Twice-shy network operators reject high spectrum prices

Ofcom announced its long-awaited consultation on the auction of so-called 4G spectrum earlier this year. When the UK last auctioned frequencies for mobile phone networks, it enjoyed a £22.4bn windfall. That is unlikely to happen again

Ofcom announced its long-awaited consultation on the auction of so-called 4G spectrum earlier this year. When the UK last auctioned frequencies for mobile phone networks, it enjoyed a £22.4bn windfall. That is unlikely to happen again, say Stefan Zehle and Graham Friend of telecom auction specialists Coleago Consulting.

The German auction that concluded on 20 May 2010 provides a benchmark against which the UK can measure its expectations.

On offer were a total of 358.8MHz of paired and unpaired spectrum in the 800MHz, 1.8GHz, 2GHz, 2.6GHz bands. The auction was a combinatorial simultaneous multi-round auction. All blocks were auctioned simultaneously, and bidders could place a single bid for packages of spectrum blocks rather than bid individually on separate blocks.

The overall result was an average price of US$ 0.19 (€0.15) per MHz per head of German population ($/MHz/pop is the most common measure of price for spectrum). There were significant variations in price across the different bands but overall the prices were reasonable.

The key drivers of these reasonable prices were firstly the available frequency bands and secondly the number of bidders.

In the Indian 3G auction, which closed only a day earlier on 19 May, bidders paid up to US$4/MHz/head. This was mainly because there were many more bidders than available spectrum, and only three out of seven or eight bidders ended up with spectrum.

The number of bidders vs available spectrum also resulted in high prices paid in the auctions in the USA in 2007 and in Canada in 2008. It's down to economics class 101, supply and demand.

After the merger of Orange and T-Mobile the UK now also has four mobile operators, same as Germany. However in the UK we may a new entrant, ie BT, bidding for spectrum. This may increase prices.

In Germany, all eyes were on the "digital dividend" 800MHz band. This is spectrum freed up by the switch to digital rather than analogue broadcast television. The propagation characteristics of this sub-1GHz band are better for long range coverage outside urban areas and better for in-building penetration is dense urban areas.

Even so, prices paid for the 800MHz spectrum were relatively low, amounting to an average of US$0.91/€0.73 per MHz per head of German population.

In the US 700MHz auction in the top 20 areas, mostly cities, the top price paid was $4.17/MHz/pop. The US average price was US$1.18, which was only 30% higher than prices paid in Germany. However, the US has a much lower population density than Germany, and generally the higher the population density the higher the prices paid.

The reasonable prices paid for the 800MHz spectrum in the German auction are due to three factors:

• Only the four incumbent operators (Vodafone, Deutsche Telekom, Telefonica O2, E-Plus) bid for the spectrum. There were no new entrants.

• Unusually, the auction of the 800MHz spectrum was designed to provide mobile broadband coverage in rural Germany. This clearly depressed prices. Operators must first build coverage for 90% of the population in villages with a population of not more than 5,000 inhabitants; in phase two to cover towns from 5,000 to 20,000 residents; in phase three towns up to 50,000 and only in phase four can the spectrum be deployed in larger cities.

This is the opposite of normal practice where mobile operators roll out networks first where there is the greatest density of population in order to maximise the return on their investment, so, urban areas. The value of the spectrum to an operator will be lower if they have to invest in what they would normally regard as marginal or even uneconomic areas first.

• Vodafone and Deutsche Telekom were restricted to a maximum of 20MHz each out of 60MHz on offer. This meant the fight over the remaining 20MHz was between Telefonica O2 and E-Plus. In the event Telefonica O2 placed a higher value on the spectrum and outbid E-Plus, obtaining 20MHz; E-Plus had to content itself with spectrum in the 1.8, 2.0, and 2.6 GHz bands.

Prices paid for the higher bands were only a fraction of the 800MHz band. Since E-Plus did not buy any 800MHz spectrum, paid only US$0.06/MHz/pop (€0.05/MHz/pop) compared to the three others which paid an average of US$0.22/MHz/pop (€0.17/MHz/pop). Yet E-Plus gained two 10MHz blocks in the 2.0GHz band, the next most valuable band.

Prices paid for the 2.6GHz band with an average of US$0.03/MHz/pop (€0.02/MHz/pop) were low, but in line or even slightly above other recent 2.6GHz auctions in Europe. Some of these achieved only the reserve price of less than $0.01/GHz/pop.

The 2.6GHz band is the primary band for the deployment of LTE (4G) and is ideal to provide capacity for mobile broadband in densely populated areas. The coverage roll-out conditions do not apply to the 2.6GHz band and Germany has many cities with high concentrations of populations. This makes this spectrum potentially more valuable in Germany than in many other countries.

The difference between paired and unpaired spectrum

Paired spectrum (also called FDD - Frequency Division Duplex) is spectrum where one channel is used for the traffic from the network to the mobile and another (paired) channel from the mobile to the network. This is how all mobile networks work. It was fine for voice because essentially the traffic is the same in each direction. There is plenty of kit that uses this spectrum right now for mobile broadband.

Unpaired spectrum (TDD - Time Division Duplex) is where the up and down traffic is shared in the same channel. This is good for data traffic because more traffic travels from the network to the mobile rather than the other way around, a bit like ADSL. There is not much kit out there to make use of this though, but that's likely to change, notably through the use of more mobile broadband dongles.

Read more on Voice networking and VoIP