Do Unwanted Children Get Less Schooling in Matlab, Bangladesh?

Susan Christiansen, Johns Hopkins University
David Bishai, Johns Hopkins University
Abdur Razzaque, ICDDR,B

This paper explores whether children resulting from unwanted pregnancies experience lower levels of education in Bangladesh. Previous studies have either not controlled for selection bias or used retrospective wantedness measures, which are subject to post-hoc revision bias. We use pre-pregnancy statements of fertility intentions from a 1990 household survey of 7,942 women in Matlab, Bangladesh to avoid these potential biases. To test the effect of infant wantedness on child education, we ran OLS and Heckman regression models to predict years of secular education based on prospective wantedness while controlling for demographic and socioeconomic variables. The Heckman model controls for selection bias by predicting whether a woman would have a baby on the basis of her residence in the Matlab treatment area. Using the Heckman model we find boys resulting from unwanted pregnancies experienced a decrease in education by 0.263 years (p=0.05) and mixed effects of wantedness on education for girls.