3. Incremental optimised deprival valuation (ODV)

Core approach

6. Vodafone’s view is that fair return to the Government for the renewal of spectrum is the market price of spectrum. As the ODV method is likely to provide the best approximation of the market price, this in turn will ensure that the Government receives a fair return.
7. Vodafone does not believe that ODV will be simple to administer. ODV requires modelling of an optimal network for a generic operator in New Zealand. A considerable amount of work will be required to develop this model. There is a significant amount of detail that is not covered in the MED paper which will need to be agreed with operators for ODV to work.

Definition of optimal network

8. The following parameters are key to the definition of an optimal network

• Technology
• Network traffic
• Network configuration 

Technology

9. The Management rights are being renewed for the period 2011 to 2031. So the technology used in the optimal network should be the technology that is likely to be used between 2011 and 2031. Systems beyond IMT 2000 will be the predominant technology in the 2011 to 2031 period (details outlined in Section IV).
10. Based on the information currently available to Vodafone the following parameters define Systems beyond IMT 2000.

Table 1: Systems beyond IMT 2000 parameters

11. The minimum bandwidth and spectrum efficiency parameters have the greatest impact on spectrum value using ODV.
     

Network traffic

12. We propose that an “S curve” is used to forecast network traffic. These curves are typical of what is found empirically when tracing technological adoption through time. They are used in TSLIRIC models to forecast traffic in other countries e.g. UK, Netherlands and Sweden . Therefore the rationale for the shape of the curve is well established.
13. However, calibrating an S curve to 2031 is difficult.
14. The number of operators determines the network traffic. Currently there are two operators. We think that a three operator assumption would be useful for sensitivity analysis, but less likely than a two operator scenario.
15. The two operator view is supported by Telstra’s public statements, analysis by CSFB and Citigroup Smith Barney, although Goldman Sachs differs.
     

Network configuration

16. We suggest using the actual site numbers to define the optimal network configuration.
     

Key parameters for applying ODV

17. Our work to date shows that the following parameters are key to the definition of the optimal network

• Spectrum holding
• Discount Rate
• Cost of Network

Spectrum holding

18. The ODV methodology gives no guidance on the choice of spectrum holding. The amount of spectrum holding that the model incorporates will be critical to the
    valuation as it represents the assumption about scarcity and scarcity in ODV will drive the eventual value.
19. The problem with the model being used to choose the spectrum holding is that it determines scarcity which in turn determines price. But in a market the price determines the holding. This is a circularity problem.
20. We suggest then that the spectrum holding be set at either the hol ding that has resulted from an actual market which is the existing holdings. Or the holdings that would result from a theoretical market which would be equal portions for each operator.
21. A third method for setting the spectrum holding is to use the network model to generate the required spectrum at the end of the period and this is the spectrum holding that a rational operator would choose to purchase at the beginning of the period. This gives a reasonable spectrum holding based on the traffic forecasts used in the model.
22. However, this third method assumes implicitly a zero price for spectrum, results in a high spectrum holding, and therefore generates a low price for spectrum. This is an example of the circularity problem mentioned above.
23. Vodafone proposes that either the actual or the holding that would be generated from a theoretical market be used. This is the only way we can see the model simulating a market outcome.
     

Discount rate

24. Vodafone believes that a commercial risk adjusted discount rate which would be applied to capital budgeting of network investments should be used to value the spectrum. We think that over such a long period, a commercial rate is most appropriate because it represents a risk adjusted return on a network investment, and this corresponds exactly to an investment in spectrum as they are complementary investments.
     

Network costs

25. While it is possible to work out the total network cost, we think that the network changes that would result from depriving a network of spectrum would be confined to the radio network. We have consequently assumed that the core would remain unchanged with changes in spectrum.
     

Benchmarking

26. Vodafone does not support the Ministry’s suggestion to use benchmarking to verify the spectrum price calculated by the ODV method. Vodafone analysis shows that benchmarking gives a wide range of answers and the answers do not seem to reflect the New Zealand market situation.
27. In general we are cautious about international benchmarking. It is always difficult and contentious and influenced strongly by the assumptions that underlie the approach. In our view benchmarking can be useful or indicative at a high level, but is unlikely to be helpful for making decisions in detail.
28. As set out below, Vodafone has used two benchmarking methods to estimate the spectrum price in New Zealand using overseas spectrum prices. Both show that the benchmarking method proposed in the discussion paper is not likely to give a renewal price that will represent the New Zealand market.
     

Simple benchmarking

29. The simple benchmarking method uses the following formula to work out the spectrum price in New Zealand.

Table 2: Price of 2x15MHz of spectrum in New Zealand based on 2G and 3G prices paid in other countries ($NZ, million)

30. The parameters used for calculating the above figures are outlined in Appendix A.
31. From the data in Table 2 you can see that the benchmarking method has a very wide range. This makes the method unhelpful as a way to test an ODV result. It is also obviously strongly influenced by the high prices paid in some European countries for 3G, which reflected market conditions at the time but which do not seem directly relevant to determining the market value of 900 MHz spectrum in New Zealand in 2011.
     

Regression benchmarking

32. The second benchmarking methodology we used was a linear regression model to forecast the expected price of spectrum. We regressed spectrum cost against population, population density, level of competition, spectrum bandwidth and licence period.
33. Figure 1 shows how the actual and benchmarked spectrum prices compare.

Figure 1: Results of regression benchmarking approach

34. The parameters of the regression analysis are outlined in Appendix B.
35. As can be seen from the chart, the model performs fairly well for Australia and Germany, but is inaccurate for other countries. It actually suggests that spectrum should be subsidised in New Zealand (i.e., have a price less than zero), which just underlies the unhelpfulness of the approach.
36. There are certainly other benchmarking methods that could be developed. But in our view they are unlikely to help resolve the fundamental uncertainty about the value of the renewing spectrum in New Zealand.
37. Vodafone proposes that benchmarking is not used to verify the renewal price derived by the ODV method.