The Indian Techonomist

New policy for data communications: a proposal

Copyright (C) 1996, Rishab Aiyer Ghosh

Currency note:unless otherwise stated, $ means US Dollars. US$ 1 = Rs 36 approx. (January 1996)

In 1995, the cable television industry in India, by subscription revenue alone, was worth Rs 21 billion, spread between 45,000 networks. The corresponding figures for 1991 are as low as Rs 500 million and 7,000 networks. This rapid, runaway growth was almost entirely due to the unregulated nature of the market.

India, at the threshold of the information age, must follow one of two paths. The one, of continuing, restricting regulation and high entry barriers to data communications businesses leads to China, with a tiny information industry of which limited access to the Internet is but a symptom. The other, of deregulation and low entry barriers, follows the example of Hong Kong, with perhaps the most bustling Internet industry in the region. It is also in keeping with India's traditional freedoms, and will make India as important to the Internet as it has become, in less than five years, to global broadcasting.

This note argues that Indian datacom should follow the path of freedom, a choice that should be obvious for economic reasons alone, but does not seem so to those who formulate our policies. This is largely because there is insufficient understanding of the distinction betweens the natural business models of datacom and telecom.

Telecom is hardware, datacom is software

Telecom has long been thought a "natural" monopoly as it appears, at first glance, to make little sense to duplicate the extensive physical infrastructure it involves. New experiences and new thinking around the world are changing this assumption. None of this has any bearing, however, on datacom.

Datacom uses the same physical infrastructure as telecom: the same cables, the same telephone exchanges, the same satellite and fibre-optic backbones. It uses all this, for it is built above it all, in a layer almost entirely composed of software. Rather than people using lines to transmit speech, machines use them to transmit data; all that differentiates the two is the software that allows the data to be transmitted correctly, to the right recipient.

If telecom is about railway tracks, datacom is about the goods that are shipped over them. Although one might argue that the first is a "natural" monopoly, the second is quite an ordinary business, like manufacturing soap. Therefore, not being a monopoly, datacom requires little regulation; indeed, regulation tends to be counter-productive, working against the competitive market's pull towards consumer satisfaction.

Much cheaper to make than soap

But there is something special about datacom, especially in the form of the Internet. It is an extremely cheap business to get into, with a very high growth rate at the "cottage industry" end. Although it is important to allow the building of datacom infrastructure (i.e. the overlaying of data on telecom infrastructure), the true potential of the Internet and the greatest benefit to consumers (i.e. better service and lower prices) lie with the "Internet Service Providers" (ISPs), the tiny firms that have triumphed the world over.

Obviously, such a situation must be encouraged (or at least not discouraged) in India, by keeping the non-productive costs of entry, such as licence fees, very low. It is also important to ensure an atmosphere of fair competition, which would naturally benefit consumers. What follows in this proposal, then, is an illustration of the categories of providers, the relationship between them as it should be, and a sustainable, practical licensing and regulation policy for the Department of Telecommunications (DoT) that would best serve the growth of Indian datacom.


1. Definitions

2. Goals

3. Regulation

4. Licensing

1. Definitions

Gateway providers: provide physical (telecom-level) or virtual (datacom-level) connectivity to the Internet outside India. For the moment, the role of both physical and virtual provider is monopolised by Videsh Sanchar Nigam Limited. There is no reason why any other organisation should not provide virtual connectivity to the public, using VSNL for the physical link (such as a satellite uplink).

Network providers: lease telecom capacity from the telecom provider (or VSAT providers) and build regional or national public-access data networks ("backbones"), using the Internet protocol. In the US, this would be MCI, Sprint, AT&T, and a handful of others. Network providers may cater to individuals and small businesses, but usually sell capacity in bulk to large organisations, other networks, or service providers.

E-mail providers: an anachronism created by India's present licensing policies, which charge enormous amounts based on traffic (e-mail messages), while internally paying fixed costs for capacity leased in bulk. All commercial e-mail providers would offer Internet services (which includes free e-mail) if forced to by competition. As such, this category is redundant.

Service provider: is the "retailer" of Internet access. Typically fairly low-cost operations running out of home-offices, there are tens of thousands in the US, and nearly a thousand in Hong Kong-whose population is the size of Bangalore's. Service providers buy capacity off networks or gateways, which they sell to individuals and smaller businesses. They are able to offer high-quality, flexible and customised services. They don't usually sell bulk capacity to businesses, although there is much cooperative sharing of capacity with other network and service providers, as seen in America's Commercial Internet Exchange (CIX).

Content providers: create material distributed electronically for a premium, though often for no charge; or provide electronic environments for people to interact on matters of interest (like a live newspaper letters' page). These are treated as electronic publishers, receiving all the protections of Article 19(1)(a) of the Indian Constitution and do not require licensing or regulation in any form.

Bulletin Board Services: are essentially content providers who allow users to access the content through (data) phone calls, rather than through the Internet. BBSes, as Bulletin Boards are called, should not be included in any licensing or regulation regime; they are usually free or barely break even, are run by hobbyists, and are the first introductions to datacom for most people who later go on to use the Internet. In the opinion of former Chief Justice P N Bhagwati, following the Supreme Court's ruling a year ago against restrictions on free speech in the electronic media, the present licensing policy (Rs 1.5 million per annum) for BBSes would be deemed unconstitutional. It is true that many BBSes today go beyond their usual services to provide commercial e-mail, but this is bound to stop once Internet access is widely available.

2. Goals

Telecom Secretary R K Takkar has gone on record to affirm that datacom licences should not be an exercise in revenue generation for the DoT. The sole purpose of licensing is to ensure that the prospective service provider is "earnest about that business [and will] be able to provide a public service"1-which it certainly will not be able to do if it has to pay fees so prohibitive, they can be a major cost of operation. Large fees do not ensure quality of service or responsibility to consumers, which result from free and fair competition. Licence fees simply limit providers to a few large companies and add to their overheads, thus reducing competition and giving them an incentive to form cartels for profit maximisation.

However, a degree of responsibility can be ensured by fees related to costs, rather than revenues, as with the earnest money for tenders, set at a small percentage of the tender value as a guarantee of the bidder's commitment. A further advantage of low cost-based licence fees is that, unlike totally unpredictable revenues, the costs of creating and operating a datacom service are easily estimated, as shall be seen in section 4 on calculating licence fees.

Some responsibility is ensured by a reasonable licence fee; but free and fair competition can only be safeguarded by clear regulations enforced by a regulator. Suggestions for these regulations, not very different from those governing competition in basic telephony in the UK or New Zealand, are provided in the next section. There is already (or will be) a Telecom Regulatory Authority of India; there is no reason why these regulations in datacom, together with consumer complaints, cannot be handled by the same body.

3. Regulation

Equal access to telecom infrastructure: All and any licensed gateway, network and service providers should be provided access to telecom infrastructure such as leased lines and satellite links at equal rates. This includes access to overseas satellite links provided by VSNL, for which providers should pay no more than other bulk users of bandwidth; if VSNL wished to charge providers a higher, reasonably related rate, it should do the same for a licensed, financially self-sustaining subsidiary providing Internet access under the VSNL brand name.

Equal access to gateway providers: All and any licensed network providers as well as service providers should be given connectivity to the Internet through gateways at equal rates for equal bandwidth.

Equal access to networks: All and any licensed service providers should be provided access to network providers' infrastructure at equal rates for equal bandwidth. Ideally (as implemented elsewhere in the world) these rates should be at parity with the networks' other buyers of bulk bandwidth (such as large businesses). If that is not deemed feasible by the DoT, then a reasonable ceiling (25%) should be set on the premium service providers have to pay above the rate for other bulk users.

Network interconnectivity: In order to avoid unnecessary duplication of infrastructure while existing capacity lies unused, networks should be permitted to connect to one another. They should be encouraged to form cooperative arrangements that reduce the role of payment in the sharing of spare capacity (such as America's CIX).

Retailer interconnectivity: Service providers should be allowed to resell capacity to other service providers, and should be able to form cooperative national or regional networks.

Rationalisation of DoT leased line tariffs will be necessary as a first step for much of this to happen. At present, networks that connect, even at one location, have to double their rental for their entire leased capacity across the network. This is extremely counter-productive, and debilitating for the datacom industry. Apart from increasing pointless duplication of networking resources, it leads to ridiculous (not to say wasteful and expensive) situations where traffic between points on two networks, both located in New Delhi, is routed via Bombay and New York. The DoT should radically simplify this tariff structure, eliminating these anomalies. If it insists on holding on to the existing revenue that results from such contorted pricing (quite forgetting that revenue potential would be much larger in a simpler system, with higher demand), it should at least eliminate such sloping tariff scales in the case of cooperative arrangements between service providers, and between networks.

4. Licensing

The purpose of deregulation and de-licensing is not to increase the profit margins of big companies, who would build networks anyway, passing on licensing overheads to consumers. De-licensing is suggested primarily in order to allow the entry of the hundreds, eventually thousands, of small operators who drive the datacom industry and, providing competing alternatives to consumers, bring prices down. That the resulting rapid growth will also increase the earnings of large companies is only incidental. Therefore, following the licensing policy outlined here, there need be no significant revenue loss to the DoT in terms of fees, and there will be considerable revenue gain in terms of telecom infrastructure usage.

Distinctions are made between gateway, network and service providers for the purpose of competition regulation only; there is no reason to formulate three separate licensing tariff structures, when a simple uniform one will do.

License fees should be a small fraction of costs, not revenues, as described in section 2. Of course, this should not mean evaluating costs and licence fees on a case by case basis; rather, as costs are not hard to calculate, and are nearly uniform across the market, a fixed fee structure can be publicly announced. There are two basic costs for datacom providers: the capital costs of acquiring hardware such as computers and communications equipment, and the much larger operating cost of leasing bandwidth. The latter typically accounts for 70% to 80% of total datacom costs in the first year, and close to 100% consequently, at least in the present Indian scenario.

First year capital costs may range from under Rs 500,000 for a minimal service provider to well over Rs 50 million for a nation-wide network. However, there is much room for efficiency and cost-cutting here, so stipulating fees based on an assumed capital cost would not work. Also, capital costs are by definition at start up time. If licence fees are to ensure reliable operation of service, they should be based on recurring operating costs, or more specifically, this being the Department of Telecommunications, on the costs of leasing telecom infrastructure.

The cost of leasing telecom bandwidth, unlike other costs, is uniform. It is invariably available from a few sources: the DoT itself; private VSAT providers; and VSNL for physical overseas links to gateway providers. Bandwidth is also not amenable to cost-cutting, unless uniformly applied across the market: you either have a 64 kbps line, or you do not. It is proposed, therefore, that annual licence fees be set at a small percentage of the costs of planned telecom bandwidth for that year, such as 2.5% (anything above 5% would have a negative impact on consumer prices, or quality of service).

This scheme has many attractions. It immensely simplifies the fee structure, while remaining completely fair and greatly reducing barriers to entry. A small, minimal service provider could start up for as little as Rs 5,000, while a large nation-wide network would pay several million rupees. A large network, incidentally, would not pay much less (and may even pay more) under this scheme, as compared to the old, more restrictive licence structure for e-mail providers. So if the DoT, despite Mr Takkar's personal opinions, continues to look upon datacom licensing as a source of revenue, it should remain satisfied.