Dear Sir,

I am a final year PhD student in the Department of Computer Science in

University College Cork and I read the Review Document with interest.  I would like to suggest an auction mechanism that I believe may be well suited towards the distribution of local radio licenses in Ireland.

 

Combinatorial auctions sell multiple distinguishable items and permit bidders to offer bids on any subset of these items. Such auctions enhance economic efficiency when there are tangible benefits in securing items that complement one another. The simplest way to illustrate the benefits of such auctions in the case of radio licensing is through examples.

 

Example: Suppose three bidders (x,y,z) are interested in licenses for two adjacent locations (A,B). Bidder x is interested in location A, bidder y is interested in location B and bidder z seeks both locations

AB but is uninterested in having just one location. If bidders are allowed bid on combinations of items, bidder z is happy to participate in the auction. On the other hand, if bids on combinations (or package bids) are not permitted he runs the risk of securing the rights for just one location and thereby making a loss, this is called the 'exposure problem'. This incites depressed bidding for bidder z, thus reducing revenue and militating unfairly against z. [Milgrom - Putting Auction Theory to Work, 2004]

 

Package bids allow bidders to express their complementarities more fully and exploit the complementarities between locations most effectively. A license in Leitrim alone may be uneconomical for example, but bidders from Donegal or Sligo may submit another bid that includes Leitrim. Furthermore the Sligo bidder may submit a bid for all three counties in an attempt to take control of the northwest. He may also look westward and attempt bids for Sligo with Roscommon for example. The winners are determined using the nationwide revenue-maximising solution.

 

Such auctions have been used very effectively by the FCC (Federal

Communications Commission) in America for the sale of spectrum rights to mobile phone operators. There are other issues associated with such auctions that require mention also. The number of possible combinations of items increases quickly with the number of items so the number of possible bids could explode if not capped. Limiting the number of bids from a bidder solve this problem but (slightly) reduces overall economic efficiency. Bidders typically also require some training (1/2-1 day usually) so that they understand the implications of bids. Calculation of the winners can be computationally difficult (hence the interest of myself, a Computer Engineer), but this shouldn't pose a problem if the number of bids in total is less than 10,000 or so.

 

I believe combinatorial auctions may be ideal for the sale of local radio licenses in Ireland for the reasons given above. The stronger the perceived complementarity between items in an auction, the more suitable it becomes for combinatorial (or package) auctions.

 

Please feel free to contact me if you wish to discuss the topic further.

 

Regards,

A. Holland