10.22.10

Spectrum crisis or spectrum bubble?

Posted in Regulatory, Spectrum at 8:43 pm by timfarrar

I vividly remember attending the Cisco analyst conference in the fall of 1999, soon after I moved to the US, and being confronted by a rosy picture of ever increasing demand for telecoms equipment because of the dramatic growth in Internet traffic. I was moved to ask the question of whether in fact we were in a bubble, and if Cisco’s revenue trajectory would therefore end up resembling its logo. Perhaps unsurprisingly, I was never invited back to that particular event.

This evening I took the time to read Andrew Odlyzko’s seminal paper on Internet traffic growth in the 1990s. This paper explores the myth that Internet traffic was doubling every 100 days, which was a key contributor to the telecoms bubble of the late 1990s, and the subsequent loss of hundreds of billions of dollars of investors’ money. One of the highlights mentioned in Odlyzko’s paper was former FCC Chairman Reed Hundt writing in his book You Say You Want a Revolution that “[i]n 1999, data traffic was doubling every 90 days ….”. In reality, Internet traffic was growing rapidly, nearly doubling every year, but the growth rate from a short period of initial growth in 1995-96 had been erroneously equated to a long term growth rate. It was of course in everyone’s interests to promote this myth – it pumped up valuations, allowed startups to raise billions of dollars, and enabled politicians and others (such as Cisco) to claim credit for developing “the new economy”.

Thus it is particularly striking that we now see the current FCC Chairman using a flawed study to promote the “transformative moment we must seize” because “we are likely to see a 35X increase in mobile broadband traffic over the next 5 years” (based ironically to a significant extent on Cisco forecasts). We also see startups such as LightSquared and Clearwire raising (or at least trying to raise) billions of dollars on the back of their spectrum assets to build out new networks and realize the “extraordinary opportunity of communications technology” that the FCC Chairman envisages.

Of course it is true that mobile broadband demand is growing rapidly, just as Internet demand was growing rapidly in the late 1990s. However, perhaps we should pause and ask whether we are in danger of repeating some of the mistakes of a decade ago? When cellular operators, spectrum owners, equipment suppliers and regulators all have a vested interest in talking up the spectrum crisis, there has to be a danger of disappointment. I’ve already commented that spectrum might not be such a good investment and that the Clearwire auction looks to be going badly. There are even rumors that Clearwire is preparing for a round of significant layoffs. As a result I have to wonder whether we will look back at this moment in six months or a year, and conclude that yesterday’s FCC spectrum summit marked the point at which the spectrum bubble began to unwind.

Analyzing the spectrum “crisis”: can the FCC add up?

Posted in Regulatory, Spectrum at 2:41 pm by timfarrar

Yesterday the FCC announced with much fanfare the release of a new technical paper giving its “Mobile Broadband Spectrum Forecast”, which was trumpeted by the FCC Chairman as putting “the importance of this debate in perspective by assessing the economic value of this spectrum, which it estimates to be as high as $120 billion”.

The FCC paper is a fairly unsophisticated analysis, and its results contrast significantly with an Ofcom study released in April 2009 which found that under most scenarios for data growth the UK was unlikely to experience a spectrum crisis until 2020 or beyond. Although the UK is of course a very different market to the US, it was a surprise that the results were so dramatically different, which prompted us to take a closer look at the FCC calculations.

Unfortunately it appears that the FCC analysis contains a significant error in its calculations. Specifically, while the study assumes that all future cell sites are deployed to add capacity, it neglects to take into account that a significant fraction of current cell sites were deployed purely for coverage in rural and suburban areas, and are unlikely to ever need more spectrum beyond that available today. As a result, following the FCC’s methodology, these cell sites should not be included when calculating how much new capacity and spectrum is needed (as the FCC indicates elsewhere in the paper, upgrading to more spectrally efficient technologies would provide significant gains and presumably sufficient capacity when data usage does increase in rural areas).

UPDATE: Assuming somewhere between 20% and 50% of current cell sites (as of 2009) have been deployed for coverage rather than capacity (which seems reasonable, as according to the FCC Competition Report, cellular voice services are provided to 60M people in rural census blocks covering 2.3M square miles, and 76% of these regions have two or more providers) then the FCC’s own analysis shows that the incremental spectrum demand by 2014 is reduced from 275MHz to between 117MHz and 227MHz and the cost of deploying extra cell sites to serve this demand (the FCC’s definition of “economic value”, assuming a cost of $550K per cell site) is reduced from $120B to between $33B and $85B. Only 60K-155K additional cell sites are needed to meet demand, compared to the 99K that the FCC’s model estimates will be deployed anyway in the next 5 years. Put another way, the annual growth in the number of base stations would increase from 7% p.a. to between 10% and 15% p.a., which is not significantly out of line with the 12% growth in base stations that occurred in 2009, according to the FCC’s own figures.