Caste in Class: Evidence from Peers and Teachers

Differences in academic achievement across Indian castes are both large and persistent. I make use of rich individual data to explore how class caste composition affects academic progress as well as the mechanisms in place. Benefiting from exogenous assignment of students to classes and teachers, I find that a one-percentage point increase in the proportion of low-caste class- mates leads to a fall of around 2% of a standard deviation in the mathematics score and to much smaller effects in English. This phenomenon is mediated through lower effort exerted by the students, which itself emanates from the students' worsened perception about the extent to which their teachers value them. This non-cognitive channel, which has not been previously identified in the peer effects literature, suggests that the use of a fairly malleable input such as more open and receptive teachers among low-caste students would be an appropriate policy response.