Looks like that the Indian spectrum auctions have become mega-events of “Spectrum Display” than “Sale”. Why? A huge amount of spectrum is offered in these auctions, but most of it is put back into the inventory to be offered again in the next round. This phenomenon is true for all auctions held so far, except for the years (2010, 2014, 2015) when the bidders were forced to bid even at high prices to ensure the continuance of their business (in 2014 & 2015, when the 2G license expired) and preserve the value of their existing investment (in 2010, when the auctions were catalyzed by the subsidized entry of new operators). See chart below.
But why is this so? In order to understand this, I read through all the past recommendations of the TRAI (Telecom Regulatory Authority of India) and the follow-up action of DoT (Department of Telecommunication) in setting the “Reserve Price” for the past auctions. Focusing on “Reserve Price” is important, as the government’s ability to sell spectrum is strongly dependent on it. Optimally set “Reserve Price” ensures wider participation, investing in infrastructure, and helps preserve the competitiveness of the market. Doing so is beneficial for all — consumers, operators, and the government, and the loss of it CANNOT be compensated by any amount of revenues the auction ends up generating. Why? As any disproportionate amount of money the operators pay for spectrum is done with the sole objective of increasing the barrier for its competitors to enter the market. Now, this discussion is particularly important today when the GOI is lining up for the next round of spectrum auction for 5G, the success of which strongly depends upon our acknowledgment of past mistakes and not repeating them again. Hence this note.
Reserve Price Models
In this section, we will go through all the auctions of the past to understand how the process of setting reserve prices became too simplistic, trivial, and confusing— defeating the whole objective in the first place.
2010 Auctions
The process of setting reserve price for any auction starts with TRAI first, and then it is up to the DOT to accept or reject it. For the 2010 auction, TRAI set a reserve price of Rs 1050 Cr for 2x5 MHz of 2100 MHz spectrum. Not sure what pricing model TRAI used, but the DOT raised it to Rs 3500 Cr without assigning any reasons. However, the auction resulted in a price of Rs 16,749 Cr (4.8 times increase). We all know the reasons why — a) Scarcity — 3 slots were chased by 8 to 10 players; b) Survivability — without the 3G overlay, the operator’s 2G investments would have added little value.
2012 Auction
Then in 2012, the TRAI used the 3G auction prices to estimate reserve prices of all bands. The pricing model used was too simplistic. See under.
This raises some serious questions —
a) Why no ground-up valuation of the spectrum was done for arriving at reserve price? b) What was the need for pricing the 700 MHz band when even the band plan for it was not finalized and there was no ecosystem yet, which the TRAI itself had acknowledged (page 72, clause 2.130)? c) Why a multiplying factor of “4” was used for 700 MHz when for the same category of bands (800 & 900), the TRAI recommended a factor of “2”? d) Why the reserve price of 2100 MHz was not scaled down by 20% as was done for the 1800 MHz band?
The last question is specifically important, as this has set the precedence — making the whole process of estimating reserve prices too trivial and totally useless. Actually, in many situations, the process even contradicts TRAI’s own recommendation reproduced under for easy reference (Clause 3.83, Page 100, dated 23rd April 2012).
“A study of various auctions held globally in last 3 to 4 years reveals that the reserve prices are generally around 0.5 times the final prices. However, in the context of the Indian telecom sector, where the demand for spectrum is considerably higher, the authority has decided to use a factor of 0.8 to determine the reserve price”
Now one needs to ask these important questions as well.
A) When the final prices of 2010 3G auctions were available, and the same was used to arrive at the prices of other bands like 1800 MHz, and which was finally scaled down by a factor of 0.8 for estimating the reserve price, then why didn’t the TRAI apply the same scaling factor for estimating the reserve price of 2100 MHz band as well? B) Why didn’t the TRAI in the later years change the factor of 0.8 to 0.5 (used internationally) when the spectrum was left unsold in large quantum (indicating no scarcity), as the factor of 0.8 should have remained only if the demand for spectrum was high? This is as per TRAI’s own recommendation, No?
This is a perpetual problem both with the regulator and the licensor, i.e first they define policies by laying out some caveats, and then later pick the same policy without taking into the caveats which were used to define the policy in the first place.
Nevertheless, DOT asked TRAI to review the prices of the 800 MHz band, and TRAI responded saying that the factor can be reduced to 1.3 x of 1800 MHz since 2x5 MHz is not available in most circles. Then DoT cut the reserve price of 1800 & 800 MHz bands by 30% and 50% respectively. But in spite of that kind of reduction, the 800 MHz had no takers. It is worthwhile to note that as they (DOT & TRAI) reduced the prices of the 800 & 1800 MHz band, they totally forgot to align the 900 & 700 MHz band prices too, thereby disturbing the balance between bands.
2013 Auctions
This auction happened within 4 months after the last auction of 2012. Here again, the DOT cut the price of 800 MHz by another 50%. For the 1800 MHz band, it did so by 30% in those circles that did not sell in the last auction. The 900 MHz band price it kept at 2 x that of the 1800 MHz band. Again, the auction was a miserable failure with MTS taking few circles of the 800 MHz band.
2014 Auctions
In this auction, the TRAI decided to do an independent assessment of the value and reserve price for the 1800 MHz spectrum than just linking it to the 2100 MHz price of 2010 as was done in 2012 (Clause 3.22, Page 56, TRAI’s recommendation dated 9th Sept 2013). However, it did NOT totally disregard the past 1800 MHz auction/reserve prices and calculated based on 3G auction prices of 2010. Instead, it adopted an innovative approach. It did a ground-up valuation of 1800 MHz using various pricing models, and without considering the last auction/reserve prices of 1800 MHz as one of the many inputs into the model. Then the 80% of the valuation (emanating out of the model) was compared with the last auction/reserve price of 1800 MHz. The final reserve price was pegged as the “Minimum” of the two figures. See Chart below.
A similar approach was used for valuing the 900 MHz band. See Chart below.
But DOT later raised all these prices (both 1800 & 900 MHz bands) by 25% without assigning any reasons.
2015 Auctions
In this auction, the TRAI used a similar approach to valuation as it did in the last auction, but with a caveat. The 1800 MHz reserve price was set Maximum of the two figures a) The last 1800 MHz Auction Price; b) 1800 MHz 80% Valuation emanating out of the model. See Chart below.
For the 900 MHz band, the Valuation was set Minimum of the two figures — a) 2 x 1800 MHz Average Valuation; b) Average Valuation of the Model. See Chart below.
For the 800 MHz, the reserve price was just set as 80% of the Valuation emanating out of the model. See chart below.
For the 2100 MHz also, the reserve price was set as 80% of Valuation emanating out of the model. See chart below.
The DOT later raised the prices for all band circles than what the TRAI recommended. By how much the DoT raised the prices can be seen from the tool embedded here.
The Problem
The basic problem in TRAI’s valuation method is again the use of caveats of “Maximum” and “Minimum”. Instead, the TRAI should have just followed a simple approach of using the 80% of the “Median” values that emanated out of the models as the “Reserve Price”. The use of “Maximum” and “Minimum” though can be justified on a contextual basis, however (as stated earlier), the context was totally forgotten in the later years, thereby translating the valuation process into a simple exercise of pegging the indexed value of the last “Auction Price” as “Reserve Price”.
If the whole exercise of estimating “Reserve Price” just translates into indexing the last “Auction Price”, then what is the point of going through the elaborate process of consultation and waste so much time and effort? The DoT is fully equipped to do it by themselves, No?
2016 Auction
By this time TRAI started lacing the output of their pricing models with more caveats. This made the overall process of estimating reserve prices more confusing and redundant. Why? The final output of TRAI’s reserve price model in most LSA was just the indexed value of the last auction price. For example -
- 1800 MHz — RP became the indexed value of the last AP in 16 of the 22 circles.
- 800 MHz — RP became the indexed value of the last AP in 8 of the 22 circles.
- 900 MHz — RP became the indexed value of the last AP in 6 of the 22 circles.
- 2100 MHz — RP became the indexed value of the last AP in 18 of the 22 circles.
Strangely, TRAI adopted a different approach for setting up the “Reserve Price” for the 2300 & 2500 MHz bands. It reduced the indexed value of the last auction price of 2010 by 20% in order to arrive at its reserve price. Why? For the purpose of consistency, shouldn’t the RP be just the indexed value of the last AP? See Chart below.
Also, TRAI picked up their old recommendation of 2012 and pegged the 700 MHz band price as “4" times as that of the 1800 MHz band. Why? That too when the calculations were wrong (Check out — Is TRAI’s 700 MHz Price Outcome of an Inadvertent Error? & Pricing Rationale of 700 MHz) and price of 800 and 900 MHz bands are much lower and when estimated based on ground-up valuation? This was the most irresponsible part of the TRAI’s recommendation, as it destroyed the band forever, and kept it lay unused, which otherwise could have been used for expanding broadband connectivity to rural India.
And surprisingly DoT accepted all these prices (of all bands) without raising any objections.
2021 Auctions
In 2021, the TRAI following a similar approach it did in 2016. This resulted in the following.
- 1800 MHz — RP became the indexed value of the last AP in 18 of the 22 circles.
- 900 MHz — RP got Reduced to 80% of Average Valuation (No Auction Price of 2016).
- 800 MHz — RP got Depressed as in Many Circles Spectrum Offered Did not Sell.
- 2100 MHz — RP became the indexed value of the last AP in 11 of the 22 circles.
- 2300 MHz — RP became the indexed value of the last AP in 16 of the 22 circles.
- 2500 MHz — RP became the indexed value of the last AP in 14 of the 22 circles.
And DOT accepted all these recommendations — no questions asked.
Summary of Issues
To an attentive reader, the issues should have been clear by now. a) Why did TRAI and DOT mix the concept of “Reserve Price” and “Valuation of Spectrum”? This is in spite of the fact that TRAI had on multiple occasions recognized that “Reserve Price” should be always less than the “Valuation” by a certain percentage. Then why both DoT and TRAI used the indexed value of the last auction price as reserve price? b) Why TRAI in 2016 picked up an outdated recommendation of 2012 to estimate the reserve price of the 700 MHz band? Why did the industry remain a mute spectator and did not object to this gross mistake of TRAI? Strange, isn’t it? c) Why did the TRAI in 2016 scaled-down (by 20%) the indexed value of the 2010 auction of 2300 & 2500 MHz band in order to arrive at its reserve price? This is when, as a practice, they always had been using the indexed value of the last auction price as reserve price? d) Why did the TRAI not used the scaling factor for arriving at the “reserve price” from “valuation” as 0.5 (instead of 0.8), when the demand for spectrum went down (unsold spectrum)? This is when they in their recommendation (mentioned above) had acknowledged that internationally 0.5 is used to scale down the valuation to the reserve price?
And many more of such issues….
Conclusion
By now you might have concluded why the Indian spectrum auctions have translated into “mega-events” to “displaying spectrum” than “selling”. This has resulted in a huge quantum of spectrum lying unused with the government which otherwise could have been put to good use. If the regulator has no issue with such wastage when why emphasize mandatory rollout criteria in all spectrum sold through auctions? It is analogous to putting a high-end security apparatus at the gate of a house that has no boundary wall. It is high time now that we identify and correct these contradictions/mistakes so as to make the next auction of 5G successful, thereby motivating the operators to take a sufficient spectrum for deploying meaningful 5G services in the interest of the nation and its consumers.
(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.
