Tenure of rights (ARCHIVED)

1. Management rights are for a fixed terms of up to 20 years.¹² The history of the development of the New Zealand management rights approach is set out in the MEDPIB 35 document, and we do not propose to revisit that here, except to recap the NERA proposals which underpin much of New Zealand's ground breaking work on spectrum property rights, and where the MED acknowledges that NERA advocated that:

Duration of rights - On economic grounds alone, spectrum rights should be granted in perpetuity although the efficiency loss associated with fixed-period rights is likely to be small if rights for future periods are issued well in advance of the termination of existing rights. If the Government chooses to issue fixed-period rights these should take the form of 20-year rights with new rights issued, via an auction, no fewer than 5 years in advance of the time they take effect.

2. In our opinion, the 20 year maximum and fixed term right provided for in law does not represent the most efficient model for the development of a real market in spectrum. It is better than the 15 year term offered in Australia, but in our view it is not so much the length of tenure that is the issue, but the fact that it is for a fixed non-renewable term.
3. In Australia, this issue of tenure was considered by the Bureau of Transport and Communications Economics (BTCE) in its review that preceded the deployment of the Australian rights regime.¹³ It was more recently reviewed by the Productivity Commission.
4. While BTCE examined all sides of this issue, it recognised that a major argument against fixed terms is that there could be costs associated with the disruption caused at the end of the term - such as a loss in business during a changeover or additional costs associated with the changeover - which might outweigh any benefits of fixed terms.
5. The potential risk of a need to shut down a telecommunications network worth potentially billions of dollars at the end of an arbitrary licence term does nothing for the long term interests of end users or investors. The loss of quality in the network over the last few years of a fixed term licence as the owner reduces capital investment to factor the risk of losing its licence (or not having its licence renewed) is not in the long term interest of end users either.
6. Indeed, the decay in quality of a telecommunications network as capital maintenance dries up suggests that the last 5 or so years of a licence can be expected to become a self-fulfilling prophecy of business decline. NERA recommended that if a 20 year term was put in place, then this uncertainty in the sunset years of a fixed term needed to be addressed and resolved and so it advocated that the reallocation/renewal process be undertaken 5 years before the expiry of the licence. With all respect to NERA, while this might be a fine theoretical solution to the problem of fixed tenure rights, as a matter of practical application, even if the licence changes hands with 5 years of notice of a change in ownership, the disruptive effect on business, on investor confidence, and on consumers is still likely to be significant.
7. Further, if trading is considered, the lack of a full 20 year term following any secondary market trading [say] midway though a licence term implies that licensees buying spectrum in the market will have reduced payback periods for their investments. As the payback period shortens the further into a term the trade is made, the prospect of trades actually taking place that might improve the overall efficiency of spectrum use will decline. Such a fixed term, then, can be seen as a brake on innovation, a brake on trading, and a brake on improving the efficiency of spectrum use.
8. There are also practical issues that must be confronted with fixed term licences. Part of the idea of property-like rights is the capacity of rights holders to engage in trading, and in particular to engage in subdivision and aggregation of rights. We suggest that there may be a limitation on trading by these mechanisms in licences because of an absence of usable tenure. We believe that incentive to trade increases as tenure increases.
9. We note here that in real property, titles are typically for perpetual terms and the market is vibrant and active. There are exceptions, for example, property in the Australian Capital Territory. The Territory was established as the seat of Federal Government and so it is all Federal land, and private home ownership is on the basis of 99 year Crown lease over land. Since the establishment of the Territory was in the early 1900s, renewal of these leases has not been an issue until recently.
10. The very first leases in the territory will not be subject to renewal for some years, however, there has already been some public rumbling about the issues raised about the end of the term - whether the term will be renewed and whether there will be a renewal charge and what that charge will be. There is uncertainty created about this issue in the market. This example illustrates that even with a larger timeframe the fundamental issues are not removed, rather they are simply delayed.
11. An important distinction with spectrum licences as opposed to much real property is that networks utilising spectrum may take decades to develop and as such the potential loss or risk at renewal is measured by the cost of the dependant network, including its customer base rather than the value of the spectrum. As such related market and financing risk may also arise from any regulatory risk that spectrum licences or rights may not be renewed or renewed on materially different terms.
12. One of the fundamental considerations in developing a framework for disaggregation of rights is to consider how that framework will also support re-aggregation of rights in different configurations over time. In our opinion the New Zealand model of spectrum management rights is deficient in this area for it was designed not with trading as a goal, but with private band management as its goal. This contrasts with the Australian model that was built from the ground up to support and promote trading, including disaggregation and re-aggregation in all dimensions.
13. In the temporal domain, there are difficult issues that emerge to constrain the secondary market trading that flow from fixed term non-renewable rights and the potential non-alignment of expiry dates in adjacent licences. These issues emerge from a theoretical understanding of multi-dimensional rights.
14. Spectrum rights are multi-dimensional and at the highest order have the following six axes:

• x, y, and z cartesion co ordinations for defining points and areas and spaces in three dimensional space:

1. x denoting longitude;
2. y denoting latitude;
3. z denoting height (altitude or "ceiling"), and we note that the legislation specifically provides for a height dimension by stating that Parts 3 to 12 and 16 of the Act do not apply above 50km);

• frequency;
• time; and
• technical conditions (and we visit the implications of this for trading below under Variation of Technical Conditions). Note that technical conditions assume a collection of parameters.

15. Each of these axes has it own subtleties, rules and concepts, and indeed they can be broken into further dimensions, each with their own axis.
16. In the practical implementation of tradeable spectrum management rights, in a multi-dimensional rights system comprising n axes, an aggregation can only take place between two adjacent rights if boundaries defined in n-1 of the axes are identical, and the remaining axis defines full mutual adjacency.
17. To illustrate how this principle works, consider two adjacent rights, A and B. Both A and B are defined to cover all of New Zealand, to the 50 kilometre height, and so the x, y and z axes are all identical. Both rights have the same term (so the temporal axis is identical) and the same technical conditions. Rights A and B are frequency adjacent. In this case, all of the conditions for an aggregation in frequency are met and a new licence created post aggregation will be for a single entity of logical multi-dimensional space.
18. Consider, on the other hand the situation where all of the conditions in the previous example are the same, except that the end-dates of the licences are different. These licences cannot be aggregated easily into a single logical multi-dimensional space because fewer than n-1 of the axes of definition are identical. Such an aggregation cannot logically lead to the definition of fewer than two discrete spaces, and so two licences will always be the result unless there is an extremely complex rights recording and management regime in place. We specifically recommend against introducing such complexity.
19. The solution to this particular example is to define one of the licences for the combined bandwidth for the duration of the shortest term, and then create a residual licence with a future start-date, and expiry at the end of the latter expiry date for the remaining spectrum. Whatever the case it still leads to more than one licence defining a logical spectrum space.
20. Our investigations identified one such issue where aggregation has not been able to occur in relation to two adjacent properties with terms that differ by only a few weeks. On its face, this is an inefficiency though we recognise that a reasonable solution may not have been obtained for other reasons. 
      

Recommendation 2

Irrespective of any move towards perpetual rights or a statutory presumption of renewal (our recommendation below) the MED establish a formal rigorous procedure and rules to facilitate aggregation of adjacent rights.

1. As more spectrum management rights are created over time under the current rules, the probability that adjacent rights with fixed terms and different expiry dates will rise, and since these right cannot be logically aggregated into new licences, the existence of non-aligned expiry dates is in our view a constraint on trading. The problem does not emerge with perpetual rights or rights that enjoy a presumption of renewal.
2. The Productivity Commission reviewed tenure arrangements in its 2002 report on Radiocommunications. The Commission found that:
    
   The Commission favours the introduction of perpetual spectrum licences as a medium-term objective.¹⁴ 
   
3. However, the Commission noted a number of transitional issues to be dealt with and in the short term recommended that the re-auction of spectrum for expiring spectrum licences take place at least 3 years before licence expiry.¹⁵ The Commission in its report did not consider the practical concerns for managing aggregation and disaggregation of licences with different expiry dates that we have highlighted above.
4. In its submission to the Productivity Commission, NECG noted:
    
   As in property law where the Crown grants freehold interests in land but retains the ability to completely resume it and regulators have the right to shape planning and management of that property asset, the Commonwealth and the regulator could still be able to plan and manage the appropriate use of spectrum through appropriate and publicly transparent processes.¹⁶
5. Unless there are fundamental and compelling public policy reasons to the contrary, we urge the MED to reconsider the benefits of perpetual rights. While we do not believe a compelling public policy case exists which can not be overcome by tools such as compulsory resumption (discussed below), we do recognise that governments are reticent about deploying perpetual rights.
6. While the theoretical case considered by NERA, BTCE and the Productivity Commission presents the issue as a contrast between fixed term rights and perpetual rights and argues in favour of perpetual rights, there is also a half-way house that is worthy of consideration, and that is a fixed term with statutory presumption of renewal, except in tightly prescribed conditions.
7. This model might be implemented as an initial 20 year licence to preserve all existing rights, with a statutory presumption of renewal for 5 years, and then repeating every five years, but with some very limited and tightly prescribed justifications for non-renewal, such as a major change of band use mandated by the ITU, or some compelling public interest that requires resumption of the spectrum, say for public safety or national security. Such a resumption might need to be exercised at least five years before the next expiry to provide business adequate time to vacate, and it may be accompanied by provisions for compensation on just terms for the loss of infrastructure.
8. The presumption of renewal is used in Canada subject to the following:

• No breach of licence condition occurs;
• A fundamental reallocation of spectrum to a new service is not required (e.g. a reallocation by the International Telecommunication Union);
• An overriding policy need does not arise (e.g. a spectrum reallocation to address a national security issue);
• A consultation process would commence no later than two years prior to the end of the licence term (if the Department foresaw the possibility that a licence would not be renewed); and
• The imposition of any renewal fees and/or amendments to licence conditions for the initial licensees in the subsequent term would also be addressed in a consultation process which would commence no later than two years prior to the end of the licence term."¹⁷

29. Elsewhere in this report, we acknowledge the possibility raised in our interviews that some sort of regular fee for spectrum management rights might have some benefit of forcing licensees to treat spectrum assets as a recurrent accruing expense, rather than a one-off capital expense and so provide a balance-sheet incentive for licensees to regularly review their spectrum holdings and release the surplus to market. We note that a 5 year presumption of renewal would provide a trigger point to require additional fees to be paid for the management right to be renewed, and that such a process is likely to increase the efficiency of holdings by compelling licensees to regularly assess the utility of holdings, justify additional expense and fund an accrued liability on the current balance sheet.
30. In terms of our concern regarding trading of rights and especially aggregation, a model that provides for statutory presumption of renewal is able to deal with the issue of alignment of expiry dates. The date in fact becomes arbitrary and of no practical concern unless the Government exercises its discretion not to renew, in which case, our proposed threshold of a minimum 5 years notice to quit is able to be implemented with a combination of written notice to the new licensee, and some form of restricted term renewal.
31. We believe that there are significant theoretical and practical problems with the current model of fixed term rights with no renewal. While perpetual rights are widely recognised in theory as optimal, we appreciate and accept Government reticence to move in this direction and so we offer here a model that deals with our concerns as they affect trading while avoiding the full perpetual rights threshold that we feel may be unpalatable. 
     

Recommendation 3

The MED consider the practical issues surrounding of the current fixed tenure arrangements for management rights, and move to introduce either perpetual rights as recommended by NERA and favoured by the Australian Productivity Commission or alternatively a mechanism that provides for a statutory presumption of renewal on a five year term (guaranteed periods may apply to initial allocation) and a minimum of 5 years notice of any potential decision not to renew. 

32. We acknowledge a criticism of perpetual rights often raised by spectrum regulators that perpetual rights constrain the ability of the Government to satisfy changing international allocations especially for safety of life and other essential services. We believe this is a valid concern, but does not invalidate the goal of perpetual rights. Rather it suggests that Government should have a mechanism to recover rights should public policy reasons justify such a course. We should not abandon perpetual rights just because something may happen in the future and only in some limited circumstances - rather we should develop mechanisms that deal appropriately with these exceptions.
33. In Australia, this issue is dealt with in the Radiocommunications Act 1992 through provisions relating to the resumption of spectrum licences. There is a power for the ACA acting with due process to resume spectrum from a licensee with or without the licensee's agreement. Under the Australian Constitution, Governments may not resume without compensation on just terms and so the Australian Act sets out a schedule which defines a procedure for establishing a fair compensatory value for resumed spectrum.
34. We see a need for some sort of compulsory acquisition process where the Government has a clear public policy requirement to resume a perpetual right. We see compensation being appropriate and we understand the need for a mechanism to help assess that. Below in this report we propose that a market exchange could be a source of transparent and public pricing information to facilitate this process.
35. A specified resumption right may be seen as desirable with or without perpetual licences as the Government moves to place larger amounts of spectrum previously held by Government under its management rights into private management rights with long (twenty year) tenures. 
     

Recommendation 4

In conjunction with any move to perpetual rights or a presumption of renewal, a compulsory acquisition process with appropriate public interest criteria and a basis for just and fair compensation be implemented. 

36. In our interviews, we also encountered some concern that perpetual rights, while widely accepted and welcomed, may present an issue regarding incentives to release surplus spectrum. It was felt by one interviewee in particular that the introduction of an annual fee for perpetual management rights akin to the rates levied by a local council on land might provide an economic point of focus that compels rights holders every year to review the operational expenditure component of spectrum holdings. This view resonated with our own thinking about market incentives. Our model described above proposes a slightly different implementation to the one suggested by our informant, with a fee at the 5 yearly renewal, but it still addresses the issue raised.
37. The competition benefits of such a scheme are discussed below in the Competition Safeguards section. 

¹²S.34(g)(i) of the Act.

¹³Australia, Bureau of Transport and Communication Economics (1990), Management of the Radio Frequency Spectrum - Occasional Paper 102, AGPS, Canberra.

¹⁴Australia, Productivity Commission (2002), op cit, p.121.

¹⁵ibid, p.126.

¹⁶The Productivity Commission's Review of Radiocommunications Acts and the Australian Communication Authority - A Submission by Network Economics Consulting Group, 2002, page 12.

¹⁷Industry Canada, Spectrum Auctions Homepage, Frequently Asked Questions found at: www.strategis.gc.ca.