This paper conducts an econometric analysis of data for a sample of over 4000 children in India, between the ages of 1-2 years of age, with a view to studying two aspects of the neglect of children: their likelihood of being immunised against disease and their likelihood of receiving a nutritious diet. The starting hypothesis, consistent with an universal interest in gender issues, was that girls were more likely to be neglected than boys. The analysis confirmed this hypothesis. In respect of vaccinations, the likelihood of girls being fully vaccinated, after controlling for other variables, was five percentage points lower than that for boys. In respect of receiving a nutritious diet, the treatment of girls depended very much on whether or not their mothers were literate: there was no gender discrimination between children of literate mothers; on the other hand, when the mother was illiterate, girls were five percentage points less likely to be well-fed relative to their brothers and the presence of a literate father did little to dent this gender gap. But the analysis also pointed to a broader conclusion which was that all children in India suffered from sharper, but less publicised, forms of disadvantage than that engendered solely by gender. These were the consequences which stemmed from children being born to illiterate mothers and being brought up in the more impoverished parts of India. (228 words)

This paper analyses inequality and poverty in India within the context of caste-based discrimination. It does so by decomposing the difference between (caste) Hindu and Scheduled Caste (SC) and Scheduled Tribe (ST) households in: their average household incomes; their probabilities of being in different income percentiles; their probabilities of being at different levels of poverty into: a "discrimination effect", which stems from the fact that a household's income level, into which its (income-generating) profile translates, depends on whether it is SC/ST; an "attributes (or residual) effect" which stems from the fact that there are systematic differences between SC/ST and Hindu households in their (income-generating) profiles. The results, based on unit record data for 28,922 households, showed that at least one-third of the average income/probability differences between Hindu and SC/ST households was due to the "unequal treatment" of the latter.disparities between castes. The first group typically estimates earnings functions for SC and non-SC workers, with a view to seeing how much of the earnings difference between the two groups can be explained by differences in worker qualities (for example, Banerjee and Knight, 1985). The second group is concerned with measuring the degree of economic disparity between, on the one hand, SC and ST persons/households and, on the other hand, non-SC/ST persons/households (Gang et al., 2002;Deshpande, 2000a;Saggar and Pan, 1994).The novelty of this paper is that it attempts to combine the two strands of research by analysing inequality and poverty in India within the context of caste-based discrimination. It does so by decomposing the difference between (caste) Hindu and SC/ST households in: (i) their average household incomes; (ii) their probabilities of being in different income percentiles; (iii) their probabilities of being at different levels of poverty into: (a) a "discrimination effect", which stems from the fact that a household's income level, into which its (income-generating) profile translates, depends on whether it is SC/ST; (b) an "attributes (or residual) effect" which stems from the fact that there are systematic differences between SC/ST and Hindu households in their (income-generating) profiles. 3 This is accomplished by, first, estimating an income-generating function and using the Blinder-Oaxaca decomposition methodology (Blinder, 1973;Oaxaca, 1973;Oaxaca and Ransom, 1994) to estimate the size of the "discrimination effect" in determining average inter-group income differences. Then, a multinomial model for the likelihood of being in different income quintiles (Diamond et al., 1990) is estimated and the Blinder-Oaxaca decomposition methodology is extended to multinomial probability models to estimate the size of the "discrimination effect" in determining average intergroup differences in the probability of being in different income quintiles. Third, following a suggestion by Ravallion (1996), a multinomial model for the likelihood of being at different lev...

No abstract

No abstract