The German spectrum auction: Failure to negotiate?

Telecoms.com periodically invites expert third-party contributors to submit analysis on a key topic affecting the telco industry. In this piece Stefan Zehle, CEO of Coleago Consulting, reviews the recent German spectrum auctions and questions the sums paid by the bidders.

The German spectrum auction ended in round 181 on the 19^th of June 2015 with Vodafone, Telekom and Telefonica paying a combined €5,081 billion. The average price paid across the 700, 900, 1800, and 1500MHz bands is equivalent to a reasonable 0.23€ MHz per unit of population (pop) and certainly well below prices paid at the AWS-3 auction in the USA which closed in February 2015 with an average price equivalent to 2.71 US$/MHz/pop.

However, the blocks acquired by the operators at the end of round 181 are exactly the same as the standing highest bids in round 34 but a with a total price of only €1,993. This means the three operators between them spent around €3 billion more than might have been necessary to achieve the same or a similar outcome.

The auction included the renewal of 900MHz and 1800MHz bands, as well as two new bands, 700MHz and 1500MHz. Germany is the first European country to auction the second digital dividend 700MHz spectrum (DDII), which consists of 2x30MHz. Its propagation characteristics make it ideal to provide coverage in rural areas as well as coverage consistency in urban areas and in-building coverage.  The 700MHz band is currently used by broadcasters and will only be available for mobile from 2017 and nation-wide in mid-2018. There is a 98% population coverage obligation attached to this band.

The renewal of 900MHz and 1800MHz is crucial for business continuity and the 2x5MHz block sizing prepares the bands for moving from GSM or 3G to LTE. As a competition safeguard the bidders were capped at 2x15MHz of the 900MHz spectrum, so at the very least each bidder was guaranteed one 900MHz block.

The 1500MHz band is unpaired (TDD) spectrum, intended to be combined with existing LTE bands to increase the downlink capacity. It could also be used as a stand-alone small cell solution or for eMBMS (LTE broadcast).

Unlike many other countries, Germany does not seek to extract the maximum cash from the sale of spectrum. German policy makers realise that the economic and social benefits which flow from operators building excellent mobile broadband networks are far greater than any short term financial gain from the proceeds of a spectrum auction.

The latter point is also evident from the relatively low reserve prices. The reserve price of €75 million for one 2x5MHz block (10MHz in total) equates to €0.093 per MHz per head of population (€75 million / 10MHz / 80.77 population = 0.093 €/MHz/pop). Many countries set far higher reserve prices. For example, the average reserve price for the 1^st digital dividend 800MHz auction in Europe during 2009-2012 was 0.50€/MHz/pop, and much higher in some countries, notably Italy with 0.80€/MHz/pop.

Exhibit 2:        Spectrum on offer and prices paid in German auction

Source:            Bundesnetzagentur and Coleago calculations

Exhibit 3:        Spectrum auction – MHz obtained and price paid

Source:            Bundesnetzagentur

From the outset, the stage was set for moderate auction prices. Telefónica O2’s acquisition of E-Plus in 2014 reduced the number of operators in Germany from 4 to 3. Consequently only the three incumbent operators – Telefónica O2, Telekom, and Vodafone – entered the auction.

A simultaneous multi-round ascending (SMRA) auction as the one Germany can be viewed as a negotiation between bidders, answering the questions of where do we settle, who pays how much and how much to we all pay. The information provided by the auctioneer during this well designed auction showed round by round the standing highest bidder and the amount bid, thus giving bidders the maximum of information to allow such quasi negotiation to unfold.

Initially this worked very well with the six 700MHz blocks being divided equally between the three bidders at the reserve price. However, the 10 blocks of the 1800MHz spectrum cannot be divided into three, and one bidder would have to end up with more than the two others. A 4-3-3 block assignment outcome could have been expected. In round two, Vodafone had for the first time standing highest bids on 4 out of 10 blocks, i.e. 2x20MHz. This is significantly above their current 1800MHz holding of 2×5.4MHz, and signalled a strong desire to change the status quo in 1800MHz.

In round 31 Vodafone increased bids to five of 10 1800MHz blocks – at this point the “negotiation” between  bidders seemed to fail.

Amazingly, in round 31 Vodafone increased this to 5 blocks. At this point the “negotiation” between bidders seemed to fail and it set the scene for the 1800MHz block to become the focus of a bidding war. Vodafone is likely to have had a high valuation for the 1800MHz spectrum. Acquiring 5 blocks (2x25MHz) as they did leaves Vodafone still behind Telekom (2x30MHz). However, it puts Vodafone ahead of Telefonica. For its part, Telefonica was defending its existing large holding in 1800MHz.

The bidding pattern indicates that Telekom and Telefonica would have been happy with 3 blocks each. They are likely to have accommodated Vodafone with 4 blocks and the auction could have concluded as early as round 32. The outcome in terms of spectrum acquired by each bidder would have been the same as the final outcome, except that Telefonica would have ended with three 1800MHz blocks instead of two and Vodafone with four instead of five.

The round 32 total bid amount was €1,954 million. Vodafone’s total round 32 bid amount was €657 million. However, bidding continued, and the final round 181 outcome saw Vodafone paying €2,091 million and gaining only one additional 2x5MHz block in 1800MHz. This means the marginal price paid for the additional block is €1,414 million, a chunky 1.77€/MHz/pop.

Exhibit 4:        Vodafone round 32 and final outcome compared

Source:            Coleago calculation based on Bundesnetzagentur data

It is likely that the marginal price paid for the 5^th 1800MHz block is well above the marginal value to Vodafone of that 5^th block. While the aim of an auction is to acquire spectrum most operators focus on value creation. For example, if operator A values 5 blocks at €100 and could acquire these 5 blocks for €90, then €10 of value would have been created.  If the same operator A values 4 blocks at €80 and could acquire these 4 blocks for €50, then €30 of value would have been created. The latter is the better outcome for A because it creates more value.

The fact that the 1800MHz bidding war spilled into the 700MHz when all was settled in round 1, is particularly annoying for shareholders. Had demand moderation prevailed bidders would have paid only €450 million instead of the €1 million they now have to pay for the same spectrum.

Demand moderation could have saved the industry a collective €3 billion. Given the narrowing free cash flow margins, this is not a trivial sum.

The result of the German multi-band spectrum auction breaks new ground because prices for 1800MHz spectrum ended up 44% above 700MHz and 25% above 900MHz whereas hitherto sub-1GHz spectrum was always deemed to be more valuable than spectrum above 1GHz. Other than the auction dynamics, there are reasons why this may no longer be the case:

The deployment of LTE in the 800MHz band already provides LTE coverage and better in-building penetration. Hence additional 700MHz spectrum has a lower “coverage value” compared previously acquired 800MHz spectrum.

With increasing mobile broadband traffic densities in cities and the difficulty to find suitable sites, the value of 1800MHz spectrum relative to sub-1GHz spectrum is likely to have increased.

European operators have GSM and LTE deployed in the 1800MHz band. With a total of 2x75MHz, the 1800MHz band is the widest IMT band, and thus offers the best opportunity to create wide LTE channels.

Other EU countries will also auction to 700MHz spectrum and will look at this result with great interest.

Technical Note

Spectrum below 1GHz propagates further than higher band spectrum. An operator would need around three times as many radio sites to provide the same coverage with 1800MHz versus 700MHz or 900MHz. Spectrum below 1GHz also penetrates walls better and hence provides better in-building coverage, i.e. were most usage takes place. Furthermore, within urban areas, sub-1GHz spectrum ensures that there are no LTE (4G) coverage blackspots. Hence deploying LTE in sub-1GHz spectrum produces a more consistent user experience.

Stefan Zehle is an expert in telecoms strategy, including marketing and forecasting as well as business planning and modelling. The scope of Stefan’s work encompasses mobile and fixed telecoms services and telecoms and IT products. Coleago Consulting Ltd is a boutique specialist telecoms management consulting firm staffed by highly experienced industry experts, advising telecoms operators and regulators around the world.