No abstract

Research on caste-based inequalities in India has generally focused on differences between large categories such as the Scheduled Castes, the Scheduled Tribes, and the remainder of the population. We contribute to the literature on horizontal inequalities in India by looking within these groupings, and studying differences between the individual jatis that comprise these categories. Using census data, we find evidence of persistent inequalities in educational outcomes between the jatis, suggesting that socio-economic hierarchies have proved to be stable throughout the post-Independence period. Indeed, the evidence points to divergence: communities with more education in 1961 also had higher educational attainment in 2001. Also, while numerically larger Scheduled Caste communities witnessed greater improvements in educational levels compared to smaller ones, this was not true for the Scheduled Tribes. This may be the result of their greater political mobilization.

Indian data are mostly published at the state or district level. Multi-year analyses of these data are made difficult by the many changes in state and district boundaries that have occurred since the first comprehensive census of independent India in 1961. Between 1961 and 2001, the number of states and union territories in India increased from 26 to 35, and the number of districts increased from 339 to 593. There were several changes in both names and boundaries. We document these changes and use them to construct regions of amalgamated districts with constant boundaries. There are 232 such regions for the entire period 1961-2001. These can be used to construct panel data sets that cover this forty-year period.

Many school children in ethnically diverse countries live in multilingual environments where the medium of instruction in school differs from their mother tongue.We conduct a field experiment in a multiethnic village in northern India to study the relative importance of linguistic distance and neighborhood isolation in the acquisition of language and mathematics skills in elementary school. The village has two sizable ethnic groups speaking languages very different from the medium of school instruction. Students attempt a set of mazes and take tests in reading and mathematics. We find that performance in language-dependent tasks relative to mazes is lower for children whose mother tongue is different from the medium of instruction only if their ethnic group is also residentially segregated. Language policy in most countries has traditionally favored either mother-tongue instruction or assimilation into mainstream language and culture. Our results suggest that there may be value in a more nuanced approach that flexibly combines some instructional support in a mother tongue for those who need it, with policies that encourage contact with speakers of dominant languages to benefit from peers and social networks.

Abstract