The effect of class size on student achievement: evidence from Bangladesh

Abstract

This study examines the effect of class size on student achievement in Bangladesh using national secondary school survey data. A Ministry of Education rule regarding allocation of teachers to secondary grades is exploited to construct an instrument for class size. This rule causes a discontinuity between grade enrolment and class size thereby generating exogenous variation in the latter. It is found that OLS and IV estimates of class size effects have perverse signs: both yield a positive coefficient on the class size variable. The results suggest that reduction in class size in secondary grades is not efficient in a developing country like Bangladesh. Last, as by-product, some evidence is found suggesting that greater competition among schools improve student achievement.

Acknowledgements

Thanks to Geeta Kingdon, Stefan Dercon, Katy Graddy, participants at the Econometrics Society South-East Asia Regional Meetings, 2002 and VI ERC/METU International Conference on Economics for their suggestions and comments on an earlier version of the paper. I am also grateful to Mizanur Rahman of BANBEIS for his help in compiling the dataset used in this study.

Notes

The justification for this is that the literature routinely substitutes STR as a proxy for class size. For example, Case and Deaton (1999) use district average of STR as a proxy for class size.

The index measures the additional number of schools serving children within a union in the sample and hence is a proxy for school competition within a given geographical area. A union is a magisterial unit in Bangladesh which consists of several villages but smaller than a thana and hence, smaller than a district.

There are altogether 64 magisterial districts in a total of six divisions in Bangladesh.

For the teacher allocation rule to serve as a valid instrument, the rule must be binding on all schools. Schools recruiting additional teachers (not guided by the MoE rule) are required to bear all the expenses arising from such additional unauthorized recruitment of teachers (Mia, 2001). This implies that private (unaided) schools have no such explicit incentive to abide by the MoE rule and consequently dropped from the analysis.

As such, there is a loss of approximately 9% of the original sample schools. This may create some selection issue if non-respondent schools were systematically different from those for whom responses are known. However, no particular factor could be found which could generate the missing observations in a non-random fashion.