To madrasahs or not to madrasahs: The question and correlates of enrolment in Islamic schools in Bangladesh

• Main Authors: Asadullah, M.N., Chaudhury, N.
• Format: Article
• Published: Elsevier 2016
• Subjects: HB Economic Theory; HC Economic History and Conditions; L Education (General)
• Online Access: http://eprints.um.edu.my/18346/; https://doi.org/10.1016/j.ijedudev.2016.01.005
• Institution: Universiti Malaya

This paper provides the first comparative assessment of the market share and socio-demographic correlates of children's enrolment in madrasahs in rural Bangladesh using data from a purposefully designed household and community surveys and census conducted in 12 districts. We find that unrecognized madrasahs do have a large presence in rural areas in terms of total numbers but their enrolment share is small. Recognized madrasahs on the other hand has a much larger share in overall student enrolment, particularly in secondary education. Sample households primarily report religious concerns as the motivation for sending children to madrasahs. Yet, only in 7% of households do all school-aged children attend madrasahs implying that religious preference matters but is not the single most important motivation for madrasah education. Therefore we formally investigate the individual, household and community related correlates of madrasah enrolment among 6-18 years old children using Probit regression model. We find a relatively weak effect of gender implying that rural madrasahs attract both boys and girls. Older children are more likely to be enrolled in madrasahs. A statistically significant and negative relationship is found between madrasah enrolment and household income as well as household's access to electricity. Among community-specific factors, children living in locations with fewer non-madrasah schools are more likely to attend madrasahs. Presence of NGOs and availability of satellite dish connection also significantly reduce enrolment in madrasahs. We additionally use nationally representative household survey data to validate some of these findings.