Statement issued by the Direction General of Telecommunications

28 September 2001

Contents

Chapter 1 Introduction and summary

Chapter 2 The current spectrum management regime

Chapter 3 Developing a spectrum trading environment

Chapter 4 Conclusions

Annex 1 Response to questions

Annex 2 Spectrum management

Chapter 1

Introduction and summary

1.1 Oftel welcomes the opportunity to respond to this very important and fundamental review of radio spectrum in the UK. Such a review is timely given the major changes in digital technology which have massively increased the potential importance of radio spectrum to the UK economy. These changes make it vital that the outcome of this review forms a key input to the legislation to establish OFCOM as the regulator for a converging telecommunications, internet and broadcasting markets.

1.2 Markets for electronic communications services are part of a rapidly changing, fast moving world. The prospects are for the rate of change to increase affecting all stages and inputs to the process for delivering these services.

1.3 Developments in digital communications technology means the same or equivalent products and services can increasingly be delivered over different electronic networks – based on radio or wire technologies. This means such products and services previously sold and delivered through one network are available through several networks, bringing both the products and services and the wire and radio delivery networks into competition with each other. The increase in network flexibility to offer different services from digitalisation, and the increased competition it can bring therefore represents a very important opportunity for consumers to get a better deal from electronic communications services – from new broadband services, to lower prices, to higher quality of service.

1.4 To achieve the change to more competitive digital communications, there needs to be radically improved access to radio spectrum. Greater access to spectrum for operators/suppliers enables the promotion of competition through new entrants to the market challenging existing players to serve customers.

1.5 The step change in the flexibility and usefulness of spectrum and the demand for spectrum resulting from the ‘digital revolution’ has profound implications for the role of regulation. It means that the focus of existing spectrum management which is primarily, on issues of interference, standards, and licensing – and which has been appropriate in the past – will no longer be the most appropriate focus of regulation. Instead the prime need will be for rapid access to spectrum and flexibility in use of spectrum to meet continually changing consumer demands. If current restrictions on access to spectrum are maintained, they will act as a barrier to entry with the effect of denying consumers the benefits of the new digital world.

1.6 The removal of barriers to entry is, and has been, a critical mechanism by which the Director General of Telecommunications (DGT) has pursued his statutory duty under the Telecommunications Act 1984 to promote competition in telecoms. The DGT also has a statutory duty to advise the Secretary of State on competition issues in relation to radio spectrum. The proposed new EU regulatory framework will also allow flexible forms of spectrum trading to be introduced for the first time, which will be key in developing the changes in spectrum management that are required to make Britain the best place to do e-commerce. The outcome of this review, and the opportunity for legislation under the Communications Bill therefore represents a good opportunity to establish the policy and strategic framework of the next decade. This opportunity will not be repeated.

1.7 Recognising the changes in electronic communications markets , the Government has proposed the setting up of a regulator , OFCOM, to facilitate the move towards convergence. A key function for OFCOM will be decisions on the use of radio spectrum.

1.8 The need for far-reaching changes in the way access to spectrum is managed is therefore essential. Making the most effective use of radio spectrum for the benefit of consumers is a key driver for the development of converging telecoms, Internet access and broadcasting services.

1.9 The current practice is to manage spectrum on a largely centrally planned basis with very limited scope for market mechanisms to work. This approach will not be able to respond effectively to the much greater demands that will be placed on it and, if it is not replaced, will act as a barrier both to entry and to the development of competition.

1.10 The need to meet consumer demands will be best met through the widespread use of market mechanism ie spectrum trading. Oftel recognises that changing the basis on which spectrum is used to a market based approach will be a difficult task and will not be readily achieved in a short timescale. However the potential benefits are so great that the work must be given a high priority. Issues of standards and interference will not disappear under spectrum trading but such issues need to be addressed in an innovative way that is consistent with effective spectrum trading to ensure problems can be overcome.

1.11 Market trading mechanisms will therefore be needed to free up the use of spectrum to meet consumer demands. The main elements needed are summarised below:

• auctioning off existing unallocated or unassigned spectrum to the highest bidder when a selection process is required (subject to appropriate constraints to address any competition concerns);

• spectrum users deciding what use spectrum is put to, as far as possible, within a defined spectrum property right;

• allowing public policy users of spectrum (such as MoD and broadcasters) to sell off or lease their excess capacity, and keep a substantial part or all of the funds raised – so such users have an incentive both to only use the spectrum they need and to enhance their ability to fund new services;

• allowing spectrum to be re-sold by those who have acquired it (subject to any possible competition concerns), again without constraining the use to which the spectrum is put subject to conditions set out in a spectrum property right;

• limiting restrictions on use to issues of preventing spectrum interference, international treaty obligations, and national security and encouraging voluntary rather than mandatory standard setting;

• pressing for greater flexibility in international agreements over spectrum use; and

• publishing a challenging yet achievable timetable, with milestones, as to when this regime should be introduced.

1.12 The review commissioned by the DTI and Treasury provides the opportunity to establish an appropriate policy base for the above actions to be taken. This can be done by feeding into the proposed new OFCOM Bill, and any future work in the setting up of OFCOM arising from such legislation. The creation of OFCOM in combination with new European legislation gives Government a unique and timely opportunity to achieve substantial increases in the level of competition in electronic communications services for the benefit of UK consumers.

1.13 Oftel’s response to the review is structured as follows :

Chapter 2 analyses the current basis for spectrum management and its limitations in a digital world.

Chapter 3 sets out the key steps involved in moving to a spectrum trading environment.

Chapter 4 summarises Oftel’s view on the need for spectrum trading in combination with auctions.

Annex 1 gives Oftel’s response to Review questions which fall within its remit of promoting competition.

Annex 2 sets out in more detail some of the issues in relation to spectrum management.

Annex 3 reproduces an analysis commissioned by Oftel of the economic issues in relation to spectrum trading.

1.14 Oftel has sought to provide a substantive response to the Review, reflecting the importance of spectrum management in general, and spectrum trading in particular, to the future of electronic communications services and consumers. Clearly there is more work needed to develop the policy foundations to underpin spectrum trading. To meet its statutory duties to promote competition Oftel will therefore continue to participate in this work with the Radiocommunications Agency (RA) and other relevant parties in the run up to OFCOM.

Chapter 2

The current spectrum management regime

Strengths and weaknesses

2.1 Oftel agrees that the current regime of centrally managed spectrum has worked well in the past, as shown by the durability of the 1949 Wireless Telegraphy (WT) Act, which has had relatively few modifications until recently. However the economic importance and the demand for spectrum is now such that it is becoming increasingly difficult to cater for all reasonable demands for spectrum, using the existing regime. This was recognised by the implementation of the 1998 WT Act. The 1998 WT Act gave the RA the powers to auction spectrum and to charge for other licences at a level which was not dependent on the cost of administering such licences, as was the case before the 1998 Act. Oftel welcomed the move to a more market based approach that the 1998 Act allowed.

2.2 A clear line can be drawn from the RA’s Future Management of the Radio Spectrum in 1994, through to Spectrum Management into the 21^st Century in 1996, to the 1998 WT Act. In 1998 the RA also published Managing Spectrum through the Market, which received a positive response to spectrum trading, and served as a platform to seek regulatory changes at a European level, to allow spectrum trading in the UK. In April 2000 the UK RA was also the first administration to hold a 3G auction. These actions can be seen as a major endorsement of the policy of using market based mechanism to manage radio spectrum.

2.3 Oftel welcomes the use of auctions by Government to assign radio spectrum. However auctions do not reveal, if the purpose for which the spectrum is designated is best (the allocation), if the use to which the spectrum can be put is mandated (which is the current case). This stage of deciding what the allocation (ie used for mobile or broadcasting say) for a particular radio band is, is still decided by the National Radio Authority (NRA), based on its judgement (after consultation with industry).

Administrative pricing

2.4 Under the current system of administrative pricing there is an inherent difficulty in a regulator determining the true commercial value of spectrum for a particular service, and charging on this basis. This is the difficulty at arriving at such a calculation given the complex technical and business variables involved in deploying modern radio communication services. The current administrative pricing regime also does not tell us if the service currently operating in the band is the most valuable one, as it uses the value of spectrum to the incumbent service, not the service that would value the spectrum most. This does not encourage spectrum to flow from low value uses to high value ones. The ability to ensure spectrum is used for the most valuable purpose is key to economic efficiency, and this requires a flexible way of the market deciding what spectrum is used for, and by whom. Those that value spectrum the most will be those who will offer the highest value services to consumers (either directly or to a service that is required by others to offer services to consumers). Prices are a vital element in the mechanism by which economic efficiency can be achieved. If prices are somehow set at the wrong level, the market receives the wrong signals as to how much of a service should be produced.

2.5 Even assuming that the NRA correctly determines what the most appropriate use for a given band should be, this is still only a static snapshot. That is to say it may well change with time, and the ability to change use is very slow under the current process. It is this problem of change of use of licence conditions that needs to be addressed, and cannot be efficiently done under the current regime. This lack of ability for users to determine what service should be provided can lead to problems associated with 'licence scarcity', rather than actual spectrum scarcity. The value of a spectrum licence may be artificially high because the amount of spectrum allocated to a particular use is artificially restricted, or artificially low because the services which operators are allowed to offer under the terms of the WT Act licence are not those on which users place the most value (discussed in more detail below).

2.6 An added difficulty is that the current system does not encourage users to share spectrum. For example a broadcaster may be able to share spectrum with a commercial user in certain areas. Such sharing arrangements may involve a cost (either from a co-ordination overhead, or some re-engineering of radio sites), which currently the broadcaster could not recover, because they cannot charge such a commercial user. It seems clear that trading and allowing change of use will incentivise such commercial deals. This can only benefit consumers in making spectrum more easily available to producers of radio based services.

Spectrum flowing from low to high value services

2.7 The current regime does not encourage users of radio spectrum to fundamentally review their strategic options with regard to the asset of radio spectrum. For example a public user of spectrum might be assigned 100 MHz of spectrum which it uses fully. Over time it may be that the user could, with extra cash, use less spectrum by investing in new equipment. A trading environment would encourage this, as the public user would be able to sell or lease this excess spectrum. The current regime involves long negotiations as to how much spectrum should be handed over and when. Without the incentive of profit (or looked at another way opportunity cost by forgoing that profit) such efficiencies, and the ability to offer new and innovative services to market more rapidly, are lost. One could argue that the current regime may actually encourage such hoarding behaviour, if the administrative pricing tariff is not set sufficiently high. This is especially true if the administrative pricing is based on the value to the existing user, and not on the price placed on it by the highest value service (ie if the spectrum could be used for any service, what would the most valuable one pay?).

2.8 The central allocation system makes the free flow of the vital spectrum resource slower than it might otherwise be. Government needs to consult widely on what spectrum should be used for, and the process of allocation needs to be consulted on also. Such consultation processes pose potential difficulties, as it can be hard to balance conflicting needs and wants of radio users. This can be further complicated in that individual organisations are incentivised to say what they think is most likely to get them the desired allocation or assignment. In the new fast moving environment of telecommunications, the speed with which such decisions may be required, and the potential complexity of them from technological and business viewpoints, makes such a central process increasingly difficult. The requirement for speed of process was recognised in the RTTE Directive regarding type approval of telecommunications devices (including radio), which was felt to be too long under the previous regime. Oftel believes that these difficulties with the current regime could disadvantage consumers in terms of reducing choice and competition.

Spectrum efficiency (technical versus economic)

2.9 Oftel’s view is that one of the key issues that has been highlighted in the questions raised by the Review is the definition of spectrum efficiency. Oftel agrees that it should relate to economic efficiency, and that the technical definition is a secondary issue. The technical definition can be thought of (crudely) as getting the maximum number of users of any given service in any particular area per unit of spectrum. This technical metric has tended to be important when spectrum was granted on a first come first served basis, or via a 'beauty contest'. In a market based scenario, users of spectrum are able to balance the extra costs that may be associated with acquiring more spectrum, with the extra costs that may be associated with engineering their radio systems to offer the same capacity with less spectrum.

2.10 Oftel would argue that it is the economic benefit that needs to be considered, as this is what will benefit consumers, as opposed to how intensively the spectrum is used as related to any technical efficiency measure. Economic efficiency occurs when resources are allocated in an optimal way so that the resources of the economy are employed in producing goods and services of the maximum possible value to consumers. If undue preference is given to technical spectrum efficiency this may lead to opportunities to create wealth (and hence benefit to consumers) being reduced or lost. This economic benefit is sometimes expressed as the sum of benefits created for consumers and producers.

Spectrum efficiency and OFCOM’s statutory duties

2.11 It should be incumbent on OFCOM to manage the radio spectrum in the best interests of all consumers/citizens, not just the interests of particular individual or groups of radio users. It is however debatable whether there needs to be any separate specific duty in relation to the spectrum in addition to those duties enunciated in the White Paper. Those duties were formulated to provide a general framework for the whole of the communications sector including the radio spectrum. If it is considered that these duties are not sufficiently wide to ensure the best use of the radio spectrum, any statutory duty should focus on the economic efficiency as opposed to the technical efficiency of radio spectrum usage. Economic efficiency involves maximising the benefits generated from the allocation and use of spectrum (these benefits are not only about wealth maximisation but can include defined social goals). This duty contrasts with one that would focus on allocating/using spectrum efficiently from an engineering/technical perspective, the ultimate objective of which is to maximise the number of users of any given part of the radio spectrum.

2.12 Technical spectrum efficiency is not an end in itself and is only one of the factors that needs to be addressed in managing the radio spectrum. It is the value or wealth created by the service that needs to be the primary consideration, which can also be applied to public policy goals. It is not economically efficient to give spectrum to the most technically spectrum efficient system if it has no customers. Better a system which uses more spectrum (perhaps because such a system is cheaper to build), that offers service to many customers. If too much weight is attached to technical spectrum efficiency then the potential benefits of services of high value to consumers may be lost because it does not meet the technical spectrum efficiency metric of other services of lower value to consumers.

Spectrum scarcity or licence scarcity?

2.13 Without a dynamic trading process for spectrum, ie one that allows change of use within your spectrum property right, one may see distortions in valuations placed on spectrum. For instance, the UK 3G auction raised a substantial amount (£22.5 billion), but it is not clear whether similar sums would have been raised had it been possible to provide 3G services using spectrum outside the designated bands, or whether instead the amounts raised reflect an artificial restriction on the number of licences. In a traded environment with flexible use, other possibilities such as refarming, or using nearby non-core spectrum, or waiting for extra spectrum to be internationally identified, are not subject to regulatory uncertainty.

2.14 This suggests that under the current regime it is not the spectrum itself that is valued, but the service for which the spectrum is licensed. This can lead to distortions which may impact on UK competitiveness, and the cost and timely availability of services to UK consumers. This was highlighted in the Consultation, which shows a wide variation in valuations of units of spectrum, from £1k per MHz to £21,000k. This may to some extent be due to public policy goal requirements for certain service, but without a market evaluation of the true costs of such goals it is hard to determine if this is an appropriate use of radio spectrum.

2.15 This problem of licence scarcity is a direct result of the difficulty radio users have in changing the designated use of their radio spectrum. Thus they cannot follow market patterns in the use to which they put spectrum. Telecommunications markets can be fast moving, with new information entering the market place, which might modify the behaviour of participants. The high values attached to some parts of the spectrum are likely to be a signal that more spectrum should be allocated to the uses in question – but this is prevented under the current regime. This can give rise to significant inefficiencies in the UK economy. Even if the figure for this loss due to inefficiency is low now, it can only be expected to rise in the future in the absence of trading. This is due to the ever increasing demand for spectrum, and the ever increasing rate of change of radio technology, which requires spectrum use to change rapidly also.

2.16 A flexible spectrum trading environment would make access to spectrum in bands where equipment is available at a cheap enough price to provide a commercial proposition, the main consideration. This could reduce the cost of entering the market that might be faced today, with spectrum use constrained to particular services and standards. The current regime tends to mean that higher value services are not allowed to compete with lower value ones for spectrum, but have to compete with operators wanting to offer the same high value service. This artefact of the current regime would be ameliorated in a flexible spectrum trading environment.

Chapter 3

Developing a spectrum trading environment

3.1 It is Oftel’s experience that markets are the most efficient way of allocating/assigning scarce resources, and that in principle spectrum should be treated much like any other key input and its value determined by the operation of market forces. However, Oftel recognises that in certain circumstances there may be market failures that need to be corrected, and it is for this reason that regulators exist.

3.2 Oftel agrees with the sentiments expressed in the consultation document, that only a system which is more market based will deal with issues of speed and complexity that are facing NRAs today, and that such problems will only be exacerbated in the longer term as the demand for spectrum increases. Oftel believes that spectrum should be considered in no different a manner than any other scarce resource, that must be valued before being assigned to public policy goals. It is to that end that Oftel would suggest moving to a system that uses a mixture of auctions and flexible spectrum trading (ie allows change of use). Central to realising the benefits such a system could have to consumers in terms of promoting competition and choice is in defining the spectrum property right in such a way, as it allows flexible trading. It will also be necessary to ensure the minimum number of constraints on this trading environment, as these constraints will add to the overall cost of transactions and limit the benefits of trading.

3.3 To yield such benefits in promoting competition and choice, and so benefit consumers, will require that the following issues be addressed:

These issues are discussed below.

The use of auctions to assign spectrum (issue 1)

3.4 Auctions have the advantage of being transparent and easy to award (unlike the complex process of deciding a beauty contest winner), and ensure (all other things being equal) that the spectrum goes to the company that values the spectrum the most, and can hence make the best use of it in economic terms. This should be the most efficient use of the spectrum, which will benefit consumers who will have access to the services they most value.

3.5 Oftel considers that while spectrum auctions deliver a fair and transparent result in ensuring that those that value the spectrum most get it, this is a static valuation. That is to say it may change with time, in which case some other market mechanism is required to ensure radio spectrum is efficiently allocated or assigned, or bidders may have miscalculated. Spectrum trading is rightly identified as a key issue to ensure that spectrum is managed in the best way for the benefits of all (including consumers). We should rely on the ability of the market to select what spectrum is used for, and what service runs in that band as far as is possible. It is this combination of spectrum auctions and flexible trading that will deliver the best solution for the UK as a whole (including consumers).

3.6 Oftel does not believe that auctions raise the prices of goods and services to consumers as they are a sunk cost. For example, if one rents out a house, the rent charged must be the market rate, ie comparable to similar properties in that area, and is not related to what it cost. Similarly, the price charged for a service like mobile telephony will reflect the degree of competition in the market for that service. The amount which a mobile operator would be prepared to bid in a spectrum auction would then reflect expected future levels of prices and profits, and the possibility of exploiting any market power. Competition law of course exists to tackle excessive market prices.

3.7 It seems clear that the freer the spectrum market for trading is, the more efficient it will be, and the greater the benefits delivered will be. However, there are some important issues that need to be addressed such as ensuring sufficient spectrum remains for critical public services, such as defence and emergency services, as well as for other public services such as broadcasting. This is discussed in more detail below (Public Policy uses of radio spectrum).

Spectrum trading (issue 2)

3.8 Currently spectrum licences are for individual transmitters, which have to be notified to the RA. In some cases the sheer number makes this impractical and this is done via an exemption such as for GSM handsets, or for certain types of low power devices operating in deregulated spectrum bands. For the full benefits of a spectrum market to be realised will require a move away from an apparatus licensing regime to one that is based on a spectrum property right. That is where the owner of such a property right licence has the ability to use transmitting equipment which fits within the bounds of that property right. If individual radio transmitters need to be licenced, or an amendment is needed to a Statutory Instrument when users want to change the use of their spectrum, then this will be a significant transactional cost, which may mean that the services of highest value to consumers are not delivered due to this cost. It may be possible to issue spectrum property right licences using the existing WT Act (1949), by having licences that do not refer to particular radio equipment types. But the concern would remain about the flexibility of such a system, if change of use required modifications to a Statutory Instrument. The important issue is that spectrum licences have the properties that allow flexible spectrum trading, and do so with regulatory certainty and transparency.

3.9 Spectrum trading will allow spectrum to be acquired more quickly. Markets will decide on the number of players not central administrations, and will give operators the flexibility to perhaps initially make smaller investments in spectrum, as opposed to the large one required in say 3G to ensure being able to enter the market for the next 5 years or so. A spectrum trading environment would lessen the risk in entering such markets for new entrants. It would also allow them (and incumbents) to grow their business by acquiring spectrum as their demand increased in a shorter time frame than they currently do, as they are currently forced to go to a ‘pot’ administered by central government.

3.10 It has been suggested that a market for radio spectrum trading might be inefficient due to holders of key pieces of small spectrum holding such deals ‘to ransom’. However, this is no different from deals such as land say, and can and are resolved by the market except in very exceptional circumstances, where some power such as compulsory purchase may be required. The advantage would be that operators would be free to use spectrum from another part of the band. At the end of the day any commercial deal depends to a large extent on the relative strengths of the negotiating positions of the parties involved, and the information they hold about each other.

Public Policy uses of radio spectrum (issue 3)

3.11 Under the current regime spectrum is apportioned to such users as defence, and broadcasting, as well as services such as radio astronomy, because of the public interest in having these services. Without support from central Government such services in the form we know them would not exist. In moving to a spectrum trading environment it is legitimate to ask what will happen to such services. Oftel believes that a spectrum trading environment will reveal the real cost to the UK economy of supporting such goals and that spectrum should be treated no differently to any other scarce resource required by organisations carrying out public policy goals, such as land, or staff, or IT. That is, the cost of delivering such services should include the spectrum cost.

3.12 A possible scenario is that such users keep their current spectrum to provide the services they are required to do under their individual mission plans. If such organisations were able to trade their spectrum to commercial organisations (which would require change of use permission or property rights), then they would be able to determine how much spectrum was worth to them. That is not to say that they would instantly sell off all their spectrum, as then they would no longer be able to carry out their core mandated functions. In the case of users such as the MOD, as a Government Department, they would presumably require the approval of the Secretary of State to sell spectrum. However, they would be able to decide if, with an injection of cash, they might require less spectrum, or if they could share certain spectrum with commercial users under given conditions (say take back clause in times of national emergency for MOD). The benefit of such a system would be that the onus would be on the current users of such spectrum to decide how much spectrum they need, not on a third party such as an NRA.

3.13 It is this functioning of opportunity cost that is probably the main driver that would force non-commercial users to use spectrum efficiently in an economic sense. Facing up to the opportunity cost of hanging on to spectrum that they no longer really require for their core function, or identifying a service perhaps they could purchase from a commercial organisation, should help ensure such users fundamentally review their need for spectrum. Such an ability to trade spectrum in this manner may also add useful revenues to such users without affecting their ability to deliver their core functions. This could apply to users such as broadcasters who may be able to share with certain types of other users (say on site professional radio). Such sharing would not be costless and may require investment in new planning tools, or re-engineering of radio sites. This cost would need to be offset by the price a commercial user would be prepared to pay.

3.14 If such users required extra spectrum then it would either need to be acquired by purchasing it on the open market, or compulsorily purchased (with suitable compensation). This is akin to other assets such as land which may be required for public policy purposes (for defence, or transport infrastructure). This would make the cost impact of the requirement more transparent, as the public monies required to purchase spectrum could presumably only be made available after some form of cost benefit analysis. The current situation of making spectrum available from a centrally controlled pool is a very blunt instrument for carrying out public policy goals.

3.15 A trading environment would not mean an end to public policy use of spectrum. It would mean that such goals would need to be costed as the market would reveal the opportunity cost of spectrum (or compensation would be required to purchase new spectrum). This however is the same for all other resources, and is reflected by the fact that all Government departments are required to operate within given financial budgets, that are constrained by tax revenue and the level of Government borrowing.

3.16 Freeing up spectrum to commercial use in this way has the potential to lead to a growth in the UK economy (due to efficiency gains), which will make more tax revenue available to spend on public policy objectives. For opportunity cost to be an effective incentive to encourage more economically efficient use of the radio spectrum, will require that public policy operators are allowed to keep a substantial part of any monies raised in such selling or leasing of excess spectrum. If too much is clawed-back using some form of windfall tax the incentive to trade will be reduced or lost, and hence the greater benefit of spectrum used for its optimum economic purpose is lost. Similar reasoning could be applied to commercial users of spectrum who have not paid the full economic value for their spectrum licences (ie awarded spectrum via a beauty contest or first come first served) .

3.17 Spectrum is an asset and as such should be accounted for as any other public asset is. This would be consistent with resource accounting procedures for Government assets.

Spectrum property rights (issue 4)

3.18 Central to the functioning of any market will be the definition of the property right traded, in this case radio spectrum. The right should be defined in such a way so that the geographic boundaries of the right, as well as the maximum power of the transmitter, and the shape of the emission outside the given piece of spectrum, can be easily enforced. This is important so that the cost associated with trades is reduced. Uncertainty about how such a property right is defined has the potential to significantly increase the complexity of such deals (and hence their cost). The owner of such a right should be free to sell such a right on, as well as be able to lease out such spectrum to others, and use it for any service that fits within that property right.

3.19 A spectrum property right would then have the characteristics and terms of:

• A geographic area/volume
• A frequency range of use
• The acceptable Electromagnetic Compatibility (EMC) emission limit it is able to radiate outside its spectrum right
• The name of the owner of the legal title that is lodged in a central register administered by or on behalf of OFCOM
• A general proviso not to breach international treaties to which the UK is bound
• Provisos to cease transmission if safety of life or national security issues are raised (with compensation if the user is operating within their licence terms)
• A requirement that the sale of licences to users with dominance/SMP in relevant markets (as listed before hand by OFCOM) needs prior approval

3.20 The above bullets would seem to cover the main issues that could be used to define a spectrum property right. Once a property right is clearly defined the market should deliver the best outcome, even when negative externalities (see Annex 3) are present. That is in the case of radio spectrum the actions of one user can have a negative impact on another user in a separate frequency band. If a clearly defined property right for spectrum is defined, then a spectrum market will yield an efficient outcome (and would remove the problem associated with spectrum negative externality). This is supported by the Coase Theorem (see Annex 3 for background economic analysis), which was specifically developed for the case of radio interference.

Arbitration, enforcement, and the central spectrum register (issues 5 – 7)

3.21 As well as a clearly defined property right there would need to be a system of enforcement, and a central registry of ownership and subsequent transfers of ownership. The central register would be very similar to the land registry in recording the ownership of such licences. This would help provide certainty to buyers regarding legal title, it would also help potential buyers determine who to contact regarding pieces of spectrum. With regard to enforcement one can conceive of three main types. The first type an adjudication function, which could be set up to resolve the impact of interference on neighbours and calculate the appropriate recompense. This could perhaps be a voluntary arbitration service run by or on behalf of OFCOM. Such an arbitration body could act as a backstop to encourage issues to be resolved by the parties involved.

3.22 The second type would be to deal with ‘pirate’ broadcasters, who have no intention of operating within the law, and seek to transmit covertly and without notifying any authority or other legitimate users of the radio spectrum. They would have no form of spectrum rights in the area/spectrum they broadcast. Such transgressions would require a more formal enforcement role as undertaken now by the RA’s Radio Investigation Service.

3.23 The third type of enforcement might be to enforce regulations regarding UK international treaty obligations (eg change of use not compatible with these obligations), or enforcement of competition regulation. It might also include breaches such as not keeping the central spectrum property register up to date. These type of breaches are characterised by non-compliance with terms required by the NRA to allow flexible spectrum trading to function, as opposed to breach of contract between private parties.

Competition issues (issue 8)

3.24 Under the current regime spectrum trading is not allowed, which means that spectrum cannot be directly sold and bought by companies. However, it is possible to sell companies that own a spectrum licence, or ask the Secretary of State to transfer the ownership of a spectrum licence. If a spectrum licence is transferred via the purchase of a company then it may become the subject of merger regulation where any potential competition issues can be addressed. In a spectrum trading environment, where spectrum may be sold by itself, merger regulation will not necessarily apply, because two enterprises do not need to merge to transfer spectrum property rights.

3.25 In such a spectrum trading environment consideration needs to be given as to whether any extra regulation over and above the standard competition rules that apply, need to be implemented. Any such regulation will need to be crafted in such a manner so as to ensure that competition aspects that may arise due to spectrum are properly addressed, but that the regulation is not overly burdensome, and hence reduces the benefits that might accrue to consumers and competition from such trades.

3.26 One possible solution would be to require that any operator who is designated as having dominance / significant market power (SMP) by OFCOM, has to notify OFCOM of the intent to purchase spectrum. OFCOM would then have the opportunity to either approve such deals or not. One possibility is to give OFCOM a time limit of one month say, after which a deal is deemed to be approved if OFCOM raises no objection.

3.27 OFCOM’s process for approval of such deals will need further consideration. Any process of analysis should be based on previously published guidelines, that have been consulted on that take into account the current state of the market, and the relative importance of the spectrum being sold. Such an analysis could be similar to that undertaken in mergers to see if allowing such a purchase to go through would be in consumers interests and not anti-competitive.

3.28 In the past some spectrum licences have been awarded on favourable terms to encourage extra competition. In moving to a new spectrum trading regime, some consideration should be made as to if and how such pro-competition terms should be preserved. One such example under the current regime was the 3G licence reserved for a new entrant, this licence was larger than the others (ie had more spectrum). It might for example in the 3G case it might be specified that Hutchison 3G could only sell its spectrum to another new entrant. However this would need to be explored in more detail.

3.29 Any detailed conclusions will need to await further consultation on spectrum trading rules, and would be subject to views expressed in that process. Oftel’s starting position would be that it would seek to keep the necessary safeguards to ensure spectrum trading did not undermine the competitiveness of the market, by having the appropriate level of regulation.

International regulation and its constraints on trading (issue 9)

3.30 Due to the international nature of radio it is necessary for administrations to meet regularly to decide on how to allocate/assign spectrum to particular services (mobile, fixed, broadcasting, etc) and on what terms spectrum can be used (exclusive, shared). It may also set maximum powers and type of radio emissions that can be used. The main fora for such negotiations are the ITU – Radio sector, and the CEPT for Europe. European Community Decisions also need to be considered, which primarily focus on encouraging the use of certain standards, such as those for GSM and UMTS. The main purpose of such harmonisation is to either reduce interference, and/or to encourage particular radio systems.

3.31 Normally the ITU level concentrates on the least restrictive option to minimise radio interference (and usually allows any use as long as no significant interference is caused to the protected service). Many such international constraints relate to types of service that will have ‘primary protection’, that is the maximum level of interference such services can expect from neighbouring countries. This level will depend on what services are being used in those countries. For example a band may have a primary use of broadcasting, and a secondary use of mobile. The broadcast users in this scenario will get more protection from mobile users than vice versa, as they are primary users.

3.32 Some international agreements relate to encouraging particular standards to benefit consumers from the ability to roam and/or economies of scale in production. There are different levels of harmonisation that can be required, going from mandating a particular standard (GSM say – which interestingly is the de facto world 2G standard without any ITU allocation) to having only the requirement that it be used for a particular type of service, mobile , fixed, satellite, aeronautical etc.

3.33 It would seem that the flexibility the UK has to allow trading with change of use is reasonably high. The normal international treaty obligations allow for uses of spectrum to occur on non-interference basis. This means that the UK is not normally treaty bound to use the services dictated by international regulation. In any case the classifications are normally quite broad (mobile, fixed, broadcasting etc), and there is always the possibility of having UK specific footnotes to give the UK greater flexibility if required. Oftel agrees with the Review’s foreword, that policies should be recommended which seek to progressively minimise such constraints through unilateral and international action.

3.34 The UK RA has a well respected international section, who with their good offices, should be able to influence future debates, to allow modifications to regulations if these are required. A practical scenario might be that an owner of spectrum decides to change its use to something not listed in international regulations. Such a service could begin on a non-protected basis, as long as it does not cause interference to services used in other countries that are listed. This could be an interim situation until the NRA was able to negotiate changes to regulations to allow such changes (with suitable inputs from the UK spectrum owner concerned).

3.35 OFCOM will need to ensure that as far as possible the spectrum choices that are best for the UK are allowable under the framework of international agreements to which the UK is treaty bound. These decisions should also encourage other administrations to adopt similar services as those used in the UK so that greater economies of scale can be achieved in radio equipment.

3.36 In some cases the UK may be required to use particular standards, such as via EC or CEPT Decisions. In some cases even where a detailed standard is used, such as for 3G, the spectrum may only be identified for use by such services by the ITU. That is to say not all countries need to make spectrum available for such services. Some may choose to do so in all or part of the bands identified, others not to do so at all. In fact the ITU ‘family’ of standards for 3G includes a number of different radio technologies, including DECT, UMTS, and CDMA2000.

3.37 As long as individual radio users were made aware of any such treaty obligation constraints, they could themselves determine what impact this has on the value of their spectrum when trading it, and how to best engineer their radio systems to overcome problems such as causing harmful interference to our continental neighbours. Whilst international regulations will to some extent be a restriction, spectrum users following market principles will still deliver the most economically efficient outcome. A restricted market is better than no market.

Spectrum trading and standards (issue 10)

3.38 One of the arguments used against allowing spectrum trading with change of use is the benefit to strategic trade policy of mandating standards for certain frequency bands. The importance of standardisation for certain radio services has become clear with the introductions of services such as GSM (2^nd generation digital mobile services). Both consumers and operators have gained from economies of scale from common standards and the ability to roam (use foreign networks when abroad). It appears to Oftel however that such reasoning would only apply to a small number of services, and in any case that the market could be relied upon to use the most appropriate equipment in spectrum they ‘own’, as discussed below. As with all services, to be commercially viable, it will require that operators are able to purchase equipment from manufacturers at prices low enough to offer services at prices consumers are willing to pay.

3.39 Oftel would argue however that it is not the mandating of standards for certain spectrum bands that has led to this success, but commercial self interest of operators and manufactures, who realise that economies of scale of lower unit equipment costs, which boosts market sizes. However, it is clear that mandating only one standard can cause legal problems and be regarded as a technical barrier to trade, as was suggested with regards to the Council Decision on UMTS, which selected one particular variant of the ITU IMT-2000 family of 3G standards. It is also somewhat axiomatic that regulators will find deciding between such standards almost impossible, unless there is a clear steer from industry. If such a clear position exists the market will follow that steer. If there is not, then regulators are in an impossible position, and are likely to be no better placed than market participants (and normally worse placed).

3.40 As is discussed in Annex 2 and Annex 3, mandating of standards is not without costs from the overhead of agreeing standards, and in the possible restriction of innovation. It should be left to individual organisations to participate in standards fora, and to adopt such standards in spectrum they have voluntarily agreed. Operators and manufacturers will participate in standards fora if they see the commercial benefit of doing so. Even now, normally, individual manufacturers cannot know how successful they will be in winning contracts for such equipment once a standard is agreed, or indeed how lucrative participating in standards will be (ie if their IPR is selected to form part of the standard). The same may hold true for operators who may not yet have acquired such spectrum via an open competition. The main driver for standards is probably when operators have been awarded spectrum, and this is more quickly done via trading. It would allow spectrum to be acquired more easily and rapidly than via a process that requires NRAs to consult on making bands available (and clearing incumbents), and on open and transparent award procedures. Such processes under the current regime can take a number of years, and introduce a great amount of uncertainty.

3.41 Mandating spectrum for standards may be required in some very exceptional circumstances, but in the main, the benefits will flow from operators and manufactures getting together where things such as roaming are required, or economies of scale can be realised, which outweigh the extra burden associated in agreeing standards; and the possible limiting effect they have on innovation and new services.

3.42 The RTTE also makes clear that "Member States may restrict the putting into service of radio equipment only for reasons related to the effective and appropriate use of the radio spectrum, avoidance of harmful interference or matters relating to public health." This seems to make the case for mandating standards for particular frequency bands less viable, and while recognising that harmonised standards are of benefit, they should be voluntary, not mandatory. This issue is more fully explored in ERC Report 102, which concludes "-- reference to standards may cause problems with WTO rules, as well as with EU legislation, if not properly justified. Therefore there is a need to justify each of its provisions by at least one essential requirement/legitimate objective", it also makes the point that only those parts of the standard that relate to this essential requirement can be mandated. It is clear that the mandating of standards for certain frequency bands must be "without prejudice to the possible use of such frequency bands for other similar systems with similar performance capabilities". This would seem to imply that as long as undue interference was not being caused, there should be very little constraint on what spectrum users should be allowed to do with their spectrum, except in very exceptional circumstances, which can be objectively justified.

3.43 A flexible spectrum trading environment will not mean that operators will be able to offer any service in any frequency band, because of the requirement to offer commercially viable and competitive services. Some bands will be more suitable for certain types of services, due to propagation characteristics, etc. Also economies of scale may make bands too far from ‘core bands’ for certain standards too expensive to offer services. Thus the extra cost of making equipment outside ‘normal’ bands for some services, will act as a constraint on what services can be offered at a commercially viable rate. However, market participants are best placed to make such a judgement, and this is likely to change over time. The rate of change is likely to increase with the ever faster speed of technology development.

Chapter 4

Conclusions

4.1 Any future spectrum management regime ought to have promoting economic efficiency of the use of spectrum as its primary aim as this will benefit consumers. This does not mean that only commercial users will be catered for, as public policy goals can still be incorporated within an economic framework. This is done by making Government aware of the opportunity cost of public policy uses of spectrum. This would be consistent with resource accounting procedures for Government assets. For instance it would not make sense reserving spectrum worth £10 million for a service whose value to the UK was estimated by Government to be worth only £5 million.

4.2 A combination of auctions to get spectrum into the private sector (if a selection process is required), and flexible spectrum trading (with the ability to change the service that can be provided in the spectrum) will give the best solution for the UK in terms of competitive advantage, and benefits for consumers. It will do this by encouraging spectrum to more freely flow from low value uses to higher value ones.

4.3 Public service users and Government users (ie broadcasters and Defence etc) should be allowed to trade their spectrum to the commercial sector. Such public users are best placed to decide how much spectrum they need and what can be traded away (sold or leased), and still carry out their core functions. For such a system to work such users must be allowed to keep the monies generated from such deals to give them an incentive. Without this incentive there will be no reason to trade or try to lease out spare capacity to other users. Of course, it would also be necessary to ensure that managers face the correct internal incentives to further the goals of the organisation.

4.4 From an economic point of view windfall gains are unlikely to cause concern. Indeed the realisation of windfall gains can be a sign that markets are working efficiently. However, if it is decided that windfall gains should be recouped by the authorities, for example, for distributional reasons, then it should be done in a way which does not distort incentives to trade. Such windfall gains could also arise from commercial users of spectrum, who were originally awarded spectrum via a beauty contest, of a first come first served procedure. This is discussed further in Annex 3.

4.5 Extra ex ante regulation may be required where an operator with Significant Market Power (SMP) proposes to acquire spectrum and this would have the effect of strengthening its position in the market. Such deals may need to be notified to the competition authority for approval. However this is only a preliminary view, and will likely need to be the subject of further consultation.

4.6 Technology neutrality is the best approach to regulation, and standards should not be mandated in frequency bands to promote strategic trade policies, except in very exceptional circumstances. It is still possible for Governments to support standards by participating in standardisation fora, the use of funds for research, or by using tax breaks to incentivise such participation by industry. It is interesting to note the success of the world-wide web and the internet without the need for Governments to mandate any standards.

4.7 A challenging yet achievable timetable should be set, with milestones, as to when this regime should be introduced. This could be done within the context of the legislation for, and the setting up of, OFCOM, eg that the future OFCOM legislation allows flexible spectrum trading to be carried out, as outlined in this response, so that spectrum licences with the properties outlined above (spectrum property rights) can be issued.

Annex 1

Response to questions (where Oftel has a view)

Economic gains from efficient use of spectrum

1) How best can Government assess the economic gains from enabling more efficient use to be made of spectrum?

It is not easy to predict in advance the gains that might derive from more economically efficient use of the radio spectrum. RA studies show the extent of the value of radio spectrum to the UK economy of £20 billion per annum. Even a few % increase in efficiency will yield a very significant figure. In any event the figure of £20 billion is likely to grow and the negative impact of inefficient allocations on the UK economy could be very high. Such negative impacts could arise from reduced competition in some telecommunications markets due to licence scarcity and the extra time it might take for new and innovative services reaching consumers without flexible spectrum trading. However spectrum trading will reveal the size of the trading market, which could be regarded as an indicator of inefficiencies under the current centrally planned system.

2) How could information from market transactions and economic impact studies best help inform the design of spectrum management policies?

The ability to trade spectrum flexibly (ie with change of use allowed as far as practicable) will reveal the market value of radio spectrum. This will be particularly useful in costing public policy requirements such defence and broadcasting. This will allow Government to better quantify the cost of such goals. For example when spectrum is reserved for public uses, the market will be able to reveal the value of spectrum close by. This will help ensure that the value of spectrum reserved for such purposes is implicitly taken into account by Government in assigning/allocating spectrum. That is that the public policy goal that spectrum is required to fulfil is at least worth the opportunity cost of that spectrum (the money that could be had from selling it on the open market).

Alternatively if such public users have the ability to trade spare spectrum capacity to commercial organisations, the users themselves will be able to dynamically regulate the amount of spectrum they require. For instance new radio systems may be come available that use less spectrum, but require higher investment. The ability to trade spectrum may give such public service type organisations the cash required to invest in new radio systems.

Economic principles of spectrum management

3) How far can the over-arching principle, that spectrum users should bear the opportunity cost of their usage, be applied in practice?

The principle of users bearing the opportunity cost should always be applied. For commercial use of radio spectrum, where there are no public policy goals, it seems reasonable that such users should pay the market rate for the spectrum they require. Thus spectrum should be treated as any other scarce resource that is required by the public and private sectors.

4) How can the trade-offs between competing economic and social uses of spectrum be more clearly articulated in the principles governing spectrum management?

We would argue that by having a market that reveals the true commercial value of radio spectrum, Government will be better able to assess the costs involved in public policy objectives. If the market reveals that the value of a particular piece of radio spectrum is £1 million, then the benefit of the social objective should at least be that value of £1 million.

Legislative basis for spectrum management

5) To what extent would a separate spectrum management duty for OFCOM be helpful, and how could this best be articulated in a new statutory framework for communications regulation?

It should be incumbent on OFCOM to manage the radio spectrum in the best interests of all consumers/citizens, not just the interests of particular individual or groups of radio users. It is, however, debatable whether there needs to be any separate specific duty in relation to the spectrum in addition to those duties enunciated in the White Paper. Those duties were formulated to provide a general framework for the whole of the communications sector including the radio spectrum. If it is considered that these duties are not sufficiently wide to ensure the best use of the radio spectrum, any statutory duty should focus on the economic efficiency as opposed to the technical efficiency of radio spectrum usage. Economic efficiency involves maximising the benefits generated from the allocation and use of spectrum (these benefits are not only about wealth maximisation but can include defined social goals). This duty contrasts with one that would focus on allocating/using spectrum efficiently from an engineering/technical perspective, the ultimate objective of which is to maximise the number of users of any given part of the radio spectrum.

6) What additional statutory alternatives to apparatus licensing could assist OFCOM in meeting its spectrum management objectives?

Oftel believes that licences should be available that allow owners of radio spectrum to trade it as freely as possible. Central to that idea is that the property right conferred by the spectrum licence should be as clear as possible, and permit the users to change its use (ie from mobile to broadcasting) with the following caveats:

1. that the radio emissions remain within the pattern allowed by the spectrum licence;
2. that safety of life services are not unreasonably affected;
3. that there are no issues raised regarding UK treaty obligations, or with regard to national security; and
4. that competition concerns are addressed.

7) How far can new modes of licensing, based upon access to defined spectrum rather than defined wireless apparatus, assist in enabling more efficient use to be made of spectrum?

The ability to have a spectrum licence that allows change of use, and the ability to freely trade licences is of fundamental importance to the benefits that can be delivered from auctions and spectrum trading. Any restrictions to the functioning of such a market, in terms of what is or is not allowed, or that increases the transaction costs, would have a significant impact on the efficiency of such a market.

The modes of trading will need to be considered in light of public consultation, but Oftel would hope that such modes of trading will allow change of use of spectrum, and the ability of public operators (ie MOD/broadcasters etc) to sell or lease spare capacity to commercial users.

Regulatory framework for spectrum management

International dimension

8) How can the UK’s stance towards international spectrum management policy best reflect the opportunity costs of different spectrum uses?

The UK should try to ensure that international regulation is based as far as possible on the market principles that should be used in the UK (auctions and trading with change of use). It should also try to ensure that issues relating to the value of radio services are fully explored in such fora. The ITU has a responsibility to consider economic as well as technical issues when deciding on the use of radio spectrum.

9) What scope is there for greater autonomy in domestic spectrum policy within the constraints imposed by the UK’s international commitments?

Oftel agrees with the Review’s foreword, that policies should be recommended which seek to progressively minimise such constraints through unilateral and international action. It would seem that the flexibility the UK has is reasonably high with this regard. The normal international treaty obligations allow for uses of spectrum to occur on non-interference basis. This means that the UK is not normally treaty bound to use the services dictated by international regulation. In any case the classifications are normally quite broad (mobile, fixed, broadcasting etc), and there is always the possibility of having UK specific footnotes to give the UK greater flexibility if required. The UK RA also has a well respected international section, who with their good offices, should be able to influence future debates, to allow modifications to regulations if these are required. A practical scenario might be that an owner of spectrum decides to change its use to something not listed in international regulations. Such a service could begin on a non-protected basis, as long as it does not cause interference to services used in other countries that are listed. This could be an interim situation until the NRA was able to negotiate changes to regulations to allow such changes (with suitable inputs from the UK spectrum owner concerned).

One area that might be difficult to overcome would be that of satellite services. Frequencies allocated to the mobile/terrestrial transmit end of a satellite service might not be useable in the UK as they would interfere with the receiver in the satellite (whose footprint may cover the UK). It might still be possible however to use the satellite transmit bands (mobile receive), as this would only effect satellite mobiles operating in UK territory.

10) How should the UK Government judge the trade-off between a more liberal approach to spectrum management and one in which technology standards and spectrum access are mandated as part of a strategic industrial and trade policy?

As mentioned in the main text, Oftel does not agree with mandating standards as a way to promote strategic trade policy. Standards will only be successful if the industry supports them, and consumers will pay for such services. The industry is best placed to make those choices. Regulators are not best placed to pick winning technologies. There may also be issues to do with technical barriers to trade in mandating standards. Support for industry initiatives such as ETSI and 3GPP (responsible for 3G standardisation) is the best way forward.

11) If there were greater latitude in international allocations and/or the UK’s implementation of such decisions, to what extent would market mechanisms result in harmonisation of equipment and transmission standards?

As discussed in the text, the market would decide to the extent that harmonised standards are cost effective. The process of standardisation is not without costs, from people participating in standards meetings, to slowing down innovation by having to have changes agreed by other parties. There are benefits from economies of scale, and the ability to roam in some circumstances, but these are not sufficient in all cases. Some forms of standardisation only relate to very general characteristics, others to the definition of complete systems. The market is best placed to decide to what extent spectrum should be harmonised.

Oftel would argue that harmonisation of standards is not an end in itself, and should only be done when the benefits outweigh the costs.

National dimension

12) Within the current and proposed statutory framework, what improvements (if any) could be made to the institutional arrangements for spectrum management in the UK?

Oftel understands that currently base stations need to go through a procedure known as ‘Site Clearance’. This is so all potential base station transmitters above certain powers can be checked for possible interference impacts with other Government departments as well as some commercial organisations that sit on a panel. It is also not clear why all users of spectrum should go through this procedure if they operate in frequency bands that are exclusive to them. Oftel understands that this procedure looks into the interference mechanism of ‘blocking’ which can be caused by transmitters well away from the receive frequency band. However it is possible to engineer receivers with improved blocking performance, or reduce the problem by adherence to industry site engineering procedures (such as ETSI ETR 053 on Radio Site Engineering). This need to co-ordinate each and every base site seems to be a substantial overhead under the current system. Oftel would ask that some thought be given to the fundamental purpose of this panel, and if any way could be found to stream-line its workings. It would appear to Oftel that under a spectrum trading regime, if operators are within their spectrum property right they should not need to get such clearance. It may be required to keep a list of such base stations for international treaty obligations, as some operators may need to ensure that the field strengths laid down on neighbouring countries in certain radio channels is below an agreed figure. This would allow the NRA to ex post check such base stations. Oftel’s concern would be if such a regime would act as a disincentive to flexible spectrum trading and hence reduce the benefits that might accrue to consumers.

13) To what extent would greater transparency of specific data on current and prospective spectrum uses support efficient spectrum use? What are the key issues and trade-offs pertinent to the provision, by RA, of an on-line database containing spectrum-utilisation details? How far is transparency compatible with commercial confidentiality and public safety and security considerations?

Under a spectrum trading regime, sufficient transparency would be necessary in order to establish the identity of a party causing interference and to demonstrate a right to compensation in the event of arbitration being necessary. However, there also needs to be flexibility in order to keep up with changes of use.

Oftel believes that a central register of spectrum property rights should be set up by or on behalf of OFCOM. This would give certainty to those buying spectrum that those they are dealing with have legal title. It would also facilitate putting potential buyers in touch with owners of particular frequency bands.

14) To what extent could intermediaries play a valuable role in buying rights to manage a particular frequency band and then selling access to parts of this spectrum to users on a commercial basis?

In a number of services the individual users of spectrum are too small to warrant individual assignments (such as some taxi firms using private business radio or some fixed point to point radio links). Under the current regime the NRA administers the radio spectrum and sets exact criteria to users about how they should share radio spectrum. This normally involves very detailed frequency planning, and the NRA is required to guarantee that the assignment will be able to support a given grade of service. Under a spectrum trading environment this role does not need to be undertaken by the NRA. It can be envisaged that private spectrum management organisations (SMOs), will buy the spectrum required to run such services, and issue individual licences to small users. Such SMO’s are likely to be more focused on their customers needs and to offer a wider variety of ‘licence products’ due to their market focus.

Such licences will contain the planning conditions required to have different users share frequencies. These SMOs could offer bespoke services tailored to individual requirements (at suitable prices). Thus a market based mechanism, where SMOs buy spectrum and charge individual licences to make fair returns, is possible even for services that require careful planning to share. Of course such SMOs could also benefit from allowing different services to share such bands (say mobile and fixed links), or take advantage of systems that can share because they ‘point in different directions’ (eg have large back to front ratios in their antenna patterns). Such SMOs could also specialise in purchasing spare capacity from other users whose core business is say mobile, but are prepared to accept other uses for a price (ie a percentage of the SMO’s revenue say).

Such new business models may yield up unexpected resources of spectrum, and new assignment methods that can more accurately model interference scenarios. They could also free up the significant financial resources for large IT investments that may be required for such detailed spectrum planning. It may also encourage current radio users to implement interference reduction techniques (which may be expensive) to allow SMO’s to exploit their spare capacity. We might see a virtuous circle of investment to allow sharing, which generates SMO revenue, and encourages more investment to allow sharing. This would make it easier to deliver services to companies and consumers based on radio spectrum, as well as allowing easier access to spectrum for new and innovative solutions. Under the current system the NRA is not incentivised to do commercial deals to maximise profit (and hence wealth and value to consumers), but to make spectrum available for users and uses it believes to be the most appropriate. This will rule out some of the business models discussed above.

Spectrum use: marketed and non-marketed outputs

15) To what extent is the review’s distinction between radio spectrum used for marketed and non-marketed goods a helpful one?

The key point may not be the distinction itself but the recognition that suppliers of non-marketed goods should also be made aware of the opportunity cost of the spectrum they use.

Issues in non-marketed uses of spectrum

Defence

16) What opportunities exist for commercial and other civil users to share spectrum with the defence establishment in the UK?

Oftel would assume that there is scope given the amount of spectrum allocated to defence, but that the best way to show this is by allowing such defence users to sell or lease their spare spectrum capacity, and keep the proceeds from such deals. This would make defence users themselves undertake root and branch reviews of their use of spectrum. They could use less spectrum by more investment, using revenues from selling under-utilised frequencies. A market based solution would remove the need for OFCOM to have to make decisions about who should use what spectrum, and if they should have more or less over periods of time.

17) What further incentives could be introduced to encourage more intensive and efficient use of spectrum allocated to defence?

Oftel believes that by having spectrum trading with change of use, such users will face an opportunity cost if they do not make the best use of spectrum themselves (either by investment in systems, or selling spare capacity). This opportunity cost is likely to be the most effective way, as third parties deciding how much spectrum defence needs will be very complex. Even the use of administrative pricing will be complex, due to the difficulty in setting charges for spectrum that do incentivise such use. There is also an argument that unless spectrum can be sold or leased to commercial users it may simply be considered a fixed overhead by military users who take account of this when bidding for funding from central government although there may still be some incentive to trade-off spending on spectrum against spending on other assets.

Broadcasting

18) How can the Government’s commitment to value the spectrum used by broadcasters be implemented in a way which encourages spectrum efficiency?

Oftel believes that the same issues that would incentivise military users would do so for broadcasters, namely spectrum trading and change of use spectrum property rights. Such users are best placed to know what the value of spectrum is to them. If a commercial organisation can make a better offer for it (to buy exclusively or share) then broadcasters will do such deals. The requirement remains that they be allowed to keep the funds raised from such deals. Such broadcasters would have legal obligations to provide broadcast services, and are best placed to know how much spectrum they really need to do that.

Due to the shared nature of broadcaster spectrum (ie channels accessed from a central pool), it may be possible to envisage some form of SMO scenario with broadcasters owning shares in such an organisation. The SMO could have a primary duty to ensure sufficient spectrum for broadcasters to deliver their services, but also to maximise other revenue sources via spectrum sharing arrangements. This would be similar to the current arrangements for broadcaster infrastructure (masts and transmitter equipment etc.).

19) What further incentives might be introduced, consistent with wider broadcasting policy and with previous Government about television and radio franchise fees, to encourage greater spectrum efficiency by commercial broadcasters? Are there differences in the approach to incentives before and after the proposed switchover from analogue to digital terrestrial TV broadcasting?

Again flexible spectrum trading with change of use that forces an opportunity cost on broadcasters. They would then face the same incentives as mentioned in the questions relating to defence users of spectrum. These organisations would have to ensure that the correct internal incentives applied to its managers in selling or leasing spectrum for commercial use.

20) What incentives might be introduced, consistent with wider broadcasting policy and the Government’s approach to the funding of the BBC and its public service remit, to encourage greater spectrum efficiency by the BBC?

As above, spectrum trading and change of use of spectrum, along with keeping the revenue from such deals.

Aeronautical and maritime services

21) How far could changes in spectrum use charges be reflected in air traffic control fees which are passed on to users of UK airspace and landing services?

In principle it would seem desirable for aircraft operators (and perhaps ultimately passengers) to be made aware of the opportunity cost of the spectrum used for air traffic control purposes, particularly given the level of congestion at certain times and locations.

Emergency services

22) How far can spectrum pricing influence emergency services to make efficient use of spectrum over time?

Oftel believes that the same mechanisms of spectrum trading with change of use, and keeping the proceeds of such deals will yield the best results. Emergency service operators are best placed to make decisions about how much spectrum is worth to them, and if proceeds from selling of or leasing such rights could compensate them. They would face the same choices that the military users would, in that they could use the proceeds from such deals to invest in new technology so they use less spectrum, or perhaps move to services provided by third parties. As in other cases, these organisations would have to ensure that the correct internal incentives applied to managers in selling or leasing spectrum for commercial use.

23) What scope might there be for sharing of spectrum access with other users to enable more efficient spectrum use?

Again commercial pressure should be the mechanism used. Emergency services are likely to be tempted to do such deals if the revenue from them is more than the overheads of co-ordination, or re-engineering their radio networks. The advantage of a trading system is that OFCOM would not need to become involved, and it would be left to private negotiations between the parties.

Spectrum pricing and auctions

24) Where should the balance lie between administratively-set incentive prices and competitive auction of spectrum licences? To what extent could the two approaches be combined to encourage spectrum efficiency?

Auctions have the advantage of being transparent and easy to award (unlike the complex process of deciding a beauty contest winner), and ensure (all other things being equal) that the spectrum goes to the company that values the spectrum the most, and can hence make the best use of it in economic terms. This should be the most efficient use of the spectrum, which will benefit consumers who will have access to the services they most value.

Oftel believes that administrative pricing has a number of difficulties in promoting the economically efficient use of radio spectrum. It requires civil servants to calculate a value of spectrum based at present on the use to which it is being put by its current user. There is no guarantee that the service currently being used is the highest value, so it will not encourage low value services to migrate out of spectrum to make way for high value services.

This ability of the market to assign spectrum to the highest value use quickly, is key to deriving the maximum benefit to the UK of the scarce resource of radio spectrum. This impact of opportunity cost to spectrum users (if they don’t sell on spectrum to others for more than it is worth to them) is a key driver that could act on commercial and non-commercial users alike.

Administrative pricing is also extremely complex to calculate. It relies on very involved analysis of how operators deploy their radio equipment, and what value they would place on more spectrum to provide more capacity, as opposed to other solutions involving more infrastructure investment (such as smaller cell sizes in mobile systems). The values chosen historically have been quite low and do not appear to have had any measurable effect on the demand for spectrum by users subject to administrative pricing. This could be overcome by incremental price rises, but again these would need to be agreed, and it would not be clear when the price was correctly set.

Oftel would also argue that current regime of administrative pricing (even with its limitations) is a very interventionist type of regulation and as part of OFCOM would prefer a lighter regulatory touch be used if possible. A system of spectrum trading would obviate the need for such complex calculations and allow the market to decide, especially if change of use is allowed. It would allow the optimum distribution of spectrum between various competing services to be arrived at by the market. This would be in-keeping with Oftel’s desire to intervene as little as possible in market regulation (subject to Oftel’s high level objectives).

Oftel would favour the use of auctions and flexible spectrum trading to encourage economic efficiency. Administrative pricing has proven complex as the fee is set in negotiation with the existing users, and its value to them, rather than its value to the highest bidder. This can lead to charges which are set too low to have a significant effect on licensee’s behaviour. The use of auctions and spectrum trading, with change of use, will be useful in any interim period, in revealing the true market valuations of spectrum. If spectrum close by is valued by the market in such a way (ie via auction or secondary trading), then this should feed into the valuations used under administrative pricing. This would mean that the more widely such market based valuations are used, the more accurately an administrative price can be determined.

25) What factors should determine the choice of frequencies subject to auction of licences?

Oftel believes that there should be a presumption in favour of auctions where spectrum demand exceeds supply.

Spectrum trading

26) Which areas of spectrum use are most amenable to and which areas offer the greatest potential efficiency gains from the introduction of spectrum trading?

Oftel believes that all areas would benefit from a flexible spectrum trading environment (ie that allows change of use), due to the extra speed and certainty with which users would be able to change the use to which spectrum was put. It would appear to Oftel that the flexible trading of spectrum from uses such as defence and broadcasting might give the greatest potential efficiency gains, due to the amount of spectrum they hold.

27) To what extent would a move to licensing of spectrum access, as opposed to wireless apparatus licensing as now, facilitate spectrum trading?

The use of apparatus licensing would act as a significant disincentive to trade spectrum. This is because change of use of radio spectrum could not easily be facilitated. This ability to have change of use is key to reaping the benefits of spectrum trading, as it allows high value users of spectrum to obtain it from low value users. It also makes the sharing of bands between different services easier to organise. It would appear to Oftel that if spectrum trading is introduced with apparatus class licensing, then much of the potential for increased efficiency will be lost.

28) What changes to the terms and conditions of licences for the operation of wireless equipment and/or for access to spectrum would facilitate spectrum trading?

A spectrum property right type license would have the general characteristics and terms of:

• a defined geographic area;
• a defined frequency range of use;
• the acceptable EMC emission limit it is able to radiate outside its spectrum right;
• the name of the owner of the legal title that is lodged in a central register administered by central government;
• a general proviso not to breach international treaties to which the UK is bound;
• provisos to cease transmission if safety if life or national security issues are raised (with compensation if the user is operating within their licence terms);
• a requirement that sale of licences to users with dominance in relevant markets (as listed before hand by OFCOM) needs prior approval; and
• length of licence (if not a permanent right).

29) If new modes of licensing spectrum access (rather than equipment operation) were introduced, how could rights to spectrum usage, such as interference standards and length of licences, best be defined to facilitate spectrum trading?

Rights should be defined as shown above. Standards do not need to be mandated for spectrum, and operators will be free to implement any standard their property right will support. The length of licence will need to be long enough so as to ensure owners will invest in deploying services that may take a long time to adequately pay back such investment. It seems clear that the longer the licence is the less will be the possible starving of investment that may happen towards the end of a licences lifetime. One option would be to have permanent rights. Oftel is still undecided on the length of such licences but less than say 25 years may lead to lack of investment. The longer the period the less impact this ‘dead-time’ will have on investment.

30) What market infrastructure, such as spectrum registers and dispute resolution procedures, could facilitate spectrum trading?

Oftel believes that a central register of spectrum property rights should be set up by or on behalf of OFCOM. This would give certainty to those buying spectrum that those they are dealing with have legal title. It would also facilitate putting potential buyers in touch with owners of particular frequency bands.

It would also seem appropriate to set up an arbitration panel that would act as a backstop for disputes between legitimate users of radio spectrum. The panel should have the ability to arbitrate on if compensation needs to be paid for interference caused by one party to another. It is not clear to Oftel if this panel should be mandatory or not, that is to say should spectrum right holders have the ability to use law courts to seek redress.

31) What lessons can be learnt from the experiences of other countries (such as Australia, New Zealand and the United States) in introducing different modes of spectrum trading?

It does not appear that spectrum trading has had a large impact in New Zealand, although there have been some deals. This may be due to the small population and low population density making the ability to get equipment outside certain bands less economically viable.

It would appear that trading has not had a great effect in Australia, but that may in part be due to the large number of their equivalent of apparatus licensing.

It is not clear to Oftel what extent spectrum trading has been implemented in the US. There are various regulatory issues such as spectrum caps, and must carry obligations of radio broadcasters on cable that may have an impact on any such market. Although there appears to have been a large degree of consolidation and exchange of spectrum licences to form quasi-national mobile coverage in the last year or so. These deals appear to need the approval of the FCC.

As described in Annex 3, one positive lesson is that spectrum trading has not had any obvious damaging effects despite some predictions to the contrary.

The boundaries of spectrum regulation

32) What factors should guide regulators in setting the boundaries of licence-exempt spectrum use?

The role that SMOs can play in administering licence exempt spectrum is discussed in Annex 3. It is suggested there that the market can play a major part. However, it should be noted that this is one area where Oftel believes that further consideration of the extent to which market forces are allowed to operate freely needs to be given.

Annex 2

Spectrum management

Purpose

A2.1 The purpose of this Annex is to highlight some of the issues that make spectrum licensing different from Telecommunications Act licensing, and give some background about Oftel’s role. The Annex also serves to provide more detail on some of the issues discussed in the main text (such as standardisation) where arguments can be more fully explored. The Annex also sets out Oftel’s view of the key issues associated with complex task of spectrum management which is being examined in this Review.

Oftel’s role

A2.2 Oftel is the telecommunications regulator within the UK and is charged with enforcing Telecommunication Act licences, and has concurrent Competition Act powers with the Office of Fair Trading (OFT). Oftel seeks to promote the interests of consumers and to ensure that reasonable demands for telecommunications services are met as far as possible, in the UK.

A2.3 Oftel’s goal is to achieve the best deal for consumers in terms of quality, choice and value for money, and this strategy is underpinned by four high level regulatory objectives:

• effective and sustainable competition;
• well informed consumers;
• adequately protected consumers; and
• prevention of significant anti-competitive practice.

A2.4 With regard specifically to radio spectrum, the Director General of Oftel has a duty to advise the Secretary of State as to competition issues surrounding the use of radio spectrum (section 51 of the Telecommunications Act 1984).

A2.5 Oftel is one of the organisations that is intended to be part of OFCOM as outlined in the recent white paper, along with the Radiocommunications Agency, which manages radio spectrum within the UK. As such Oftel has an interest in the future regulation of radio spectrum in the UK as it will be part of that future authority, whose duties will include spectrum management.

The problem of radio spectrum regulation/licensing

A2.6 The main difference between licensing radio spectrum and other telecommunication services is that a radio spectrum licence gives access to a scarce resource which is not the case with other telecoms licences. This places special requirements on its licensing. If it issued to one user then generally it is denied to another. A normal telecommunication Licence issued to one company does not reduce the number of such licences available to other applicants, that is there is no opportunity cost to be evaluated.

A2.7 As a limited resource, when a spectrum licence is issued other than by means of an auction or trading on an open market, a detailed analysis needs to be done to ensure that the limited resource of spectrum is given to the user best able to use it. This level of detail is greater than say for a normal assignment of a general Telecommunications Licence. Because spectrum is a scarce resource the organisation to which it is assigned has the ability or opportunity (in some cases) to extract a scarcity rent value from that spectrum, which raises questions about where such a scarcity rent value should sit (with company shareholders or with the Government)? Tied in with such considerations is the fact that spectrum can also be used to bypass some existing telecommunication areas such as the local loop and leased lines, and so inject extra competition where another fixed infrastructure provider would find it difficult to make a business case.

A2.8 Spectrum Licensing is further complicated by the fact that users in one piece of spectrum can interfere with those in another. The negative impact that such interference causes (negative externality) can be viewed as the interfering system being too ‘noisy’ or the victim system being too ‘sensitive’ or not ‘immune enough’. Which ever point of view is taken there are engineering cost implications in reducing emissions outside the assigned spectrum or increasing the immunity of the victim receiver.

A2.9 Another layer of complexity is added in that many users of radio spectrum are small businesses who are not given separate or exclusive radio channels (because it would not be efficient to do so). They share them with other such users, who may be their direct competitors ( the most common example being mini cab firms). In such circumstances the licensing regulation needs to ensure sharing is possible. An analogous problem might be the highway code and its enforcement to allow all road users to travel safely (ie all drive on the left). Because of the complexity of such common use of spectrum third party control is required. This role is normally carried out by the RA, but could in theory be given to an independent third party with guidelines on how such spectrum should be used (preference to small businesses say). Such organisations are known as Spectrum Management Organisations (SMO’s), and could also be run on a purely commercial basis.

Background

A2.10 The overall economic management of radio spectrum should be to maximise the net social benefit in the use of radio spectrum, that is by calculating the net benefit to consumers and suppliers of such goods and services, and also taking into account net third part benefits. Such third party benefits can be negative because radio systems can cause interference to other radio systems/services. Such issues however need to be ranked against other goals such as public policy and international treaty obligations. Oftel would argue that the best way of making such decisions is to allow a spectrum market to decide who has what spectrum and for which purposes it is used.

A2.11 Markets are generally the most efficient way of allocating scarce resources, with appropriate regulation to guard against potential failures. Even if not used for all radio users (such a public services such as defence) it will reveal the value of the asset being given. This value represents the opportunity cost, that is (crudely) the money lost for other purposes by giving this spectrum for a particular use. This is helpful in two ways. Firstly it means that the value of the social objective should be at least as high as the opportunity cost. Secondly it allows the public policy centred user of the spectrum (say defence) to make value judgements as to if the opportunity cost of holding such spectrum (the money it could be sold for) is worth it. For instance such users may sell of under utilised pieces of spectrum to invest in other areas (more tanks, or troops say), or to release equity by leasing out spare capacity. A military user may not have a requirement for spectrum in city centres but mainly in rural ones, and could allow such commercial use in urban areas. Such a lease might include the ability to claw back this spectrum in times of national emergency, or for pre-determined training purposes.

A2.12 This does not remove the difficult issues of public policy, such as how much spectrum should be given to services with no direct commercial benefit (say radio astronomy or emergency services). But it does allow the opportunity cost to be calculated and this can be weighed against the public policy benefit (ie revealing the value of the spectrum used). This revealing of value forces public welfare use to be quantified to balance out this opportunity cost, which should lead to a more consistent approach being taken in the assignment for spectrum. It also allows such users to realise some of the value of this resource, via say spectrum trading, which would make such deals attractive. The setting up of such deals however may involve costs to such public users in terms of co-ordination of radio systems, or perhaps added engineering. The money raised by such deals would help offset these expenses.

A2.13 This process of markets would ensure spectrum was used by those best placed to use it, and could remove a large element of the need for civil servants to make judgements as to the relative commercial merits of various users of radio spectrum.

Issues surrounding spectrum management

Issue 1: Promoting consumer and producer benefits eg economic spectrum efficiency

A2.14 Most customers of any service are prepared to pay more for that good or service than the actual price at which it is offered. The difference between the amount a customer would be prepared to pay and the price is the consumer surplus. The difference between the cost of producing a good or service and the revenue from sales is known as the producer surplus. Maximising consumer and producer surplus (along with benefits to third parties known as externalities) gives the net social benefit of using the scarce resource of radio spectrum.

A2.15 By allowing spectrum to be used to provide commercial services consumers and producers benefit. The extent to which they benefit will depend on the relative supply and demand of the services and the costs involved in delivering those services. Such considerations can be important in deciding on the number of operators who should be licensed in a given band (such as UMTS).

A2.16 The idea of economic spectrum efficiency can be applied to public policy objectives too, as these will have some notional value, that can be compared with the cost of forgoing its use by commercial users (opportunity cost). It should be this concept that is primary in the minds of those determining who uses radio spectrum and for what purpose. It is Oftel’s belief that markets should reveal this as far as far as possible, and that the fewer constraints there are on such markets, the greater the gains will be for the UK as a whole.

A2.17 It should also be recognised that the principle should not be to maximise this for radio spectrum users alone, but for all. Radio spectrum is a vital resource that impacts on the economy as a whole, and provides benefits to the economy as whole, and should not be viewed in isolation.

Issue 2: Promoting wider competition

A2.18 The services provided using radio spectrum can sometimes be done using other technologies. An example would be broad band wireless access at 28 GHz, whose services might compete with xDSL technology over the copper local loop.

A2.19 Under the current regime, when assigning spectrum the impact on consumer surplus for services as a whole should be considered, not just those via radio. For instance even though the maximum value (to potential operators) of spectrum may be for a particular service, due to lack of competition in a another telecommunications service sector (such as local loop) it may be appropriate to place restriction on the use of new spectrum. One example is the uses to which BWA spectrum at 28 GHz can be put (ie the regulatory constraints imposed). A restriction has been placed on such operators using it for third party backhaul because the need to promote competition in the local loop as a whole is greater than the narrow requirement to assign radio spectrum efficiently for a specific use.

A2.20 In a spectrum trading environment the market itself can in general deal with such issues dynamically, by deciding what spectrum is used for. In the absence of regulation, a lack of competition is likely to be reflected in excess profits. If an area of the market has high profit margins, then this will be an encouragement for new entrants to come into the market (all other things being equal). They will be able to do this by acquiring spectrum in an open market. This should be a quicker process than Government deciding that a market is not competitive and deciding to make spectrum available to new entrants for a particular type of radio service.

Issue 3: Social welfare

A2.21 There may be public policy goals which outweigh other factors in allocating spectrum to particular services. Such users would be MOD and emergency services, where public policy objectives might again outweigh allocation to other service choices. Such decisions would need to be evaluated against the opportunity cost involved in denying spectrum for such other uses. Putting a value on the opportunity cost of such spectrum use forces the public welfare consideration to be quantified, if only to say it is at least as much as the opportunity cost. The net impact of such considerations may be to modify the amount of spectrum given to such uses (more or less depending on the values calculated). This would reveal the opportunity cost of the service being used for a public policy goal. Under the current regime it is very hard to quantify this opportunity cost as there is no spectrum market to reveal the value.

A2.22 With a flexible form of spectrum trading such users could to an extent be self regulating, in deciding how much and in what way they were prepared to give up spectrum to the private sector. Such users would presumably have their public policy objective as a legal requirement to maintain or provide or carry out certain functions. Under such constraints they would not be able to give up more spectrum than was the absolute minimum to carry out their public policy objectives. If they required spectrum in the future they would be required to purchase it on the open market, or in exceptional circumstance it would be compulsorily purchased (with suitable recompense to the owners).

A2.23 This would shift the onus away from spectrum regulators deciding how much spectrum non-commercial users require, to the organisations themselves, who are best placed to judge. The current method of awarding spectrum for ‘free’ is a very blunt tool to give assistance to support organisations that require spectrum to carry out public policy goals. Spectrum trading would be a refinement to help determine the exact costing of such goals, which is important in weighing competing requirements.

Issue 4: Techncial spectrum efficiency

A2.24 Technical spectrum efficiency is mainly perceived as fitting the greatest number of users with a given part of the radio spectrum. The usual metric is number of users per km² per MHz. It is achieved by careful frequency planing which uses a knowledge of the appropriate isolations required between radio users to prevent harmful interference. Such frequency planning tools are normally sophisticated IT systems with detailed topographic and radio engineering parameter information.

A2.25 It should be borne in mind that in general as technical spectrum efficiency increases, the extra cost involved may not linear. After a point there may be diminishing returns, so that the extra costs involved in using less spectrum to provide a given service are larger than the opportunity cost of the spectrum saved. Spectrum trading should be used (when possible) to determine this trade-off.

A2.26 However, technical spectrum efficiency is not an end in itself, and increasing it has a cost impact on radio systems, which may be passed on to subscribers. Oftel would argue that the primary importance should be in producing the wealth generated from using spectrum, economic spectrum efficiency. For instance, one would not prefer system X to system Y, simply because X was able to accommodate twice the number of users in a given spectrum band. Consumers might value the services provided by System X less than those provided by System Y. Worse still forcing an existing system to be more spectrally efficient in a technical sense would, in all likelihood, cost more, the larger the number of subscribers it had (eg they would have more handsets and base stations to modify). Oftel believes that technical spectrum efficiency is a secondary objective to economic efficiency.

A2.27 It is also interesting to note that the ITU, which is responsible for the co-ordination of radio spectrum on a world-wide basis notes in its mission statement that, "--the mission of the ITU Radiocommunications Sector is, inter alia, to ensure the rational, equitable, efficient and economical use of the radio-frequency spectrum ---". The mission statement further notes that there is a, "growing recognition of the economic value of frequency spectrum and the application of economic principles in the management of that resource, recognising the rapid market-driven and user-oriented technological development."

A2.28 An argument for using technical spectrum efficiency might be when spectrum is not charged for, and as a means of limiting the amount required. But even here some consideration would need to be made with regard to the costs placed in providing services in limited spectrum, and such an analysis would be equivalent to Issue 1 (economic spectrum efficiency maximisation).

Issue 5: Standardisation

A2.29 Standardisation can come in many forms and degrees. In terms of radio spectrum usage it could be that the minimum level of standardisation might be that required to reduce interference. For example all users of a particular band are limited to a certain transmitter power, and the emission mask ( the ‘shape’ of the radio transmission/frequency profile) may be mandated. This sort of standardisation is common in deregulated bands where different user share bands without any formal co-ordination. It can also go as far as having devices so highly specified that a radio from one user’s system will work seamlessly with the radio network of another user, as happens with GSM.

A2.30 Standardisation can yield benefits in enabling large numbers of users to share spectrum without costly co-ordination procedures, particularly where such users are low power and use very localised radio services. It can also give significant benefits in economies of scale of production, ie the unit cost of making 1 terminal is much higher than the unit cost on 10 million terminals.

A2.31 There may be significant negative elements associated with harmonisation (which generally increase with the level of harmonisation required). Mandating standards can stifle innovation and force operators to use equipment that is not the most suited for their requirements. Harmonisation should only be adopted if the benefits from economies of scale, and or the benefits from having national and international roaming, (or any other benefits), outweighs this extra burden, and the market itself could not reach such a conclusion.

A2.32 Technical solutions such as ‘software radios’ which can adapt to any standard may reduce this problem, but such solutions are not likely to be commercially viable in the next few years. Certainly multi-band radios as for GSM in 900, 1800, and 1900 MHz bands have eased the problem of requiring only one specific band for a given radio service to benefit from economies of scale from production.

A2.33 It is hard to calculate the possible future benefit of having spectrum mandated for a particular service, and such practices as mandating standards have sometimes been regarded as technical barriers to trade. It is also possible that mandated standards may fail in the market place (CT2, DSRR etc.) which means that there has been an opportunity cost wasted since the market was not able to exploit the given piece of spectrum with the mandated standard.

A2.34 It is not mandating certain standards in certain bands that leads to commercial success, rather it is having standards that are supported by a large number of manufacturers and potential operators. Such co-operation is best left to voluntary standards drafted by industry bodies. Market participants are best placed to know what the market requires (as opposed to regulators). There is always the danger that regulators will pick an inappropriate standard, which will not be supported by the market place. If standards are commercially attractive they will be implemented by operators. In a spectrum trading environment they will be able to purchase small amounts of spectrum to ‘test the market’. They will obviously try to obtain spectrum as close to the ‘core band’ agreed by the standards body as possible. If this is the highest value use of such spectrum they should be able to purchase it from the incumbent. If not they will look around for something cheaper.

A2.35 It is still possible for central government to encourage such users or uses of spectrum, either via tax breaks, or funding into research, or by support of standardisation fora. It is also possible to support any initiatives by purchasing such spectrum in the market place, to be made available for such services, if it was felt the market could not deliver. In the final resort some form of compulsory purchase may be made, analogous to land compulsory purchases, may be possible after a proper consultation process, and adequate financial recompense to the spectrum owner.

A2.36 In general technology neutrality should be used, unless there is a good reason to justify imposing a standard, such as economies of scale or increased spectrum efficiency, and the market itself could not come to such a conclusion. That is harmonisation should be kept to the minimum required by international agreement unless there is demonstrable justification to do otherwise. This issue should be left to markets, and intervention should occur only after a market failure, rather than in anticipation of it. In practice however this form of intervention (ie mandating standards) is likely to be very much the exception, rather than the rule.

What about GSM?

A2.37 Many point to the success of GSM and ask if it could have been successful without the Council Recommendation that set aside the spectrum 890-915 and 935-960 MHz for GSM within the EU. In fact the Recommendation did not specify the amount of spectrum from this band, only that it should use these bands. When GSM began much smaller initial allocations were made in the UK, which had until this year analogue mobile systems in these bands. The UK pioneered an extension to this frequency range (known as DCS 1800 at the time). This was mandated via a CEPT Decision in 1995 (the European equivalent of the ITU – see below), comprising of c. 40 countries. This designated the bands 1710-1785 and 1805-1880 MHz to DCS, however only 2 x 15 MHz of this band was required by 1 March 1997, and a further 2 x 5 MHz by 1 January 1998. This in total is only 2 x 20 MHz out of 2 x 75 MHz.

A2.38 At the time many outside the UK doubted the commercial viability of DCS, but Europe has a number of 1800 operators now. It was not the CEPT Decision or the Council Recommendation that made GSM fill the bands which were in part reserved for them. In fact until relatively recently even in the UK the 1800 bands were not all used for GSM, and as mentioned above, the 900 MHz bands were used for TACs as opposed to GSM. Oftel would argue that the success of GSM in Europe, and beyond (Far East and even the USA with GSM 1900), is in fact due to the support by industry in standards fora. The key drivers for participating in such fora, and deploying equipment standards compliant equipment being the value of roaming customers, and the economies of scale possible. An added benefit of participation to operators being open standards, that allow them to buy equipment from many vendors (ie not locked into one supplier). It is interesting to note that GSM has half a billion users world-wide, and most of the digital mobile subscriber market, and yet has no ITU world-wide allocation.

A2.39 It seems clear that if GSM had not been a commercial success operators in the UK would have kept their existing TACs networks, and would perhaps have asked to implement another 2G standard in these bands. Its success was built on the fact that it was not organisations such as the CEPT who developed the GSM technical standard (which is primarily controlled by administrations), but that ETSI took over this work in 1988 from CEPT, and ever since has considerably improved and updated GSM (eg new GPRS data services etc). ETSI is primarily driven by commercial organisations such as operators and manufacturers, with a weighted voting system depending on the commercial size of the organisation (although NRAs have a sizeable share of voting rights).

A2.40 In Oftel’s opinion technology neutrality is the best approach to regulation, and standards should not be mandated unless this is required to create a level playing field for competition, say in mandating an interface standard to allow access to the network of a dominant operator. Spectrum trading will place the decision about what standard is used in what band in the hands of the spectrum owners, who are best placed to make such a judgement. It is unlikely that those with a commercial interest will make choices that destroy some of the value of their asset (spectrum). If they do the ability for others to buy up undervalued spectrum should correct this.

Annex 3

Spectrum trading paper by Tommaso M Valletti, Imperial College and CEPR, London

A3.1 Oftel has commissioned this research into the issues behind spectrum management, which serves to underpin some the economic issues discussed in this submission. The views expressed are those of the author, but Oftel agrees with the author’s strong support for spectrum trading. However, Oftel believes that further work is required on some of the issues raised in the paper, such as creating permanent spectrum rights, and the ability of the market to deal with deregulated spectrum. Most importantly from Oftel’s view as a competition regulator, we believe that spectrum trading may require new ex ante competition rules, and not be left solely to existing competition regulation. However this is an initial view that will need to be reviewed in light of the forthcoming RA consultation on spectrum trading, and responses to it.

Spectrum Trading

Tommaso M. Valletti^*

Imperial College and CEPR, London

This paper was commissioned by Oftel - the UK Telecommunications regulator. The views represented however are those of the author’s.

21 June 2001

1 The benefits of a market for spectrum

Auctions of 3G spectrum licences in UK and Germany have shown that market players are prepared to pay high values to obtain the rights to use spectrum. However, the way in which spectrum has been sold – limiting the quantity of relevant spectrum available for auction and prohibiting the use of spectrum, in neighbouring frequencies, for 3G purposes by current or future users – has artificially forced up the prices of these licences.

Perhaps, if the service was also feasible using portions of neighbouring spectrum not originally allocated for 3G services, some operators would be willing to buy it. However, this may not happen for different reasons. One reason is a more technical one, ie that some bands are reserved and cannot be used to avoid interference problems. Another reason may be that a neighbouring band may be allocated for completely different purposes, using different technologies and supplying different services. Such neighbouring bands may have been assigned to other firms using different methods, like a lottery or on a first-in-first-out basis, or may be in the hand of the military. Even if they wanted, neighbouring operators could not offer a 3G service if it was not permitted by their licence conditions. If that is the case, the licence prices then reflect a licence rent, not a spectrum rent, since a scarcity problem has been created artificially.

This should make clear that the focus of most of the recent discussion (auctions vs beauty contests) misses an important point. This is because licences are assigned to particular operators within the context of a prior allocation of spectrum across major types of usage. If this spectrum allocation is done suboptimally (as is almost certainly the case), then spectrum assigned for some specific purpose, such as 3G mobile services, is unnecessarily scarce. Optimality and scarcity here have an economic meaning, ie they reflect the interaction of demand and supply side considerations. An allocation that is efficient on a purely technical ground (spectral efficiency) need not be optimal from an economic point of view if consumers have a low willingness to pay for the services supplied. There may be another technology with lower spectral efficiency (less customers potentially accommodated in a given band) but higher economic efficiency if such a technology is able to supply higher-value services to the end customers.

The old model of spectrum management makes a distinction between spectrum allocation and spectrum assignment. Typically, spectrum is first allocated to a certain use, and then some mechanism is employed to assign licences. This distinction does not make much sense from an economic point of view. Individuals or companies that have a need for spectrum are different economic agents with different willingness-to-pay. This abstract figure can encompass 3G operators with alternative business plans, or a broadcaster that uses the spectrum to provide digital TV. Since the point about a market system is to create an environment in which scarce resources end up in the ownership of the agents that value them most highly, it is important that we introduce the price mechanism in the market for spectrum in the most flexible way. This means that individuals or companies should get property rights and be allowed to decide about the use they intend to make of their spectrum band, as long as they pay for it. Another consequence is that the number of licensees would not be determined by the regulator, but would arise endogenously from the working of the market place. This should not be a source of particular concern since transactions would be subject to competition policy in order to prevent harmful spectrum concentration. Also notice that – in case competition policy is not particularly effective – the attempt by an incumbent to exclude rivals from supplying a competing service would be made more difficult with spectrum trading. In fact, the incumbent would have to hoard much more spectrum in order to exclude rivals from all bands which could potentially be used to provide a substitute service.

Under a market system the opportunity cost of spectrum, in any one market, is evaluated by market participants and not by regulators. This is what makes a price mechanism particularly appealing. A competitive market system ensures that the resulting allocation of resources is efficient. Efficiency means that consumers get their preferred goods and services, and firms employ the most efficient techniques to supply those goods and services. Efficiency is the objective that regulators should promote, since society well being (given by the sum of consumer and producer surplus) is maximised at efficient allocations.

Alternative resource allocation mechanisms, in which some central agent decides who gets what, have generally failed. This is not because the central agent is corrupt or inept (even if this may be true in practice), but because of the informational burden associated with finding the efficient allocation. The central planner would need to know all consumers’ needs and all production possibilities of the firms. This is not realistic, and the central planner could not simply rely on asking the true needs and true capabilities to the other agents, since there would be no incentive to revel the truth. On the other hand, under a decentralised price system, each producer should only know the prices and its own individual technology in order to select its preferred (ie profit maximising) production plan, and each consumer should know the prices and his own tastes in order to buy his preferred consumption bundle. Knowing only the prices and local information such as own costs and tastes is enough for producers and consumers to make the choices that are required for coherence and efficiency. This is a fundamental theorem in economics (so fundamental that it is in fact known as the first – or fundamental – theorem of welfare economics) that provides the intellectual basis for a laissez faire approach to market economies. So long as government ensures a level playing field and enforces the rules of a competitive economy, no further intervention is required or appropriate.

There is an overwhelming consensus that scarce resources are better allocated by competitive markets than by regulators and, in this respect, spectrum should not represent an exception. There are various ways of employing the price mechanism, ranging from auctions to secondary markets and, more in general, spectrum trading. Unfortunately, there is a discrepancy between the current perceived spectrum shortage and the current system of spectrum allocation.

Competitive markets do work. Households and firm demonstrate this every day through their purchases and sales of goods and services and of labour. Of course, we are not denying that sometimes there are failures that could be corrected by the appropriate intervention of a regulator. However, the standard practice of competition authorities is typically to intervene only under extreme circumstances, recognising both that most failures are fixed by markets themselves and that attempts to correct such failures are not costless. This principle should be the starting point to envisage a market for spectrum. It would represent a dramatic change in the current spectrum policy, where an efficient market-based approach is not the default rule but rather the exception. We should reverse the reasoning: rather than starting with a list of potential market failures, to justify direct intervention, let the market decide how to allocate resources and intervene only in the presence of some serious market failure.

The benefits of spectrum trading are evident. Trade occurs only when there is a ‘double coincidence of wants’: when a seller is willing to sell, and he finds a buyer who is willing to buy, the act of exchange indicates that some extra value is created by the transaction. The absence of a market for spectrum would not allow for such extra value to be created and society would be worse off. It should be noted that even if initial rights are allocated via a well-designed auction, the absence of a market for spectrum would still cause potential losses, as the following example illustrates:

Example 1. There are two buyers, with respective valuations v₁ = 3 and v₂ = 2. If an auction is used (for instance a second-price sealed-bid auction or equivalently a simultaneous ascending auction), then buyer 1 bids 3, buyer 2 bids 2, the object goes to buyer 1 who pays 2. After the auction is closed, a third potential buyer appears, with a valuation v₃ = 4, hence efficiency requires that the object should go to buyer 3 who has the highest valuation. In the presence of spectrum trading, buyer 1 would resell the licence to buyer 3, say at a price p = 3.5, and buyer 3 would be happy to buy the licence (any price comprised between 3.01 and 3.99 would work as well). In the absence of spectrum trading, the licence could not be resold after the auction and society would lose the extra value created by the exchange, amounting to (v₃ – p) + (p – v₁) = 1.

In the previous example, the ability to transfer the licence clearly increases efficiency ex post. This may be due to various reasons. Entrepreneurs that appear late in the market, technological developments that were not anticipated ex ante and are captured by new operators, changes in consumer tastes, etc. It may also be a new technology that is patented later in time but was not allowed under the initial licence conditions. In the latter case, a flexible market for spectrum would allow buyer 1 either to sell the licence to buyer 3, the patent holder, or to buy the patent from buyer 3. In any case, we are in the presence of the important result of Nobel laureate Ronald Coase. Once property rights are defined in a clear way, and assigned to one party, trading will always ensure that the efficient allocation is reached. This proposition holds true even in the presence of a trade externality, since bargaining leads to an efficient allocation independently of the allocation of property rights. An appropriate application of the Coase theorem is the case of negative externalities caused by interference that we analyse below.

The current system looks orderly, simply because it is over-conservative and gives an excessive weight to interference concerns. It may have worked well in the past when the range and number of users was small. It has produced an order that – given current market conditions – is deceptive. If too little use of radio waves is made, there is obviously no problem to be immediately detected. However, the opportunity cost of keeping the current system is very high since entry and trade opportunities are severely curtailed by it. Those who oppose changes in the old system point that a market for spectrum would be too risky and bring too many market failures. We now turn to analyse these concerns. To be sure, there would be losers with the adoption of a new system of spectrum management. Inefficient incumbents that are insulated from entry, simply by lobbying the NRA and its central allocation system, would not be protected any more. Society as a whole, however, would unambiguously gain from introducing spectrum rights.

2. Potential market failures: how serious are they?

2.1 Technology I

Interference between adjacent bands is a serious problem. If the generator of the interference does not take into account the negative externality it imposes on other licensees when making its private choices, it may operate in a manner that does not maximise society’s well-being. However, if property rights are unambiguously defined and assigned to one party, the market would provide a solution to this problem. For instance, imagine a situation with some operators using low power devices in neighbouring bands. At some stage, one operator would like to use part of those frequencies for high power TV that would most likely interfere with neighbours. Such operator would not be allowed to supply its service unless the neighbours agree to be sufficiently compensated. Compensation may be of the extreme form of having to buy the entire neighbouring spectrum if TV emissions disrupt completely the other services. Clearly, this could happen only if the value created by having TV alone using that whole chunk of spectrum exceeds the value from the alternative use by the remaining licensees. This consideration would apply in many other cases. For instance, an operator in the TETRA band would want to use it for GSM, or sell it to another use. Under the current system this simply cannot happen. With spectrum trading, if switching from TETRA to GSM increased interference with users of neighbouring spectrum bands (not necessarily occupied by GSM operators), then the latter would have rights to compensation. The TETRA operator would switch to GSM only when it is efficient to do so, ie when the net gains – after paying any compensation – are positive.

These are precisely applications of the Coase theorem: trade would occur only if it increases efficiency, taking into consideration all the operators affected by the exchange. The key to achieving efficient allocations lies in identifying the gains from new services and the losses from discontinuing existing ones, and in the ability to execute trades in a timely matter. On the other hand, inefficiencies such as the ‘tragedy of the commons’ problem arise when there are rules preventing private ownership.

Regulation would still play a role also in this environment, but intervention would be kept at a minimal and non-intrusive level. Firstly, technical bodies – such as ETSI – would harmonise the practices of the various states relying on statistical models, developed by engineers, to quantify the level of signal interference experienced by a licensee. Secondly, the NRA should intervene in case negotiations among the parties fail. This arbitration task could be taken by a small committee of trained people, able both to understand the nature of the interference in question and to compute the amount of compensation required. It is important to notice that whilst the existence of an arbitration body is necessary to enforce property rights in a credible way, it should not need to intervene in the vast majority of cases since parties would have an incentive to negotiate a solution without going to court.

As a practical suggestion, the three-stage test on interference practised in New Zealand can be an appropriate starting point. The test checks if:

1. The emission is in the licensee’s band and area. If so, then no further action is required; otherwise it considers if
2. The emission is below background noise. If the emission is below acceptable levels of interference then everything would be fine; otherwise this would imply that
3. A third party user suffers from interference.

Evidence would have to be supplied by the third party, making the monitoring role of the regulator a feasible one. Once evidence is supplied, the original licensee’s source of emission would have to be blocked unless the two parties agree on some form of compensation. In case negotiations fail the arbitrator would have the power to resolve the dispute. Notice how a framework like this one allows dealing with negative externalities in a changing environment.

This approach can work in a market for spectrum because it is not too difficult to identify the source of interference problems. When this is not true, a case could be made for a third role for the regulator: as a co-ordinator rather than an arbitrator. The Coase theorem works well when there is efficient bargaining among the parties involved. This is realistic in the presence of a few parties, but could become more problematic as the number of parties increases. However, it is debatable if the role of co-ordination should be taken by the regulator or if a solution would be found in the marketplace anyway. If multi-party bargaining fails, it means that potential gains from trade are left unexploited, giving incentives to some agent to emerge as a broker and fill the role of co-ordinator. For instance, in the case of shared bands (such as PMR) a spectrum manager could co-ordinate the various users, and only the manager would be liable to the three-stage approach described above. Notice that the co-ordination role would also be diminished if property rights were clearly specified. To continue with the PMR example, signals could contain headers that would allow identifying each particular licensee, without needing a broker at all, unless the market deems this to be an efficient market structure.

Rather than delay entry by over-regulation, until too many questions are answered to the satisfaction of regulators and incumbents, rules should encourage parties to discover in the marketplace how to best satisfy consumer needs. Risk would shift from central agencies to private investors. New intermediaries could arise if needed. For instance, third party certification would allow entrants to contract for interference problems. Monitoring could be contracted to specialists, hence there is no particular need for having a central agency that monitors spectrum usage ex ante, ie before interference eventually occurs. Insurance companies could test systems for harmful emissions, insuring users and companies against damage. What a regime like this one needs is the clear definition of rights and of adjudication of disputes. An expert court could be instrumental to expedite the resolution of technical issues, while a specific regulatory agency is not necessarily needed.

2.2 Technology II

Spectrum is used in network industries, where technology adoption plays a delicate role. Whilst in markets without network effects it seems to be unambiguously desirable to allow multiple competing technological systems, this is less so in markets with network externalities: there are both advantages and disadvantages to having multiple systems rather than a single standard. On the one hand multiple systems bring product variety that is beneficial to consumers since it is more likely to find one’s preferred ‘variety’ in the presence of multiple varieties than in a market with a single type of product. On the other hand a single standard ensures compatibility that is desirable in many cases, and it may also lead to greater benefits from economics of scale in manufacturing.

The presence of (strong) network externalities typically leads to ‘tipping’ markets, where the winning technology takes the whole market. The theoretical literature does not provide an unambiguous answer to the question of whether the prevailing technology will also be the best one. Advocates of government intervention argue that imposing a single standard makes it possible to realise network externalities faster, and reduces the technological uncertainty among consumers. This is why GSM was adopted as a common 2G standard in Europe. Advocates of free markets point out that letting systems compete is the best guarantee to promote better technological systems (possibly a voluntary standard), and reduces the risk of being locked into an inferior technology promoted by the government (mandatory standard). For instance, because there was no mandatory standard in the US, it was this market that generated the new CDMA technology, which is the key building block for third generation mobile telecommunications.

The dispute can be rephrased as follows: one side believes standards generate markets, while the other side believes that markets generate standards. This debate does not preclude spectrum trading at all, but asks if some restrictions are needed to induce firms to employ particular standards. Since standard bodies often emerge, even on a commercial basis, and it is also questionable whether the regulator would be able to pick the best standard in the first place, prima facie evidence is in favour of allowing firms flexibility in their technological choices. In addition, spectrum trading would make less risky innovative activities. If a company decides to launch a new service based on a new technology (compatible or incompatible with existing ones), it can buy the spectrum it needs and market its product. If the product is successful, the company may want to expand by buying additional spectrum. If the product does not prove popular with consumers, the company could resell rather quickly the spectrum and reduce the losses from potential failures.

It is true that a standard is beneficial in industries with strong network externalities, but it is much less clear if there should be an interventionist approach de jure rather than waiting for the industry to deliver a de facto standard. Restrictions bear the risk of preventing spectrum from being used to deliver services that the public is most willing to buy, and to employ the technologies that firms believe to be the most appropriate. This trade-off does not have an easy solution. A mandated standard would not be subject to unintended consequences when the underlying technology is stable, but in that situation the firms would adopt the same standard anyway. On the other hand, an interventionist approach is much riskier in the presence of fast technological progress since it may lock an industry into an inferior technology. This is a serious hazard in a dynamic perspective, since a mandated standard may kill promising research trajectories at an infant stage if they fall outside the standard specifications. Of course, without any intervention there may be another type of risk, namely that firms are unable to co-ordinate, and waste too much time in a war of attrition, waiting to see if some uncertainty is resolved by the trials of competitors. In this case the intervention of the regulator could be helpful to create a critical mass of adopters. However, this is an action that can be done without affecting the terms of licence trading, for instance by promoting and sponsoring industry fora and research. The regulator should bear the burden of demonstrating why it believes that in certain cases industry ‘self-regulation’ on standard adoption is doomed to fail. Only in such cases one could use less flexible forms of licensing, for instance, apparatus licences or spectrum licences with more restrictive conditions.

2.3 Market power

There may be concerns that spectrum trading allows some operators to acquire excessive market power by buying an excessive amount of spectrum or by buying spectrum with the only purpose of taking such vital input away from potential competitors. This is a legitimate concern but it does not make the market for spectrum any different from other markets. Spectrum trading would be subject to ordinary competition policy. To give an example – in the presence of a market for spectrum broadcasting companies would be allowed to offer telephony services and vice versa. If a TV network buys some spectrum that is also appealing, say to a mobile operator because the TV network is able to supply better entertainment programs using that frequency band, then there should be no particular concern. If on the other hand, the true intention of the TV network is to pre-empt or predate against a mobile company, with the fear that the latter may enter the entertainment business, then the competition authority would intervene to stop the TV network from buying such spectrum.

As it is standard in competition policy, any analysis would have to start with the definition of the relevant market. One cannot say whether there is a spectrum market per se without having more details about the case under consideration. The demand for spectrum is in fact a derived demand. The relevant spectrum market will, therefore, generally be as broad as the relevant retail market for services provided over it. The relevant test is whether the price of one particular spectrum band is constrained by the price of some other band, including all the substitute goods at the retail level. If the spectrum is used as an essential input for the delivery of a particular service, then the relevant market would have to include demand-side and supply-side possibilities for that type of service. For instance, if an ex PTT acquires spectrum in order to supply broadband fixed wireless access, then the analysis would have to concentrate on the economics of the local loop. The fact that the ex PTT has also got spectrum to broadcast TV would probably not be relevant. On the other hand, if there are ‘spectrum managers’ then the total amount of spectrum they hold could be, in principle, a relevant indicator. To the extent to which the competition authority has enough technical expertise, there is no need to have spectrum specific regulation related to anti-trust concerns.

What is problematic in the old model of spectrum assignment that is currently in place, is that incumbents have an incentive to combine the interference problem with the market structure problem. The old model is inherently hostile to new competitors, in that it gives incumbents an opportunity to deter competition by leveraging interference concerns, and to eliminate entry that is automatically limited by the way spectrum bands are allocated. Policies that keep interfering transmissions at a minimum imply that spectrum is under-utilised. Current rules err on the side of having too little rather than too much interference. With such rules in place, equipment does not need to upgrade performance, as there is no demand for equipment to receive additional signals. Incumbents then seize on the least discriminating apparatus to test the effect of new transmissions and interference becomes a self-fulfilling regulatory reality.

Clearly, incumbents like all this since it is a system that excludes entry into their markets. Some commentators have argued that the block allocation system has historically served spectrum-based industries (in particular AM radio, TV broadcasting, and cellular radio) as a cartel enforcement device, limiting service competition by denying licences to newcomers and policing technical operations so as to lower industry output. The problems of interference are unlikely to be underestimated, as incumbents know how to be heard by the NRA. On the contrary, losses from deterring entry will be under-reported, and are under-valued, since potential winners are not well identified. Entrants wishing to compete or develop new technologies must apply for permissions, and assume all the burden of making a public interest showing. Even more fundamentally, consumer interests are generally not represented at all.

2.4 Social objectives

We argued in section 1 that the ability to transfer the licence and – more generally – to allow spectrum trading would increase efficiency ex post, after some initial allocation of property rights has been achieved. The downside is that secondary markets may decrease efficiency ex ante. For instance, an initial auction mechanism could have reserved a licence to new entrants, or introduced bidding credits for them, if it was believed that such entry was beneficial for society in terms of innovation capabilities, product variety and market competition, as the following example illustrates:

Example 2. There are two buyers, with respective valuations v₁ = 3 and v₂ = 2. These valuations are as in example 1 except that buyer 2 is a new entrant and the seller deems that the benefit to consumers from having a new entrant rather than an incumbent is quantifiable as 1.5. Thus the total benefit to society of allocating the object to 2 is 2 + 1.5 = 3.5. However, a straightforward simultaneous ascending (English) auction still assigns the object to buyer 1 since the individual valuations of each buyer remain unchanged and do not reflect society’s valuation. However, this does not pose a great problem. A simple modification of the auction restores efficiency. Each buyer is assigned a number that represents the social benefit if the object is allocated to him. His bid is computed as the sum of his actual bid plus the social benefit. In the case of example 2, this means that the bid of buyer 2 is augmented by 1.5. Then, it is easy to see that in equilibrium, buyer 1 still bids 3, buyer 2 still bids 2, but now the augmented bid of buyer 2 is 3.5. The object goes to buyer 2 who pays the bid of buyer 1 minus the social benefit, that is 1.5.

If a secondary market is allowed, then a new entrant may sell if its private value for the licence is below the private value of an incumbent that did not receive the licence. In example 2, the possibility of spectrum trading would make the ‘modified’ auction completely useless. After the auction is closed, buyer 2 would then sell its licence to buyer 1 at a price comprised between 2 and 3, and society would lose the extra value of having buyer 2 in the market, which was the main reason for modifying the auction in the first place. The additional value to society created by the entrant would not be taken into account in his private decision.

The example makes the point that social preferences expressed in auction rules, eg promotion of a new entrant, would be undone by trading. Of course, it is possible to reconcile the two things with some form of intervention. For instance, in example 2 buyer 2 can be allowed to resell the licence only to another entrant and not to an incumbent. Alternatively, if buyer 2 sells the licence it would have to repay its bidding credit of 1.5, so that it would sell only to somebody with a valuation higher than 3.5, which would be socially efficient. Adjustments of this kind could work, and can be implemented via restrictions on the licence conditions (type of use, etc.). However, once more we should posit the more fundamental question as to whether some specific categories of users and/or uses should be protected. The downside of such restrictions is that it is likely that the resulting spectrum allocation would be suboptimal and extra gains for society would be foregone. For instance, many licences contain rollout obligations. If unrestricted trading is permitted, these obligations would be lost. Hence, one would have to restrict, somehow, the terms of trade. However, these additional restrictions may prevent more efficient operators to emerge, say in densely populated areas, if they have no interest in supplying services to rural areas. A better alternative would be to try to disentangle rollout obligations from the main licence conditions. The very same social objectives could be achieved by a system of explicit subsidies, which could be allocated by an auction mechanism in rural areas alone. If too many restrictions are ‘bundled’ together in a certain licence there is little scope for encouraging more flexible arrangements such as spectrum trading.

A central problem is represented by the spectrum given to the military. Almost everywhere the spectrum is allocated and no fee is requested. This does not encourage at all an efficient use of the spectrum. Perhaps there is somebody else in the economy that may be prepared to pay a high price to use the same band. Since spectrum is a scarce resource with alternative uses, we should ask what society gains would be in the two situations. Security, national defence, scientific research and so on clearly have a value, and the military or the researchers could get some bidding credits accordingly. However, they would still have to bid against potential rivals using their budget. If they are not prepared to do so then an argument could be made on economic efficiency grounds that their spectrum should be returned. Since it may be difficult (but not impossible) to calculate the precise value to society of keeping some frequencies with the military, an alternative approach is to find a lower bound to it. Imagine, under a free competitive market for spectrum, there is a quantifiable demand coming from the private sector for a particular portion of the spectrum allocated to the military. It would then be in the hands of the government to decide if the benefits to society of not allowing the transaction (ie keeping the spectrum with the military) are at least as high as the sum offered by the private sector. This would be a budgetary decision contained in a political party’s platform over which citizens could have a chance to express their views, ie vote in a general election.

The main point made in this section is that spectrum is a very blunt and opaque tool to achieve social objectives. This is not equivalent to saying that some social objectives should not be promoted at all. On the contrary, governments do this regularly by putting a cap on the resources that can be spent over a particular class of services, while the decision on the mix of expenses is typically decentralised. The spectrum should not be an exception. In fact there is no conflict generated by the co-existence of public subsidies and privately owned spectrum. Police departments or emergency organisations compete for all sorts of inputs; there is no reason why they should be exempted from market participation in this particular case. Efficiency can only be achieved in an effective way if government would buy or rent the spectrum it desires to use at the market price rather than procure it by force, in the same fashion as it does for many other vital inputs – from labour to jets.

As we have just said above, the presence of a market for spectrum would reveal the opportunity cost of the current use of spectrum, say for national defence or emergency services. The aggregate value of the spectrum of the next best alternative could then be computed and should form (part of) the budget allocated to that service, that government may want to adjust upwards or downwards. The user (or department) would then decide whether to keep the spectrum, buy additional tranches, or sell it in order to free resources that could be used elsewhere. This solution would be very likely to achieve additional efficiency gains on top of the direct ones obtained by trading, for instance the MOD would have incentives to use less spectrum-intensive technologies. Under the current system, one could not simply ask if a certain band is really required by a user who received it in the past for free, or for a nominal sum, since the huge informational asymmetries would not elicit truthful responses. There is no way to know the efficiency properties of the status quo, unless one decides that users should confront a price system. A further downside of the current system is that a government department is not rewarded for saving bandwidth. On the contrary, if spectrum-saving technologies are expensive, it achieves the opposite effect.

2.5 Transition problems

Spectrum trading requires an initial allocation of property rights. Since licences offering similar services have been assigned using very different mechanisms (first-in-first-out, beauty contests, auctions, etc.) there may be a concern that such a disparity has to be addressed before a market for the spectrum is activated. From an economic point of view, the two issues are quite separate. If a licensee has received a license at a price below the market-clearing price, this firm will receive a windfall gain. However, efficiency gains from further trading are available independently from the way the licence was obtained in the first place. No delay in the adoption of spectrum trading should be advocated to resolve the problem of windfall gains. The solution to windfall gains has to do with corporate taxation and political decisions about equity rather than efficiency. On the other hand, delays in the introduction of spectrum trading would impose costs of missed opportunities on society.

3. How a market for spectrum could work

Spectrum trading should be as simple and flexible as possible. If constraints are imposed then some potential trade would not occur, and its associated value would be lost. In practice, this means that:

• Property rights and liability rules should be defined as clearly as possible in terms of frequency bands, geographic area covered, and duration.
• Licensees should have the flexibility to determine the services they want to provide with their spectrum, using the technology they deem to be the most efficient.
• Property rights should be transferable for sale or lease, in whole or in part.
• Property rights should also be given in perpetuity, with the understanding that compulsory purchase (with compensation) may be required under some extreme circumstances.
• As we pointed out in the previous section, there could be some potential market failures but they could be dealt with ex ante (interference limits, eventual standard adoption) and ex post (competition law).
• A public registry should be created to ensure transparency for private wireless users, effectively displaying information on opportunities and easing entry into unoccupied bands.

This approach is the appropriate benchmark. The next question is how to get there from the status quo – in a radical way or in a gradual one. A gradual approach may be required to avoid the case where lack of experience may disrupt services. This is a legitimate concern. However, it should also be pointed out that a gradual approach may just hide the vested interests of the major players that want to protect their own businesses from the threat of entry. At present, in some frequency bands, spectrum is the only bottleneck that impedes entry of new operators. A gradual approach that does not remove as soon as possible this constraint would be very costly in terms of opportunities forgone. Bearing in mind this crucial observation, a realistic option could be to follow something similar to the Australian experience.

Radiocommunications systems in Australia are licensed by the Australian Communications Authority using one of three licence types (www.aca.gov.au):

• Class licensing where anyone can use a particular type of device provided they comply with the terms of the class licence. Individual licensing is not required, and hence trading is not an issue.
• Apparatus licensing. This type of licence normally specifies the type of service that can be delivered under the licence as well as conditions including sites of transmitter, power, etc. Apparatus licences have been tradable in Australia since 1995. It is important to note that an apparatus licence traded still retains the original licence conditions (eg type of service, site, transmitter power) unless the ACA agrees to change the conditions of the licence. However, changing the conditions is a separate exercise not related to trading (ie an existing licensee is equally able to apply for a change in conditions of a licence). Apparatus licences are usually annual, but may last for a maximum of 5 years. They are commonly renewed on expiry.
• Spectrum licensing. It was introduced in 1992 with the passage of the Radiocommunications Act, but in fact the first spectrum licences were not issued until 1997. They are normally issued following an auction (though there are provisions, which have been used, allowing them to be issued for a pre-determined or negotiated price). Spectrum licences are technology neutral, the ACA does not specify the use to which they can be put, the power of transmitters or the sites at which transmitters/receivers can be deployed. Limits are imposed on emissions at the boundary of the licence in order to manage interference with a licensee’s neighbours. Spectrum licences are tradable in whole or in part, can be sub-divided or aggregated, provided technical conditions of the licence (eg out of band and out of area emission limits) are not breached. Spectrum licences are not renewable (unless it is in the public interest to do so) and are issued with a maximum term of 15 years. On expiry, the licence would be reallocated via a price mechanism.

The largest element of spectrum authorisations, in terms of both total number and spectrum amount, is still the apparatus licence; however, in terms of value this would not be true and spectrum licences would represent the biggest portion. Some apparatus licences in low value bands are also being moved to class licences. Trading among spectrum licences has taken place, and almost exclusively in terms of whole licences (ie licences have not been subdivided so far). As far as interference is concerned, there have been no problems. The ACA maintains a searchable register of licences to facilitate trading. It is also worth noting that an Australian merchant bank has just established a private on-line spectrum trading desk (with some ACA encouragement).

The Australian approach has much merit, although it has given too much weight to apparatus licences that may still represent an unnecessary burden. The downside of apparatus licensing is clear since it would dampen all alternative uses that do not conform to the licence requirements. Apparatus licences are often advocated when there is almost a one-to-one correspondence between some piece of equipment and the type of service that can be delivered with it. Apparatus licences would then reduce transaction costs and the user would be able to buy a single bundle (equipment plus spectrum). The question that one should ask is not whether this is a possibility (indeed it is a relevant one), rather why the regulator should intervene, substituting his judgement to the solutions that would be delivered by the market. Equipment manufacturers may themselves want to tie some particular frequencies into their equipment, however this may still leave the door open to some other user that needs the spectrum as an input for completely different purposes. This perspective reduces quite a lot the scope for apparatus licences in favour of more flexible spectrum trading.

3.1 Class licensing

Class licensing is typical for low power transmissions, where users are not paying for spectrum. A zero price for a resource should not be a surprise in economics; as long as supply exceeds demand there is no scarcity problem. The problem arises when a zero price is given to a resource that is scarce. Demand will exceed supply and some other form of rationing should occur, for instance allocating the resource on a first-come-first-served basis. Deregulated spectrum is then akin to common land. Everything works fine at first since land seems so abundant, but at some point too many people will try to exploit the land, resulting in the so-called ‘tragedy of the commons’.

To avoid this devices are permitted to roam across designated unlicensed bands, and interference is limited only by virtue of strict power limits, effectively granting users de facto private rights over adjacent space. The regulator regulates equipment manufacturers but leaves users free to police their own local radio waves. There are pros and cons with this solution. On the one hand, competition is enhanced. Where spectrum is not scarce, unlicensed bands may provide valuable service to the public. On the other hand, if co-ordination among users is important to efficient spectrum utilisation, this kind of open access is problematic. This possibility typically leads the regulator to mandate technical standards for unlicensed users, enforced through the equipment approval process.

There is a strong argument for class licences (open access) for very localised applications. In these situations users may impose interference on themselves (think of the interference between a TV set and Bluetooth technology). Hence the problem is internalised and causes less trouble. For instance, producers will take this into account when designing their products. Still, this observation does not prevent a market for spectrum. In fact, the presence of property rights may even spur competition for local area networks wireless standards. Technology suppliers could purchase rights to spectrum, standardising on their preferred systems.

In a way, class licensing represents the other extreme of the current spectrum regime. It gives too much protection to incumbents with licensed spectrum, while too little protection is granted for unlicensed spectrum. Such a regime considerably distorts investments.

The question is why, if spectrum sharing is efficient, it takes a government policy banning exclusive use to promote it. The usual argument is that, with large numbers of users, transaction costs and co-ordination problems make the Coasian argument for a market-based solution less appealing. This might be true, but is not then clear why a central agency could respond to market signals better than private spectrum managers. They could compete to attract traffic, charging subscribers, operators, or manufacturers for the use of the spectrum. To reduce interference – once again! – NRAs typically restrict power emissions. It is a much more appealing option to leave it to private licensees or spectrum managers to profit by discovering and implementing value-maximising traffic strategies. While there may be millions of cordless users, the owner of a band reserving spectrum space for DECT can minimise transaction costs by collecting fees from vendors only (like royalties). Non exclusive bands (shared use) are entirely possible in privately managed bands where the spectrum manager retains responsibility for all transmission. Despite the technical abundance assertion, unlicensed spectrum use is not free – which is why standards are called for (and debated fiercely). Clearly, further analysis of the costs and benefits of reducing the importance of unlicensed spectrum is needed. The preliminary arguments presented here suggest that unlicensed spectrum is more costly than alternative processes undertaken by spectrum managers who internalise costs and benefits from spectrum use decisions.

4. Conclusions

Spectrum is a scarce, valuable resource. This paper has argued that the current model of spectrum management is highly inefficient and should be replaced with decentralised solutions. The current model suppresses competitive entry, blocks efficient spectrum use, and insulates old technologies from innovative challenge. In the new system, the default rule should grant operators with the highest flexibility, with the regulator monitoring the proper working of competition rather than deciding who does what.

Policy-makers should focus on creating opportunities for radio spectrum use, making access to spectrum easier, making licences abundant and flexible. This will increase the spectrum rent (by enhancing the functionality of spectrum as a resource), but will decrease the licence rent (as competition grows).

The main implications are the following:

• Legislation should be amended to define in a clear and unambiguous way spectrum property (usage) rights. Property rights should contain limits on emissions at the boundaries of licences in order to manage interference problems and to provide guidelines in case of disputes.
• Property rights should be flexible and be given in perpetuity, implying that spectrum could be subsequently aggregated and disaggregated.
• We have identified potential market failures, but they do not represent an impediment to a market for spectrum. In particular:
• Competition policy should ensure a level playing field as in any other industries; and
• The regulator would act as an arbitrator in case of disputes over interference problems.
• Some additional sources of market failures do exist (standards, transaction costs, and social objectives) however they do not necessarily imply that less flexible arrangements should be imposed. One should not only question how serious these failures are, but also if the regulator would be able to find better solutions than the market.
• Only when additional market failures are very important, and the regulator can convince that the industry would be unable to self regulate, then spectrum trading could be constrained in some way, for instance replacing it with apparatus licences.
• These principles should be applied to all public services.