Teacher characteristics and student performance in India: a pupil fixed effects approach

This paper exploits unique data that permits the matching of students' test scores in different subjects to the teachers that teach those subjects. Within-pupil (across-subject, rather than across-time) variation is used to examine whether the characteristics of different subject teachers are related to a students' marks across subjects. There are four main contributions. Firstly the findings, using a credible methodology for identification, give only modest grounds for optimism about the effects of teacher policies. A teacher's possession of Masters level qualification and pre-service training have well identified but small effects on student achievement. While a teacher's union membership strongly reduces pupil achievement, union membership is typically not a policy variable. The bulk of the variation in student achievement is a school fixed effect and observed school characteristics explain less than 30% of this fixed effect. The second main contribution of the paper is to highlight the importance of 'controlling for' the non-random matching of students to schools and teachers. The finding that within-pupil effects of many teacher variables differ very significantly from the across school effects indicates that much of the extant achievement production function literature - which perforce relies on across school estimation - leads to incorrect inferences because it confounds the effect of unobserved school and pupil heterogeneity with the effect of teacher characteristics. This underlines the importance of finding credible sources of within school and preferably within-student variation in future research. Thirdly, the paper showcases the use of an across-subject estimator of the achievement production function which is similar to the more familiar panel data approach but which circumvents the problem of non-random attrition of students/teachers over time and the problem of non-random matching of students to teachers, and which permits the identification of teacher effects in cross-section data that are readily available. Finally, a school fixed effects equation of teacher pay shows that while teacher compensation is efficient in some respects, i.e. teachers are rewarded for characteristics that raise student achievement, it is not so in other respects. In particular, union membership is substantially rewarded when in fact it is associated with signficantly lower student achievement, raising the question whether teachers' right to unionize pits teacher interests against student interests.