The success of the telecom auctions depends on how much ‘data’ spectrum is on offer, not the ‘voice’ one
Though the Telecom Commission has done well to, by and large, accept Trai’s recommendations on lowering the base price for the forthcoming spectrum auctions, the chances of these doing well are low unless a few more steps are taken—industry wants a certain type of spectrum but not enough of that is on offer.
In a very broad sense, you can divide telecom into ‘voice’ and ‘data’ and, again very broadly, you can categorise spectrum in the same manner. So, the 1800 MHz frequency band is broadly ‘voice’ spectrum, the 2100 MHz frequency is largely ‘data’ and the 900 MHz one can effectively be used for both ‘voice’ and ‘data’. You can, of course, use even 1800 MHz spectrum for data, and some firms will bid for this spectrum primarily to do ‘data’, but there are some complications. For one, there isn’t that much of an ecosystem—most affordable smartphones are geared for 3G and work on the 2100 MHz frequency, not 1800 MHz—which means you could have a repeat of the CDMA fiasco years ago when, though the technology was good, there simply weren’t enough mobile models available for customers. Two, 1800 MHz spectrum is fragmented, not giving telcos the 5 MHz contiguous slot they need for data operations.
Given that (see chart 1) the ‘voice’ market is largely saturated, and low-paying at that, telcos are unlikely to pay much for purely ‘voice’ spectrum. Growth in subscriber minutes has levelled off from 54% in FY09 to 9% now; just a few days ago, RCom announced it had cut off 10 million subscribers. Subscriber growth for ‘data’, on the other hand, is powering ahead. For Bharti Airtel, ‘data’ subscribers have risen from 35.8 million in September 2011 to 43.5 million in September 2012 and to 50.6 million in September 2013, their average data usage has shot up from 107 MB to 187 MB and 231 MB in the same period. In terms of average revenues per user (ARPU), this has risen from R44 to R55 and R70 in the same period—in just the last four quarters, the share of ‘data’ as a proportion of mobile revenues is up from 4.9% to 9.2%.
To that extent, telcos will bid the most for ‘data’ spectrum and the least for ‘voice’ spectrum—that is, the highest bids will come in for 900 MHz and 2100 MHz spectrum and the least for the 1800 MHz spectrum which, while it can be used for ‘data’, has limitations we’ve just discussed. It’s important to keep in mind, as compared to 2010 when telcos bid very irresponsibly, the bids are likely to be more ‘bottom up’, or carefully considered for each circle and for each area of business. It’s a good idea to look at each spectrum type differently.
Take the 900 MHz spectrum that comes up for renewal soon in Delhi, Mumbai and Kolkata. In Delhi, this is held by Bharti and Vodafone who will, in any case, be loathe to lose this given the bulk of their revenues are serviced from this spectrum. The bid price will then depend on the number of other entrants there are. Theoretically, a Reliance Jio can also want to come in, as can an Idea or an RCom. This is where the business case comes in. Based on Trai recommendations, the base price for 900 MHz spectrum is R285 crore per MHz (R218 crore for 1800 MHz), which means a 5 MHz slot will cost R1,425 crore in Delhi (R1,093 crore for 1800 MHz), no small amount of money. This means the company has to generate R213 crore of ebitda per annum (R164 crore for 1800 MHz), assuming a 10% cost of capital and amortisation over 20 years. That, in turn, means R713 crore of annual revenues (R547 crore for 1800 MHz), assuming an ebitda margin of 30%. Given the Delhi ‘data’ market is R1,000 crore or thereabouts right now, this means a telco bidding for Delhi needs to be looking at capturing around 30-40% of the data market over the next 2-3 years. Given how entrenched the incumbents are, betting on this may not be the smartest strategy. And since the incumbents already have a revenue stream to defray costs against, they can afford to bid up the price to keep new entrants out.
So telcos need to be smart about their bidding—at a collective level, they bid R67,000 crore for 3G spectrum in 2010 which means they needed an ebitda of R10,050 crore each year and a revenue of R33,500 crore; what they’re getting is a fourth or a fifth of this. Two, telcos need to look for alternatives where spectrum costs are lower and where the competition will be less—1800 MHz and 2100 MHz offer the best solutions. Even if you forget the limitations of 1800 MHz, the government is thinking of charging a premium for contiguous spectrum, making it that much less attractive.
In which case, the government’s best bet is to auction less of the 1800 MHz spectrum right now and wait till it is able to get more contiguous spectrum. Meanwhile, since Trai has said there is no need to reserve spectrum for re-farming, it can release the 1900 MHz spectrum it had kept for re-farming for CDMA mobiles and swap this with the defence ministry for 3 slots of 5 MHz each in the 2100 MHz band. That way, telcos get the spectrum they want and the government the money it so desperately needs—at even half the prices they fetched in 2010, this could mean another $5-6 billion in auction bids.