Sense in sociability? Maternal education, social capital and child schooling in rural Bangladesh

Abstract

Using detailed micro-data on social relations and networks from rural Bangladesh, this paper examines the social determinants of children's schooling in a developing country. We employ a comprehensive set of controls for parental sociability and NGO membership as proxies for family social capital. We find no evidence that parental sociability bolsters child outcomes as inputs in the educational production function or indirectly via boosting maternal social knowledge, which is an important predictor of children's educational status. On the other hand, the effect of maternal social knowledge prevails even when we additionally control for neighbourhood-level unobservables and maximum education level of the household. To be precise, we find no evidence to suggest that ‘social’ parents are ‘better’ parents: social knowledge endowment of mothers is not related to their sociability. Rather, educated parents and those with higher unobserved taste for child quality have better social knowledge irrespective of their sociability.

Cet article se penche sur les déterminants sociaux de la scolarisation des enfants dans les pays en voie de développement, en s'appuyant sur une étude détaillée des relations et des réseaux sociaux dans le milieu rural au Bangladesh. Nous utilisons un ensemble de variables de contrôle pour la sociabilité parentale ainsi que l'adhésion aux ONG comme critères indicatifs des niveaux de capital social familial. Nous ne trouvons aucune corrélation qui puisse laisser croire que la sociabilité parentale améliore la trajectoire scolaire des enfants, que ce soit de manière directe ou bien indirectement en accroissant le niveau des connaissances sociales maternelles, un facteur dont notre étude démontre l'influence importante sur le statut éducatif des enfants. Ceci même lorsque nous tenons compte des ressources communautaires intangibles ainsi que du niveau d'éducation maximal du ménage. Nous concluons donc que les parents « sociaux » ne sont pas de « meilleurs » parents, dans la mesure où les connaissances sociales maternelles ne sont pas liées à la sociabilité des mères, et que ce sont plutôt le niveau d'instruction des parents ainsi que leur foi intrinsèque dans les qualités de leurs enfants qui déterminent la trajectoire scolaire de ces derniers.

Keywords:

• economic development
• maternal education
• micro-credit
• social capital
• social network

Notes

 1. An exception is Morgan and Sorensen (1999) who find only weak evidence in support of the hypothesis that connections between the parents of students is positively associated with student learning.

 2. For a critical review of these studies, see Durlauf and Fafchamps (2005).

 3. There is also a trivial case where β₄ = 0 and social knowledge plays no role in schooling outcomes. This could be true for the following reason. ‘Social’ mothers may trade child care time off to socialize with others outside the household. If the resultant time-constraint binds, it may offset any positive effect arising from greater sociability.

 4. These are good proxies for what is defined by Portes (1998) as family social capital as a source of extra-familial network.

 5. These people are often known as village ‘Matbarr’ or ‘Morol’.

 6. We also have information on the total number of kin residing outside the residential compound (bari) in the village.

 7. We did not use individual's use of ‘burqa’ as it is a choice variable whereas a bari-level measure is relatively exogenous.

 8. According to Larance, many women in the rural village know very few women outside of their 4–5 household living quarters; from such a low baseline, perhaps it is not surprising that simple weekly meetings would prompt big increases in social interactions. This change in network, if attributable to NGO membership is likely to be significant in rural Bangladesh where traditional social norms often prevent interactions that involve women and people outside of their immediate living quarters and kin groups.

 9. In a recent study using data from rural Bangladesh, Pitt, Khandker and Cartwright (2006) also find evidence that women's participation in micro-credit programmes leads to greater mobility and social networks.

10. The variable ‘maximum education amongst adult householder’ controls for within-household externality effects owing to the presence of educated adult kin; this variable has been found to be an important determinant of child schooling outcome in rural Bangladesh (e.g. see Asadullah 2008).

11. The extent of NGO activities by villages in Matlab such as credit programmes (of BRAC, Grameen Bank, Rural Development Project 12, etc.), adult literacy programmes and NFP schools is not known in detail. From the community-level data file of the MHSS, it appears that there are very few non-formal school in Matlab – only 1.6% of school-going children are enrolled in these schools (against a national-level share of 8% of total primary enrolment).

12. This is because most of the female heads are widows or wives of migrant males so that data on children's paternal background is not available for this sub-sample.

13. School participation and grade completion may overestimate educational achievement of children in rural Bangladesh where educational quality remains very low across all school types (Asadullah, Chaudhury and Dar 2007). It is for this reason that we additionally investigate the social capital effect in the context of cognitive skill and literacy status of children.

14. Nonetheless, paternal social knowledge has no effect, literacy regressions being an exception.

15. For instance, size of the coefficient on maternal social knowledge in primary school completion regression is nearly twice that of mother's education (see Table 1, model 4). Furthermore, the two variables have identical impact on the probability of the child being literate (see Table 2, model 4).

16. Given that our IV model is exactly identified, we cannot carry out a test of instrument validity. However, first stage regressions (available from the author upon request) show that the coefficient on the land dummy is highly significant and positive. This result is consistent with McKernan (2002).

17. This is also confirmed by an F-test of joint significance of these seven variables.

18. These results are available from the author upon request.

19. Indeed, women's involvement in social networks has been found to increase their control over resources (e.g. see Hashemi and Schuler 1996) which in turn can lead to better child outcomes.

20. An exception is Adams, Madhavan and Simon (2001) who study the influence of women's social networks on child survival in Mali.