Thursday, July 2, 2009

A Rational Spectrum Allocation Policy

Giving more spectrum to fewer players maximises net benefits

Shyam Ponappa / July 2, 2009

All nations are equally endowed with radio waves, or ‘spectrum’. But some do better than others at using this resource. Currently, by trying to optimise spectrum use measured by communications-traffic-per-unit-of-spectrum, our benefits are suboptimal. India’s misplaced emphasis on this kind of spectrum efficiency results in economic inefficiency. Some basic issues concerning spectrum allocation must be addressed before misdirected considerations of government revenues or fiscal deficits lead to self-inflicted damage from ill-advised auctions. India needs a more rational spectrum allocation policy for better communications, including broadband at reasonable prices. We can learn from the experience of other countries.

Inappropriate goals
Our predicament results from pursuing inappropriate objectives. Imagine building roads to minimise land use without taking into account the objective of carrying capacity or transit time for optimising throughput. While less land may be used, the roads would be narrow, cluttered with bottlenecks, and afflicted with low throughput. What if we sought to maximise net benefits instead, ie, carrying capacity balanced with costs, with well-designed, multi-lane roads, even if more land were required. If our goal were net benefits (however defined, with whatever social and economic attributes), we could maximise these, instead of some other measure that reduces net benefits.
With radio waves, our goal is apparently ‘efficient’ spectrum use measured by the technical yardstick of traffic-per-unit-of-spectrum, and not net benefits. This is analogous to the example of constricted roads, even without the problems of land availability. Because of this, our spectrum usage is not driven by net benefits. So, India has the most ‘efficient’ use of spectrum as measured by traffic-per-unit-of-spectrum, according to a report by consultants to the Spectrum Allocation Committee (SAC), which made its recommendations in May this year [see Figure 1].
Figure 1: Spectral Efficiency

However, the consultants point out that India is burdened with additional costs, as we need both more equipment and better technology. Further, by restricting spectral carrying capacity, we get less overall utility.
What if we set the right goals?
By setting different goals, eg, efficient capacity resulting in net benefits, we could capitalise on economies of scale and increase net benefits. This implies less fragmented spectrum, with a lower number of operators as in other countries. Less fragmented spectrum also allows a lower number of cells in a given area with more capacity, ie, lower costs. These are compelling reasons for more generous spectrum allocation.
How things are
In India, a communications company gets a small sliver of spectrum depending on its technology, ie, GSM operators get 8.8 MHz, while CDMA operators get 5 MHz. Once these companies get subscribers, the government provides another sliver of spectrum. Further, the amount of spectrum provided for a given level of subscribers has been reduced from one-half to one-fourth between 2006 and 2008. That’s like reducing NH1 to a single lane each way.
India has more operators per circle than anywhere else, higher network costs, and services nowhere near as good. Consequently, operators have to use more equipment inefficiently because of limited spectrum. Second, they need more advanced (ie, expensive) technology to make the most of each unit of spectrum. Third, service quality drops as the density of subscribers saturates the capacity of available spectrum; else, they must invest more to prevent deterioration. How does this compare with the rest of the world?
Other countries have three to five operators (China, Hong Kong, Indonesia, Malaysia, the Netherlands, Pakistan, Singapore, South Korea and the UK), while India has 11 to 12 operators per circle with more expected, going up to 14-16. Spectrum auctions imply more operators and even more intense, inappropriate competition.

How things could be

India’s policy of a large number of operators leads to fragmented spectrum assignment, with lower traffic capacity for a given amount of spectrum (like narrow roads in land use). Limiting the number of operators was found beneficial by Vodafone for the UK as cited by the consultants, as well as by a 2008 study in India.**

The consultants estimate that India’s spectrum usage is eight times more intense than in the UK, Hong Kong or Singapore. Operators in these countries have from 16 MHz to 26 MHz, the latter for the Netherlands, UK, Singapore and Malaysia; Pakistan has 14 MHz. The average spectrum per operator in these countries is nearly 22 MHz with far better services, against India’s average of 5.5 MHz. Advanced technology in India implies significant costs for operators as well as for users, because of more advanced handsets. The consultants suggest that advanced technology be deployed only when it is cost-effective.

Spectrum Allocation Committee Report, May 2009

The SAC has some constructive recommendations, eg, an expert group to rationalise the use of the entire spectrum, a uniform spectrum fee, and fewer operators to avoid fragmentation as in other countries.
Some recommendations, however, are seriously flawed from a net benefits perspective, ie, faster, more ubiquitous services at reasonable costs. One of these is a preoccupation with pricing and spectrum auctions, whose goal — collecting revenues — is in conflict with the goal of net benefits. Another is delinking service licences from spectrum allocation, whereas access to spectrum may be the most effective way to deliver services. Unfortunately, the committee’s charter did not stipulate net benefits and broadband as goals. Therefore, the committee did not address either.
The government needs to act on the SAC’s positive recommendations, while reviewing and addressing the basic purpose of communications, which is presumably for net benefits (‘welfare’) in the public interest.
* ‘An assessment of spectrum management policy in India’, Plum Consulting, December 2008: David Lewin, Val Jervis, Chris Davis, Ken Pearson.
** Figure 4.1 from the Plum Consulting report above, and Table 5, Optimal Number of Operators in 2010, from ‘Optimal Number of Mobile Service Providers in India: Trade-Off between Efficiency and Competition’, Rohit Prasad & Vardharajan Sridhar, International Journal of Business Data Communications and Networking, Volume 4, Issue 3, 2008:

Letters: Spectral transparency
Business Standard / New Delhi July 08, 2009

This refers to Shyam Ponappa’s ‘A rational spectrum allocation policy’ , July 2. The author finds fault with the globally accepted method of measuring spectral efficiency in terms of traffic per unit of spectrum and optimising the use of this scarce resource. A high spectral efficiency does not necessarily mean that the network is congested. Unlike roads, telecom networks are dimensioned to provide a specified ‘Grade of Service’ (GoS). GoS specifies the level of congestion on the air interface. Typically, mobile networks are dimensioned for a GoS of two per cent. The TRAI monitors GoS under the Quality of Service (QoS) regulation. The operator is required to provide enough network resources including spectrum to meet the GoS norms. Therefore, the analogy of congested roads is not quite correct.

Another contention of the author, that the number of operators is ‘limited in the UK’, is not correct either. Although the UK had a duopoly regime in the 1980s, after extensive consultations, the policy was reviewed in 1991 and a policy of open competition was adopted — this means the number of operators over there is only limited by market forces.

Also, for efficient utilisation of spectrum which is a scarce national resource, there is no alternative to a transparent bidding process or auction as recommended by the Spectrum Allocation Committee (SAC).
R R N Prasad
Former Member, TRAI / Telecom Commission

Letters: Spectrum Allocation
Business Standard / New Delhi July 14, 2009

RRN Prasad’s letter to the editor on July 8, on Spectral Transparency, makes various assertions without supporting data.

Prasad asserts: ‘The author finds fault with the globally-accepted method of measuring spectral efficiency’.

a) My objection is to the criterion of efficiency for spectrum allocation, instead of a measure of public welfare/utility/net benefits. We need reasonably-priced services, not efficient spectral use.

b) No other country uses spectral efficiency as the criterion for allocating spectrum.

Prasad asserts: ‘A high spectral efficiency does not necessarily mean that the network is congested... The TRAI monitors GoS under the Quality of Service (QoS) regulation…Therefore, the analogy of congested roads is not quite correct.’

The fact is that there is spectral congestion. The TRAI reports high levels of congestion for January-March, 2009, as in previous periods going back to 2006 at least. Details at:

Prasad asserts: ‘Another contention of the author, that the number of operators is ‘limited in the UK’, is not correct either… the number of operators over there is only limited by market forces.’

There are four 2G mobile operators in the UK, as stated in my article.

Prasad ends with a non sequitur: ‘…for efficient utilisation of spectrum which is a scarce national resource, there is no alternative to a transparent bidding process or auction as recommended by the Spectrum Allocation Committee…’

The question is, ‘efficient’ use for what purpose? What is the objective? If it is public welfare, technical efficiency in spectrum use is not getting us there.

The initial efforts at telecommunications privatisation began in 1994 in India with a disastrous bidding process. In 2000, the New Telecommunications Policy ’99 replaced auction fees with a revenue-sharing arrangement; only after that did the sector take off. Should our spectrum allocation build on international practice, plus:

a) Take into account our successful telecommunications development experience after NTP ’99, or

b) The disastrous experience of auctions?

Auctions abroad have been generally disastrous in Europe and in America. Besides, in comparing our situation with Europe or America, remember that these are completely different environments, at very different stages of economic development. The default situation there is that most things work; here, it is that few things work.

An Arthur D Little report for the GSM Association begins the section on India as follows:*

‘India presents an extreme example of detailed spectrum management or micro-management by a regulator.’

It goes on to say: ‘The current Indian approach to allocating and attributing spectrum is fraught with risks to the demand-driven development of mobile broadband services, which will likely be delayed and frustrated unless the underlying policies and the processes for resolving the kinds of disputes it provokes are substantially revised.’

* ‘Mobile Broadband, Competition and Spectrum Caps’, Dr Martyn F Roetter, Arthur D Little, January 2009.

Shyam Ponappa, via email