This ISP Policy is Incomplete

This ISP policy is a great step forward. This ISP policy is incomplete.

Why do we need licensing at all? For a brief moment, let us look back at another infrastructure debate, that surrounding Enron. The issue at stake was whether electricity from Enron was too expensive. The critics of Enron felt that the price of its electricity was too high, and supporters felt that it was a fair price. Nowhere did anyone suggest that Enron had to pay a `license fee' to the Government of India for the privilege of producing electricity in India. That should be a role model for the telecom sector.

This ISP policy has done well by issuing 15 year licenses, and requiring no ransom. However there is a principle at stake here. I know something is wrong when a firm has to approach the Goverment of India to request permission to risk its own capital in order to improve competition in an industry and benefit customers. Entry into the ISP business should be completely free; a trip to Delhi should not be required. I should be free to connect modems to the computer in my bedroom and become an ISP! The technology of doing this is so simple that given the phone lines, it takes just a few days of work to go live as an ISP.

The Internet was born because of the creative energies of people who set about connecting up to neigbouring sites and setting up ways to dial into their computers at night. Their unregulated ideas have led to the world's largest industry. A licensing environment will not produce innovation. It serves as an entry barrier against small startups, and reduces the impact of the Internet upon India's economy.

The last mile problem Dial--up internet services, today, have a tariff structure which consists of Rs.20/hour to VSNL and Rs.25/hour to MTNL. Most new ISPs will be forced to go through MTNL/DOT, who will ``control their air supply'' in terms of availability of lines, facilities like call hunting, uptime of lines, etc. MTNL will also be able to undercut them on pricing of local calls when it starts competing in the ISP area. This is going to be a thorny area for TRAI to resolve.

Cable modems could be the fastest way to bypass some of these problems and connect up millions of homes at megabit speeds. However, the cable industry is as yet quite undeveloped -- cables running on rooftops and gangland killings do not inspire confidence. So there is no alternative to reforms which make it possible for ISPs to bypass DOT/MTNL on the last mile.

So far, in India, we have paid great attention to individual users of the Internet. However, some of the greatest productivity gains for the economy will come when firms get access to 2 Mb/s leased lines for prices like $1000 per month. It is not yet clear how ISPs will obtain cheap leased lines to connect up to firms.

How will domestic routing take place? The ISP policy makes an important step forward in saying that ISPs can freely use long-haul cables laid by any vendor, not just the DOT. However, mere cabling is not the same as a domestic network which can route traffic between any two points in India.

To think this through, imagine dozens of Indian ISPs connected up to Internet providers in the US. Already, all major Internet sites run by Indian companies use Internet providers in California (the land of the best telecom regulation, and hence the lowest telecom prices, in the world). We will face a situation where gigabits of Indian traffic travels to the US, gets switched there, and returns to India. (We are already in this situation. However the traffic is only of the order of megabits so far; entry by numerous ISPs will grow this figure thousand-fold).

This is wasteful, and expensive.

Last year, when the ISP policy was being re-evaluated by the Bimal Jalan committee, Shuvam Misra (of spaceNET) and I wrote an article where we pointed out that the core problem in India's IT infrastructure is a domestic network where local traffic is switched (Designing India's National Information Infrastructure, by Ajay Shah and Shuvam Misra, Economic and Political Weekly, 8 November 1997). Our key conclusion is that we need a domestic network, which we call IndiaNET, with four features: (1) it should be vendor--neutral, public infrastructure, (2) it should offer IP hookups to anyone without any restrictions on end--use, (3) it should only address domestic connectivity using Internet protocols, leaving the job of obtaining foreign hookups to ISPs, (4) it should be a wholesale network, selling 2 Mb/s lines for an all-inclusive price of around $1000/month.

The ISP policy announced recently is an important step forward, but a critical hurdle is still going to be a lack of this IndiaNET.

Tariff structure arbitrage is a Good thing. Telecom policy formulation, in India so far, has taken a dim view of `tariff structure arbitrage'. This is the complicated name given to the simple activity of people trying to be intelligent in finding the cheapest way to get things done. For example, if voice calls are cheaper over the Internet than over dialup lines, then people will obviously prefer to make voice calls over the Internet. If it is cheaper to do international calls using ``callback services'', then anyone who is able to make dollar payments should use callback services instead of dialing out using VSNL. This is sensible, and clever, and cleverness should be applauded.

Every system of telecom prices will be evaluated in clever ways by people, and their cleverness will often surprise the people setting rates. Tariff structure arbitrage shows up anomalies in pricing, and generates pressures to correct anomalies. If clever arbitrageurs did not exist, anomalies would persist; in a well functioning economy they should rapidly vanish.

In a similar vein, the ISP policy is concerned about unfairness to basic service providers, who have been forced to put up large sums as ransom, and hence denies ISPs the ability to directly connect up with end users. It is, indeed, an anomaly if some vendors (ISPs) pay no license fees and lay fibre to the home while others (basic service providers) are saddled with huge license fees.

I believe that India's telecom system has problems on a massive scale, and that to try too hard to avoid such anomalies would generate paralysis. It is going to take massive surgery to remove these problems in the regulatory and policy framework governing India's telecom. In the short run, liberalisation is far more important than avoiding anomalies. Thus I believe that the short term solution that we should adopt is to liberalise anyway; i.e. to allow ISPs to run fibre on the last mile, even though this will hurt the interests of basic service providers. The long--term solution is to eliminate the ransoms which basic service providers have been forced to put up. But `fairness' to basic service providers -- a highly limited notion of fairness as long as MTNL and VSNL are exempt from paying ransom -- should not hinder a sensible Internet policy.

Back up to Ajay Shah's media page