1 November 2021
After 200 days and 1727 rounds of bidding, Portugal’s mammoth 5G auction finally closed last week. ANACOM’s choice of auction format significantly prolonged the auction – alternative auction formats would have closed the auction much sooner. New entrants were successful, with two bidders acquiring spectrum in both the main and new entrant phases of the award. Despite both the long auction and high levels of competition, though, the prices paid by winners remained relatively moderate.
The objective of a well-designed spectrum auction is to determine, at the lowest price possible, how the available spectrum should be divided between the interested parties. The theory is that, under such circumstances, the spectrum is assigned to the user/users that will generate the greatest economic output from the spectrum.
In recent years, spectrum auctions have been criticised for being too expensive and taking money out of an industry that is already struggling to generate a positive return on investment. However, the cause of this – at least in our opinion – is due to many auctions being poorly designed. This has been for a whole range of reasons:
- reserve prices have been set too high – often to due pressure from governments to maximise revenues.
- spectrum unnecessarily being reserved for entrants or other uses – leading to artificial scarcity for the mobile operators.
- use of inappropriate auction formats and lot packaging (how the spectrum is divided into multiple blocks).
The Dutch 3.5GHz was a welcome example of a well-designed auction. The lots were structured sensibly (three 60MHz lots plus twelve 10MHz lots), a pragmatic auction format was used (a clock auction) and reserve prices were set low. There was little to artificially inflate competition and prices. The result was that each of the three operators gained 100MHz of spectrum for a modest fee – EUR58 million each, or EUR0.03 per MHz per head of population – which is amongst the lowest prices for 3.5GHz spectrum across Europe.
That said, not everything related to the Dutch 3.5GHz auction was best practice. Due to Inmarsat’s existing use of the spectrum, the auction was delayed by several years, leading the Netherlands being amongst the last countries in Europe to award this key capacity band for 5G. Also, only 300MHz of the band was awarded, with 100MHz being kept back for private networks. This could easily have led to artificial scarcity and high prices, as was seen in the German 3.5GHz auction. However, it appears that the operators showed restraint in the auction, settling for 100MHz each.
The modest amount raised in this auction will be relief to the Dutch mobile operators – who have historically paid amongst the highest prices for spectrum across Europe. Notably, a new entrant reservation led to extremely high prices in the 4G auction in 2012.
Auction design fiasco
The Portuguese Prime Minister António Costa described the auction format chosen by ANACOM as “obviously the worst possible auction model”. There were two major issues with the format:
- Use of small increments in the bid price each round – just 1% for the vast majority of the auction
- Use of the SMRA format with many lots in individual bands.
In an SMRA (Simultaneous Multi-Round Ascending) auction, participants bid on individual lots, meaning that the price of each individual lot rises independently. It is a widely used, easy to understand auction format which is perfectly suited to many cases – the main drawback is that it can lead to long auctions.
This was exactly what happened in Portugal. The number of lots, combined with the small bid increments, created a perfect storm. The auction had 58 lots, including 40 10MHz lots in the 3.6GHz band. Therefore, in a situation where two operators were competing over a single additional 10MHz lot, it would take 40 rounds for the price of all 3.6GHz lots to rise by just 1%.
Casting the mind back to 2019, a similar situation occurred in the German 2100MHz / 3.6GHz award – although the use of higher bid increments meant that this auction lasted three rather than ten months. In such circumstances, the use of a package clock auction (as used recently in Switzerland and Romania) or an SMRA-clock hybrid format (as used recently in the UK and Australia) would have progressed these important 5G spectrum awards significantly faster, as the prices of every lot in each band would rise each round.
New entrant success
We previously wrote about the potential for new entrants to succeed in the Portuguese market. The auction result revealed that not one but two operators cleared one of the key barriers to success: acquiring sufficient spectrum to build out an economically viable network. In fact, both new entrants (Digi Romania-affiliated Dixarobil and Másmóvil-backed Nowo) acquired not only the spectrum set aside for them in the new entrant auction but also 40MHz each in the 3.6GHz band, which will enable them to offer reasonable quality 5G services.
However, this potential shift from three to five operators in the market, which would dramatically increase the levels of competition, should be taken with caution. History suggests that five-player markets tend to consolidate and consistent rumours about a merger between Másmóvil and Vodafone in Spain suggest one option to achieve this. We also note that the combined spectrum holdings of Dixarobil and Másmóvil would generate a fourth player with a strong enough portfolio to challenge the incumbents on network quality.
We also note that Dense Air acquired more spectrum to support its neutral host model, complementing licences it previously acquired on the secondary market in Portugal.
Low prices paid
Despite the large number of bidders (hence elevated competition for spectrum), the headline prices paid were not particularly high – however, Portugal’s high annual fees meant that the overall benchmark prices are moderate. This is perhaps a function of the amount of spectrum available – with 400MHz of 3.6GHz on sale and the three smaller bidders happy to settle for a combined 120MHz, moderate prices are an understandable result.
Looking at prices paid in the two key 5G “pioneer bands”, we see that competition in the 700MHz band was limited, with prices well below European benchmarks. In the 3.6GHz, with six bidders acquiring spectrum, prices were relatively high – comparable to the German 5G award, but still lagging the Italian 5G auction.
Figure 1: European 5G benchmarks [Source: Aetha]
