Evaluating spectrum auctions in India

Communication airwaves, also known as radio-frequency spectrum, are an important resource for mobile communication technologies. It is a subset of frequencies in the electromagnetic spectrum that can be used for communication, and corresponds to frequencies from 3 kHz to around 300 GHz. While spectrum is not depletable, technology effects can limit the extent to which it can be beneficially utilized at a point in time. Governments typically license the use of spectrum to regulate its application and maximize social benefit. With the exploding demand for all things wireless, radio spectrum has become a scarce commodity in many countries. In India for instance, aggressive bidding during auctions (2010, 2012, and 2015) led to dramatic increases in spectrum prices. Spectrum-management policies over time have eased pressure on operators and encouraged spectrum efficiency. In 2017, the average spectrum holding for an operator in India was 31 MHz, compared to the global average of 50 MHz.

In August 2018, the Telecom Regulatory Authority of India (TRAI) published its recommendations on the auction of spectrum across several bands, including two bands yet to be auctioned in India, 3300–3400 MHz and 3400–3600 MHz. These bands are likely to emerge as the primary bands for 5G services. The recommendations include a discussion on the availability of spectrum, roll out obligations, spectrum caps, block sizes, valuation, and the reserve price of spectrum. The latter is the chief focus of this study as well.

Spectrum pricing is an invaluable tool to promote efficiency. The International Telecommunication Union (ITU) proposes four different methodologies for spectrum valuation: (i) price from previous auctions duly indexed; (ii) estimation of the value of spectrum by assessing producer surplus; (iii) valuation of spectrum using a production function approach; and (iv) valuation of spectrum using a revenue-surplus approach. In addition to these four methods, TRAI also uses a multivariate regression technique.

According to TRAI, the reserve price should be the higher of the two – 80 percent of the average valuation of the spectrum band using the five methods or the indexed value of the price realized in the October 2016 auction. In service areas where no spectrum was offered in 2016, reserve price should be 80 percent of the average valuation of the four methods (excluding indexation). In service areas, where spectrum was offered in October 2016 but remained entirely unsold, the reserve price should be lower of the two – 80 percent of the average valuation of the five methods or the reserve price as fixed in October 2016. Only in four of the 22 circles, reserve prices are fixed using the 80 percent of the average across five methods. For most circles, reserve price is indexed to the previous auction outcomes.

The report

Broadband India Forum (BIF), an independent policy forum and think tank, in association with Indian Council for Research on International Economic Relations (ICRIER) an autonomous, policy-oriented, not-for-profit, economic policy think tank, have released a report on spectrum pricing and auction methodology that evaluates TRAI’s recommendations for the upcoming auctions and presents alternatives using cross-country experiences.

In this report, it uses the 1800 MHz band as the anchor to evaluate the recommended reserve prices. It focuses on the significant variation in the results of each method, ultimately entailing a judgment on which model or combination is the best fit to the reigning conditions. For example, the multiple-regression method yields significantly higher prices than the other methods for many service areas. In addition, use of mean to average across models when variations are significant puts a disproportionate impact of the outliers on the average value. When the data is skewed, the median is generally considered to be a better measure of the central tendency.

TRAI uses the marginal cost of funds-based lending rate (MCLR) to index values of previous auctions. MCLR or any other lending rate is primarily used to arrive at net present value of projects. While the use of base rates is not uncommon for the determination of spectrum prices based on indexation, it does not reflect changes in price levels either in the economy or in the sector. A better method would be to use a telecom price index, if available, or the consumer price index (CPI). Alternatively, TRAI could consider studying price changes in the telecom sector. Ofcom, UK, extensively researches on pricing trends for communication services using data on consumer preferences, usage levels, contract lengths, promotional prices, etc.

TRAI’s own guidelines of 2010 state that if auction of a specific band is conducted again in less than a year, it would not require a revision in reserve prices. In 2013, TRAI clarified that using reserve prices from a previous auction requires: (i) spectrum to be identical and (ii) auctions to be held very close in time so that the market and macroeconomics are not materially different. The report documents some inconsistency in the applicability of this principle across service areas.

The price of spectrum is a combination of several factors, including prevailing market conditions and the forecast. Not everybody’s forecast can be expected to be identical; expected revenue, incremental costs, regulatory conditions, and risk will be a part of the mix. The report uses revenue-per-circle as a crude proxy for value to operators. In order to test if reserve prices reflect the underlying value, it runs correlations between circle revenue and corresponding reserve price for each spectrum auction over the period 2012 to 2016. It finds a positive correlation of about 50 percent for most years, with the exception of 2015, which is 0.81. For the upcoming auctions it is the lowest, implying least association with the underlying value of spectrum.

An alternate method to measure variations between reserve price and its underlying value is to estimate the coefficients of variation (COV) and compare these. COVs measure the dispersion around the central value of a distribution. The COV for revenues is the lowest implying that the earnings across circles lie within a band while those for reserve prices and market prices are higher.

International experience with auction models reveals a mixed bag. Regulators are empowered to make a judgement in what are sometimes fluctuating market conditions. In several instances of spectrum auctions in US, UK, and Canada, the regulator reserves one or more blocks of spectrum for a new entrant. This approach is effective in increasing competition but it may result in entry by firms with higher costs and less attractive offerings than incumbent. European auctions have shown that spectrum set-asides are inconsistent with efficient allocation of spectrum. The English design sets aside spectrum for a new operator while the German design does not. The German design resulted in more competition and potentially higher revenue. Austria also adopted the German design and achieved higher levels of competition in the market with three new entrants. Other countries, such as Netherlands, Italy, and Switzerland used the English design for auction of their UMTS license and spectrum. Proposed in 2006, the Combinatorial Clock Auction (CCA) was used for ten major auctions during the period 2012 to 2016. These modern spectrum auctions allocate multiple units where bidders may value the units as complements. It overcomes the problem of “reasonable allocation at low prices”, a typical outcome of the simultaneous ascending auctions. While CCA is arguably better than the simultaneous ascending auctions, it is complex and requires a high level of bidder sophistication. It can also result in widely varying prices depending on the strategies adopted by bidders.

New trends are also emerging in the allocation of 5G spectrum. For example, FCC announced an incentive auction in order to free up more spectrum for 5G. Under this format, existing rights holders in those bands can choose either to relinquish their rights in exchange for a share of the auction revenue or alternatively receive modified licenses after the auction, consistent with a new band plan and service rules.

Since 2010, the Department of Telecommunications has consistently used auctions for spectrum allocation. Over the six auctions held during the period 2010 to 2016, the government has auctioned portions of frequencies and the average reserve price in every subsequent auction has witnessed an upward revision. The outcomes have to an extent been discouraging. There is a lack of enthusiasm among operators due to unrealistic expectations on part of the government, which in turn is an outcome of the political economy around private use of a ‘national resource’. As a corrective measure, the regulator has proposed a steep cut in the reserve price of the 700 MHz band, which saw no demand in the previous auctions, and has also recommended making the entire spectrum available for auctions.

Designing auctions is always fraught with risk and given past events in India even more so for spectrum. Experience also suggests that reliance on reserve prices may not always yield successful market outcomes. There are several other factors that influence auction outcomes, such as bidder turnout, market conditions, and choice of the auctioning agent. The auction design is also crucial. India currently follows a simultaneous multi-round ascending auction (SMRA). Combinatorial clock auction (CCA) is a popular alternative as it avoids aggregation risks and is arguably more efficient. A combination of formats can also be explored.

Spectrum auctions in India should try to balance transparency in allocation and revenue expectations for the government. Setting high reserve prices could actually be counterproductive. It could reduce government revenue and stifle sector growth. Building trust between operators and the government is crucial for long-run viability of the sector. This deficit needs to be bridged now.

“Reserve prices are only the starting point for discovering prices that the market will bear. They ought not to become an end in itself. Setting high reserve prices meant that about 60 percent of the spectrum offered in the October 2016 auction remained unsold. Unsold spectrum is an opportunity loss for the government. In 2017, the average spectrum holding for an operator in India was 31 MHz, compared to the global average of 50 MHz. For effectively embracing the 5G ecosystem, spectrum-management policies will need to ease pressure on operators while also improving spectrum efficiency. Spectrum auctions in India should try to balance transparency in allocation and revenue expectations for the government. Setting high reserve prices could actually be counterproductive. It could reduce government revenue and stifle sector growth. Building trust between operators and government is crucial for long-run viability of the sector.”

Dr Rajat Kathuria
Director and Chief Executive,
ICRIER

“We are pleased that ICRIER has come out with this report at the apt time as India stands on the brink of 5G era and TRAI has released its recommendation on auction of 5G and other bands. It is widely recognized that the telecommunications sector is one of the primary bedrocks of an information-based economy. In India, legacy issues have led to inefficient allocation, management, and usage of spectrum. A combination of over-pricing and delays in releasing spectrum has also led to inflated valuations, which have a long-term impact on investments in the sector. This report is meant to offer a practical guideline that is benchmarked to established global norms to calculate the reserve spectrum price. The report has worked toward balancing revenue generation with the infrastructure investment needs and desired consumer benefit to make broadband services available in a ubiquitous manner.”

TV Ramachandran
President,
Broadband India Forum