Back in 2010 the FCC came out with its notorious forecast that there would be a 275MHz spectrum deficit by the end of 2014, based on projections of traffic growth, cell site growth and network efficiency. I pointed out at the time that there were problems with their calculations, and in reality it turned out that at the end of 2014 there was a nearly 300MHz spectrum surplus not a deficit.
The primary reason for this error was the flawed estimate of network efficiency, which was predicted only to double from 0.625bps/Hz to 1.25bps/Hz between 2009 and 2014 (note that in practice new cellsites are also placed to meet data traffic demand, but for simplicity I’ve rolled this into overall network efficiency).
In reality, if we plug in the actual growth in reported traffic and cell sites over the period from 2009 to 2014 (as summarized by Brattle in their 2015 CTIA-backed effort to continue the spectrum crisis narrative) the implied network efficiency (i.e. traffic per MHz per cell site) was 525% of the 2009 value by 2014, not 200% as the FCC predicted (i.e. a 425% improvement in efficiency, not the 100% predicted).
The chart above combines this data with Brattle’s predictions of traffic and cellsite growth from 2014-2019 and highlights that once again the prediction is for very modest network efficiency gains over this period, to only 139% of the 2014 figure in 2019, and Brattle use this figure to assert that there will once again be a spectrum deficit of over 300MHz by 2019.
Put another way, Brattle assert that the efficiency gains between 2014 and 2019 (39%) will be less than one tenth of the improvement that occurred between 2009 and 2014 (425%). Of course, in the real world, massive efficiency gains are already flowing from the deployment of MIMO technology, and Gigabit LTE is only just getting started.
Sprint stated in its October 2016 spectrum leaseback transaction that it is already achieving average spectrum efficiency of 1.6-1.95 bps/Hz, well above the Brattle predictions for 2019, and that it expects the average downlink efficiency in its Band 41 spectrum to reach almost 13bps/Hz by 2019 after the introduction of massive MIMO.
Indeed, the only way Sprint was able to justify the high valuation placed on its spectrum (based on a calculation of how much capex Verizon would be able to avoid by purchasing Sprint’s spectrum) was to assume that massive MIMO would not be feasible in paired spectrum bands. In practice massive MIMO is already being addressed in paired spectrum, and will simply be a bit less efficient than in TDLTE implementations. And if even a fraction of the efficiency gains set out by Sprint are applied to the FCC/Brattle model, then it is clear that there will continue to be a large spectrum surplus not a deficit.
So returning to my title, if spectrum is like oil, then MIMO is just as revolutionary for the spectrum market as fracking has been for the oil market. And as we are seeing in the incentive auction, where Verizon declined to participate, and AT&T has said it will be spending less than $2.4B, there are very similar implications for the price of spectrum as we’ve seen for the price of oil.