Effects of a Sibship Extension to Foster Children on Children's School Enrolment: A Sibling Rivalry Analysis for Indonesia

Abstract

In this paper, we question whether and how the sibship extension to foster children, a common practice in many developing countries, affects children's school enrolment status. Based on Indonesian data, we show that children's school enrolment increases with the number of foster grandchildren in their sibship. Given that grandchildren fostering is associated with positive transfers from biological to host parents, this result reflects either a release of liquidity constraints in the host household or the uncommon preferences of host grandparents. Conversely, the proportion of other foster children in a sibship does not have any effect. The latter result's explanation depends on the reason children other than grandchildren are fostered.

Acknowledgements

I am grateful to Thomas Barré (CES, University of Paris 1), Sylvie Lambert (LEA-INRA), Abla Safir (LEA-INRA, World Bank) and three anonymous referees for their helpful comments and suggestions on a previous version of this paper. All remaining errors are mine.

Notes

1. These results should be read with caution, however, due to the endogeneity of the gender composition of a sibship (Morduch, 2000).

2. Child fostering consists of the transfer of a child from his biological home to another home while both of his parents are alive (Isiugo-Abanihe, 1985). In this sense it is not an orphan placement.

3. Author's calculations based on Demographic and Health Survey reports.

4. Author's calculation obtained from the Indonesian Family Life Survey (IFLS) 1993.

5. In contrast, in West African countries, where child fostering is not only driven by poverty issues but also by customs and traditions, and where support from laterally extended family members is as institutionalized as support from vertically extended family members, the host household selection is likely to be more difficult to control for, leading to a biased estimation of the effect of interest.

6. Isiugo-Abanihe (1985) identifies five reasons for child fostering: children are fostered to cope with economic and demographic shocks; children are fostered to enhance their schooling and social mobility (educational fostering); children are fostered to ‘establish and strengthen social, economic or political alliances’ (alliance fostering); children are fostered to provide domestic services and/or emotional support (domestic fostering); and children are fostered to enforce kinship ties between kinsmen who share kinship obligations and assistance (kinship fostering). As we will see, patterns of child fostering in Indonesia suggest two reasons underlying the fostering of a child: poverty issues (including crisis, domestic labor and kinship fostering) and schooling.

7. If a foster child arrives with very few basic skills, he could profit enormously from learning. If the marginal returns of such a catch-up are higher than those of an additional year of education for biological children, then foster children have expected returns from education that are no lower than those of a biological child. On the contrary, every else being held constant, they have higher returns.

8. This contrasts with what is observed in other countries, notably African countries. For instance, in Cameroon, for a similar age group, only 34 per cent of foster children are foster grandchildren. Author's calculation from Cameroon DHS 2006.

9. This is not surprising, since in the case of foster grandchildren, having biological children involves having aunts and uncles of the same age group.

10. For instance, other foster children live in households hosting on average more orphans than the other households. This might be due to the fact these households are wealthier than the others and thus can afford the costs of rearing additional children. If so, other foster-children should also benefit from living in these households. They may also host more orphans than other households because of higher labour needs. If so, other foster-children are probably sent to these households for labour purpose.

11. When one parent is absent, the value of his education is unavailable. We replace this value by zero to control for the presence of both or one parent.

12. The twelve Province dummies are North Sumatra, West Sumatra, South Sumatra, Lampung, Jakarta, Yogyakarta, West Java, East Java, Bali, West Nusa Tenggara, South Kalimantan, South Sulawesi

13. The correlation between these two shocks is near zero.

14. Domestic work demand, in particular demand for the care of young children, is controlled through the inclusion of the number of young children resident in the household.

15. Computing the net amount of transfers received both by the mother and by the father leads to a similar conclusion: we do not observe any difference between each sub-samples of children.

16. Note that, regarding this last measure, the missing values also differ significantly between each sub-sample. We will therefore introduce this variable by coding the missing values as ‘zero’ and control for the latter.

17. I thank an anonymous referee for having suggested that the release of liquidity constraints could also explain the results obtained.