How to design an auction

The 3G auction of 2010, held in the wake of the 2G scandal of 2008, was a blessing, for it achieved two objectives critical to emerging market economies such as India: it ensured transparency in the process of allocation of scarce spectrum, and also resulted in revenue maximisation. Because of prevailing spectrum constraints in certain service areas and expectations of a data-led boom, the 3G auction produced healthy bids with the government collecting Rs 67,719 crore, albeit with significant inter-service area variations. Delhi and Mumbai by themselves accounted for 40 per cent of the winning bids. The current 2G auction that ended last Wednesday (November 14) seemingly achieved the transparency objective, but fell way short of the government target of Rs 30,000 crore by a massive 69 per cent.

The less than enthusiastic response from service providers in the current auction is because of the high reserve price set by the government. There were no bids in Delhi, Mumbai, Rajasthan and Karnataka circles. Only in Bihar did the winning bid surpass the reserve price, and that too by just over 9 per cent. That the auction ended in just two days compared to the 35-day bidding for 3G spectrum in 2010 is further proof of the high reserve price.

Economic theory asserts that in a well-designed auction, resources get assigned efficiently to the parties that value them the most, obviating the need to ex ante fix the "market" price. That is the job of the auction and not the administrator. It is worth noting that the Telecom Regulatory Authority of India (TRAI) had recommended a reserve price of Rs 18,000 crore for 5 MHz pan-India 2G spectrum. Questions were raised around the logic of fixing such a high reserve price. This was eventually reduced to Rs 14,000 crore, thanks to the intervention of the Empowered Group of Ministers (EGoM). Clearly, the 23 per cent reduction was not enough, although TRAI could argue that they were simply following the Supreme Court's instruction of conducting a "3G-type" auction. The final reserve price recommended by TRAI was, however, over 8 per cent higher than even the 3G price, reflecting mechanical indexation to arrive at the reserve price.

That the high reserve price deterred bidding is obvious, but significantly it also created strong incentives among a subset of bidders to coordinate strategy. Of the five players in the auction, Airtel,Vodafone and Idea are incumbents with renewal of their telecom licences due in a few years, while Videocon and Telenor are new entrants, seeking fresh spectrum without any legacy issues. The latter two were thus bidding to acquire spectrum to commence operations.

Significantly, the reason the incumbents were able to successfully follow a strategy of not bidding is because the reserve price was fixed too high and it discouraged new bidders. It is reasonable to conclude that the high reserve price helped produce a coordinated equilibrium, where not bidding was better than bidding in certain service areas for the incumbent operators. If the reserve price had been fixed lower, not bidding by incumbents would have introduced the risk of losing spectrum to the new entrants.

For the incumbents, there was a lot riding on this auction. The Department of Telecommunications (DoT) had decided that the charge for renewal of expiring licences and the payment of one-time fee for excess spectrum would be based on the price discovered in this auction. The higher the bid, the higher would be the associated payments to be made by the incumbents, translating into several thousand crores. For example, in Delhi and Mumbai, where a large chunk of the excess spectrum resides and where licence renewal is round the corner in 2014, at a reserve price of approximately Rs 550 crore per MHz, a holding of 10 MHz of spectrum translates to a fee of Rs 7,590 crore per operator. It is not surprising that the dominant strategy for the incumbents was to not bid.

There are several lessons one can draw from this experience. Paul Klemperer, an authority on auction design from Nuffield College, has famously stated an auction's design is not "one size fits all". It must be tailored both to its environment and to the designer's objectives. In India's case, the twin objectives should be to ensure transparency and integrity of the process. To focus chiefly on extracting maximum rents from bidders sends the wrong signals and may result in stunting growth. The sluggishness of 3G in some European countries such as the UK and Germany circa 2000 was partly attributable to the high cost of spectrum. In India too, 3G services are in their infancy, although part of the reason is also the underdeveloped ecosystem for data services.

What's even more important is that clubbing multiple policy objectives, such as renewal of licences, payment for excess spectrum and re-farming of spectrum, on the outcome of a single auction, created strong incentives among incumbents to game the system.

Once the dust settles, the government will need to decide the way forward. Its stated objective is to conclude the next auction before the end of this fiscal. One option will be to pool all available spectrum (unsold and additional that is available) and fix a significantly lower reserve price for the auction. A simple way to do this will be to inflate the 2001 2G auction prices to 2012 levels using either the GDP deflator or the nominal interest rate. After all, it is only a reserve price; the true market value will be discovered during the auction. A lower reserve price will guarantee larger participation and reduce gaming possibilities, since coordination among a larger number of players is more difficult. If there is apprehension that too many players will enter in an already crowded market, a suitable framework for mergers should easily address that concern.

The temptation to use telecom as a cash cow should be studiously avoided, since pass through in the form of higher consumer tariffs is inevitable in the medium to long term, eroding the significant gains Indian consumers have enjoyed.

Alternatively, the auction could start with the old reserve price and be reduced in steps until all available spectrum in the circle has been sold or the minimum reserve price discussed above is breached. This is referred to as the Dutch auction, in which the auctioneer begins with a high asking price that is lowered until some participant is willing to accept the auctioneer's price. The winning participant pays the last announced price. Although the Dutch-type auction is attractive in theory, with an inbuilt disincentive to cooperate, the ascending bid, the open auction conducted in India has also been the popular approach for spectrum assignment across the world. In addition, there is the familiarity factor, which tilts the scales in its favour. At the same time, however, we must be willing to reduce the reserve price and let the market discover the price, instead of imposing an unrealistic one from above.

The writer is director and chief executive, ICRIER

Please read our terms of use before posting comments
TERMS OF USE: The views, opinions and comments posted are your, and are not endorsed by this website. You shall be solely responsible for the comment posted here. The website reserves the right to delete, reject, or otherwise remove any views, opinions and comments posted or part thereof. You shall ensure that the comment is not inflammatory, abusive, derogatory, defamatory &/or obscene, or contain pornographic matter and/or does not constitute hate mail, or violate privacy of any person (s) or breach confidentiality or otherwise is illegal, immoral or contrary to public policy. Nor should it contain anything infringing copyright &/or intellectual property rights of any person(s).
comments powered by Disqus