Spectrum auctions that allow bidders the flexibility to combine small spectrum lots need measures to protect competition in the downstream market. Being able flexibly to build spectrum packages that best suit the specific needs of a bidder has clear efficiency benefits, but also opens up the opportunity for bidders to buy up spectrum solely to deny their competitors access to an essential resource. A common approach is to impose spectrum caps that limit the amount of spectrum any individual bidder can buy, or to reserve spectrum for particular bidders in order to guarantee new entry. As discussed in this article, both approaches pose some difficulties, which have led to a novel way of safeguarding competition that is being considered1 for the upcoming spectrum auction in the UK.
Why do we need competition safeguards?
When allocating licences for the use of radio spectrum, whether by auction or other means, most regulators will consider the implications for downstream competition. Certainly it would not be ideal to end up with a single organisation as the sole supplier of, for example, the latest mobile telephony services for the next 15 years – monopoly prices anybody? And of course the benefits to consumers are not simply limited to lower prices – competition can also result in an increased range of products available to consumers and can encourage firms to invest in the development of new and higher quality products. This means that promoting (or at least maintaining) downstream competition is a key objective in spectrum awards.
Unfortunately, market incumbents2 generally are not too keen on effective downstream competition; they do not want to share their customers or reduce their prices. Indeed, the long term benefits of keeping competitors out of the market may generate sufficient incentives to buy up spectrum in an auction simply to deny access to this vital resource to competitors, and large incumbents often have the financial means to pursue such a strategy. If this is seen as a likely outcome, smaller players and potential entrants could be put off participating in the auction altogether. Why bother investing the necessary time and money if they have no hope of winning any spectrum?
The regulator therefore needs to take measures to prevent this kind of abusive behaviour, offering new entrants a realistic expectation that participation in the auction could be fruitful, but without overly restricting the incumbents’ potential for acquiring spectrum themselves, or the flexibility to put together larger packages of spectrum that is a key benefit of using an auction with a flexible lot structure in the first place. This is where competition safeguards come in.
Caps and reservations – some examples
A regulator is selling spectrum in two bands; Band A and B, each with 50MHz of spectrum split into five generic 10MHz lots. Bidders can place bids for spectrum packages, specifying the number of lots in each band they wish to acquire, allowing them to flexibly build a package that best suits their requirements. Suppose that the regulator wishes to ensure that at least five bidders can have access to some spectrum. Some (but by no means all) of the options that might be considered are:
- a cap across all of the available lots – bidders cannot win more than a total of 20MHz in both bands;
- a cap on a subset of the frequencies on offer – bidders cannot win more than 10MHz in Band A, but are free to bid for as much of Band B as they like; or
- different caps on mutually exclusive subsets of lots – a cap of 10MHz for Band A and a cap of 10MHz for Band B (note that this is different to cap of 20MHz across both bands).
Multiple caps that apply to overlapping subsets of lots – bidders cannot win more than 20MHz in total, and no more than 10MHz of this can be in Band A.
Consider the spectrum release described in Example 1. However, suppose that there are already three operators in the downstream market, and the regulator wishes to promote the entry of at least two more. If 20MHz of spectrum is considered sufficient for a new entrant to be a viable competitor in the market, the regulator could reserve 40MHz for new entrants. This might be done by specifying specific frequencies on which the incumbents cannot bid (say, for example, the lowest 20MHz Band A and the lowest 20MHz in Band B, or the lowest 40MHz in Band A or Band B).
Consider the scenario in Example 2, but suppose the regulator is concerned that a “strong” new entrant will buy up the entirety of the reserved spectrum, leaving nothing for a second new entrant. In this case the regulator might:
- reserve 40MHz of spectrum for new entrants (either generic lots or specific frequencies); and
- prevent new entrants from bidding on more than 20MHz of the reserved spectrum.
The strong new entrant would then be able to acquire the spectrum it needs, without being prohibited by the (stronger) incumbents and without being able to use its own position to strategically block smaller competitors from entering the market.
Spectrum caps impose upper bounds on the amount of spectrum that bidders can win. They offer a way of ensuring that the incumbents cannot buy up too much of the available spectrum and that there is sufficient left for new entrants.
Depending on the quantity and characteristics of spectrum available, there is potentially a wide variety of ways in which spectrum caps can be applied. For example, one could impose a cap across all of the available spectrum or only on a subset of the frequencies on offer. Where the spectrum can be split into categories, different caps could be applied across various subsets of the available frequencies, and we might even see some of the spectrum included for multiple caps.3
Spectrum caps may be applied on a bidder-by-bidder basis, typically to take account of existing spectrum holdings that may complement any additional spectrum won. Players with existing spectrum licences (the incumbents) will likely need to win less additional spectrum in order to roll out their services after the auction. With this in mind, tighter spectrum caps could be applied to the incumbents, offering even greater scope for entry to the market and allowing for a more even distribution of spectrum holdings post auction.
Another common approach, setting spectrum reservations, allows for a pre-specified amount of spectrum to be allocated to bidders who satisfy certain criteria, e.g. those who qualify as new entrants. Reserved spectrum works in a similar way to spectrum caps in that it ensures that some spectrum cannot be allocated to the incumbents and is available for new entrants to compete for amongst themselves. Indeed, spectrum caps can imply an effective reservation of spectrum if the total amount that can be acquired by incumbent bidders under their caps is less than the available spectrum. One of the main differences between spectrum caps and reservations is, however, that the latter create an asymmetry between entrants and incumbents that can be abused strategically: those who can bid for reserved spectrum may be able to drive up the prices paid by those who cannot without fear of retaliation, as they can switch between reserved and unreserved spectrum, but the other bidders are prevented from doing so.
Pick ‘n’ mix
Of course, there are circumstances in which it might be desirable to combine the use of spectrum caps and spectrum reservations, particularly if the asymmetries between bidders are more complex than simply incumbent versus new entrant.4 Suppose a regulator has decided to reserve some of the spectrum for new entrants, but wishes to ensure that at least two new entrants (of potentially differing financial power) will have access to the amount they need. This could be achieved by imposing further restrictions, such as a cap on the amount of reserved spectrum on which each of the new entrants can bid. The addition of the cap would ensure that any differences between competing new entrants could not be abused for strategic purposes.
Plans for the upcoming multiband auction5 in the Netherlands include setting a spectrum reservation in the 800MHz and 900MHz bands for applicants who qualify as new entrants. The quantity of reserved spectrum (a maximum of 2x15MHz) will be determined based on the number of potential new entrants and preferences expressed for the 800MHz and/or 900MHz spectrum in their applications. In addition to this, the auction will impose a cap of 2x10MHz on the amount of reserved spectrum that each new entrant may acquire.
The difficulty in getting it right
The decision on exactly how spectrum caps or reservations should be set up will depend largely on the value of, and demand for, the spectrum and its potential usage in the downstream market. Suppose spectrum in Band A will allow new entrants to roll out services to compete with the incumbents, but Band B is of no use to them. Then applying a cap for Band B, or reserving some of Band B for new entrants, would be ineffective as the new entrants would not care how much of that was acquired by the incumbents. If, on the other hand, spectrum in Band A and Band B is substitutable, an overall cap or reservation might be a more sensible approach.
One must also take a realistic view on the prospect of new entry. If there are simply no organisations wanting to enter the market (even if they were given the spectrum for free), then as much as the regulator wants to promote competition, there is little they can do. It would be inefficient in that case to impose protective measures for potential new entrants who do not exist, leading to unused spectrum that could be better utilised by the incumbents for improving their services.
It is important that the regulator gets the safeguards right, as the consequences for not doing so are potentially serious. Setting spectrum caps at the wrong level risks preventing some or all bidders from building the spectrum packages they need – if caps are too tight, it might not be possible for bidders to acquire enough spectrum to deliver high quality services; if caps are too loose, they may not provide much of a safeguard. Similarly, spectrum reservations that are not correctly balanced could either leave new entrants unable to win sufficient spectrum, or result in reserved spectrum going unsold (due to lack of demand from new entrants) when it could have been utilised by the incumbents.
Outcomes that leave firms unable to compete and/or result in valuable resources going unsold represent a serious inefficiency, leading not only to lost revenues for the government but more importantly to lower social welfare. It is important then that whoever sets the competition safeguards has a clear and realistic understanding of the downstream market and the likely demand in the auction, in terms of the number of firms likely to bid and the minimum spectrum requirements of each participant.
A novel approach
Simple spectrum caps and reservations are not the only ways of safeguarding competition. Indeed, alternative methods may provide greater flexibility for achieving pro-competitive objectives, although potentially at the cost of some additional complexity. For example, Ofcom are proposing the use of “spectrum floors” for particular bidders in the upcoming UK 800MHz and 2.6GHz auction.6
The basis for this approach is that the provision of LTE services (the most likely use of the spectrum) in the UK is considered to require access to a minimum amount of spectrum in certain bands (including some of the spectrum already held by some firms in the UK telecoms market). The application of spectrum floors in combination with spectrum caps not only ensures directly that, where possible, no bidder acquires too much spectrum, but also that a sufficient number of winners will end up with the necessary resources for competing effectively in the downstream market and making use of new technologies.
Spectrum floors essentially set a flexible spectrum reservation for the relevant bidders (those who have qualified and chosen to benefit from the spectrum floors), as there are different ways in which a bidder could obtain sufficient spectrum to achieve the floor. The reservation is then being applied in a way that best fits with the demand from other bidders. The objective of ensuring entry (or sustainability of existing competition) is achieved in a way that minimises the restrictions imposed on (and distortion of) the auction outcomes and the opportunity cost of not awarding the reserved spectrum to bidders not amongst those benefitting from the spectrum floors.
The flexibility of the model, and the way it can respond to the demand expressed during the auction, removes some of the risk of inefficient outcomes and unsold spectrum through misspecification of simple spectrum reservations. It offers benefits to both bidders, who are less likely to face losses through an inefficient allocation, and to the regulator, upon whom the burden of setting the safeguards at the right level is significantly reduced.7 However, it adds complexity for the auctioneer, and works only with certain auction formats (in particular combinatorial auctions with a winner determination process that allows the incorporation of constraints on permissible outcomes).
Competition safeguards are a way of providing incentives for smaller players and potential new entrants who might otherwise be discouraged by their limited chances of obtaining any spectrum to participate in an auction, thus promoting competition in the downstream market. These measures, however, must be carefully designed, as misjudgements could have serious and detrimental consequences, condemning the downstream market to poor quality services and limited innovation. This article attempts to give an idea of the complexity involved in some of the issues regulators need to think about when approaching these options. The trade-offs that exist and the correct choice obviously depend on the specifics of the award under consideration.
Print This Post
- As of August 2012. [↩]
- For simplicity, unless stated otherwise, we generally distinguish between new entrants, who are not currently operating in the market and may face financial restrictions on their ability to compete in the auction, and incumbents, who already operate in the relevant market and hence may have an advantage due to their financial power, existing customer base, infrastructure and complementary licences for spectrum in other bands etc. [↩]
- Spectrum categories would generally consist of subsets of the available spectrum, with spectrum in each of the categories being considered to besubstitutable. For example, many auctions offer spectrum in several frequency bands that are the basis for defining lot categories. Other forms of categorisation might be considered if there are value differences across frequencies within a particular band, or if there is substitutability across multiple bands, for example. Specific frequencies might be included in multiple categories – the 700MHz band could form a category on its own, but also be included (in order to enforce a spectrum cap, for example) in a “Sub-1GHz” group along with the 800MHz and 900MHz bands. [↩]
- In reality there is often no such clear-cut distinction. There may be incumbents of different sizes, where the smallest face similar risks to new entrants, or there could be an entrant with existing operations in foreign markets that give it a significant financial advantage over the smaller domestic new entrants. [↩]
- As of August 2012. [↩]
- As of August 2012. [↩]
- For more information on the rules of the UK 800MHz and 2.6GHz Auction, see the Ofcom website. [↩]