Why setting the floor price for digital
dividends auction is the right approach


This article is by Flavio Menezes,Professor of Economics at the University of Queensland. It first appeared on The Conversation and is replicated here with permission.

analysis Last week’s directive from Communications Minister Stephen Conroy to the Australian Communications and Media Authority (ACMA) to postpone setting the reserve prices for the upcoming digital dividends auction will help ensure the sale’s success. To the untrained eye, Conroy appears to have intervened in the affairs of a statutory authority. However, it was a warranted and sensible answer to developments in the sector.

The Minister’s decision will most likely lead to the efficient sale of a public-owned resource and enable the realisation of the asset’s true market value. At stake is access to the 700MHz and 2.5GHz frequency bands of the spectrum, freed up with the switching of television from analogue to digital. Mobile carriers are particularly interested in accessing the bands to expand their fourth generation (4G) services nationally, and given the trends in the demand for mobile data traffic, access to the spectrum could be worth up to $3 billion at its sale in April 2013.

Internationally, sales of spectrum have not always generated the expected returns, which is why the design of the auction is crucial.Unlike with the allocation of the 2G and 3G frequency spectrum, this time the ACMA has decided to replace the simultaneous ascending price auctions with a combinatorial clock auction. The decision was made following industry consultation, and the ACMA’s auction design captures many of the insights from a substantive amount of research and practice over the last 20 years.

The adopted auction design includes two stages.

The first stage is the clock auction, where the lots of spectrum are divided into several categories of identical generic items. Each category is associated with a “clock” and bidders must submit the number of lots they want for each category at the current round price. The price for each category in which demand exceeds supply rises until there is no longer excess demand in any of the clocks. This method allows bidders in effect to bid for a package of lots across different categories. (Bidders will need to assemble the package of lots that best suits their needs.)

The second stage takes place when there is no excess demand in any of the clocks. This stage involves bidders making their best and final offers, through a sealed bid, for all the different combinations of spectrum they want. Sealed bids have to conform to bidders’ preferences revealed through their bids in the clock round. The winning bidders at the end of the second stage are those that make the highest value combination of bids. All bids are considered from both phases of the auction, but bidders can win in only one of their bids.

An innovation in this design is the pricing rule, which adapts insights from the notion of Vickrey auctions in an attempt to ensure that bidders reveal their true preferences for the lots they want. In a single-object Vickrey auction, the object is allocated to the individual with the highest bid and the price paid is equal to the second highest bid.

This leads bidders to reveal their true valuations in the auction, as their bids only affect the probability that they win the auction but not the price they pay if win. Truthful revelation of demand is required for efficiency—so that the auction allocates the object to the bidder that values it the most. William Vickrey won the 1996 Nobel Prize in economics, alongside James Mirrlees, for his contribution to auction theory.

The pricing rule in this auction adapts Vickrey’s insights to take into account the fact the lots being auctioned are both substitute and complements. Under these circumstances, Vickrey prices are no longer guaranteed to be efficient.

Consider a simple example, where there are two objects for sale: A and B and three bidders 1, 2, and 3. Suppose that bidder 1 only wants object A and bids b1A = $5 for it. Bidder 2 only wants object B and bids b2B = $3. However, Bidder 3 bids b3A = $2 for object A, and b3B = $2 for object B, and b3AB = $5 for a package containing objects A and B. Vickrey prices are pA = $2 and pB = $2.

However, these prices are not appropriate as they do not take into account the value of Bidder 3’s package bid of $5 for A and B, which is greater than the sum of the Vickrey prices.

The solution used in the ACMA design would take into account the value of the package bid and in this instance these bids would result, for example, in prices equal to pA = pB = $2.50, and in the allocation of the two objects, A and B, to players 1 and 2, respectively. These prices are efficient as Bidder 3 is not willing to pay more than $5 for the package. In the actual ACMA auction, an optimisation method will be used to calculate such prices.

While similar dynamic auctions have been used successfully during the last 20 years in Australia and around the world, there have been some well-known failures and this is what the Minister seems to be trying to avoid. A key concern for the government is that there might not be enough competition in the auction, with one of the country’s three major mobile players Vodafone Hutchison Australia (VHA) looking unlikely to take part. (VHA CEO Bill Morrow has repeatedly signalled the company’s possible lack of participation.)

In the worst-case scenario for the government’s revenue, the sum of the other two major players Optus and Telstra’s bids at the opening round of the clock stage might not outstrip supply. If this happens, the ACMA will allocate lots at the chosen reserve prices. This is why reserve prices have to be set strategically: to substitute for the possible lack of competition in the auction. In the hypothetical example given above, if Bidder 3 were not to participate in the auction, objects A and B would be allocated to bidders 1 and 2, respectively, at the reserve prices. A reserve price of zero would then mean zero revenue.

Two decades ago, in the sale of radio spectrum in New Zealand, the government used a second price auction with zero reserves. In one auction, the highest bidder bid NZ$100,000 and won the licence for NZ$6 and in another auction, the winning bidder bid NZ$7 million and won the licence for NZ$5000.

The more recent 3G auctions in Switzerland also provide a powerful warning. There were four licences for sale and a bidder could only bid for a single licence. Joint-bidding agreements were allowed and low reserve prices were set in expectation of strong competition. However, the field shrank from nine bidders to just four in the week the auction was scheduled to begin. This resulted in bidders having to only pay the reserve price—prices that (per capita) were one thirtieth of the UK and German auction prices and one fiftieth of the predicted revenue.

By setting an appropriate reserve price, the Australian government will avoid such disastrous outcomes.

Flavio Menezes does not work for, consult to, own shares in or receive funding from any company or organisation that would benefit from this article, and has no relevant affiliations. This article was originally published at The Conversation. Read the original article.

The Conversation


  1. Has anyone considered the likelihood of NBNco participating in the auction?
    Doesn’t it seem likely considering what they said about problems with future speeds?

    • I’m sure that has been considered, which is why they are making the auction something a little more complex than a ‘hands up’ bid.

      One would assume NBNCo would only be interested in the long distance, high data rate spectrum, specifically the 700MHz band. This is the optimal band for non-urban wireless propagation as it travels much, much farther than the GHz bands, however it is not suitable for urban (high subscriber number) environments.

      To quote the case for the US auction “In this case lower is better since 700-MHz signals propagate better, spreading farther and penetrating buildings more easily than higher frequencies. This greater range means that each 700-MHz cellular antenna can service a larger footprint, which means fewer cells (those interlocking service areas that a “cellular” network is made up of) will be required overall. That should, theoretically, make it cheaper—$5 billion cheaper according to some estimates—to build a national wireless network.”

  2. If they don’t sell all the spectrum, I wonder if emergency services will get the extra they asked for?

Comments are closed.