1. Case-by-case review

Issue 1.1

What significant changes in cellular technology and in the cellular market do you foresee within the next 10-20 years?

15. The introduction of Wi-Max, Wi-Bro, world phones and combo phones will mean data and voice traffic will float between networks and services as and when the required data rate is available. Cognitive radios will further add to the ability of devices to actively select the mode (UMTS, EVDO, WIMAX etc) and band or a combination of several networks and bands simultaneously in order to achieve higher throughputs that customers may require. As illustrated in Graph 1 below, it is expected that as technology continues to develop, then the maximum data rates capable from each technology will evolve towards similar maximum levels assuming the same spectrum bandwidth is available for each technology.
16. Therefore, it is envisaged that no single band or technology will be more important than another. They will each have relative importance depending on location, the availability capacity of other bands and the relevant technology constraints.
     

How might these changes affect the future use of the 800 MHz and 900 MHz
bands?

17. We consider that the 800-900MHz bands will be best suited for rural areas, because, given their ‘bandwidth limited’ capacity ceilings and propagation characteristics, they provide better coverage at lower data rates. Graph 2 below shows how the higher data capacity requirements for urban and metro areas would be supplied by a mix of cellular technologies (e.g. Wi- Max, Wi-Bro, UMTS, EVDO), across all frequency bands (ie 800-900MHz, 1800MHz, 2.1GHz, 2.3GHz, 3.5GHz and even 5.0GHz bands), with each band or service sharing the data rates required by the customer.

Graph one: Overall data trend

Graph two: Overall data growth trend

Top

18. Therefore it could be argued that as emerging technologies become mainstream across several other frequency bands the value of 800-900 MHz spectrum is diluted somewhat by its limited bandwidth capacity.
19. It is illustrative to compare Telecom’s current use of the 850 MHz spectrum with the possible future uses.  

Current use:

AMPS A band (10 MHz FDD)

• AMPS A band can be utilised for seven CDMA 1.25 MHz carriers.
• Telecom currently uses up to three carriers for 1X voice and data traffic plus an additional carrier for EVDO.
• Some sites may be temporarily expanded to six carriers to support peak traffic periods (eg Christmas holiday sites and National Field Day events etc).
  

AMPS B band (10 MHz FDD)

• AMPS B band can be utilised for up to seven CMDA 1.25 MHz carriers.
• It is currently used for 333 AMPS/TDMA carriers of 30 kHz bandwidth each.
• Telecom currently supports approximately 300,000 subscribers on this AMPS/TDMA network, however these subscribers are scheduled to be migrated to CDMA before the AMPS/TDMA network shuts down in April 2007.

20. Possible future use of Telecom’s 850 MHz spectrum:

AMPS A band (10 MHz FDD)

• AMPS A will continue to be utilised to provide up to seven CDMA 1.25 MHz carriers for ongoing 1X and EVDO capacity.
• A band would support two times today’s traffic levels, assuming:
  • Significant data growth and fixed to mobile substitution;
  • All mobile traffic is on 850 MHz;
  • Continuing need to manage customer requirements at peak traffic periods eg Christmas holiday sites and National Field Day events etc.

AMPS B band (10 MHz FDD)

• After AMPS/TDMA closure, the AMPS B band will be available to support one or more of the following:
  • Further CDMA traffic growth (seven additional carriers x 1.25 MHz).
  • Development of higher data rate services through:
• EVDO Rev-B (multi-carrier up to 15 x 1.25 MHz, expected to be available in 2008)
• EVDO Rev-C (Flexible bandwidth from 1.25 MHz up to 20 MHz, still to be standardised, possibly OFDM, MIMO).
  • [ ] TCNZCI
  • Ease a transition should Telecom decide to upgrade the existing CDMA network to another technology, e.g. LTE in the longer term. We would need to migrate CDMA customers to a new network before removing CDMA carriers and deploying new equipment. This is the issue stopping the adoption of 900 MHz UMTS in Europe – the bands are too congested to enable enough free spectrum to install new equipment and migrate customers over. 

   Top
Issue 1.2

What is the current level of spectrum-related investment in the 800
MHz and 900 MHz bands?

21. Telecom is concerned that when it comes to investment, there should be equal focus on ‘coverage and new build’ versus ‘value add’. In general, Telecom operates an 18 month Capital plan. Considerable investment and costs are associated with Data platforms (e.g. Push to Talk, caller tunes, video streaming etc), business as usual (maintenance and performance enhancements) and new product developments.
22. Telecom has forecast to spend a minimum of [ ] TCNZCI this year on upgrading its existing 850MHz CDMA network to meet new business and customer demands. (See Appendix 1 for a breakdown of associated costs. Appendix 2 details Telecom’s mobile capital expenditure over the last few years).
     

How might the use of these bands be optimised?

23. The use of the bands can be optimised in two senses.
24. The first is investment optimisation – what is the best way to invest in the band. Given that it is normal business practice to avoid any unnecessary capital expenditure, we ensure traffic loading is maximised within quality of service limitations before additional equipment is deployed.
25. What is the most efficient way to use this spectrum. With regard to spectral optimisation, CDMA as a technology is very spectrally efficient (given that it only requires a repeat pattern of one). We use a single block of bandwidth nationwide, with additional blocks being added as capacity demands grow. Telecom therefore believe CDMA 850MHz is fully optimised at any point in time. This is as opposed to GSM which typically requires 3 blocks of spectrum nationwide as a minimum, where lower repeat patterns could result in further spectral efficiencies at a cost of more cell sites.
     

What are your intentions in the next few years with respect to this spectrum?

26. Telecom intends to extend EVDO, deploy REV A to metro areas. We may also add Rev B as well as enhanced services such as Push to Talk, video calling, location based services and other advanced services.
     

Issue 1.3

Should rights to unused spectrum be renewed? Why or why not?

27. Telecom will require all of its existing 850 MHz management rights to meet existing data forecasts in the short to near term. We therefore do not consider any of the 850 MHz is or will be unused spectrum and we seek to renew the full 20 MHz of 850 MHz that we currently hold. We are also concerned that limiting the 850 MHz spectrum available to Telecom would jeopardise our ability to offer higher data rates, and the advanced services that rely on high data rates, to customers. In particular, any reduction in the availability of spectrum may limit our ability to provide higher data speeds to rural customers.
    Top  

Issue 1.4

Which of the three options with respect to the length of the renewal
period is appropriate?

If you do not agree with any of these options, what alternative timeframes do you suggest? Why?

28. Option A would be our preferred period. That is, renew the management rights for 20 years, with a renewal price up front. This would provide us the security of tenure we require to commit to large scale investments. It is a sufficient term to ensure return on investment and payback on any large scale infrastructure investments that we may wish to make.
29. We require further information about Options B and C (for example, a formula for how they would be applied) to be able to evaluate them. However in general, any term less than 20 years would heighten investment risk and possibly restrict the flow of funds necessary to build desirable new wireless products and services.
     

Issue 1.5

Do you have any comments on the appropriateness of applying the Covec formula to the New Zealand cellular market?

1. Telecom believes that of all the methodologies presented by the discussion paper, the Covec methodology is the only approach that meets all the Ministry’s criteria and is therefore appropriate for the New Zealand market.
2. The Ministry’s stated aims are:
   1. For the Crown to receive a fair financial return for the use of the spectrum in the future period;
   2. To approximate market value; and
   3. To use a methodology that is transparent and simple to implement.
3. Also, in May 2003 Cabinet noted that in considering whether to reallocate rights it would seek to ensure that the rollover of rights did not compromise the ability of spectrum management to:
   1. Provide for economic efficiency, ensuring spectrum is allocated to its highest value use (thus maximising the value of spectrum to society as a whole);
   2. Allow for the technically efficient use of spectrum;
   3. Provide for security of tenure to encourage spectrum-related investment;
   4. Ensure that concentration of control of the spectrum does not inhibit competition in downstream markets (ie primarily through the application of appropriate competition regulation);
   5. Provide a reliable means to resolve interference problems, which minimises transaction costs; and
   6. Allow spectrum to be easily changed over time to higher value uses, which emerge due to changing technology and demand.
4. In 2003 Cabinet also confirmed its interim decisions that:
   1. Spectrum rights would be reallocated rights five years before expiry for a further 20 years subject to a case by case review to ensure consistency with New Zealand’s international radio obligations and the general objective of maximising the value of the spectrum to society as a whole;
   2. The Crown should receive a fair financial return for the use of the spectrum in the future period; and
   3. Spectrum rights should be reallocated to existing rightholders based on price-setting formulae that estimate the market value of the rights, and that, if existing rightholders do not wish to pay this price, the respective rights be reallocated by way of auction.
5. We believe the Covec methodology achieves the Ministry’s aims by:
   1. Resulting in a price that is likely to be close to that which would result from a fair auction process (as the measure of market value) given the range of realistic input variables;
   2. Offering the greatest degree of certainty as to the variables that will influence the final price, thereby being simple and transparent; and
   3. Being technology neutral and therefore simple and transparent. Covec enables determination of a price based on revenue streams, rather than requiring agreement on future technology choices and multiple cost inputs. An alternative such as IODV may unfairly penalise specific technologies eg CDMA as their spectral requirements differ from GSM for example.
6. We also consider the Covec methodology meets the aims of Cabinet by:
   1. Adopting a long term tenure that will support investment decisions e.g. 20 years;
   2. Allowing Telecom to renew all its existing spectrum rights, which would enable Telecom to meet future capacity requirements and deploy advanced services.Top
7. Contrary to commentary in the paper produced by PWC² [paragraphs 6.8 – 6.12], Telecom believes the Covec methodology is sufficiently robust to apply to the New Zealand cellular market. As acknowledged by PWC, it is transparent to implement and has a few simple steps:
   1. Take the initial 1991 auction value of the spectrum – V1
   2. Forecast forward industry revenue growth out to 2031.
   3. Determine the average year on year revenue growth percentage (z) from 2011 to 2031 and use this as a proxy for business net cash flow growth. See Graph 4 & Table 2.
   4. Apply the formula: V2 = V1(1+z)20 where V2 is the value (price) of the spectrum in 2011 and 20 is the license renewal period.
   5. Discount V2 (the “Price” in 2011) back to the period the payment will be made via the formula: V2 / (1+r)n where r is the relevant Weighted Average Cost of Capital (WACC), and n is the number of years prior to the renewal date (2011) that the price is paid e.g. 5.
8. Taking one step at a time from paragraph 35, it is possible to consider their respective elements in detail.
   1. Step a)
      Telecom paid $17.1m for its combined AMPS A & AMPS B spectrum in 1991. Since the cellular industry was relatively immature in 1991 it is possible that this price reflected expectations, which in reality, were either exceeded or have fallen short of. However the prices paid for spectrum in more recent auctions would suggest that $17.1m is a reasonable starting point. In light of the significant hype around the 2.1GHz auctions in 2001, and price points south of $17m gained for the three 15MHz parcels of 2.1GHz spectrum, using $17.1m as a starting point for Telecom’s 850MHz spectrum seems fair and reasonable. Table 1 illustrates a high level of consistency between the 1991 850MHz auction prices and the 2001 auction results for both 1800MHz and 2100MHz spectrum.
       

Table 1: 2001 spectrum auction results

2. Step b) Top
   To forecast forward industry revenues, and therefore revenue growth, it is necessary to develop some views on the future of the New Zealand cellular industry. Typically any industry will follow a growth path similar to the S-curve presented in Graph 3.

Graph 3: S-curve for industry growth

The question is whether the New Zealand cellular industry is closer to A, B or C in its growth to maturity? The evidence suggests that the industry is likely to be at C, and possibly B at most. For evidence of C, one only has to look to New Zealand’s mobile population penetration as per Graph 4. This, combined with the likelihood that new technologies will be spread across multiple spectrum bands, lends good support to the view that cellular services on the 850MHz band are in the mature phase of their life cycle.

Graph 4: New Zealand mobile population penetration

Top

Alternatively, A or B may also be valid starting points if 850MHz cellular services substitute other telecommunications or entertainment customer expenditure, or if machine to machine devices etc provide new growth opportunities. This is unlikely to be the case as higher bandwidth service in urban areas are likely to be provided through alternative bands. The level of uncertainty over whether point A, B or C in Graph 3 is the appropriate starting point is a key reason why relying on past data points to forecast forward industry revenue growth is somewhat futile. As per Graph 5 it is possible to scope out several scenarios that are realistic possibilities for the future revenue profile of the New Zealand cellular industry. While the uncertainty will never be removed completely, the simple and transparent forecast process makes it plain to everyone how the uncertainty has been treated.

Graph 5: NZ cellular industry revenue forecasts

If one believes C is the appropriate launch point in Graph 3, then this would correspond best with a log revenue forecast curve. In essence this could be a lower bound scenario. If one believes A is the appropriate launch point in Graph 3, then the linear revenue forecast used in Graph 5 might be best. Likewise, point B in Graph 3 might map better to the polynomial forecast curve in Graph 5. Following this line of logic, three revenue forecast scenarios can be used: linear (the upper bound), polynomial, and logarithmic (the lower bound).

3. Step c)
   From the forecast lines present in Graph 5, it is possible to calculate the average growth rate (z) between 2011 and 2031. This can be done via the formula:

Table 2 presents the growth rates (z) resulting from the three forecast trends in Graph 4.

Top

Table 2: Growth rates (z) resulting from forecast trends

Used as a proxy for business net cash flows, these growth factors are probably too high. Voice margins are rapidly shrinking for New Zealand cellular operators and there are no signs that these are, or will be, replaced by good data margins. Data revenues are growing, but these high cost products do not deliver the same margins as the traditional voice services. On the contrary, competition in data services is expected to increase, further negating any potential increase in revenue that data growth may have caused. However, it could be argued that this upward bias is outweighed by the simplicity and transparency gained by using revenue as a proxy for net cash flows.

4. Step d)
   Applying the formula V2 = V1(1+z)20 where V2 is the value (price) of the spectrum in 2011, results in the price points present in Table 3.
5. Step e)
   Discounting the 2011 price back to 2007 via the formula: V2 / (1+r)5, yields the price points in the last line of Table 3.

Table 3: 2011 and 2007 price points for the 20MHz of 850MHz spectrum

 
Having calculated the values (price points) present in Table 3, the question then becomes: which forecast curve is most likely to accurately reflect the future growth profile of New Zealand cellular operators? Obviously this is a subjective question and the answer will be based on beliefs. However, as the Ministry has suggested with ODV, a weighted average approach could be applied to the three forecast methodologies. For example, using a weighting of 1/3 for each of the logarithmic, polynomial and linear forecasts, produces an expected value (price) in 2011 and 2007 as follows:

• 2011 weighted average price – $28m
• 2007 weighted average price - $17m

38. Telecom believes this approach is very simple and transparent. Further,

Telecom is convinced that it:

1. results in a fair return to the Crown;
2. benchmarks well against recent spectrum auctions; and
3. would closely approximate the price resulting from an auction process.

Issue 1.6

Do you have any comments on the approaches that were examined as alternatives to the Covec formula in valuing spectrum for the New Zealand cellular market? Are there any other approaches that should have been considered and why?

39. In general we think that the Covec model is, given the information to hand, the only model that is simple to implement, is fully transparent and results in a fair market value and return to the Crown.
40. Our view is that the Covec model can usefully be applied in the circumstances and no convincing case has been made that the alternative (IODV) will produce a better result in terms of the Government’s policy criteria.