Overestimating Wireless Demand: Policy and Investment Implications of Upward Bias in Mobile Data Forecasts

Abstract

It is now taken almost as a matter of faith among telecommunications professionals that there is a “spectrum crunch” precipitated by ever-growing demand for mobile broadband. This, in turn, has driven a strong policy consensus among industry and policymakers for the U.S. federal government to actively make available to mobile network operators (MNOs) vast new swathes of spectral resources via exclusive-use licenses. In the National Broadband Plan, for example, the Federal Communications Commission (FCC) argued that “[t]he growth of wireless broadband will be constrained if government does not make spectrum available to enable network expansion and technology upgrades.” The White House has taken up this cause as well, issuing a memorandum calling for 500 MHz of new spectrum to be made available for mobile and wireless broadband by 2020 and making new spectrum resources central to policies supporting “everything from smart phones to wireless broadband connectivity for laptops to new forms of machine-to-machine communication within a decade.” Similar efforts to free up spectrum for mobile broadband are taking place across the globe. It is indisputable that mobile connectivity, driven by new technological de-velopments in materials, miniaturization, and computing, has increased dra-matically around the world and radically transformed economic and social life. Both mobile subscriptions and the amount of data traversing mobile net-works has increased substantially in the past several years; by the end of 2014, there were an estimated 2.3 billion mobile broadband subscriptions globally, almost five times as many as in 2008. At the same time, decisions about spectrum should not and cannot be made lightly due to the physically limited nature of spectrum and the corresponding strong public interest in efficient spectrum allocation. Regulators must con-sider social, equity, public safety, environmental, and other factors in addition to economic impacts when determining whether and how to repurpose spec-trum. Furthermore, freeing up “new” spectrum is only one of several mecha-nisms for dealing with increasing mobile demand, and exclusive-use licenses are one among many different management options for spectrum bands. Alt-hough conventional spectrum wisdom highlights making new exclusive-use allocations available to MNOs, this may not be the most efficient or socially beneficial decision in all cases. It is exactly these kinds of concerns that led to calls in the National Broadband Plan for the U.S. government to make spectrum allocation and licensing data more accessible, as the “complexity of the [existing] system and the resulting lack of transparency and usability create impediments to public policy and limit the emergence of new technologies that could employ such data to optimize use of the spectrum automatically.” In this paper, we examine the provenance, reliability, and uses of mobile demand forecasts and find that several highly visible spectrum demand esti-mates over the past several years have exceeded actual traffic, which may have biased spectrum allocation decisions and policies in socially suboptimal ways. These reports include Cisco Visual Networking Index (VNI) esti-mates, which are commonly cited in government and industry reports, includ-ing the National Broadband Plan, as well as estimates endorsed in ITU’s WRC-07. We explore potential technological, economic, and sociological fac-tors that may drive these biases, as well as possible solutions and policy reme-dies. This paper is not meant to imply any deliberate misrepresentation from any party, although this is certainly possible when billions of dollars are at stake, as with spectrum policy. Rather, the paper investigates various potential sources of bias and methods to mitigate such risks, in the hopes that it might inform businesses, regulators, and the general public about how to better allocate and manage finite spectrum resources.Neither does this analysis imply that making any new, exclusive-use availa-ble for mobile broadband is the suboptimal policy decision. In the near future, such reallocations provide benefits to both MNOs and the general public through deficit reduction and economic multipliers. Rather, it suggests addi-tional caution and scrutiny are warranted for long-term spectrum policy deci-sions, rather than a blind faith about ever-increasing mobile data demand. In these cases, especially, policymakers should push for new, higher quality data before committing to long-term policies as well as favor flexibility and revers-ibility in their decision-making.