition to a new market-based spectrum management regime.

While there will not be a single licensing framework in light of the need for multiple regimes (some administrative, some unlicensed, and some exclusive licensed) and the difficulties of effecting a comprehensive transition in a timely manner, policymakers should strive to limit the number of diverse regimes and to harmonize existing regimes to the extent possible. Limiting the number of licence types, adoption of common terminology and process rules should help in this goal. For example, in the United States, the bifurcation of spectrum management responsibilities across federal government uses (responsibility of NTIA) and commercial/private and lower levels of government uses (responsibility of FCC) is not desirable. Similarly, incremental reform steps such as in Canada that proceed band-by-band often not as part of an overall plan or road map (while better than no reform at all) may result in a proliferation of heterogeneous licensing frameworks (e.g., different rules for spectrum that was acquired via auction as opposed to that acquired through administrative assignment).

Finally, as will be discussed further below, tying the definition of the licence to well-defined and unchanging technical/physical characteristics should limit future debates over interpretation of spectrum access rights.

5.4.8 Service and Technical Neutrality, as Far as Possible

The licensing regime should aspire to service and technical neutrality, as much as possible. As explained earlier, legacy licences adopted restrictions on the choice of technology and services that could be offered, in part, to help manage interference. Prior agreement on the technology facilitates agreement on the appropriate propagation model so that licensees may more easily prevent or resolve interference disputes. For example, when one cellular operator seeks to expand his capacity by deploying smaller cells, this will have the effect of altering the flux of energy across his network footprint which will also change the characteristics of the power that leaks into adjacent geographic territories and bands. When other licensees adopt similar strategies (which they are more likely to do if they face similar technical and market options), the interference impacts are more likely to be symmetric and off-setting.Footnote 117

In short, from an interference management perspective, the rationale for retaining non-neutral technical or service-based restrictions in licences is to serve as short-hand for the sorts of considerations that are addressed by the electrospace notion of interference discussed earlier.

The cost of not allowing service or technical neutrality is that it reduces the flexibility of licensees to adopt the lowest cost available technical option — although imposing service/technical neutrality may contribute to the realization of scale and scope economies and hence the costs of various options if applied over a sufficiently large market. In Canada, which like Australia is more likely to be a "technology-taker," it is unlikely that service/technical homogeneity constraints would be of much help in facilitating the realization of scale/scope economies which are generally determined on a global scale. The restrictions on service/technical neutrality may also limit the range of services providers may offer; denying consumers enhanced services or opportunities for providers to differentiate their offerings. Since regulators are presumably less knowledgeable about changing technical possibilities and market needs and regulatory changes are slow and costly, these restrictions are poorly matched to the needs of robust, competitive markets. For example, the restriction against 2G providers using 2G spectrum to offer 3G services was intended to bias technology upgrades toward the GSM-approved W-CDMA "UMTS" technology, whereas locales such as the U.S. that allowed 2G providers to use their spectrum to offer 3G services provided a market for CDMA2000 technology.

It is worth noting that service and technical neutrality are distinct but related concepts. For example, sacrificing service neutrality may make it easier to realize certain types of technical neutrality while sacrificing technical neutrality may make it easier to support service neutrality. For example, managing interference among fixed Tx's with known locations is much easier than for mobile Tx's, which implicitly requires larger guard space. Thus, it is relatively easier to allow secondary use or technical flexibility in a band where services are limited to fixed as opposed to mobile services. Alternatively, technical limits (e.g., requirement to use the OFDM modulation specified by WiMAX) may allow greater flexibility in mobile or fixed data/voice services. Since service restrictions have obvious implications for technical flexibility and vice versa, the two concepts are closely coupled. The limits inherent in the definition of the electrospace considerations discussed earlier demonstrate that even in a fully liberalized regime, spectrum management and regulation will continue to be necessary and so full technical or service neutrality are not achievable except as ideals.

The reasonable goal is to eliminate all arbitrary service and technology-based restrictions. For example, restrictions on services that are motivated largely by desire to control industry structure (e.g., limit the number of facilities-based competitors) or to implement spectrum access subsidies (e.g., increasing the supply of spectrum for targeted uses such as public broadcasting) should be avoided as already discussed. To the extent that such policies are warranted, they ought to be applied explicitly and on an appropriate intermodal basis (e.g., it makes no sense to promote wireless broadband separately from wired broadband services).

Eliminating service and technology restrictions as much as possible from licences will make licences more fungible and contribute to the liquidity of secondary markets for spectrum.

5.4.9 Long Period but Fixed Term Limits

The choice of the term duration for tradable and flexible licences involves making compromises between two desirable outcomes:

  • firstly, uninterrupted investment incentives for firms which avoid the emergence of disincentives to invest towards the end of a licence period for fear that investments in fixed assets whose lives go beyond the term of the licence will be stranded;
  • secondly, flexibility available to the regulator to influence both spectrum use and the definition of spectrum user rights.

It must also be recognized that the nominal term of a licence may understate the licensee's legitimate and legally enforceable rights to continued use of the spectrum. As an extreme, traditional apparatus licences in the U.K. are annually renewable, but the custom and practice of almost automatic renewal impose serious but vague constraints upon the regulato's freedom on action — vague because the extent of the rights has not been legally tested. New tradable licences formed out of existing licences in the U.K. are intended to be 15–20 years and renewable. In Canada, licences have similar terms — 10-year duration and a 10-year extension with a reasonable expectation of renewal.

The trade-off noted above between investment incentives can be resolved by giving the regulator, in the case of long or infinite licences duration, powers which can be exercised within highly specific conditions, to re-acquire licences by compulsory purchase, subject to the payment of compensation. These powers might be exercizable in cases of:

  • a security, defence or other national emergency
  • a need to redefine spectrum user rights to accommodate a new technology such as ultra wide band
  • a change in international agreements or treaties, e.g., a change in the ITU allocations

The mandatory acquisition of the spectrum under this last heading would have to be accompanied in all cases by i) a cost-benefit analysis and ii) a demonstration that the benefit of the technology could not be attained by ordinary commercial transactions among rights holders — in other words, by means of the so-called Coasian bargaining,Footnote 118 by means of which advantageous exchanges are normally effected; a necessary condition for this would be presence of large or prohibitive transactions costs.

We have considered both very long-term licences, which safeguard investment in assets over a term of 30–40 years, and long-term licences for, say, 15–20 years, which give the regulator the opportunity to recover spectrum at intervals when it is desirable to do so.

On balance, very long-term arrangements are more favourable on the grounds that they give better investment incentives and involve a clearer definition of licensee's rights. However, departures from this may be appropriate in specific bands.

5.4.10 Spectrum Usage Rights (SURs) Technically Defined

As noted earlier, it is desirable that spectrum usage rights and the interference protection that is their focus be defined clearly and as simply as possible with reference to the technical and physical characteristics of the spectrum. In short, such rights should be defined with reference to the behaviour of the transmitters, Tx, that give rise to the radiated energy that causes interference at victim receivers. Attempts to proxy for such technical rules via service/usage restrictions or by limits on the specific technologies that may be used are, at best, approximations. Such approximations are not robust to changing market and technical innovation over time and so result in ambiguous and uncertain interpretation of the interference protection rights granted by the licensing framework.

Technical experts in Australia (Michael Whittaker of FuturePace), in the U.K. (OFCOM), and the U.S. (Robert Matheson of the NTIA) agree that adopting a framework akin to the electrospace notion of spectrum rights, which divides the spectrum into technically-defined "volumes" within which the radiant energy may be characterized, is the right direction to go. As we explain further in Annex 4, however, the precise details of how this should be done are not yet settled. Therefore, while we think it is appropriate for policy-makers in Canada to commit strongly to moving in such a direction, we do not believe it is yet clear precisely which of the several approaches under discussion are best suited. OFCOM has been refining its approach in real-time, prompting further technical comments from those engaged in the debate worldwide. Canadian policy-makers should endorse the general approach but also participate in the ongoing debate to ensure an adequate understanding of what emerges and develop their own consensus view on what is the best approach from these discussions.

The benefits of adopting technically sound definition of property rights are several fold:

  1. Such rights are amenable to clear and simple expression in regulation with reference to industry/regulatory sanctioned standards (e.g., with reference to ITU Recommendations and metrics).
  2. The above will reduce threat of ex post litigation over interference by simplifying process by which source and responsibility for mitigating harmful interference may be assigned.
  3. The above will provide a degree of future-proofing for rights since basic physics of radio propagation are unchanging while the technologies and services that use it do change. These latter changes render service-specific or technology-specific licensing terms inflexible and uncertain over time.
  4. The above limit and focus licensing regime on interference protection rights, thereby limiting scope for spectrum management to stray beyond its core responsibility and thereby result in excess regulatory distortions to wireless equipment and service markets (which are harmful to competition and economic efficiency) and distortions to other important regulatory policy debates that should not be spectrum-focused (e.g., broadcasting content and access to information or broadband policy).

For further discussion of the details of the on-going debate over how best to define the electrospace technical-approach to licensing, see Annex 5.

5.4.11 Shared Information (Spectrum Database)

The ability of potential sellers and buyers (and regulators) to keep track of current licences is an important component of tradable markets facilitated by a database which should include publicly available information on the terms of all licensed spectrum and any sub-leases that are also in force. The database should register the location and frequencies used by transmitters and receivers which may be entitled to interference protection (see further discussion below). Knowledge of the location of existing Tx's and Rx's (where feasible) will allow potential purchasers of rights to accurately model the existing interference environment they are seeking to enter and to enable them to properly assess the rights they seek to acquire.

The database will also provide information to regulators if called upon to adjudicate spectrum disputes and to enable them to track and assess the usage of spectrum in differing bands.

Finally, the database could include additional tools to enable visualization and linkage with other data. For example, data on spectrum historical occupancy/usage and visualization tools to interpret alternative propagation models could assist in users interpreting the complex data of overlapping spectrum usage rights.

The spectrum auction and trading system operated by Cantor-Fitzgerald, the Wall Street brokerage, provides an example of the sorts of capabilities that are needed at a minimum.Footnote 119 Cantor Spectrum & Tower Exchange provides an open or closed transparent forum for both primary (auction) and secondary (post-auction) market spectrum transactions in both public and private marketplaces.

  • Sellers/Lessees can review FCC licensee information obtained by the exchange and see a snapshot in real-time. An example screen is shown below in Figure 6.
  • Qualified licence sellers/lessors or public sector entities offer radio frequency spectrum and digital sub-channel capacity in a multi-dimensional format showing coverage area, population, frequency range, radio service rules, terms and conditions, channel, time slot, etc.
  • Buyers/lessees search for specific assets (or receive electronic notification), and can easily evaluate and bid on them.

This type of system helps facilitate the critical matching function that liquid markets depend on. The absence of similar capability could be one reason for the anaemic use of such secondary trading markets as currently exist in Australia and elsewhere for portions of the tradable spectrum.Footnote 120

Figure 6

Figure 6

Source: Cantor-Fitzgerald Darrin M. Mylet, VP Wireless Services

5.4.12 Enforcement Framework

To date, there has not been much detailed discussion in academic or policy circles regarding how best to enforce and resolve spectrum access disputes or, if reforms such as those described above are adopted, interference disputes. Much of the focus on defining appropriate property rights correctly up front is to make subsequent enforcement relatively simple and to minimize the need for adjudication of interference disputes. Nevertheless, such adjudication will arise. As a general matter, it seems desirable to promote negotiation as a first step approach to resolving such disputes. This is in keeping with letting the market do as much as possible. However, when negotiation fails to reach a mutually acceptable settlement, it will be necessary to move the disputes to some formal adjudication process. Most simple matters may be readily addressed by appropriate arbitration potentially sanctioned by public and/or industry oversight. The resolution of spectrum disputes does not seem well-suited for general courts or litigation, so if a regulatory agency (rather than an independent arbitration or industry review board) is responsible, it should operate under clear, open procedures to ensure a fair and predictable process. As noted earlier, enforcement will require a level of technical expertise to assess and evaluate claims of harmful interference and to devise appropriate mitigation strategies. This seems like a natural function for the independent spectrum administrator (independent of other policy concerns as noted earlier).

A second problem that will need to be addressed is the question of equipment certification. This arises naturally in the context of unlicensed or equipment class licences. For example, in the U.S., the FCC tests and certifies new devices as compliant with appropriate rules such as Part 15 rules. In contrast, in the European Union, manufacturers self-certify their equipment as being in compliance with rules. The U.S. approach allows manufacturers more latitude in designing devices that may result in interference which will need to be subsequently mitigated at the cost of more extensive government testing before allowing the devices to be commercially sold. In contrast, the European approach of self-certification economizes on such testing but holds manufacturers to a higher standard of liability should harm result. In the future, the question of optimal certification policies likely will be both more complex and important as new systems incorporating technologies such as software radios or dynamic spectrum access technologies will challenge traditional notions of what constitutes operator-controlled versus end-user controlled radio equipment and who is the "manufacturer" responsible for compliance. A likely outcome would appear to be a need to modify sole reliance on self-certification as a way to protect against the deployment of bad devices as well as modifications to traditional hardware certification processes to accommodate the challenges of new software-intensive architectures. Additionally, as radio systems get more complex, the need to gracefully respond to systems not behaving as anticipated will be more necessary and important.Footnote 121

5.4.13 Spectrum Commons and Licence Exempt

The bulk of this report focuses on the reforms needed to support market-based allocation of exclusive-use spectrum. While this is the mode of access most desired by industry and appropriate for the majority of high-value, scarce spectrum below 3 GHz, it is not the only framework needed. Provision should also be made for spectrum commons or "public spectrum parks" that will allow shared access by so-called "unlicensed," "licence-exempt," or "class licensed" devices. In such spectrum, subject to an access protocol (which may be defined by industry), compliant devices share access rights as co-primary spectrum users. While easements provide one mode of access for such technologies, the problems with gaining political approval for such easements and the desire to relax the technical constraints associated with acceptable easements provides a strong justification for dedicating some additional spectrum for commons use below 3 GHz.

While the bulk of newly allocated or reformed commercial spectrum ought to be subject to a flexible exclusive licensing framework, a portion of spectrum also ought to be allocated as a commons. This commons will provide a safety net and cross-check on the performance of secondary markets for licensed spectrum. After transitioning to a market-based regime, we may fail to see much significant trading and may not be able to conclude from that simple observation whether there is a problem or not. If we see new entrants crowding into the unlicensed commons spectrum then that might suggest that access via the licensed regime posed barriers that may be due to market inefficiencies. Alternatively, if we see the commons spectrum relatively uncongested than we may conclude that the lack of trading reflects an equilibrium outcome. By providing wireless businesses with a diversity of regulatory options for accessing spectrum, the risks of regulatory bias are reduced. The goals of technical and service neutrality are enhanced by reserving a portion of spectrum for commons access.

Creating a commons could be accomplished as part of the auction process with a portion of the proceeds reserved to purchase public-access spectrum. Additionally, the protocol used to manage access to the commons could be modelled on the class licensing regime in Australia, the Part 15 framework in the U.S., or the licence-exempt framework in the U.K.

Finally, it is expected that between the commons and licensed models, there may be many hybrid models such as private or semi-private commons in which co-primary rights are shared among a limited number of operators. Such a model may prove useful in managing access to 3G spectrum by mobile operators that may wish to pool their spectrum. It is expected that to the extent such models prove useful, they may be implemented voluntarily by licensed providers. However, commons spectrum will provide a useful incubator and test bed for the evolution of such shared models. Licensed users may find sharing unlicensed spectrum a useful option to manage peak traffic loads and a way to economize on the need to provision strong-interference protection for only a portion of their spectrum needs.

Footnotes

Footnote 96

Markets are capable of accommodating general restrictions on anti-competitive behaviour and particular consumer protection measures

Return to footnote 96 referrer

Footnote 97

These will be recognized as the themes criteria of technical, assertive and dynamic efficiency.

Return to footnote 97 referrer

Footnote 98

This is known as the 'first basic theorem of welfare economics.' It is notable that the Policy Objective identified in the Spectrum Policy Framework for Canada (2007) takes the form of a goal 'to maximize the economic and social benefits that Canadians derive from the use of the radio spectrum resource.' p. 11.

Return to footnote 98 referrer

Footnote 99

In 1997, Jerry Hausman estimated that the ECC's delays in licensing mobile telephony cost the USA billions of dollars. J Hausman, "Valuing the effect of regulation on new services in telecommunications." Brookings Papers on Economic Activity, 1997, pp1–57.

Return to footnote 99 referrer

Footnote 100

A parallel might be the situation with respect to telecommunication liberalisation in about 1990, when several countries had opened their markets to competition with apparently favourable consequences, but the conclusive evidence in favour of the policy available today had not yet accumulated.

Return to footnote 100 referrer

Footnote 101

For a discussion of the degree to which transactions costs will deter the emergence of efficient outcomes through private bargaining and trading, see M Cave, C Doyle and W Webb, Essentials of Modern Spectrum Management. Cambridge University Press, pp.124–8.

Return to footnote 101 referrer

Footnote 102

Thomas W. Hazlett (2004), Property Rights and Wireless License Values, AEI-Brookings Joint Center Working Paper, no. 04–08, 2004.

Return to footnote 102 referrer

Footnote 103

For a discussion of the merits of this regime, see G. Faulhaber, "The future of wireless telecommunications spectrum as a critical resource," Information Economics and Policy, 2006; or, W. Lehr, "Economic Case for Dedicated Unlicensed Spectrum Below 3GHz," (2004) available at: http://itc.mit.edu/itel/docs/2004/wlehr_unlicensed_doc.pdf

Return to footnote 103 referrer

Footnote 104

Commission staff working document, Communication from the Commission to the Council, the European Parliament, the European Economic and Social Committee and the Committee of the Regions on the Review of the EU regulatory framework for electronic communications networks and services {COM(2006} 334 final), 28 June 2000, p.12.

Return to footnote 104 referrer

Footnote 105

A coverage obligation on a provider of wireless services would not normally fall into this category.

Return to footnote 105 referrer

Footnote 106

This supposition underlies recent audits of public sector spectrum use in Australia and the U.K. and President Bush's Executive memorandum to improve the efficiency of spectrum use by the federal government.

Return to footnote 106 referrer

Footnote 107

A fundamental account of this proposition is found in P. Diamond and J. Mirrlees, "Optimal taxation and public production," American Economic Review 61, pp. 8–27, 1971.

Return to footnote 107 referrer

Footnote 108

OFCOM, Spectrum Framework Review – the Public Sector. Proposals to extend market mechanisms to improve how spectrum is managed and used, July 2007.

Return to footnote 108 referrer

Footnote 109

Exclusivity refers to the fact that the licensee assigns primary rights to a single private entity. Flexibility refers to the rights that the licensee may have to change the assignment of the rights (e.g., sub-lease those rights, transfer them temporarily or wholly to another entity); the technology and uses to which the spectrum access rights are put. The PCS licences awarded at 1900 MHz in the United States to mobile telephone service providers are an example of exclusive, flexible-use licences.

Return to footnote 109 referrer

Footnote 110

Commons/Shared Access spectrum refers to the situation where multiple users share co-primary rights to access the spectrum. The rights may be restricted to a limited community (restricted/private commons) or may be open to all users who comply with the commons access regime (protocol). The 2.4 GHz ISM band used by WiFi is an example of a commons where radio access is allowed to all. As noted above, the traditional administrative method has normally included in it bands in which the commons/shared access approach has applied.

Return to footnote 110 referrer

Footnote 111

E.g., limiting choice of 3G to W-CDMA for a new band would be a mistake since this is not required to protect interference.

Return to footnote 111 referrer

Footnote 112

Focus here will be on intentional transmitters rather than unintentional radiators such as electric motors.

Return to footnote 112 referrer

Footnote 113

Spectrum caps offer one mechanism for protecting against excess aggregation of market power. If these are adopted, they should be relatively loose and should be linked to other findings of market power.

Return to footnote 113 referrer

Footnote 114

See Evan Kwerel and John Williams, A Proposal for a Rapid Transition to Market Allocation of Spectrum, OPP Working Paper Number 38, FCC, November 2002.

Return to footnote 114 referrer

Footnote 115

Absent such an approach, incumbents may hesitate to make spectrum available for fear of being unable to acquire adequate alternative spectrum; while entrants might withhold participation in the hopes of acquiring spectrum more advantageously at a later date.

Return to footnote 115 referrer

Footnote 116

See Industry Canada, CPC 2–1-23 Licensing Procedure for Spectrum Licences for Terrestrial Services, expected release date September 2007.

Return to footnote 116 referrer

Footnote 117

In contrast, consider what might happen if one GSM licensee is to switch to a wholly different technology (e.g., CDMA) in its band. The asymmetric changes in the radiated energy could cause new and previously unanticipated changes in the energy flux to adjacent licensees (in geographic areas or frequencies).

Return to footnote 117 referrer

Footnote 118

Coase (1960) demonstrated through a series of examples that bargaining among agents could lead to efficient outcomes despite the presence of externalities if private property rights were well defined and the costs of bargaining were zero. Subsequent literature has explored a variety of factors which effect whether private bargaining can eliminate inefficiencies from externalities including the nature and impact of positive transactions costs, the impact of different structures of property rights, the interaction of taxes and bargaining, the possibility that some participants possess private information, the effect of nonconvexities, and the implications of endogenous participation.

Return to footnote 118 referrer

Footnote 119

See http://www.cantor.com/demos/9036512.html for further information about their system.

Return to footnote 119 referrer

Footnote 120

See, for example, Analysys (2004), "Study on conditions and options in introducing secondary trading of radio spectrum in the European Community, Annexes to Report," Final Report to the European Commission, May 2004.

Return to footnote 120 referrer

Footnote 121

With complex systems, it is not possible to exhaustively test all possible failure modes without prohibitive expenditures on ex ante testing. Thus, there need to be systems that minimize the risks of harm should an unforeseen and harmful behaviour result. See, for example, John Chapin and William Lehr (2007), "Time-limited Leases for Innovative Radios," IEEE Communications Magazine, June 2007.

Return to footnote 121 referrer