Don't Make Spectrum a Private Property
Research Institute of Economy, Trade and Industry (RIETI)
Dear Professors David Farber and Gerald Faulhaber,
Coincidentally I made comments on your proposal for spectrum management to both of you in Japan and the US last month. Thank you for the stimulating discussions, but I am afraid that your answers were not very persuasive, probably due in part to my poor English. So I would like to restate the questions I asked you at RIETI and FCC (Federal Communications Commission).
Spectrum Should Be Commons, Not Private Property
Your proposal, which was stated more formally in recent article, is in line with the FCC's "market-oriented" approach to make spectrum a private property to be traded by private parties. Indeed it is the logical step next to the spectrum auction that has been acclaimed as the victory of economics. I agree with you that it is better than the "Gosplan" approach of the Japanese government, which I once dubbed Spectrum Socialism. However, innovations in wireless technologies are so rapid that this year's good idea might become next year's folly, as evidenced with the tragedy of the auctions for 3G mobile phones in the EU.
You may say that the failure of 3G auctions was a result of crazy speculation, but it marked the end of legacy wireless systems that use proprietary spectrums. Much-hyped FOMA, the 3G service of NTT DoCoMo, has seen no success. Instead, wireless LANs (WLANs) are growing so spectacularly that user numbers are projected to be over 200 million worldwide in 2006.
So we should reform spectrum policy for the wireless Internet, not for mobile phones. It will change the agenda of reform drastically, because WLANs do not need "efficient allocation of spectrum". IEEE 802.11b, the most popular WLAN technology, allocates signals efficiently and dynamically in a wide band (100 MHz). There is no need to allot a fixed frequency for individual users, because spread spectrum technology made it possible to identify each signal by encoding and decoding (cf. my article). So it is better to share a wide band as commons in the wireless Internet, as you recognize.
According to Shannon's Channel Capacity Formula, if we share a wide band by WLANs everybody can use the whole spectrum. On the other hand, if we divide it into small pieces of private spectrum, everybody can use only a narrow portion of the band. Moreover, if a user buys a frequency in the Big Bang auction at high cost, she will try to enclose it by proprietary protocols and applications, as is seen with the cable Internet and instant messaging.
To avoid such a mess, as you suggested, the FCC would have to enforce "easement" to open the spectrum for competitors. However, because such open usage makes the spectrum worthless, its owner will raise the price of the frequency and prevent the "collocation" of spectrum. Enter the FCC, and there will be unlimited lawsuits and a regulatory nightmare as was seen with DSL. It is the anti-commons in which too many parties claim their rights for the same property.
There is no need to reconcile the "economists' view" that favors property rights and the "engineers' view" for commons, because the latter is far superior to the former in the Internet age. Nor is it possible to make spectrum commons by making it private property. You can't have your cake and eat it too.
Strategies for Transition to the Wireless Internet
Indeed there would be some problems for transition from legacy to the Internet. I propose a simple reform: take back the spectrum and free it without licenses. It is possible legally: the FCC will only have to tell incumbents that their licenses will expire in five years. The incumbents incur no loss, because they can do the same services over the wireless Internet.
Of course incumbents will resist it because they have invested in legacy equipment, but there will be some market-oriented solutions. For example, if the frequency allocated to taxi radio is opened for WLANs, efficiency will increase more than a thousand-fold. So as Coase's Theorem suggests, it is efficient for the government to bribe incumbents by buying their spectrum.
However, since such discretionary subsidies could induce rent-seeking, I propose a spectrum buyout to buy back the idle spectrum. Because it is an ordinary procurement process, the classic English auction can be applied. This reverse auction can be financed by the auction fees the FCC has earned or by a radio tax, as adopted in Japan and the EU. The amount should be proportional to the bandwidth so as to press incumbents to use it more efficiently or exit. It is neutral for incumbents who will pay the tax and recoup it as the price of the spectrum, and it is beneficiary for new entrants who will get new markets by opened spectrum.
You may argue that the cost of the buyout will be huge, but it will not be so. Contrary to the ordinary spectrum auctions in which prices are determined by the present value of the most efficient usage, the price of the buyout will be determined by the opportunity cost of the most inefficient user who wants to exit. My guess is that the price will be less than 1/100 of 3G auctions, but it might be necessary to have experiments to estimate the price.
Spectrum buyout is, in effect, a collective auction by millions of WLAN users. So its cost is equivalent to a private auction, in which the winner will "tax" their costs to consumers as profits. The public buyout dominates private ones because it will free us from any costs to pay for spectrum forever (cf. Kremer). The role of the FCC will be limited to the regulation of interference, which can be enforced without any property rights.
Until the spectrum is opened, the frequencies that are occupied but not used by incumbents should be made available for overlay usage, i.e., dynamic allocation of signals by sensing the used channels and avoiding them. Such technologies, called cognitive radio, are already implemented in some chip sets and under standardization in IEEE 802.11h committee. Mr. Robert Pepper, the Plans and Policy Chief of the FCC, admitted this reform and preferred to call it "underlay" because current usages are prior to WLANs.
Your conjecture that spectrum will become a scarce resource in the long run has no scientific ground. Since the innovation of wireless technology is much more explosive than the demand growth, there may be such a thing as free spectrum, if you do not confine it in the legacy regulatory regimes.
Sorry for a long comment,