It is hard to think of a more far-reaching piece of legislation than that approved by the Cabinet on the right to employment. It can do a lot of good, but has the potential to do a lot of harm as well.

A law that mandates 100 days of employment for one member of every poor family in the rural areas (and there are 40 million such families) will cost the country Rs 40,000 crore (Rs 40 billion) annually -- or 1 per cent of what GDP would have become by the time the programme gathers momentum in three to five years.

The wage will be the legally prescribed minimum wage, with the states free to pay more if they so wish.

The intention is eminently laudable (and forget for a moment about the fiscal deficit), but that is not enough to guarantee success.

For there is the risk of leakages through fake muster rolls, through the Ranvir Sena capturing the programme in Bihar, and the CPM doing the same through its cadres in West Bengal.

The end result in all cases will be the same: the money gets spent and the intended beneficiaries get only a small part of it, but it has become politically impossible to get rid of the scheme.

For the idea of a social security system will acquire much more political attraction than free power, job reservations, high cane prices and other seemingly costless (costless, that is, to the political parties) electoral promises -- and the risk is especially great at the state level if the funding is by the Centre.

Fortunately, past experience with job schemes provides good examples of both the pitfalls and the ways in which the benefits can be maximised at least cost.

For instance, the Maharashtra employment guarantee scheme has worked well by focusing not on a minimum wage but on a task rate--a day's work is defined as a unit of work (which varies depending on whether the work is road construction or digging a well).

Payment depends on the measurement of that work. The result is that the actual payout per day of work is no more than Rs 45, the scope for fake muster rolls and general carpetbagging is minimised.

The other critical element of success is to avoid targets--which forces bureaucracies to spend the money so that it does not lapse.

The point of an employment scheme like this is that the beneficiaries must come forward on their own, i.e. it becomes self-selecting. There is therefore no need for targets.

Done this way, the annual cost of the scheme in a state that has 8 per cent of the country's rural poor is less than Rs 1,000 crore (Rs 10 billion). By that yardstick, a national programme should cost no more than Rs 12,000 crore (Rs 120 billion) -- which is a far cry from the Rs 40,000 crore (Rs 400 billion) bandied about in the context of the proposed central law.

The experience in Rajasthan is that leakages under job programmes used to be as high as 30 per cent. But Rajasthan also has activists who monitor the quality of government programmes and, at least partly because of their efforts, the leakage rates are said to have come down to 20 per cent and even 5 per cent.

The all-India statistics, tallying the total spend on all government job programmes and the employment created, are less definitive, suggesting in some years that there is no leakage at all, and in other years that the leakage is 50 per cent.

What this tells us is that it is indeed possible to devise a scheme that delivers the intended result without too much leakage.

But this needs clear safeguards in the law through the definition of a task rate rather than a wage rate, and the stipulation that at least part of the cost must be from the state exchequer -- which encourages states to be responsible when it comes to spending under the scheme.

The worry is that the scheme approved by the Cabinet is not crafted with these safeguards in mind. So we may get more of the risks and less of the intended benefits.