Auctions – the worst way to award spectrum! (apart from all of the other ones)

13 May 2024

Auctions have become the most popular way to assign mobile spectrum, replacing administrative methods (e.g. beauty contests) as the default mechanism. However, there is widespread criticism of auctions – notably that they take money out of the industry – and calls to revert to administrative approaches. Would this be a pragmatic ‘u-turn’ or a step backwards?

Auctions as a way to award spectrum

Spectrum auctions are transparent, objective and, if implemented properly, should maximise social value through allocating spectrum to users that generate the highest value for society.

However, recently, there has been growing criticism of auctions as a means to award spectrum. This is not surprising as there have been several examples of auction failures.

Regulators typically set objectives ahead of auctions, which could include revenue maximisation, efficient allocation of spectrum, increase in coverage or promoting a new entrant. To achieve its objectives the regulator designs an auction and sets up key auction parameters such as the overall format, reserve prices, annual fees, spectrum caps/reservations and coverage obligations. In the section below, we address some of these parameters and how they could lead to an auction failure.

Using the wrong auction format

The first, and often the most important reason why auctions fail is a selection of the wrong auction format.

The most appropriate auction format depends on various factors. Generally speaking, regulators should favour simpler formats such as the Simple Clock Auction or the Simultaneous Multiple Round Auction (SMRA). They are relatively straightforward to understand and allow bidders to understand their exposure at all times during the auction.

In contrast, auction formats such as CCA and first-price sealed bid formats could be a recipe for disaster. They require complex bidding strategies and can lead to erratic outcomes (e.g. 2001 Danish 3G auction, 2012 Swiss multiband auction, 2013 Norwegian auction).

Setting wrong reserve prices

The key objective should be to set the reserve prices low enough to encourage price discovery and competition in the auction but not so low that the spectrum can sell at artificially low prices. Importantly, high reserve prices can lead to valuable spectrum remaining unsold (e.g. Romania 2012, Australia 2013 and various Indian auctions).

Applying the wrong pro-competitive measures (spectrum caps and spectrum reservations)

Spectrum caps and reservations can play an important role to prevent a single bidder hoarding excessive amounts of spectrum. However, as they restrict either the supply or the demand for the spectrum, they can have a substantial impact on auction results. If set at an incorrect level, they can lead to inefficient auction outcome – either high prices or unsold spectrum (e.g. 2011 Belgium auction).

Poor lot packaging

How the spectrum on offer is packaged can have a significant impact on the outcome of an auction. If the spectrum is not packaged correctly, this could lead to an inefficient auction outcome (unsold spectrum or high prices). For example, in the 2018 Italian 3.5GHz auction, the spectrum was packaged into two lots of 80MHz and two blocks of 20MHz. This led to intense competition between the four MNOs and resulted in one of the highest prices for 3.5GHz spectrum in the world to date.

The alternatives

Other than auctions, administrative allocations (e.g. beauty contests) are the other option to award spectrum.

Administrative methods can take many forms, but generally bidders are evaluated on various criteria, possibly with different weightings. These criteria typically include factors such as business plans, proposed use of the spectrum, coverage and quality of service commitments, financial capabilities, and potential for innovation.

However, many people arguing against spectrum auctions forget the downsides of beauty contests:

• They can be inherently slower – administrative processes can be very time-consuming exercises, especially when candidates need to be evaluated on a number of (potentially subjective) criteria.
• Risk of inefficient allocation of spectrum – which is much larger for administrative processes than auctions. This is especially true for complex multiband awards – in an administrative process the regulator needs to divide the spectrum into pre-defined packages, which may not be an efficient distribution.
• Risk of legal challenges – these may be related to fairness, the selection of award criteria, transparency and fulfilment of the award conditions. Unsuccessful bidders are more likely to challenge the result than in an auction.
• Corruption – administrative processes are more susceptible to corruption as they are less transparent and more subjective.
• Principle-agent problem – administrative processes typically require successful bidders to make various (often subjective) commitments. To win the desired frequencies, bidders are incentivised to over-promise, then later to seek renegotiating these criteria.

Summary

Although auctions have justifiably faced criticism, this has largely been due to poor implementation (either intentionally to increase revenues, or unintentionally). A well-designed auction should raise the minimum amount of money to determine the efficient winners of the spectrum – no more! Whilst administrative methods may appear attractive to minimise auction spend, people forget their drawbacks – a lack of transparency and objectivity. It would be better to ensure that auctions are better designed than step back into murky world of administrative spectrum awards.