Recently, U.S. mobile industry trade association CTIA published a study performed by the economic consultancy NERA. It claimed that releasing additional spectrum for full-power 5G wide-area networks would generate a $260 billion uplift in GDP, and 1.5 million extra jobs, for every 100 megahertz auctioned. It used this result to call for an extra 400 megahertz to be made available in the near-term.
However, while the NERA study may on the surface appear to adopt a sensible methodology, it misses a range of important factors that cast significant doubt on its quantitative analysis and conclusions. Below, we discuss:
- Costs and delays involved in clearing and releasing spectrum bands
- An implicit assumption that all midband frequency ranges are similarly valuable
- Technology evolution to improve network efficiency of existing spectrum
- Doubtful predictions over new spectrum’s impact on Fixed Wireless Access (FWA) deployment and value
Costs of spectrum release
Firstly, the report ignores the costs, time and other consequences of clearing any new spectrum for auction and mobile use. Several bands under discussion are currently used by federal systems, such as military radars and defense communications.
The Department of Defense projects clearing CTIA’s desired bands would cost hundreds of billions of dollars and take many years to specify, test, procure and implement suitable new military wireless equipment for alternative bands — risking disruptions that could potentially compromise national security. These costs are not mentioned or offset against claimed revenues or economic benefits from commercial services.
Furthermore, other new priorities of the current US administration, such as a proposed “Golden Dome” missile defense system, would likely lead to increased usage of existing (and perhaps new) federal spectrum bands, making any clearance less likely or even more delayed, more expensive, and perhaps strategically ill-advised. The commanding general of U.S. Northern Command recently warned Congress that his command would need “complete and uninterrupted access” to DoD spectrum bands to deliver on this new directive.
Midband miscalculations and assumptions
Secondly, while NERA’s analysis has avoided some of the errors in CTIA’s midband frequency definitions, which we highlighted in previous work, it still ignores the huge tranche of 2.5GHz midband spectrum owned by T-Mobile. It also adds in 700 megahertz+ of the 6GHz band to its calculations and comparisons of the US vs. China.
Yet contrary to CTIA’s frequent claims, this band has not yet been made available to Chinese operators, and the scope and timing of its release remains unclear. No dates were given in a recent presentation by China’s regulator at an Asia Pacific spectrum conference in Bangkok.
Furthermore, frequencies such as 7-8GHz currently being discussed as possible upper-midband auction candidates are likely to have much lower economic value (and thus auction price and user impact) than those in the 2-4GHz range discussed in the report, as they have shorter range and limited indoor reach.
The NERA analysis treats the value and utility of all spectrum as close to equal and does not make a clear distinction between different tranches of midband (for instance, in the range 3-4GHz vs 7-8GHz) and what that would likely mean for auction fees, equipment procurement, and the claimed economic benefits of greater infrastructure deployment.
Industry discussions widely suggest that 5G bands above 6 or 7GHz will likely be deployed mostly in the densest urban areas only, where capacity needs are concentrated. Worse, their poor outdoor-to-indoor performance means that only the 20-30% of outdoor usage of mobile networks would see meaningful benefits for end-user applications. To not account for these limitations for value in the report is to ignore the role that the laws of physics — and the realities of where mobile services are actually used — play in determining the economic value of spectrum.
These limitations of inherent value are already starting to show up in spectrum pricing — which the report asserts is correlated to underlying value perceived. Yet, the report conveniently ignores available data on pricing for mid-band spectrum. For example, a recent spectrum auction in Hong Kong yielded far lower valuations per MHz-pop for 6GHz allocations, compared to a set of lower frequencies in the 2.3GHz range, sold a couple of months earlier.
This casts significant doubts on NERA’s calculations of both consumer surplus and its estimates of “supporting industries that rely on mobile connectivity.” Unless the FCC actually mandated high levels of population coverage as a license condition, perhaps including indoor reach, many new spectrum bands may end up confined to very limited geographical deployment, reducing any economic benefits for equipment suppliers, construction firms and others.
The report instead scales up and adjusts CAPEX and OPEX spend reported for C-band deployment, to its hypothetical 400 megahertz of extra spectrum, without considering that footprints may look very different.
This has already been observed for 28 GHz and other higher bands — U.S. carriers have sizeable allocations of mmWave frequencies that are currently mostly unused, and yet also suitable for very high-density deployments. The spend has been a fraction of that seen for C-band networks, because they proved less useful and were just deployed across small areas.
Technology improvements
Thirdly, NERA’s study underestimates the impact of the evolution of wireless network technology, especially against a backdrop of falling traffic growth rates. It asserts: “Without additional spectrum, networks would eventually become congested.” This is far from being a cast-iron truth.
While the report discusses the potential for network densification (building a denser grid of cell-sites), it completely overlooks the contribution of other new technical enhancements, such as using newer radios with better “beam-forming,” as well as using AI to optimise various parameters and functions of the network to yield higher spectral efficiency.
Numerous recent and ongoing developments could each yield 20-30% to achievable throughputs with existing spectrum. Recent innovations include using automated tilting antennas, or better in-radio estimation techniques for user mobility or radio interference. AI specialists such as GPU vendor Nvidia, or Open RAN software supplier Cohere Technologies, frequently assert they can enable significant capacity gains.
These could improve the capacity of existing cell-sites and existing spectrum — potentially enough to meet even bullish demand forecasts in the study.
If NERA truly believes that incremental network capacity directly translates to economic benefits, it is surprising that it advocates for new (and expensive) spectrum and infrastructure, rather than optimising the throughput and data that can be handled by existing bands.
Industries that rely on mobile connectivity
The NERA analysis also attempts to calculate incremental value of extra spectrum to applications such as social media and audio/video streaming. It uses a convoluted series of assumptions to suggest that congested networks impede application traffic, which could “directly translate into lost sales and foregone revenue,” and then to GDP and employment impact.
There are multiple flaws in the logic and analysis, both in the suggested sequences of cause-and-effect, and the specific numerical metrics and calculations employed:
- It uses a 16% growth rate of data traffic all the way until 2040, based on an Ericsson projection up to 2030, which itself seems a significant overestimate judging from recent trends seen around the world, including 0% growth in China and 7% in the U.K. and analysis by industry commentators such as Professor William Webb. The 2040 forecast is not realistic, especially for smartphones and excluding fixed-wireless traffic.
- It also discounts the idea of densification or radio network enhancements having an impact on network capacity. This is unrealistic too — see next section.
- It focuses on spectrum and therefore peak capacity, when the limiting factor is often reliable basic coverage, especially in indoor settings reliant on outdoor-to-indoor propagation.
- It ignores application-level evolution such as better compression, which is a fast-moving space with techniques such as AV1 already gaining wide adoption.
- It neglects to mention that many applications already offer “data saving” modes, which implies it does not impact their revenue.
Furthermore, most non-mobile traffic for services such as YouTube and Netflix is delivered over Wi-Fi, typically to TVs, smartphones and laptops in the home. However, unlicensed spectrum is also often congested, so a similar analysis would yield even larger volumes of traffic foregone and thus missing value.
In short, the analysis is not rigorous or credible — as further evidenced by the entire lack of calls for extra spectrum from the largest content and application firms themselves.
FWA market value
Another section of the report examines the theoretical value of adding spectrum for the purpose of fixed wireless access.
A full analysis of this could easily absorb a full-length article of its own. However, a central claim is that “additional spectrum increases coverage by expanding the capacity of existing base stations and increasing the profitability of new ones.”
It is very hard to square that assertion with the main focus on underserved (primarily rural) areas in which broadband availability or choice is limited. The report analyses broadband coverage by population-density decile, and states that “we assume that with an additional 400 megahertz of mid-band spectrum, FWA coverage in a given country would be similar to today’s coverage of a county with 2.05 times its density.” Given that most rural cells are uncongested today, and often have not “lit up” all existing operator spectrum, it is hard to see how adding more spectrum — especially in frequency bands with limited range — would yield a significant uplift in FWA availability or usage.
Furthermore, the carriers’ reluctance to fully exploit other options such as CBRS, mmWave or even unlicensed versions of 5G for FWA also raises questions: why are they “leaving money on the table,” if any new spectrum adds huge value to the proposition? While not all of these are usable with full power, they would still be able to add incrementally to the footprint, yet are left ignored, for some undisclosed reason, suggesting to outside observers that their desires for more high power nationwide exclusive spectrum is less about increasing capacity and more about keeping the spectrum away from others.
Conclusion
The NERA report is ultimately written by an economic consultancy with a focus on spectrum auctions, rather than an organisation steeped in the details of radio network deployment, usage and technological advances.
As such, it appears that its report has avoided some critically important questions in evaluating the true potential economic benefits of spectrum, such as where deployment can be realistically expected, how such networks might be used, what the costs and challenges of moving incumbents might be, and which adjacent trends such as technical efficiency gains might offer as alternatives.
In summary, these critical omitted points render the NERA report’s analysis incomplete while overstating the economic benefits of extra 5G spectrum and leaving the consequential costs ignored and alternative mechanisms for obtaining extra capacity or coverage overlooked.
