Alan Gatherer, Editor in Chief, ComSoc Technology News
This month we introduce a very apropos article that is not only a follow up to our October spectrum allocation article but is also tremendously relevant in this lame duck period between administrations in the US as we all wonder what strategy the Trump administration will take in allocating spectrum as the explosion in new wireless services, with new business strategies, continues. Tom and Mike reveal the answer is way more complicated than you probably thought. Comments from the Donald or anyone else are very welcome.
Putting a Value on Spectrum Allocations
The evolution to 5G is expected to enable a host of new applications and services that will place increasing pressure on existing spectrum allocations. In anticipation, regulatory agencies have introduced new initiatives for increasing spectrum allocations for broadband access, along with new modes for sharing spectrum across different types of applications. At the core of all such spectrum allocations is the determination of how access rights are defined. Cellular service providers rely on allocations of licensed spectrum for exclusive use – a de facto property right. In contrast, unlicensed allocations for open access have given rise to services based on WiFi, Bluetooth, and Zigbee. This has led to a persistent long-standing controversy whenever new spectrum becomes available for broadband access: Should it be allocated as licensed or unlicensed?
Pointing to the popularity of WiFi, Bluetooth, and other technologies hosted in unlicensed spectrum, some argue that regulators should set aside more unlicensed spectrum along with liberally licensed bandwidth. (A “liberal license” for spectrum can be traded among different wireless service providers, both cellular and non-cellular.) In a recent paper , we challenge this argument, and instead argue that liberally licensed allocations provide the most flexibility in determining the uses that maximize social welfare. Here we highlight three important points.
1. Open access does not require unlicensed spectrum.
Allocating spectrum as unlicensed means that it must be used to support open access services. However, open access services do not require allocations of unlicensed spectrum. In fact, “open access” actually refers to a technology or business model, as opposed to a spectrum policy choice, that can be supported in either licensed or unlicensed spectrum. For example, a local service provider with a spectrum license (either owned or leased from a larger provider) could operate a WiFi open access service by charging an admission fee, much in the same way some airports and hotels charge for Internet access. Hence while it may be true that open access is currently an efficient way to use certain spectrum bands (e.g., at higher frequencies) that does not imply that those bands should be allocated as unlicensed. A licensed allocation allows providers to weigh alternatives in technologies and business models in terms of costs and revenues, and those choices will continually evolve (e.g., constraining open access when congestion becomes excessive). In contrast, the unlicensed assignment effectively locks in open access indefinitely due to the high transaction costs of appealing to the regulatory authority to implement changes.
2. The value of spectrum is not the value of the service(s) it supports.
Advocates of unlicensed spectrum have often equated the value of unlicensed allocations with the total amount spent on equipment and services supported by unlicensed bands. Problems with this assertion are vividly illustrated with broadcast TV spectrum. Namely, the value of that spectrum cannot be determined by adding up sales of TV equipment. Such an estimate of value ignores the opportunity costs of excluding alternative uses for the spectrum. Also ignored are possibilities for input substitutions, i.e., in the broadcast TV scenario, alternative means for delivering broadcast TV content. Finally, when deciding on the most efficient use for new spectrum, what should be compared are not total values of previous allocations, but rather the marginal values created by alternative new allocations. In other words, broadcast TV programs are popular, but that does not mean TV channels are an efficient use of that spectrum.
Routinely, regulators rely on estimated economic values of their predicted uses when pursuing spectrum set-asides. Of course, policy makers do not actually know what applications will be most efficient in the future or how they will be best supplied to users. They rely on advocates for competing approaches to inform them of the options, even as such arguments are tinted by the financial motives of the parties. The proffered evidence used in such debates looks much like Figure 1. Displayed are estimates of the economic gains from (in clusters, left to right) unlicensed spectrum, liberally licensed spectrum, and TV band spectrum. (The * denotes the studies not commissioned by interest groups lobbying for particular spectrum allocations. One of these, Innovatio, features the value of wi-fi as calculated in an intellectual property verdict rendered by federal courts.)
It is clear that the variance in the estimates shown in Figure 1 is high. While the mean value for “licensed applications” is much higher than for “unlicensed applications”, this does not itself imply that licensed spectrum allocations are superior to unlicensed. Indeed, the highest estimated spectrum values are purportedly found for TV channels -- as studies undertaken on behalf of broadcasters seem to show. The problem with most estimates provided in this way is that they typically ignore the relevant incremental margins, substitution possibilities, opportunity costs and administrative expenses. In contrast, liberal licenses allow market forces to determine prices for spectrum access, a direct measure of marginal value.
Figure 1. Rival Estimates for Spectrum Values 
3. Free access to scarce spectrum is economically inefficient.
Unlicensed spectrum provides free open access, meaning the price of access is set to zero by regulation. While free access to a resource makes sense when that resource is abundant, it is economically inefficient when the resource becomes scarce. In the case of spectrum, this can lead to excessive congestion, as has become evident in densely populated locations. Moreover, prices reveal important information, and policies that allow providers of different types of services to bid for resources advance the efficient deployment of wireless services. Those include services in bands that support plug `n play devices, “spectrum parks” providing free public access, and “innovation commons” intended to facilitate the introduction of new technologies and services.
The lack of pricing information for unlicensed bands has given rise to several disputes among proponents of different applications. One concerns how the use of the 5.9 GHz band is regulated. While it has been set aside for the use of RF devices associated with vehicle safety and driverless cars, it could also provide value for users of wi-fi devices accessing the adjacent 5.8 GHz band. Companies like Google and Microsoft champion a reallocation to allow for the latter, while firms such as General Motors and Ford argue in favor of defending the status quo. Regulators do not know where the optimum lies. If, in fact, the rights to the 5.9 GHz band were defined such that they could be traded in the market, additional information would come to light – specifically, prices would reveal if the entire band, or 25 MHz, or some other incremental bandwidth, would be valued more highly in support of wi-fi, vehicle telematics, or some newly discovered option.
The absence of market prices for unlicensed spectrum has also given rise to the interest in LTE-U, even though LTE has been designed for exclusive use spectrum, not open access. Taking into account the marginal cost of the spectrum, as revealed by market prices, would likely remove the justification for developing LTE-U entirely. If a market were used to allocate spectrum between exclusive rights and open access, then there would be no direct incentive to mix LTE with WiFi. Instead the amount of spectrum for WiFi open access (more appropriate for short-range access) and LTE exclusive assignments (more appropriate for wide-area coverage) could dynamically shift according to demand.
It is inevitable that current spectrum allocations will become obsolete with the advent of new science and entrepreneurial evolution. With administrative allocations, reconfiguring spectrum usage to adapt to such changes will require lengthy and contentious rule makings. The expense of this overhead is generally omitted from spectrum allocation policy evaluation. Yet, it constitutes an essential component of the consumer welfare analysis. Given smart reforms, prices can assist policy makers in moving away from the high transaction costs and frustrating rigidities that needlessly restrict the dynamics of market evolution in wireless devices and networks.
- Hazlett, Thomas W. & Michael Honig. 2016. Valuing Spectrum Allocations, Michigan Telecommunications & Technology Law Review (forthcoming).
Editor-in-Chief: Alan Gatherer (email@example.com)