Despite glowing accounts of how well the Indian economy has performed in recent years, some disadvantaged groups - the Scheduled Castes (and Scheduled Tribes  - remain mired in acute poverty.

A recent study (Kijima, Y, 2006 - "Caste and Tribe Inequality: Evidence from India, 1983-1999", Economic Development and Cultural Change, vol 54) offers some surprising evidence on relative disparity in living standards (or, more precisely, in expenditure per capita) between these disadvantaged groups and others in rural India, long after the government of India introduced its policy of affirmative action.

This disparity reflects not just lower endowments of human and physical capital (for example, education and land owned, respectively) but also lower returns on them among the SC and ST households.

While there has been some reduction in the expenditure disparity over the period 1983-99, its decomposition into two components, viz, (i) lower endowments, and (ii) lower returns, is worrying.

The SCs were less worse-off than the STs in both 1983 and 1999. However, the sources of their disparities differ. While the SC households were more deprived (relative to the non-SC/ST households or Others) due equally to lower endowments and lower returns, the STs' deprivation resulted largely from lower endowments (about two-thirds). What is indeed surprising is that the relative importance of these sources has remained unchanged over the period 1983-99.

Among endowment differences, education had a significant role. It accounted for about 53 per cent of the disparity due to this component among the ST households, and for about 89 per cent among the SCs in 1999. In the second component (that is, differences in returns), locational disadvantages (for example, the lack of easy access to markets) had an important role. Among the STs, about 50 per cent of this component was due to such disadvantages, and, among the SCs, it was about 38 per cent.

A more recent analysis based on the 61st round of the NSS [R Gaiha, G Thapa, K Imai and Vani S Kulkarni (2007) "Endowments, Discrimination and Poverty", (in preparation)] throws further light on the plight of these disadvantaged groups.

The overall incidence of poverty in rural India was about 25 per cent. About 47 per cent of the poor belonged to the SC and ST households, with the former accounting for about 28 per cent of the poor.

However, the most poverty prone were the STs, as about 44 per cent were poor, followed by the SCs (with one-third being poor). Also, a measure of the intensity of poverty (that is, the expenditure-poverty gap) suggests that the STs were the poorest (with a gap of about 26 per cent), followed by the SCs (with a gap of about 21 per cent), and Others (with a gap of about 18 per cent).

As there has been a debate about the appropriateness of the poverty cut-off point, regardless of the poverty cut-off point and the measure of poverty used (that is, whether the head-count index, expenditure-poverty gap or a distributionally sensitive index which weighs the income gains of the poorest more), the STs were the most deprived, followed by the SCs, and Others in 2004-05.

This follows from the fact that the cumulative expenditure distribution curve for the STs lies above that for the SCs, and the latter lies above that for Others.

Let us consider the targeting of two major anti-poverty interventions, Food-For-Work Programme and PDS:

Of the FFWP participants, a large majority (about 63 per cent) were non-poor. It is therefore not surprising that the average per capita expenditure of a participating household (Rs 484) was well above the poverty cut-off point.

Of the beneficiaries of PDS, a larger majority (about 74 per cent) were non-poor. This is further corroborated by the considerably higher per capita expenditure of the PDS beneficiaries (about Rs 569), relative to that of FFWP participants. So both interventions were mistargeted.

The targeting of FFWP among the STs was better, as the share of poor participants was about 55 per cent. The average per capita income of a beneficiary household (Rs 396) was thus higher than the poverty cut-off point but not by a large amount.

The targeting of PDS among the STs was less accurate, as the share of poor beneficiaries was less than half (about 45 per cent). As a result, the average per capita expenditure of a beneficiary household (Rs 436) was considerably higher than the poverty cut-off point.

Among the SCs, the share of poor beneficiaries of FFWP was a little over one-third (about 35 per cent). The per capita expenditure of a beneficiary household (Rs 504) was thus markedly higher than the poverty cut-off point.

The share of poor SC beneficiaries of PDS was about one-third (33.55 per cent) pointing to greater mistargeting. Not surprisingly, the average per capita expenditure of a beneficiary household was high (Rs 490).

Among Others, the proportions of poor beneficiaries of these interventions were even lower. A little over a quarter (about 27 per cent) benefited from FFWP, while barely 21 per cent did from PDS. Not surprisingly, the average per capita expenditures of FFWP and PDS beneficiaries among them were much higher (Rs 526 and Rs 616, respectively).

These results point to generally unsatisfactory targeting of two major anti-poverty interventions. The poor belonging to any of the groups, the SCs, the STs and Others, did not participate in them as much as relatively affluent sections did.

Recent reviews of the NREG scheme are not reassuring either except that some point out that awareness of employment as an 'entitlement' among the disadvantaged has grown.

Whether this would translate into greater benefits from anti-poverty interventions and compensate them - especially the SC and ST households - for the inequities of endowments and returns is optimistic but not unlikely.