COMMENTARY

2008 Telecom Licensing Policy: Conceptual Issues

The Department of Telecommunications 2008 licensing policy increased the number of operators per circle, and accelerated subscriber growth and teledensity. It also raised interesting conceptual questions about the optimal number of licences that should be issued, the advantages of auctioning licences and effects on service quality and penetration, as well as the appropriate licensing regime for 2G and higher generation spectrum.

ndias telecom industry landscape witnessed a game-changing event when new licences were offered by the Department of Telecommunications (DOT) at a price equal to that discovered in the auction for the fourth cellular licence in 2001. This took the total number of players in every circle from six or seven to 11-13. This article looks at the impact of the 2008 policy and highlights some important conceptual questions that need to be considered. The aim is not to provide detailed answers but to present a framework for discussion. The departure in the approach of the New Telecom Policy (NTP) 2011 is also briey outlined. (Controversies related to alleged criminal conspiracies are consciously not addressed.)

base increased from 99 million in nancial year 2005 to 846 million in FY10, making it the second largest wireless market in the world, next only to China (Figure 1, p 24).2 Increased affordability (due to continuous decline in tariffs, handset prices and reduction in subscription costs) and greater availability (rapid expansion in coverage and the wider distribution network) of mobile services fuelled this rapid growth. There have been several reasons why the mobile subscriber base has witnessed buoyant growth. These include the reduction in the lifetime subscription cost, GSM launch by Reliance Communications, Tata Docomos per second billing launch and the new entrants rolling out.3 Declining Average Revenue Per User: However, despite the phenomenal growth in the mobile subscriber base, the average revenue per user (ARPU) for both GSM as well as CDMA4 service providers has taken a beating. Due to intensifying competition, service providers have been forced to slash tariffs to the extent that Indian subscribers now have the cheapest call rates in the world. With 11-13 operational service providers in almost all of the 22 telecom circles in the country, cut-throat competition prevails in the Indian telecom industry. Also, with a rise in multiple SIM card owners, talk-time usage gets split between various SIM cards, thereby reducing ARPU.5 Of course, part of the subscriber growth may have been spurious, with operators showing greater growth to qualify for the next tranche of spectrum. In order to identify fake subscribers, the Telecom Regulatory Authority of India (TRAI) has since April 2011 switched to the practice of counting only those subscribers who show up in the Visiting Location Register (VLR). The earlier practice of counting anyone who was listed in the Home Location Register (HLR) appears to have overstated subscribers by 20-25%. To the extent that spurious subscribers increased after the 2008 licensing, the actual rates of growth may be lower and ARPUs higher. The 2008 policy got mired in controversy related to irregularities in the process used. Eighty-ve new licence holders were issued cancellation notices by the DoT for not rolling out networks as per their licence agreements. Many of the new

Key Effects of the New Licence Issuances

Average number of players per circle Subscribers (in millions) HHI Teledensity Rural share in total (%)

7 261.1 1,590 25.0 24.0

11 846.0 1,394 70.9 33.4

Source : Telecom Regulatory Authority of India (TRAI) Performance Indicators Report.1

Acquisitions: Many new entrants were the beneciaries of foreign direct investment (FDI) at valuations that were many times in excess of the subsidised fees paid for the licence. For example, Norwegian telecom major Telenor acquired 67.5% of Unitech Wireless for Rs 6,120 crore, implying a valuation of Rs 9,067 crore. The acquisition was made soon after Unitech acquired the licence, implying a 5.5 times appreciation. Other acquisitions exhibited similar growth multiples. Subscriber Growth: The telecom services sector witnessed sustenance of the hypergrowth since 2005. The mobile subscriber
vol xlvI no 53

Rohit Prasad ([email protected]) is with the Management Development Institute, Gurgaon.

december 31, 2011

23

COMMENTARY

In the case of mobile services, economies of scale 1,000 80 846 68 are driven by trunking efGrowth (RHS) 800 60 ciency. As the spectrum 58 50 58449 57 600 45 Subscriber numbers (LHS) 40 deployed increases, the 392 400 capacity of a network to 261 20 166 200 carry trafc increases in 99 57 0 0 greater proportion com2004-05 2005-06 2006-07 2007-08 2008-09 2009-10 2010-11 Source : TRAI Performance Indicators Report. pared to the proportion of entrants are facing nancial difculties. increase in spectrum. In addition, there are Airtel led the way to gradual recovery of operational efciencies related to general, price levels with a 20% rise in per second administration and marketing costs. call rates in July 2011 (see Table 2).6 Other The presence of economies of scale companies were quick to follow suit. poses a ticklish question for the votaries of competition. Allowing a few rms to domConceptual Issues inate a market would lead to greater efThe 2008 licensing raises some interest- ciency in production, but at the risk of ing conceptual questions, in particular the increased mark-ups over marginal cost on three enumerated below: account of market power. There is an (1) Is there an optimal number of licences upper bound on the number of operators that the licensor should issue? who can be let in. It is given by the mini(2) Should spectrum be auctioned? Does a mum amount of spectrum necessary to low price of spectrum lead to a low price operate and the total spectrum availabifor service and greater penetration? lity. Should the licensor allow in as many (3) Can the regime for 2G spectrum be operators as feasible and allow the forces delinked from the regime for higher of competition to decide who and how generation spectrums? many survive (subject to a oor on the number of operators)? How should the Optimal Number of Operators: The com- rms remaining after consolidation be monly held view is that a competitive market regulated in order to prevent the exercise structure is the most effective for ensur- of market power? ing low prices and high quality. However, Baumol (1982) introduced the notion of in industries such as telecom services and contestability as a more fundamental conelectricity distribution, economies of scale ception of industry competitiveness comand scope are large enough to warrant low pared to market share. An industry is said levels of competition, even monopolies, to to be perfectly contestable if three condiminimise unit costs. Telecommunication tions are satised: new rms face no discarriers face huge initial costs, including advantage relative to existing rms; there for example, laying down copper lines are zero sunk costs; and third, the entry from the central ofce to each subscriber lo- lag is less than the price adjustment lag. cation in the case of basic xed line services, Table 2: Airtel Valuation Ratios constructing cell sites and base transceiver Parameters (Key Valuation Ratios) stations (BTSs) in case of mobile services, Enterprise value/revenues Enterprise value/earnings before interest, taxes, and laying optic bre cables to connect depreciation, and amortisation (EBITDA) access networks to backbone networks. Market cap/sales In contrast, once the network is opera- Enterprise value/subscriber (Rs) tional, the marginal cost of providing service Source: Airtel Financial Statements. to each additional customer is often negligi- The mobile industry is not perfectly conble. Given the enormous xed costs and neg- testable, but there exist several avenues ligible marginal costs, the carriers long-run for increasing contestability. These inaverage costs within the dened geographi- clude mobile number portability, mobile cal area may well decline with each increase virtual network operators, spectrum and in the size of the network. In other words, it tower sharing. So at what level should the is often cheaper for an operator to provide regulator focus on increasing contestabilservices to the one-millionth customer than ity without increasing the number of to the one-thousandth customer.7 participants?
Figure 1: Subscriber Base (in million) (%)
74

Similarly, while competition can unleash innovation, beyond a certain level, it can degenerate into price war and kill business model, technical, and operational innovation. What is the number of operators beyond which competition turns counterproductive in this way? Finally, the option of unbridled competition leading to an optimal market structure is predicated on the presence of a facilitative mergers and acquisitions (M&A) environment (Prasad and Sridhar 2008b). What should be the features of such an environment? Impact of Spectrum Cost on Price of Service: In our discussion of spectrum cost, we will distinguish between two types of cost: upfront payment (bundled with the licence fee in India) and spectrum usage charge, charged as a percentage of revenue, increasing in amount of spectrum held. Economic theory distinguishes between sunk costs, xed costs, and variable costs. A sunk cost is a cost which cannot be recovered after it is incurred. A xed cost cannot be changed in the short run but can allow redeployment in the face of emerging business developments. Variable costs can be changed in the short run and are incurred in proportion to the level of production. Sunk costs do not affect the price level in the market since they have already been incurred. Based on this principle, economists have argued that the cost of spectrum has no impact on the price or quality of services. They have argued that operators should be charged the full value of the spectrum and that an auction is the best mechanism of revealing that value. Is there merit in their view?
2006-07 2007-08 2008-09 2009-10 2010-11

8.1 20.1 7.9 40,154

6.1 14.8 5.8 26,629

3.5 9.4 3.2 13,820

3.0 7.1 2.8 9,909

3.2 9.4 2.3 12,034

To gain some perspective, let us start with an episode from a different industry. There are many similarities between pharmaceuticals and mobile telecom services in that both industries involve high xed costs and negligible marginal costs. Many years ago, drug companies in the United States (US) protested against regulation that required drugs to be tested for their suitability
vol xlvI no 53
EPW Economic & Political Weekly

24

december 31, 2011

COMMENTARY

for children with the argument that the regulation would increase the pre-launch costs and thereby the price of drugs. Many articles in the popular press purported to point out the fallacy of their position: since research and development (R&D) on already discovered drugs was a sunk cost, in other words, a cost that could not be transferred, it would not affect pricing decisions. While correct, this did not tell the whole story. For while R&D costs already incurred were a sunk cost, future R&D was not. It was a cost whose level would be determined by the rm on the basis of a calculation of prot potential. An increase in the R&D cost structure for rms might inhibit future R&D, leading to less competition in the market for already developed drugs and higher prices (or lower price reductions) than would have prevailed in the absence of the proposed regulation. In the telecom industry as well, the view that spectrum cost does not affect price of services needs to be tempered. The wrinkle is caused by the fact that

also a possible decline due to reduction in costs caused by consolidation of spectrum, with the spectrum vacated by the failed rm going to some incumbent. How these two opposing forces play out will depend on the specic features of the situation. On the other hand, under certain conditions, an increase in spectrum cost may not affect prices or quality at all. If we start with very low level of spectrum costs that allow rms to make supernormal prots, then increasing the cost of spectrum will, up to a point, reduce the prot level but not change the market actions of players, including the build-up of physical infrastructure and pricing strategies. In such a situation, the government would be justied in trying to increase the revenue generated from telecom. Thus the issue of the impact of spectrum cost on price and quality cannot be decided without reference to the specic features of the industry: the number of operators, the contestability of the industry, the degree of fragmentation of spectrum, and the nature of demand.

The general principle for efciency is to increase the sunk cost, i e, the upfront cost of spectrum without affecting competition, and to reduce the level of the variable cost to avoid distortions in efciency. Delinking 2G from Higher Generation Technologies: One of the oldest discussions in the regulatory landscape is on whether 3G and other higher generation services are an extension of 2G services or stand-alone offerings. One view is that voice is a basic service and hence 2G spectrum, which caters mainly to voice services, should be treated differently from 3G, 4G and BWA spectrums. This view is similar to the view in the fuel industry that kerosene, diesel and LPG should be treated differently from petrol as they cater to the common citizen. The rst observation is that the impact of spectrum cost of price must be examined in light of the principles outlined in the previous section. There are situations where higher spectrum cost need not affect either price or quality of service. Further, an untargeted subsidy with respect to voice services ends up beneting many who do not need the subsidy. Second, the view that 2G spectrum is used mainly for voice and higher generation spectrum mainly for data is misleading in the current context. Given the paucity of spectrum and of the demand for data, operators are using their 3G spectrum mainly to augment their capacity to provide voice services. Also 2G spectrums are being used to provide data services using enhanced data for global evolution (EDGE) and GPRS technologies.8 During the period when 2G spectrum is used with 2G technology, and 3G with 3G technology, voice and data usage in 2G and 3G spectrums is likely to be of similar orders of magnitude. In the near future, we are likely to see migration of 2G spectrum to higher generation technologies. In fact, with the increasing popularity of smartphones, operators are likely to service each subscriber with a combination of 2G and 3G spectrums. Given all these factors, different regimes for 2G and 3G spectrums may create arbitrage possibilities (for example, with respect to the differing spectrum usage

Figure 2: Key Reasons for Mobile Subscriber Growth (in millions)

20 18 16 14 12 10 8 6 Lifetime prepaid scheme reduced to Rs 199

RComs GSM launch and Vodafone Vodafone and Ideas new circle launches circle gathering steam. steam

Tata DoCoMo's Tata Docomos GSM launch characterised by launch characterisedby per second billing per second billing

Dec-07

Dec-08

Aug-07

Aug-08

Aug-09

Dec-09

Apr-08

Apr-09

Feb-08

Feb-09

Feb-10

Apr-10

Jun-07

Oct-07

Jun-08

Oct-08

Jun-09

Oct-09

Source: TRAI, Cellular Operators Association of India and Unnati Research.

sunk costs are incurred in stages. First spectrum is acquired and then the physical infrastructure is built. It is possible that if the cost of spectrum increases, operators may reduce spending on physical infrastructure (like pharmaceutical companies cut future R&D expenditures), thereby reducing competition. The price and quality of service may be affected as a result. In the extreme case, unexpectedly high spectrum costs can cause exit. This can reduce the number of rms and have two conicting impacts on prices (with quality being negatively correlated with price in each case). It could lead to a possible price rise due to increase in market power, but
Economic & Political Weekly EPW

The second component of the cost of spectrum is the spectrum usage charge. Spectrum usage charges are not sunk costs, but variable costs like sales taxes. Like all such taxes, they have the effect of increasing prices. Further, spectrum usage charges in India increase at higher holdings of spectrum. This makes them more distortionary than the ordinary sales tax, even one which increases with increasing revenue, because they penalise increases in revenue that are accompanied by an increase in the usage of only one of the inputs, i e, spectrum. Hence the production decision is biased in the direction of increased use of physical infrastructure.
vol xlvI no 53

Jun-10

december 31, 2011

25

COMMENTARY

charges), administrative bottlenecks, and conceptual inconsistencies.

Conclusions
The introduction of several new operators in 2008 led to a phase of hyper-competition and predatory pricing. The phase of hypercompetition is drawing to a close with the withering away of some entrants and the gradual increase of call charges. Even today, call charges remain much below levels before the 2008 licensing. This either reects the possibility that the hyper-competition phase has unleashed operational and business model innovations, or that operators were guilty of exercising market power pre2008. From rst principles, one may surmise that the quantity and quality of innovation in the hyper-competition phase was low. Hence the 2008 policy may be credited with reducing market power, especially with regard to the low and middle-tier consumers. For higher tier consumers, operators continued to use their lower price elasticity of demand to charge high rates for roaming. This could perhaps be condoned in the context of hyper-competition. However, the 2008 policy also created a high cost structure due to spectrum fragmentation. If fewer operators were introduced, the costs of physical infrastructure would have been lower, especially in capacity-constrained urban areas, and higher charges for spectrum could have been levied without changing overall capital costs. The level of competition would have been what it is today; prices would have been higher than those observed in 2008-10, but much lower than in the pre-2008 phase. Letting in a large number of operators to the limits of feasibility, and putting in place a facilitative M&A environment to throw up efcient operators may not work in telecommunications. To let in serious operators and prevent speculative activity, the licence has to bind licensees to rollout obligations without the possibility of early exit. Therefore mergers can only take place after multiple rollouts have already been effected. This leads to tremendous wastage of resources. Thus, the licensor has to control the number of companies and not rely on the M&A environment to do so. An alternative to the 2008 policy would have been to let in two or three operators

with experience of Indian markets and the nancial muscle to take on incumbents. The start-up spectrum could have been provided at a price indexed to technological improvements or determined by an auction open to all new applicants, including dual-technology operators. The remainder of the spectrum available for release could have been auctioned with all licence holders, including new ones, eligible to participate. Measures to promote contestability, like the facilitation of spectrum sharing and incentives for mobile virtual network operators, could also have been introduced. Spectrum usage charges could have been indexed to the usage charges set for 3G spectrums to avoid arbitrage possibilities, i e, operators servicing subscribers with 2G spectrum and booking the revenue to their 3G businesses. The licence fee, being a variable cost, could also have been converted into a constant low levy, indexed to administrative costs. Such a regime would have yielded greater revenue for the government, unleashed forces of constructive competition and beneted the consumer in a sustainable manner. The 2011 policy makes important conceptual departures from the 2008 policy. It recognises the presence of economies of scale and allows spectrum sharing. It cedes the role of efcient management of spectrum to market forces through its nod to spectrum trading. It recognises that high spectrum cost is mitigated by lower physical network costs and therefore aims to index spectrum cost to market prices in an environment of adequately sized spectrum blocks. Since operators are expected to make higher prots after consolidation, NTP 2011 abolishes the practice of discriminatory pricing in the form of high roaming charges. All these steps mark a new beginning for policy thinking on the mobile industry. However, important lacunae remain. These include the continuance of a high licence fee and spectrum usage charge regime, which reduces efciency, as argued above. Further a clear road map on M&As appears to be absent. While spectrum sharing and trading are good options for quasi-consolidation, ultimately an enabling environment for M&A is necessary. The government must not focus so much
december 31, 2011

on revenue maximisation that it renders M&A unviable even though legally feasible. Finally the move to all Internet Protocol Next-generation Networks requires the freeing up of spectrum currently in the control of ministries other than communications and information technology. No institutional mechanism has been proposed to avoid the delays observed in the case of 3G spectrum. The conceptual issues related to the 2008 policy strike at the core of several burning issues today: the importance of revenue maximisation, the desirability of a level playing eld and the role of competition in the context of economies of scale. If this article can shed some light in the midst of the sound and fury, it will have served its purpose.
Notes
1 TRAI provides performance indicators on a quarterly basis, on the following web link (last accessed 8 December 2011): http://www.trai.gov.in/reports_list_year.asp In particular, see TRAI (2011): The Indian Telecom Services Performance Indicators: January to March 2011, Delhi. At: http://www.trai.gov.in/ WriteReadData/trai/upload/Reports/55/Indicator_Report-Mar-11.pdf TRAI (2008): The Indian Telecom Services Performance Indicators: January to March 2008, Delhi. At: http://www.trai.gov.in/WriteReadData/trai/upload/Reports/42/reportQE3july08.pdf Compiled from various TRAI Indian Telecom Services Performance Indicators reports, various years. GSM stands for Global System for Mobile Communications. CDMA stands for code division multiple access. SIM stands for subscriber identity module. Retreived from (last accessed 8 December 2011): http://www.airtel.in/wps/wcm/connect/about+ bharti+airtel/Bharti+Airtel/Investor+Relations/ Results/PG_Results?countrytabs=2 For more on economies of scale and scope in telecommunications, the reader is referred to Nuechterlein and Weiser (2005) and Prasad and Sridhar (2008a, 2009, 2011). EDGE stands for Enhanced Data for Global Evolution; GPRS stands for general packet radio service.

2 3 4 5 6

References
Baumol, W (1982): Contestable Markets: An Uprising in the Theory of Industry Structure, American Economic Review, 72(1): 1-15. Nuechterlein, Jand P Weiser (2005): Digital Crossroads (Cambridge, Massachusetts: MIT Press). Prasad, R and V Sridhar (2008a): Optimal Number of Mobile Service Providers in India: Trade off between Efciency and Competition, International Journal of Business Data Communications and Networking, 4(3), 69-81. (2008b): M&A: New Frontier of Policy, Mint, 13 August. (2009): Allocative Efciency of the Mobile IndusEfciency try in India And Its Implications For Spectrum Policy, Telecommunications Policy, 33(9), 521-33. (2011): Toward a New Spectrum Policy Framework for India, Telecommunications Policy, 35(2): 172-84.
vol xlvI no 53
EPW Economic & Political Weekly

26