Perhaps the best example in recent times of why the government should not be allowed to collect taxes, much less spend them, is the case of rural telephony.

Since 2002, the government has levied a cess of 5 per cent on telephone bills to fund rural phones through the Universal Service Obligation Fund, the idea being that rural areas don't have enough of a market to justify telecom firms setting up networks.

The USO Fund collected a whopping Rs 7,254 crore (Rs 72.54 billion) till April but had, and I'm going to write this slowly, disbursed only Rs 1,815 crore (Rs 18.15 billion) of this. This, however, is only one part of the horror story brought out by the telecom regulator, Trai's latest recommendations on rural telephony.

Based on numbers provided to it by the USO Fund, Trai says a total of Rs 30,457 crore (Rs 304.57 billion) will be needed by 2010 to have around 28 million rural phones (as compared to 13.6 million currently), a figure that gives us a rural teledensity of around 4 per cent.

This is the problem -- there is no evaluation of whether a Rs 11,000 subsidy per phone is worth it. This is what Trai now explores, after years of being told the regime, which extends a subsidy to only fixed-line phones (and now fixed-line phones powered wirelessly) is not just unfair, it's stupid since mobile phone costs are much lower.

So, Trai's paper recommends the USO be given for mobiles as well, but the real innovation lies in the approach -- let the USO subsidise part of the cost of building the infrastructure required (a tower could cost anywhere around Rs 50 lakh (Rs 5 million), with another Rs 10 lakh (Rs 1 million) of electronics on it). This dramatically changes the economics, and, according to Trai, could get us a rural teledensity of 15 per cent by 2007.

Here's the plan. If a Bharti comes up and sets up a tower in a rural area (any place with less than 5,000 persons), the USO will give it a Rs 12 lakh (Rs 1.2 million) subsidy, and if Bharti gets a Hutch and an Idea to share this tower (they too will put up electronics worth Rs 10 lakh each for their subscribers), Bharti will get another Rs 24 lakh (Rs 2.4 million) on behalf of the two - Trai wants three operators to share a tower to ensure competition.

Effectively, Bharti's cost comes to Rs 24 lakh while Hutch and Idea's cost will be Rs 10 lakh each. Presumably Hutch will be the lead partner in another area, and Idea in yet another - so, the effective investment per tower will be around Rs 15 lakh (Rs 1.5 million) per telco.

According to Trai, and one cellular firm I spoke to, at such costs, even if they each get 400-500 subscribers who pay around Rs 200 per month, they can break even in 3-4 years. (Since urban teledensity is only 30 per cent and offers a lot more scope for expansion, hoping three operators will collaborate in rural areas appears dicey - a better idea may be to give the entire Rs 36 lakh (Rs 3.6 million) to one firm, with a proviso that it will have to allow two others to share the tower.)

Of course, for a user to get on to the Bharti network, he needs to buy a cellular phone, which could cost around Rs 1,500-2,000. If, as Trai says, this method could give you 100 million extra rural subscribers in two years, this means consumers will have to shell out another Rs 15,000-20,000 crore (Rs 150-200 billion) for their phones, apart from the Rs 200 or so of recurring expenses per month by each.

In which case, the cost works out to around Rs 30,000 crore (Rs 300 billion) {Rs 20,000 crore (Rs 200 billion) for the phones, Rs 7,200 crore (Rs 72 billion) subsidy at Rs 36 lakh (Rs 3.6 million) apiece for the 20,000 towers required to cover 70-80 per cent of rural areas, and another Rs 2,000 crore (Rs 20 billion) or so for leased lines) - this is what the USO will spend to get a teledensity of just around 3-4 per cent under the business-as-usual scenario!

Will rural folk spend so much money? They could, since, if you look at the NCAER's The Great Indian Middle Class, the 72 million households in India that earn over Rs 90,000 annually on average are evenly divided between urban and rural areas.

Indeed, the NCAER's The Great Indian Market forecasts that, by the end of the decade, roughly 11 per cent of demand for cars/jeeps will come from rural areas, as will 48 per cent of motorcycle demand and 43 per cent of demand for regular-sized colour TVs [Get Quote].

The problem, as in other cases, however, can come from the government-owned Bharat Sanchar Nigam. For the system to work, for Bharti to take its signal from the tower in the village to its base-switching centres and mobile-switching centres around 150 km away, it needs optic fibre lines. Trai's suggestion is that since BSNL's 550,000 km of optic fibre reaches most tehsil headquarters, BSNL should provide the connectivity.