Expert opinion on the Indian spectrum auction

Regulatory insight: If supply increases, prices will (quickly) stop going up


With about 2355MHz of downlink spectrum being available across 22 circles that make up the country's telecoms map, this year's Indian auction was branded as a super auction — indeed, initial forecasts predicted auction proceeds in the range of INR1.5 - 5 trillion (the equivalent of up to EUR67 billion).

However, after just five days and 31 rounds of bidding, the auction ended with total prices summing to 'only' INR658 billion (about EUR8.8 billion). Importantly, none of the spectrum available in the 700MHz and 900MHz bands, and only very limited amounts in the 2100MHz band sold, as shown in the picture below.

Figure 1:
Comparison of spectrum available and spectrum sold (downlink MHz) and spectrum sold [Source: Aetha India Auction Analysis]
Spectrum supply versus spectrum sold

At first glance, this is a surprising outcome, as these frequencies are some of the most coveted spectrum bands available for mobile operations:

  • The 700MHz and 900MHz bands are 'low-frequency' bands, offering superior propagation to achieve strong coverage deep indoors and in rural areas.
  • The 2100MHz band is a key capacity band for 3G as well as 4G — adding this band to the portfolio would provide operators with significant extra capacity as well as the flexibility to react to market requirements moving from 3G to 4G.
  • The 700MHz and 2100MHz bands were available in blocks of (2×)5MHz, a size suited to providing data services (as 5MHz is the building block of 3G and 4G carriers), whereas some other bands were only available in smaller chunks

This article focuses on what we believe to be the two key reasons that explain this outcome. We also review the implications for regulators setting spectrum prices ahead of award processes.

Recent market transactions have eased operators' spectrum shortage and reduced willingness to pay

Over the last 12 months, changes in the regulatory framework in India have enabled a lot of market activity, as operators have been looking to strengthen their spectrum portfolios. For example, RCom and Aircel announced their merger in September after RCcom had already concluded a merger with Sistema earlier in the year. Adding to this is the fact that RCom and RJio are also 'virtually merged', according to the RJio chairman. Further, market leader Airtel acquired 2300MHz spectrum in July from Aircel (such that Aircel would avoid hitting spectrum caps when merging with RCom).

All these transactions will have had a direct impact on the competitiveness of the auction. With operators strengthening their spectrum portfolios over the course of the year, their immediate need for additional frequencies was greatly reduced. Their larger starting holdings before the auction, in combination with the availability of cheaper substitute bands during the auction (see here for a discussion of the emergence of higher-frequency bands as reasonable substitutes for data capacity) will have greatly reduced the demand for additional spectrum in bands below 2300MHz.

If willingness to pay goes down and spectrum prices stay unchanged, market forces will lead to spectrum remaining unsold

When setting starting prices for spectrum blocks, Indian authorities base their decision on a number of complex approaches. However, one key principle underlying the reserve price setting appears to be that starting prices for spectrum in a band should not be set below the final prices fetched in previous auctions.

This principle seems to go counter to the fundamental logic of basic economics. Of course, the move towards data services and the growth in data consumption increases the need for additional MHz of (downlink) spectrum. However, the increase in demand has to be considered relative to the proportional increase in supply. If supply increases significantly more than demand, the market-clearing price should be (significantly) below previous auctions.

In the current auction, new spectrum meant that the overall supply increased by approximately 70%. In combination with the market transactions described above, this theoretically provided the operators with the opportunity to increase their mobile data spectrum by a factor of 5 — 10 over the course of a single year. Despite the phenomenal growth rates in data consumption that are often quoted, it is highly unlikely demand growth matched such a large increase in supply.

Further to this, we must consider the financial implications for Indian mobile operators. Given the commitments from spectrum purchases in recent auctions, some operators are facing annual payments for past spectrum purchases of over 15% of revenues. If prices remained unchanged and spectrum holdings continue to increase, this would imply a proportional further increase in annual spectrum payments. With competition already putting pressure on EBITDA, it is clear that these numbers hardly add up. Added to this is the backdrop of spectrum prices that are already (much) higher in India than in other markets. This is despite ARPUs in India being a fraction of those in Western markets.

Figure 2:
Example comparison of 700MHz prices in India and Western markets (in EUR/MHz/pop) [Source: Aetha Spectrum Fee Database]
700MHz prices worldwide

It follows that, without fundamental changes to price levels, some of the low-frequency spectrum in India, especially the 700MHz spectrum, may remain unsold for a number of years. The main downstream impact of this reduced spectrum availability will likely be on consumers being offered a reduced quality of service. As much as lower prices for spectrum may mean less revenue for the government, the overall socio-economic benefits of high-quality mobile (data) services may be significantly more valuable to the economy.

For other regulators, this may be a cautionary tale as a similar approach to setting reserve prices can be observed in other countries. Recently, the Thai regulator announced that it believes that in order to ensure 'fair competition', prices in future 1800MHz awards should be based on those fetched in the extremely competitive 2015 auctions. Given the events in Thailand over the last year (where a winning bidder of the 900MHz auction failed to pay for its licence), this may be a dangerous approach to take. A very different approach to spectrum pricing is taken in some Scandinavian countries (as also apparent from the above benchmark) — where overall spectrum costs are much lower, but data consumption and network quality are much higher...

Posted: 14 October 2016

More on the Indian auction

All information provided in this article is based on Aetha's detailed analysis of the Indian spectrum auction (Aetha Indian Auction Analysis) and our extensive in-house database of spectrum fees. If you are interested in purchasing the underlying spreadsheets and supporting information or would like to discuss any other issues with respect to the management or valuation of spectrum, please do get in touch with us.

Contact for more information

Marc Eschenburg

Partner

Contact Marc

Marc Eschenburg

Senior Manager

Contact Matt