LOOKING BEYOND UNIVERSAL PRIMARY EDUCATION

Abstract

This paper addresses gender equity in parents' educational investments in children in rural Bangladesh. The authors analyse correlates of time spent in school, studying outside school and work, using a dataset on time-use patterns of school-going children. The authors find that time spent at work varies inversely with the amount of time spent studying at home. Time spent in school is insensitive to factors such as poverty and gender. Time spent studying outside school is strongly influenced by household decisions that favour boys, who appear to have about 30 minutes more discretionary study time than girls.

Keywords:

• time-use survey
• gender differences
• investment in education
• Bangladesh

Notes

1. Gross enrolment rate, in secondary school, is the number of students (total of boys and girls) enrolled in secondary school, regardless of age, expressed as a percentage of the population (total of boys and girls) in the theoretical age group for secondary education.

2. See: http://www.ungei.org/infobycountry/bangladesh.html (accessed 25 August 2011).

3. Details on the survey and the data are available at http://dataportal.popcouncil.org

4. The overall rate of non-response and refusal was low. There was intense scrutiny during fieldwork and every questionnaire was cross-checked by a supervisor in the field to ensure rapid correction when necessary. A random sample of respondents were cross-interviewed by a data-quality enforcer. Consistency checks were conducted onsite and reconciled by supervisors.

5. Since we observe time spent studying at home only for children who attend school, we have to correct for sample selection bias. Hence, we include children who do not attend school and use the distance to secondary school in the division as an instrument. We test for the selection effect and reject the test at a 10 per cent level of significance.

6. The models shown include all respondents who have been recorded as being currently enrolled and attending school, whether or not they have reported going to school on the day in question. Our results are not different from coefficients estimated from models where only those who have reported attending school are included.

7. There are other studies that use richer datasets. For example, Edmonds (2006) uses a rich dataset from Nepal to study sibling differences in child labour in Nepal. In the Nepal dataset, there is information on time spent during the seven days preceding the survey by each member of the household in 16 different activities. Edmonds (2006) has found that older girls tend to work more than their brothers. We are unable to address such questions due to limited information in our dataset.