3. Future demand for 900 MHz spectrum

18. The Ministry has requested comments on the optimal quantity of spectrum required by current rightholders and new entrants in the 800 MHz and 900 MHz bands over the period 2011-2031.
19. This is a useful question, but it is not easy to answer with any degree of certainty. This is for two reasons.

• It is not a question of “requirements” in any meaningful sense. There is only an increase or reduction in costs associated with having access to less or more cellular spectrum rather than some “required” or “optimal” amount of spectrum. The “optimal” amount of spectrum depends on the price, which at this stage is still unknown.
• As we explain further below, demand for spectrum depends on a number of factors that can not be easily predicted twenty five years out. This underlines the importance of leaving decisions about spectrum assignments to those who are taking these risks (i.e., operators), within a framework of rules that encourages those who value the spectrum most highly to get access to spectrum that they will use.

This is not really a question of “requirements”

20. Generally speaking, for any particular technology, it is cheaper to operate with more spectrum rather than less: an operator needs fewer cell sites and less radio equipment. The value attached to spectrum in the model the Ministry is developing to help set renewal prices is precisely this cost saving.
21. So there is a cost for us of building and running our network with 21 MHz of 900 MHz spectrum as at present. Vodafone has built and configured its network to optimise the value of the spectrum which it has acquired over time. If we do not have access to 15 MHz of this spectrum, because we decide not to renew it, for example, our costs of operating will go up.
22. This cost increase is what we would be prepared to pay for continued access to the 15 MHz of spectrum that is being renewed. This cost increase, or at least the cost increase for a theoretical operator, is what the Ministry is attempting to assess in its renewal price modelling.
23. So there is no 900 MHz spectrum “requirement” in any meaningful sense, nor can it sensibly be said that we have more 900 MHz spectrum than we “need”. There is only an increase or reduction in costs associated with having access to less or more spectrum.

• If the value of this spectrum to us due to these cost savings is greater than the value that any other operator places on this 900 MHz spectrum, then we should retain access to the 15 MHz of renewing spectrum.
• If the value of this spectrum to others is greater than its value to us, then, mimicking the market price, the Ministry will set a renewal price higher than we are willing to pay, we will not renew this spectrum, and others will buy it at the subsequent auction.

24. In fact, we think that Vodafone is likely to be the highest value user of this spectrum. The only exception to this might be if access to 900 MHz spectrum is absolutely required for a new mobile entrant, i.e., without access to 900 MHz spectrum there will never be another entrant. We come back to the se points below when discussing how to maximise the value of this spectrum to society.

Future demand for 900 MHz spectrum is highly uncertain

25. The Ministry has asked for estimates of future spectrum demand by incumbents and new entrants in the 900 MHz bands. As noted above, in simple terms we can say that more spectrum is better, but of course it depends on the price that a firm has to pay to get access to the spectrum in question.
26. In our view, it is not possible to determine an optimal allocation of spectrum for existing operators or entrants over the next twenty-five years. Predicting future demand is a highly uncertain exercise with many inputs, including:

• Future services – sensitivity analysis shows that there can be wide variations in demand for spectrum depending on what services are offered. Given New Zealand’s service-agnostic approach to spectrum allocation, new services that use 900 MHz spectrum could also affect demand.
• Market position – the success of an operator will partly determine how much traffic it carries and therefore its demand for spectrum.
• Network topology – how an operator configures its network will affect its demand for spectrum.

27. We did provide some data on future traffic expectations in our submission on the optimal network model. We also pointed out the difficulties associated with predicting demand.

Future services are a key determinant of spectrum demand

28. The types of services that an operator provides are a key determinant of demand for spectrum. The amount of spectrum a wireless operator providing mobile voice services will want as compared with an operator focusing on mobile data services will be significantly different.
29. Vodafone is continuing to expand its product and service range to compete across the whole telecommunications market, i.e., including broadband, local access and toll calling as well as mobile voice and data.

• If we are successful, Vodafone’s wireless network traffic can be expected to dramatically expand to include traffic that was previously carried on fixed networks. This means Vodafone’s spectrum demand in the future can not usefully be calculated using mobile only requirements from the past.
• We are concerned that spectrum demand is being estimated using only mobile voice traffic requirements. This may be the source of TelstraClear’s preliminary view that 10 MHz of paired spectrum is sufficient for a nationwide cellular network (paragraph 13 of its submission).

An operator’s market position affects its demand for spectrum

30. The market position of an operator is another key parameter influencing demand for spectrum. The following table shows likely spectrum usage for different market positions and traffic demand levels.

Table 1: Estimated spectrum requirements for different levels of traffic and market shares

The above spectrum requirement figures were estimated using the network model developed for applying the optimal deprival valuation (ODV) methodology. Vodafone's projected 3G site count of [withheld under the Official Information Act 1982] is used. The traffic distribution is based on current Vodafone network. The technology is assumed to be the long term evolution of 3G.

31. The conclusion from the above table is that to accurately predict demand for spectrum, one needs to be able to predict both the traffic profiles of services and the market position of the operator out until 2031.

Network topology will also influence spectrum demand

32. Table 1 also shows that under the current network topology it is not going to be possible to support traffic at the levels indicated. But these are reasonable levels if wireless is used to compete across the whole telecommunications market.
33. Currently wireless operators use a macro/micro cellular network layout. So the table indicates that this topology cannot be used to deliver high bandwidth mass market services, the kind of services Vodafone is starting to provide to compete beyond mobile.
34. New technologies [withheld under the Official Information Act 1982] are being developed to address these issues. These technologies are in their early stages of
    development. It is too early to accurately assess the impact of these services on spectrum requirements.