Interspersed with its celebrated administrative failures (urban administration and public health are some examples) are India's successes. Organising the Kumbh Mela is one; the conduct of our elections has certainly emerged as another.

Until a few months ago, I would not have added the public distribution system (PDS) to this list; certainly, many bodies including the Planning Commission have been quite critical of the performance of the PDS in the distribution of subsidised food to the poor.

But a recent study conducted by my organisation, the National Council of Applied Economic Research, on the distribution of subsidised kerosene through the PDS system for the petroleum and planning analysis cell (PPAC) of the ministry of petroleum and natural gas ["Comprehensive Study to Assess the Genuine Demand and Requirement of SKO (Special Kerosene Oil)" NCAER, 2005] has yielded some surprises.

The aspect of the study that has attracted the most attention has been its estimate of diversion of subsidised kerosene from its intended beneficiaries. This issue has gained particular attention after the brutal murder of an official of the oil-marketing company, Indian Oil Corporation [Get Quote], when checking for adulteration of diesel (by kerosene) at a filling station in Uttar Pradesh.

The heart of the NCAER study was a large sample survey of urban and rural households. These households were asked questions about their use of kerosene, the price they paid, the use to which it was put, and the source of supply. The estimate of diversion comes from comparing the volumes of kerosene supplied by the government with estimates obtained by aggregating the purchases reported by households.

Press reports on the scale of diversion based on our report have been somewhat confusing. Several of these have assumed that all kerosene delivered to the PDS, and not lifted from the PDS by households, ends up diverted to non-household uses, particularly diesel adulteration.

This is not quite correct. Admittedly, the aggregate gap between annual PDS supply (11,400 kilolitres) and PDS purchase (7,300 kilolitres) is a high 36%. But the gap between estimated household kerosene consumption and PDS supply is of the order of only 18%. Fully half of the diverted kerosene ends up with households, either ones who do not possess a ration card, or at black market prices.

Equally revealing is what the household survey tells us about the reach of the PDS system. Some 94% of all rural households (up to 99% in some poorer states) use kerosene for some purpose or the other: largely for lighting, although also for cooking.

Even in urban areas, as many as 69% of households report some use of kerosene. Strikingly, 95% of these rural user households report getting part or all of their supply officially through the PDS, while even in urban areas this figure is 82%. From these numbers it would seem that the rural penetration of the PDS system is around 90% of all households, a significant administrative achievement in a poor and far-flung country. (I understand from Surjit Bhalla that the NSS indicates that most of these extended households are below the poverty line.)

Equally striking, some 90% of rural households were both aware of their entitlement and lifted the entire amount.

The main reason for this deep reach of the PDS where kerosene is concerned is, no doubt, the massive subsidy given to household uses of kerosene, estimated to be around Rs 10,000 crore (Rs 100 billion) (direct fiscal subsidy plus under-recovery by the oil companies).

This has in effect driven out any significant legal private trade in the fuel for household purposes and strengthened the incentives to engage almost exclusively with the PDS system. As with any rationed good, a parallel market does exist; this accounts for approximately 22% of total household consumption of kerosene, at an average market price (Rs 21/litre) that is almost double that of the rationed price (Rs 10/litre).

What we seem to have in the kerosene market, therefore, is a system of administrative allocation which works better across households than might be expected, but at a high fiscal and environmental cost. The question is where we go from here, and whether the same delivery machinery can be used to achieve the same reach at lower cost.

A recent pamphlet by TERI (The Energy and Resources Institute) (Neha Misra and associates 'Petroleum Pricing in India: balancing efficiency and equity', TERI 2005. www.teriin.org) argues that it would be economically and environmentally efficient to phase out production of kerosene, and replace it by a combination of electricity and LPG.

In the aftermath of the IOC murder mentioned, the ministry has floated proposals to move from the present single-price rationing system to a dual-price system, along the lines of what was done for food some years ago. Press reports suggest that this proposal is being resisted by some states, who are of course the key players in the public distribution system.

Meanwhile, the petroleum ministry has launched a pilot programme for a more intensive monitoring of PDS kerosene distribution in selected rural blocks, with a greater role for monitoring by gram panchayats.

Each of these schemes will encounter resistance and create difficulties. Since there are few alternatives for lighting in non-electrified areas, there is likely to be little price elasticity of demand and price increases will have a clear effect on real incomes for those most dependent on kerosene for lighting.

A move towards dual pricing could increase the incentives to divert kerosene from subsidised users to unsubsidised users. Reducing adulteration will require closing the gap between the price of kerosene and the price of diesel. So there is no perfect solution.

A combination of gradual price escalation for all, dual pricing for households above and below the poverty line and better administration and monitoring together may be needed to move towards the appropriate long-term goal, together with improved rural electrification. But the stakes involved are large, involving fiscal, environmental and social justice issues.

A more significant debate needs to occur than has been the case so far. Thanks to the sponsorship of the PPAC there is a slightly firmer empirical basis for this than before.