Is the Price Right?
I found Andrew Kerans, Dat Vo and Philip Conder’s paper on the Pricing of Spectrum Bands based on Physical Criteria at IEEE DySPAN 2011 to be interesting. The purpose of the paper is to find a means to price spectrum. The authors studied a number of auctions and also studied the costs of infrastructure that would be needed to deliver services in different frequency bands. I have reproduced the final graph from the paper here. The solid line shows the value they associated with different spectrum bands based on an analysis of physical criteria (infrastructure costs). Higher infrastructure costs will mean lower revenue for spectrum- which sounds logical. The revenue generated from different auctions in different bands is also shown in the figure and in the main the results are in line with the value the authors have associated with the bands. The authors suggest that this type of physical-based analysis could be used to set prices for spectrum in the future. I hope I have not misrepresented the authors in this limited summary.
Of course we all know the correlation does not equal causation. I don’t believe auctions are neutral mechanisms out of which ‘a true’ price of spectrum ‘emerges’. They are designed with specific political and economic biases. I am not picking on auctions here. Most things are designed like this but we don’t often acknowledge it. I believe that the specific mechanism used by the auction should have some impact on the revenue generated by that auction. Hence the fact that the auctions turn out to follow the value curve defined by the authors might not be so straightforward. Given the limited amounts of data (i.e. the number of auctions) they had to study this was not possible for the authors to account for auction mechanism. In addition William Webb from Neul (he is ex Ofcom) pointed out that Ofcom studies showed that the greatest factor influencing the amount of revenue generated from an auction was whether the band was a harmonised band or not. The work he mentioned was carried out by Plum Consulting and can be found here on the Ofcom website. I am always amazed at the great research Ofcom does, not to mention the accessible way the research is written up. And though I don’t find the Ofcom website the easiest of websites to navigate, it is worth a visit. In any case the harmonisation point is a much more solid point than my point about the neutrality of auctions. Hence harmonisation of higher bands may disrupt the findings.
Having said all that, I like the general approach and thinking and I like the author’s model for infrastructure costs. It could be useful for many other things. I wonder can we find a way to use it?
Btw the IEEE DySPAN proceedings will find its way on to IEEE Xplore and I will link to the paper once that is the case.