The UK spectrum auction is over. The winning bidders have been determined, the spectrum assignments within each band set and the required payments calculated. In the interests of transparency, Ofcom has now published the round-by-round bids made by each operator. This is the first time that this level of detail on the bids made during a combinatorial clock auction (CCA) has been published, and provides a unique insight into elements of the bidders' auction strategies.
The auction was a highly complex CCA with a second price rule, so both the bidding strategies and the task of determining the winning bids were far from straightforward. In this article, we examine--at a high level--some of the more interesting points that we can glean from this detailed bid data. In keeping with the auction process itself, we start by considering the bids in the principal stage of the auction (the primary and supplementary rounds) and then consider bids in the assignment stage.
The primary and supplementary rounds: significant variation in bidding strategies
Vodafone won 2×10 MHz of 800 MHz, 2×20 MHz of 2.6 GHz FDD and 25 MHz of 2.6GHz TDD spectrum. It bid strongly throughout the primary rounds (the "clock stage"), tending to reduce demand later than other operators thereby demonstrating a high willingness to pay. Given that no final price cap was in place for this auction, no bidder was guaranteed to win its final clock-stage package, but bidders did have the ability to secure this package by raising their bid sufficiently in the supplementary round. Vodafone did this, but also increased the size of its winning package by placing supplementary bids for larger packages. Through clever bidding in the primary rounds, Vodafone could also use the lack of final price cap to its advantage by bidding above final clock-round prices for some additional spectrum, comfortably securing a package of spectrum that appears to be very much in line with its goals.
Telefónica's O2 UK and Hutchison Whampoa's 3G UK (3 UK) won 2×10 MHz and 2×5 MHz of 800 MHz spectrum, respectively. Both operators bid for 2.6 GHz in the supplementary round, but these bids did not end up as part of the winning combination. O2 ended the clock stage with 2×10 MHz of 800 MHz (and one lot of low-power 2.6 GHz, which it likely never intended to win), so was faced with a difficult choice as it entered the supplementary round. When working within a fixed budget, a bidder can either put it all towards the spectrum it really wants or bid for more spectrum, which effectively reduces the amount it is bidding for its preferred package and, as a result, its chances of winning it. O2 made bids of up to almost £1.35 billion (€1.6 billion), but assigned the majority towards securing 2×10 MHz of 800 MHz, leaving only a small amount towards adding 2×10 MHz of 2.6 GHz. In the end, O2 paid only £550 million for its high-value 800 MHz, so could have comfortably won some 2.6 GHz within its budget. The operator did not know this in advance, of course, and may be an unhappy loser of 2.6 GHz spectrum in the sense that it would have been prepared to buy some at the prices others ultimately paid. Nonetheless, its apparent primary goal of winning 80 0MHz was achieved.
Everything Everywhere's (EE's) approach and result is more difficult to gauge. It began bidding on the maximum amount of spectrum that its cap would allow: 2×20 MHz of 800 MHz and 2×20 MHz of 2.6 GHz.(O2 won the coverage lot that was fixed at the top of the 800 MHz band. Vodafone was therefore likely to have preferred a lot as close as possible to the top of the band so that it would allow the operators to share antennas with lower risk of inter-modulation interference.)
At a clock price of £316 million, EE reduced demand for 800 MHz to 2×10 MHz and increased demand for 2.6 GHz to 2×30 MHz. At £403 million, it dropped demand for 800 MHz to zero and switched to 2×40 MHz of 2.6 GHz, before switching to only TDD spectrum by the end of the clock stage. This left EE with many options in the supplementary round, when it placed a large number of bids for a very wide range of packages to eventually win 2×5 MHz of 800 MHz and 2×35 MHz of 2.6 GHz. Whether this was its preferred package is not easy to tell. What is clear is that EE could have won no 800 MHz, or 2×10 MHz of 800 MHz given plausible alternative bidding strategies from its competitors. With the benefit of seeing all operators' bids, it seems that winning large amounts of 2.6 GHz spectrum at the cost of more 800 MHz was always likely. One possibility could therefore be that EE overestimated the demand from others for 2.6 GHz, which could be explained by O2 and 3 UK's supplementary round bids for 2.6 GHz not being as high as the primary rounds might have suggested.
Niche Spectrum Ventures (a subsidiary of BT) bid strongly for either 2×10 MHz or 2×15 MHz of 2.6 GHz and ultimately secured the latter, in addition to 20 MHz of 2.6 GHz TDD. How BT chooses to use this spectrum will be of particular interest to many observers. MLL Telecom and HKT were not successful in winning spectrum and their bidding strategies seem to indicate budgets of about £1 million and £10 million, respectively, to secure some TDD spectrum.
Following the principal stage of the auction, the winning packages and the majority of the required payments were determined, as set out in Figure 1. For some operators, there were small additional payments as a result of the assignment stage, which is described in more detail in the section below.
Figure 1: Operators' payments in the UK's 2013 spectrum auction, by stage [Source: Analysys Mason, 2013]
Click the image to see a larger version.
The assignment stage: network sharing and interference concerns drive bidding
Bidding was generally much more subdued during the assignment stage for individual lots. The largest point to consider for most bidders was the coverage lot in 800 MHz. This was auctioned separately from the other 800 MHz spectrum and therefore did not feature in the assignment stage. Bids in the principal stage suggested that most bidders considered the cost of the obligation to be low. The winner, O2, did not distinguish to any significant degree between its bids for the coverage obligation and non-coverage obligation lots, while Vodafone's bids for the two options suggested it had calculated a cost of between £0 and £30 million, depending on what other spectrum it won. EE appeared to be less prepared to bid for the coverage obligation relative to other 800 MHz lots.
At the assignment stage, the strongest bidder for 800 MHz was Vodafone, which appears keen to have avoided the lowest-frequency lots. This might have been the result of concerns about interference with DTT, but is more likely to have been driven by a desire to make network-sharing arrangements with O2 easier. (The relative amounts could have been different, but the overall cap of 2×105 MHz of spectrum meant that EE could win at most 2×40 MHz.)
EE also placed some lower bids, which resulted in a payment of about £8 million for Vodafone. At 2.6 GHz, BT and Vodafone both bid reasonably high amounts to avoid being adjacent to the TDD spectrum they were most likely to win. The safest way to do this was to secure spectrum in the middle of the band, which BT did through a bid of more than £60 million (although ultimately only a payment of around £15 million).
The release of the detailed bid data provides a fascinating look into the operators' strategies. While little can be determined with any real certainty, the information nonetheless holds many valuable lessons for future CCAs around the world.
Mark Colville is a senior manager and joined Analysys Mason in 2003 and has managed a wide range of projects dealing mostly with regulatory issues (across telecoms and media industries) and service pricing. He trained as a mathematician and has an MA from Cambridge University as well as a Diploma in Computer Science, also from Cambridge University. He can be reached at [email protected].