Why The Proposed “Unlicensed Auction” Is Such A Phenomonally Bad Idea — The Economics.July 14, 2011
To call the discussion draft on spectrum reform circulated by House Commerce Commitee Republicans “flawed” understates the matter almost to the point of absurdity. But chief among its many problems is the provision to auction any future allocation of unlicensed spectrum, such as the TV white spaces, aka “Super WiFi.” (Page 26) My colleague John Bergmayer has already explained one practical problem with this approach — that the beneficiaries of new technologies like TV white spaces don’t actually exist yet to bid against the existing firms like Verizon (which, no surprise loooooves the discussion draft). Also, as I observed in our press statement yesterday, I’m old-fashioned enough to think that auctioning off the right to make rules for spectrum (rather than auctioning off licenses once rules are set by a public process) is rather gauche. Yes, I know, everyone loves to talk about how the FCC and spectrum rules are up for sale, so I suppose we ought to make it official. But I am rather appalled that we have reached the point where we will explicitly auction off to “industry” the right to do their own rulemaking.
(If you think I exagerate for rhetoric effect, I suggest you read the original paper on which the bill is based: “A Market-Based Approach To Establishing Licensing Rules: Licensed versus Unlicensed Use of Spectrum.“)
But setting aside my residual qualms about simply selling the rules, the proposal is fundamentally flawed from an economic standpoint. Even if you accept the basic premise that auctions are the best way to allocate licenses among competing users, it does not follow logically that auctions are the best way to determine whether or not to designate spectrum for unlicensed access in the first place. To the contrary, this is rather like deciding that since radiation therapy is sometimes the best way to treat cancer, it must also be the best way to diagnose cancer.
Economically speaking unlicensed spectrum and licensed spectrum are two very different kinds of goods. Deciding which spectrum access is approriate for what bands under what circumstances is a very diffrerent question from how to allocate licenses once you decide to go with licensed rather than unlicensed. Indeed, the whole “licensed v. unlicensed in a fight to the death” framing that was popular when this paper was written in 2008 has pretty much died away precisely because licensees discovered that they had lots of use for unlicensed spectrum.
Last month, I testified before the House Subcommittee on spectrum as part of the lead up to this bill (video of my opening statement here). Afterwards, the Subcommitee asked I respond in writing to a follow up question that clearly had this proposal in mind. I reprint the question and why I think this is just a phenomenally bad idea that fundamentally misunderstands the underlying economics of licensed spectrum, unlicensed spectrum, and auctions below.
Some of the developers of white spaces devices include large companies such as Google and Microsoft. Some suggest that if we want to ensure that a certain amount of unlicensed spectrum will be set aside during an incentive auction of the TV broadcast band, these big companies should be required to participate in the auction and “bid” on an unlicensed band. In such a case, the winning bidder presumably would act like a guard-band manager in ensuring that unlicensed devices will be able to flourish. What is your view on the prospect of auctioning access to unlicensed spectrum?
The question proceeds from a mistaken premise. It suggests that individual large companies such as Google or Microsoft are the primary beneficiaries of the TV White Spaces. This confuses the nature of an exclusive license with the nature of unlicensed. Unlicensed spectrum acts as a resource available for all to develop. This encourages a particular type of development – notably for mass produced goods or low margin/high volume services. By contrast, licensing encourages the licensee to maximize the profitability of the spectrum, incenting the development of large scale networks that create sufficient return for the licensee. This is why there are only a handful of large licensed network operators and several dozen smaller providers. By contrast, there are several thousand wireless ISPs using unlicensed spectrum to provide broadband in rural areas unprofitable for wireless providers. In addition, there are tens of thousands of coffee shops, hotels, and other establishments offering wifi. Hundreds of manufacturers put wifi – or other protocols using the unlicensed bands — into everything from printers to refrigerators.
This is not because licensed is “better” than unlicensed or unlicensed is “better” than licensed. Rather, each is a different type of economic good. Licensing encourages the development of large, centrally controlled networks that permit the licensee to recoup the investment in the license. To further encourage licensees, they receive interference protection and the use of higher power. Because these licensees hold a scarce resource, they can invest huge sums in networks and expect return on development.
While this has been enormously successful in encouraging the deployment of our national cellular system, it comes at a social cost. The huge sums required to win licenses drive winners to develop networks where they can be certain of extracting the maximum return. This discourages experimentation, or development of less profitable services that might prove attractive to entrepreneurs. It also makes it more difficult for users to customize their network deployments, because they must abide by the equipment and network design choices of the licensee.
A good example of the different kinds of investment encouraged by unlicensed v. licensed can be illustrated by the difference in the market between femtocells and wifi for data hand off. Femtocells are designed to receive data from handsets on subsdcriber’s licensed spectrum and move the data to the subscriber’s broadband. Despite the obvious advantage of this for subscribers and network operators both, femtocells have not proven popular. In part, this is because network operators – driven by the need to recoup license costs – insist on charging subscribers a fairly high price for femtocells and treat them as a subscription service.
By contrast, the ubiquity of free unlicensed spectrum has encouraged subscribers and network operators to use unlicensed spectrum to offload data to available broadband networks. Using unlicensed for offload has become a key part of the strategy of every significant licensed wireless operator. The fact that unlicensed access is available to all without charge makes it ubiquitous, encourages economies of scale, and allowed licensed network operators to make the decision to shift data to this proven technology when it became necessary to do so.
This underscores an important lesson in spectrum management. Match the access need with the type of spectrum (licensed or unlicensed) rather than assume one type of spectrum access is “better” than the other. It is unfortunate that, in the days when advocates sought to persuade regulators to adopt auctions for distributing licenses, they characterized auctions as moving spectrum to the “highest, best use.” In fact, auctions (and other market mechanisms) encourage licensees to move to the most profitable use for licensed spectrum. But, as demonstrated every day by the billions of users of unlicensed spectrum, the availability of unlicensed access has a general resource to everyone – not just the licensee – has many powerful advantages. This is particularly true for traffic which does not require higher power or interference protection. We do not want needed licensed spectrum tied up in warehouse inventory, machine-to-machine communications, closed circuit security systems, or wireless hotspots shared by dozens or hundreds of users. It is much better to place this traffic on cheap, shared-spectrum networks that users can tweak to their individual needs. While it is certainly true that licensees could sell these services, it would add a needless layer of transaction cost, discouraging numerous beneficial spectrum uses and divert licensed spectrum and network capacity away from more efficient uses of these resources.
By contrast, the auction approach described would end the usefulness of the TV white spaces by eliminating its value as a general resource available to an unlimited number of users and developers of goods and services. Further, it would yield little by way of additional auction revenue. As was debated at considerable length in the TV white spaces proceeding, the “Swiss cheese” nature of the white spaces (holes in markets rather than national blocks) makes it particularly ill-suited for traditional auction and management. Even under traditional licensing rules, it would yield little revenue in an auction.
More critically, the usefulness of the white spaces depends on the lack of barriers to entry and the ability of all users to benefit from the availability of spectrum. This gives innovators the chance to develop new products and services without the need to seek permission. For example, many equipment manufacturers develop a wide variety of products using unlicensed spectrum for the benefit of users. These are mass produced products that any consumer may use, regardless of what wireless provider the consumer uses. The low cost encourages adoption, and greater innovation. In 2000, when wifi dongles cost a hundred dollars a piece, few users considered wifi a worth trying. As wifi chips became common in computers and other devices, consumers quickly became addicted to the ability to use their laptops untethered. Today, wifi has essentially replaced the wires and cables that used to run from desktop to printer and to every other connected device in the home or office. Had it been necessary to pay for access, it is doubtful that this highly beneficial market would have developed.
Furthermore, the concept of a “band manager” is both unnecessary for interference purposes and inconsistent with the theory of licensing. A band manager operates to prevent interference with fellow users by active management among a limited class of users. Here, the FCC has already established that the devices themselves can, and should, operate in a manner that avoids interference with licensed services. Unlicensed devices are not entitled to interference protection, making the band manager concept unsuitable.
Because the rules are already settled, an auction of the white spaces along the lines proposed would add no value. Unlike an auction for flexible licenses, where the auction places the spectrum in the hands of a licensee motivated to build a network, the TV white spaces “band manager” would simply collect money so that parties could access spectrum in a manner the FCC has already determined is consistent with the sound spectrum management. Rather than embracing the deregulatory freedom of unlicensed, such as approach would be a step backward to a regulated monopoly where the “band manager” would levy new fees and impose new restrictions above and beyond those needed to avoid harmful interference.
Finally, auctioning the white spaces to a “band manager” solely to maximize revenue raises significant First Amendment questions. The power of Congress and the FCC to regulate spectrum access derives from the need to avoid harmful interference. No one here suggests that a band manager would do anything to minimize the risk of harmful interference. This is simply a tax on wireless speech, no more constitutional than would be a special tax on newspapers. Even were the other considerations less compelling, Congress would hopefully forgo imposing a “Super WiFi Tax” simply as a means of raising revenue.
In short, unlicensed spectrum is, economically, a very different type of good from the kind of good from licensed spectrum. Licensed spectrum provides the opportunity to use spectrum at higher power with interference protection. Unlicensed spectrum provides much broader, lower cost access in exchange for lower power and no interference protection. Both types of spectrum are needed to maintain a robust spectrum environment that promotes job creation, innovation and competition among providers of all wireless services. The only justification for the proposal is that mistaken idea that a large company such as Google or Microsoft has a duty to pay for spectrum that will equally benefit Verizon, AT&T, or any other company or user that accesses the spectrum. But this assumption stands the true value of the TV white spaces on its head. By making the resource available to everyone, the FCC creates new value enjoyed by every company, innovator, retailer and user.
About Harold Feld
Harold Feld is Public Knowledge’s Senior Vice President and author of “The Case for the Digital Platform Act,” (Public Knowledge & Roosevelt Institute 2019) a guide on what government can do to preserve competition and empower individual users in the huge swath of our economy now referred to as “Big Tech.” Former FCC Chairman Tom Wheeler described this book as, “[...] a tour de force of the issues raised by the digital economy and internet capitalism.” For more than 20 years, Feld has practiced law at the intersection of technology, broadband, and media policy in both the private sector and in the public interest community. Feld has an undergraduate degree from Princeton University, a law degree from Boston University, and clerked for the D.C. Circuit Court of Appeals. Feld also writes “Tales of the Sausage Factory,” a progressive blog on media and telecom policy. In 2007, Illinois Senator Dick Durbin praised him and his blog for “[doing] a lot of great work helping people understand how FCC decisions affect people and communities on the ground.”