Spectrum AllocationMechanism for 3G Mobile Services
This article suggests a revision of the Telecommunications Regulatory Authority of India’s bidding process that allocates spectrum space to mobile phone service providers for the
provision of 3G services.
ROHIT PRASAD, V SRIDHAR
Q
In tune with the above, the Telecommunications Regulatory Authority of India (TRAI) gave its recommendations for the “third generation” (3G) mobile services in September 2006. With the objective of creating a transparent market-based mechanism for allocating the scarce spectrum resource for 3G services, TRAI recommended a simultaneous ascending auction (SAA) for all 23 circles in the country with a reserve price summing up to Rs 1,100 crore. Economists since Coase have recommended that scarce resources like spectrum must be allocated to private owners through a transparent bidding process to ensure their efficient use. However, there are instances such as the 3G auction in Europe in 2000, where a whopping $100 billion transfer to the public purse left the wireless industry limping with bankruptcy cases. In this article, we critically examine the issues in the TRAI recommendation, which are crucial for preventing the “winner’s curse” and promoting a healthy, competitive environment for sustaining the growth of this exponentially growing industry.
Reserve Price
TRAI has recommended that the 2.1 GHz spectrum be auctioned in five blocks, each with a capacity of 2 × 5 MHz. To calculate the reserve price, TRAI has used the discounted value of the simple average of the international sale price per Hz of spectrum allocated for 3G services in other countries, post-2001. Table 1 shows the international spectrum prices in the 16 countries considered in the recommendations and the year of sale. Two issues are immediately obvious from the data. Given the spread of the years in which the auctions were held in the benchmark countries, there is a need to factor in the inflation rates in those countries in order to draw a meaningful conclusion about the sale price in the current year. Secondly, countries whose sale prices are outliers must be removed from the list of benchmark countries, especially if their markets are structurally dissimilar to India’s.
The figure, depicting the spread of spectrum prices in the 16 countries, shows that Taiwan and Luxembourg are clearly outliers and hence, the inclusion of their corresponding sale price in the benchmark is questionable. The TRAI recommendation does not conform to either of these basic requirements for a meaningful benchmark.
There is a third issue, which is not immediately obvious from the data so far presented. The objective of using the sale price in other countries is to equalise the cost borne by Indian operators with that borne by their international counterparts. However, the cost borne by an operator in a benchmark country must be assessed in terms of the local purchasing power foregone in order to obtain spectrum rights. Merely converting the local sale price using the nominal exchange rate does not adequately represent the cost borne by an international operator. Similarly, after converting the international benchmark into its purchasing power parity (PPP) value in terms of international dollars, the conversion into rupees must also take the PPP of the rupee into account and not the nominal exchange rate.
Table 1: International Spectrum Prices, Post-2001
| Country | Year | Price in USD |
|---|---|---|
| Ireland | 2002 | 3.41 |
| Luxembourg | 2002 | 0.01 |
| Malaysia | 2002 | 0.39 |
| Slovakia | 2002 | 1.10 |
| Taiwan (China) | 2002 | 9.00 |
| Estonia | 2003 | 0.16 |
| Norway | 2003 | 0.40 |
| Bulgaria | 2004 | 0.94 |
| Croatia | 2004 | 0.53 |
| Hungary | 2004 | 2.82 |
| Denmark | 2005 | 2.67 |
| Latvia | 2005 | 0.03 |
| Lithuania | 2006 | 0.03 |
| Morocco | 2006 | 1.37 |
| Romania | 2006 | 0.88 |
| Sri Lanka | 2006 | 0.17 |
Source: TRAI (2006).
Economic and Political Weekly June 9, 2007
Table 2 gives the compounded annual growth rate (CAGR) of inflation (column (3)) for each of the countries using inflation data from the Federal Reserve Bank of Cleveland. The PPP conversion rate of the local currency and the nominal exchange rate using data from the World Bank are given in column (4). The adjusted price in international dollars in each of the countries is given in column (5).
As the table shows, if we adjust the sale price in the other countries to reflect their inflation rates in the period since the sale of spectrum and the current PPP of their currencies with respect to the dollar, the per unit spectrum price actually rises to $3.91 from the calculated level of $1.5. However, if we then factor the PPP of the Indian rupee when converting into rupees, the final price is Rs 36.25 per Hz.1 Hence, the price of a 2 × 5 MHz spectrum block for a pan-India allocation consisting of 23 circles, amounts to about Rs 833 crore even if we were to ignore the problem of outliers. If we remove the outliers from the calculation of the benchmark average, then the per unit price of spectrum amounts to $1.73 which is equivalent to Rs 16 per unit of spectrum in terms of PPP adjusted Indian rupees, therefore translating into a pan-India price of approximately Rs 370 crore.
Next, to arrive at the reserve price from the international price, TRAI has used a discount factor2 of about 30 per cent on their international benchmark of Rs 1,500 crore to arrive at the reserve price of Rs 1,100 crore. Applying the
10 9 8 7 6 5 4 3 2 1 0
Price per Hz in US$
same discount factor on the inflation and PPP adjusted international sale price of Rs 833 crore, we arrive at a reserve price of Rs 580 crore including the outliers and Rs 260 crore without the outliers, both of which are much lower than the TRAI recommendation.
Spectrum Allocation Mechanism
Apart from the reserve price, the most important element of auction design is the set of rules governing the number and progress of the bidding rounds. In the case of the 3G auction design, the rules for the
Figure: Spread of Spectrum Prices in Different Countries
| Taiwan (China) Ireland Hungary Denmark Latvia CroatiaBulgaria NorwayEstoniaLuxembourg Malaysia Slovakia | Morocco Romania Sri LankaLithuania |
|---|
Countries
Table 2: Purchasing Power Parity and Inflation Adjusted Spectrum Prices
| Country | Price US$ | Inflation CAGR | Nominal Exchange | Adjusted Price |
|---|---|---|---|---|
| (US$ per Unit of Local | International Dollars | |||
| Currency) to PPP Ratio | {((2) X (3)) / Col 4 } | |||
| (1) | (2) | (3) | (4) | (5) |
| Ireland | 3.41 | 1.12 | 1.16 | 3.30 |
| Luxembourg | 0.01 | 1.10 | 0.99 | 0.01 |
| Malaysia | 0.39 | 1.09 | 0.47 | 0.90 |
| Slovakia | 1.10 | 1.25 | 0.50 | 2.73 |
| Taiwan(China) | 9.00 | 1.09 | 0.26 | 38.42 |
| Estonia | 0.16 | 1.12 | 0.60 | 0.30 |
| Norway | 0.40 | 1.04 | 1.53 | 0.27 |
| Bulgaria | 0.94 | 1.14 | 0.39 | 2.73 |
| Croatia | 0.53 | 1.07 | 0.64 | 0.89 |
| Hungary | 2.82 | 1.06 | 0.60 | 5.00 |
| Denmark | 2.67 | 1.02 | 1.39 | 1.95 |
| Latvia | 0.03 | 1.07 | 0.50 | 0.06 |
| Lithuania* | 0.03 | 1.00 | 0.52 | 0.06 |
| Morocco* | 1.37 | 1.00 | 0.39 | 3.53 |
| Romania* | 0.88 | 1.00 | 0.5 | 1.78 |
| Sri Lanka* | 0.17 | 1.00 | 0.26 | 0.65 |
| India | 0.21 | |||
| Simple average | 1.49 | 3.91 |
* All auctioned in 2006.
bidding rounds do not allow enough margin for price discovery over multiple rounds in the spirit of the dictum “the market knows best”.
Consider the working of the SAA in a seven bidder circle. As per the mechanism, each bidder submits a bid in round one. The lowest bidder is put on a waitlist for the SSA and the top six move to round two. At the end of round two, the top five bidders are allotted the spectrum and the lowest is put on the waitlist. This is tantamount to a two-round auction without re-bidding. In such a scenario, the absence of gradual value discovery through observation of competitors’ bidding behaviour exacerbates the risk of the winner’s curse. In case of circles where there are only six participants, the problem takes on an even more intractable hue.
The practical demonstration of the issues that could arise with such an auction design was seen in the least cost subsidy auction to disburse the Universal Service Obligation Fund for the roll-out of mobile services in rural areas of the country. For most of the rural clusters, operators that bid for a positive subsidy in the first round (without sensing the tenor of the market) found themselves high and dry as the top four bidders had bid for zero subsidy! The ability to sense the market and re-bid in round two would have given these companies an opportunity to re-evaluate their bids in light of the others’ behaviour in round one.
Economic and Political Weekly June 9, 2007
We recommend that in all circles with more than five bidders, there should be at least four to five rounds of bidding in which at the end of each round the reserve price for the next round is set equal to the second lowest bid of the previous round, the highest bid is announced, and all bidders (including the lowest bidder of the previous round) are allowed to revise their bids. The allocation mechanism in the final round can be identical to the one proposed by TRAI. To sum up, the auction mechanism should fix a reserve price adding up to Rs 250-300 crore for pan-India presence with the bidding mechanism described above.
The above modifications in the reserve price and the allocation mechanisms are essential for the health of the industry and economic development of the country. The role of telecommunications in bridging the economic divide justifies treating it as a merit good. Every attempt must be made to give a fair chance to the operators to successfully launch the next generation mobile services.
EPWEmail: rohit@mdi.ac.in sridhar@mdi.ac.in
Notes
1 (Simple average × nominal exchange rate at Rs 45 per US$) × nominal exchange rate to PPP ratio.
2 The requisite discount rate is directly proportional to the degree of competition in the market. For highly competitive markets, a high discount rate is possible as the true value is bound to be realised over multiple bidding rounds. However, for limited competition, a low discount rate is advisable as it guards against collusive appropriation of the object at a price below its true value. In the 3G auction, the number of spectrum blocks sold to each circle is five, while the number of bidders in most of the circles is expected to be seven assuming all the current (GSM and CDMA) operators bid. Therefore, we are in a situation of limited competition and the discount rate is appropriately small.
References
Telecommunications Regulatory Authority of India (TRAI) (2006): Recommendations on Allocation and Pricing of Spectrum for 3G and Broadband Wireless Access Services. Available at http://www.trai.gov.in/
World Bank (2007): World Bank Data Bank on country GDP on nominal exchange and PPP basis. Available at http://www.worldbank.org/
Federal Reserve Bank (2007): Federal Reserve Bank of Cleveland data on inflation series of different countries. Available at http:// www.clevelandfed.org/
Economic and Political Weekly June 9, 2007
