Indian Spectrum Auctions: Key Design Issues & Way Forward

Since 2010, India has seen six auctions. This has led to huge outflows of cash from the Industry - probably highest in the world. This has impacted the industry's ability to invest in new technology and improving coverage. Some argue that "auction design" is responsible for the price of spectrum increasing exponentially. In this note, I plan to bring out these deficiencies backed up by hard data.

2010 3G Auction

The following in the "bidding matrix" of the 2010 3G auction of the 2100 MHz band.

Each rectangle in the above figure represents an operator. Each dot within the rectangle indicates an operator making a positive bid. One can clearly see the increased density of these dots during the latter part of the auctions. This looks odd, isn't it? Why would the bidders bid in every round when they didn't in the earlier phase of the auction? The culprit was the auction "closing rules" defined in the NIA (Notice Inviting Applications). This rule had designed the auction to end abruptly once the "activity factor" was set to 100%. Hence, to ensure a win, the bidders were forced to bid in every round to ensure a win (not knowing when the auction will end). This forced the price of important circles like Delhi and Mumbai every high (40% of the auction value).

In order to get a sense of the magnitude of the impact caused by the auction closing rules one needs to evaluate the money put on the table by each operator in every round of the auction, i.e the provisional winning bids. The following picture captures this for all the operators for the 3G auction.

One can clearly see the increased value of the bids placed during the latter part of the auctions which can be clearly attributed to the "closing rules". It can be clearly concluded that bids would not have increased that high had the closing rule was not set such.

2010 BWA Auction

The following in the "bidding matrix" of the 2010 BWA auction of the 2300 MHz band.

The impact of the closing rules was felt in this auction as well. However, some bidders like Infotel (now RJIO) had bid very overenthusiastically. This clearly evident by the increased density of dots in the above picture. This led to the disproportionate price increase.

2012 Auction (1800 MHz)

The following in the "bidding matrix" of the 2012 auction of the 1800 MHz band.

The auction was a subdued affair and got over in 14 rounds and most of the spectrum remained unsold. See figure below.

Why? This could be due to the following auction design issues; a) the price of spectrum was set equal to the price discovered in the 2010 3G auction, b) an unusual block size of 1.25 MHz was chosen - making most of the spectrum discontiguous - unsuitable for 3G/4G.

2013 Auction (800 MHz)

Strangely, in this auction no one bid except MTS. A total 5 MHz of spectrum in discrete chunks of 1.25 MHz was offered - resulting in some unsold spectrum, even when the spectrum price was drastically reduced (by 50% of the price set in 2012, when no one bid). See figure below.

2014 Auction (900 MHz)

The following in the "bidding matrix" of the 2014 auction of the 900 MHz band.

The basic design flaw in this auction was the insufficiency of spectrum availability with no fallback option for those whose licenses were expiring. Most bidders especially, RJIO and Tata behaved responsibly and did not contribute to the price increase (they left the auction early). But, Idea with no 3G spectrum in Delhi continued to bid and snatched 5 MHz of the renewal spectrum from Bharti & Vodafone - leading to the price increase. This is clearly evident from the auction value matrix in the figure below.

2014 auction (1800 MHz)

The following in the "bidding matrix" of the 2014 auction of the 1800 MHz band.

The basis design issue in this auction was the fragmented spectrum of an unequal block size (apart from high price). If this was not the case why would some spectrum stay unsold? See under.

2015 Auction (800 MHz)

The following in the "bidding matrix" of the 2015 auction of the 800 MHz band.

The basic design issue in this auction was unequal block sizes bids allowed between a new player (had to bid 5 MHz) and an existing player (can bid in chunks 1.25 MHz). This "unequal slot" auction design prevented RJIO to outbid Tata. See figure under.

2015 Auction (900 MHz)

The following in the "bidding matrix" of the 2015 auction of the 900 MHz band.

The basic design flaw in this auction was the insufficiency of spectrum availability with no fallback option for those whose licenses were expiring. This provided an opportunity to some bidder to park their points in this band without risking being stuck with unwanted spectrum, as the bidders whose licenses were expiring would have no option but to continue bidding. For example, RJIO parked its point from 800 MHz (due to Tata bidding lower value as explained above) here, thereby forcing Bharti, Vodafone and Idea to bid - increasing their overall outflow. See figure under.

2016 Auction (All Bands)

The basic design problem with 2016 auction was too much spectrum on the table offered at an exorbitant price. If that was not the case then why most of the spectrum remained unsold? See figure under.

Please note that 700 MHz band not included in the above figure which was left totally unsold due to the unrealistic reserve price.

Way Forward

Indian spectrum auction had design issues. These issues have contributed to higher spectrum price and unsold spectrum. Also, the reserve prices were not corrected in the future auctions to prevent these distortions from propagating and amplifying. Hence, India needs a new model for pricing spectrum based on the following principles.

The pricing model has been described at length in my earlier article titled - How to Calculate Reserve Prices for Spectrum Auctions. This model is reproduced as under for ready reference.

The prices calculated using this model is listed under.

One can see that the reserve prices calculated using the new model curate the price distortion emanating from the account of auction design issue - making the process more robust and sustainable.

(Views expressed are of my own and do not reflect that of my employer)

PS: Find the list of other relevant articles in the embedded link.