There’s been a lot of talk about whether or not the spectrum auction that ended last week was a success or a failure.
The latest auction needs to be seen in the backdrop of the one in 2015. In that auction, the centre raised Rs1,09, 874 crore. Only 11% of the spectrum, across the 800MHz, 900MHz, 1800MHz and 2100MHz bands were unsold.
Still, that auction was a question of survival for some telcos whose 20-year licences were coming to an end.
The government should have factored that into its calculations when it decided to auction 2354.55MHz of spectrum. India is divided into 22 telecom circles, and spectrum was offered in seven different frequency bands (700, 800, 900, 1800, 2100, 2300 and 2500Mhz). There were a total of 126 Circle-Band combinations on which spectrum was sold. If all radio waves were sold at their reserve price, the government stood to make roughly Rs5.6 trillion.
What do the results show?
The centre managed to raise just around Rs66,000 crore from the sale of 965MHz of spectrum.
Indeed, it is not clear if the word auction should be used to describe the process in which the Department of Telecommunications sold spectrum. Only 11 out of 126 Circle-Band combinations on offer saw even a semblance of competition among bidders. In the remaining 115, any sale that happened took place at the reserve price.
Within a band, there were different blocks of spectrum that were up on the block, and these blocks were treated as if they were identical.
In each round of bidding, telcos had to indicate their interest in purchasing a certain number of blocks of spectrum in a particular band at the price that had been fixed for the round (prices varied by circle and by band). Depending upon the demand-supply equation, prices would either remain the same or increase in each round. This process would go on simultaneously for all bands and circles until there was no band for which demand outstripped supply at the prevailing price. (This time the process went on for five business days and 31 rounds of bidding. The 2015 auctions lasted 19 days and saw 115 rounds of bidding.)
While the auction mechanism was extremely sound and well designed, it was largely rendered moot by the absolute lack of buyer interest in the auction process. The most intense auction took place for the 1800MHz spectrum in Mumbai, which fetched at least 33% more than the reserve price. There were only 10 other frequency bands across the country where we can conclusively say that the winning bid was greater than the reserve price (see table 1 below).
At the other end, large blocks of spectrum went absolutely unsold. All in all, 1,532 blocks of spectrum across different circles and frequency bands were on sale, among which nearly 600 went unsold. And these 600 included blocks with the largest reserve price attached—those in the 700MHz spectrum.
As chart 1 shows, the average reserve price for each block in the 700MHz band was significantly higher than that of other bands. This was possibly dictated by the attractiveness of this particular band in being able to offer fourth generation (4G) services.
In terms of money the government expected from the auction process, the 700MHz band accounted for more than two-thirds, accounting for over Rs4 trillion (see chart 2). The lack of demand in this segment meant that irrespective of the performance in other bands, the auction would not meet its financial objective.
The best-performing spectrum band in the auction was the 1800MHz one, where nearly 80% of all blocks on offer were lapped up. Interestingly, this is the one band where the number of blocks on offer was large, and the reserve price rather low. Apart from this, no other frequency band saw more than 50% of available blocks being bought.
There are two reasons why an auction can fail. Firstly, buyers could collude in order to keep prices low. Given the recent entry of Reliance Jio, and its battles with the incumbent players that constitute the Cellular Operators Association of India (COAI), it is hard to believe that all telcos colluded to stay away from certain parts of the auction.
That leaves us with the other reason for failure—that the reserve prices were set too high. Now, setting reserve prices for an auction is an art in itself. Set it too high, and it simply discourages buyers from engaging in a process of price discovery.
Set it too low, and it encourages collusion among buyers. In this case, with chances of collusion being low, it is surprising that DoT went with high reserve prices.
While the data available tells us little about why high reserve prices were chosen, what the data does indicate is that there was a fair amount of randomness in determination of the reserve price. For example, we would expect higher reserve prices in circles with more subscribers. While chart 3 shows that such a broad relationship exists, it is a weak one.
While no one expects a perfectly linear relationship, the degree of randomness described by the graph is too high for comfort. What it suggests is that the process for setting reserve prices was reasonably ad hoc.
With this auction over, it will be interesting to see how DoT adapts for the next one. Slashing reserve prices significantly might be an admission that the government got it wrong in setting prices too high this time, while not doing so might result in another failed auction.
This gives rise to the legitimate concern that DoT, on account of this failed auction, might put off indefinitely the next one, until market conditions have changed sufficiently for the current reserve prices to make sense.
There is another course of action available to DoT: given that most of the sale in this process happened at (or close to) a price chosen by DoT, the unsold blocks can simply be made available on tap. As long as these blocks of spectrum are available, any telecom company should be able to match the prices discovered in the auction to buy them. This can help both the government and the telcos, although it has uncomfortable parallels with the way spectrum was sold in the past, and which resulted in the so-called 2G scam.