Friday, December 30, 2011

Can Auctioning the Spectrum Go Too Far?

How should the scarce resource of the radio spectrum for communication be allocated? One of the classic triumphs of economic analysis has been that spectrum should be auctioned off, rather than being allocated by government. But some prominent economists are now raising the possibility that even if some spectrum should be auctioned off, another portion should be left simply unlicensed.

In the early days of broadcasting, back in the 1920s, one signal often struggled to overpower another. Legislation as early as 1912 sought to address this problem, and the eventual result was the creation of a government regulatory system for allocating spectrum in 1927, under the auspices of the Federal Radio Commission, which morphed into the Federal Communications Commission in 1934.

More than 50 years ago, that arrangement was famously challenged by Ronald Coase in an October 1959 article in the Journal of Law and Economics called "The Federal Communications Commission." It's available a few places on the web (for example, here and here), and is also available through JSTOR. As Coase pointed out, the situation in which users might struggle over a common resource arises often enough--for example, in the case of shared farmland. However, in the case of farmland such conflicts are resolved through property rights and ownership, along with the possibility of selling the resource to others. Thus, Coase drew together the work of earlier writers and proposed that the FCC could auction off spectrum, keep records of who owned which frequencies, and keep track of future changes of ownership.

In the 1990s, Coase's advice, which seemed outrageous to many back in 1959, had been adopted. For a discussion of U.S. spectrum auctions in my own Journal of Economic Perspectives, see this 1994 article by John McMillan. For a follow-up article on spectrum auctions around the world, including some poorly designed auctions with embarrassingly bad results, see this 2002 article in my journal by Paul Klemperer.
But my theme here is not to review the design of spectrum auctions, but to explore a counterintuitive point.
Paul R. Milgrom, Jonathan Levin, and Assaf Eilat  published in October The Case for Unlicensed Spectrum. 
They write (footnotes, citations and paragraph numbers omitted): 

"In the US, the most common approach to managing radio spectrum for commercial non-governmental use has been to assign licenses that give the licensee exclusive rights to a particular band of spectrum for a set period of time. The development of spectrum license auctions in the 1990s helped to pave the way for the growth of the mobile phone industry while generating billions in auction revenues for national governments. Yet some of the most valuable and important innovations in wireless communication, in particular the development of Wi-Fi, have taken place on bands of spectrum for which no exclusive licenses were issued. ...
While selling exclusive licenses to radio spectrum has been a valuable tool for eliminating conflicting uses and encouraging related investments, it has also contributed to concentrated market structures in wireless telephony and created barriers to entry and innovation. Leaving portions of the radio spectrum unlicensed has created multiple benefits, including encouraging the development of complementary technologies that enhance the effectiveness of devices that use licensed spectrum, triggering the development of alternative technologies that compete with licensed uses, and promoting innovative business models and technologies that have brought unexpected benefits.
There is considerable evidence that unlicensed spectrum has huge economic value. Recent past estimates, which already look too conservative, place the value created by current applications of unlicensed spectrum at $16-37 billion dollars a year in the United States alone. However, the primary benefits of unlicensed spectrum may well come from innovations that cannot yet be foreseen. The reason, as we discuss below, is that unlicensed spectrum is an enabling resource. It provides a platform for innovation upon which innovators may face lower barriers to bringing wireless products to market, because they are freed from the need to negotiate with exclusive license holders. Indeed, allocating a mix of licensed and unlicensed spectrum is attractive precisely because the two approaches have diverse advantages in terms of triggering investment and innovation.
 They argue that auctions cannot work as a way of allocating unlicensed spectrum, because the future innovators who might experiment by using such spectrum (along with the consumers who would ultimately benefit from such experiments) can never be organized in advance in a way that would allow them to participate in such an auction. While Wi-Fi is to this point the most important innovation to make use of unlicensed spectrum, they point out that similar arguments apply to Bluetooth and other "wireless personal area networks," to WirelessHD and WiGig which can transfer large amount of data over a distance of a few meters with a line-of-sight connection, and radio-frequency identification tags, the small tag that are attached to products or embedded in cards for tracking purposes. 

In juxtaposition to the work of the 1991 Nobel laureate Ronald Coase, Milgrom, Levin and Eilat point to the work of a 2009 Nobel laureate, Elinor Ostrom. "The Nobel laureate Elinor Ostrom has written extensively on the diversity of governance systems for managing common pool resources. Her work identifies several key principles: the creation of clear rules that respond to local conditions; collective decision-making that allows the participation of most community members; effective monitoring, enforcement, and conflict-resolution mechanisms; and coordination between organizations that manage common-pool resources.  As she emphasizes, these principles do not necessarily imply the creation of exclusive property rights; in many cases, alternative systems can work better."

An intriguing commonality here is that these innovations all  happened in parts of the radio spectrum that were viewed as useless for larger-scale communication--which is why they were unlicensed and thus were available to innovators. What innovations might emerge if some spectrum suitable for long-range and far more reliable communication--like that which has for decades been used for broadcast television--was instead reserved for unlicensed use and innovation?
The Milgrom, Levin, and Eilat argument is also intriguing because it points out an inherent conflict between property rights and innovation. For example, those who invent something today and seek out a patent must often be concerned that they are potentially overlapping with other patents that have already been granted to others. In some cases, the property rights of already-existing patents can choke off innovation from new competitors. As another example, when many people own property rights to many different plots of land, it may be difficult for a new use of that land to arise--whether it be a nature preserve or a natural gas pipeline--because the existing splintered property rights make it difficult to negotiate for an alternative use of the land. Property rights and market exchange are excellent at finding efficient ways for existing uses, but when it comes to certain kinds of change and innovation, they can sometimes pose drawbacks.