ABSTRACT
We investigate the effect of having opposite sex siblings on cognitive and noncognitive skills of children in the United States at the onset of formal education. Our identification strategy rests on the assumption that, conditional on covariates, the sibling sex composition of the two firstborn children in a family is arguably exogenous. With regard to cognitive skills, learning skills and self-control measured in kindergarten, we find that boys benefit from having a sister, while there is no effect for girls. We also find evidence for the effect fading out as early as first grade.
Acknowledgments
We thank Jean-Louis Arcand, Uschi Backes-Gellner, Matthias Helble, Edward Lazear and the participants of the University of Zurich Workshop in Personnel Economics and the Economics of Education for their comments and suggestions. Martina Viarengo gratefully acknowledges the support and hospitality provided by the Asian Development Bank Institute during her visit.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1 The evidence on gender peer effects is, however, largely confined to primary and secondary education and cannot necessarily be extrapolated to higher education (Oosterbeek and van Ewijk Citation2014).
2 We focus on the early waves because subsequent attrition leads to a sizable reduction in the sample size, and a shift towards children of higher socio-economic status.
3 All outcome variables are standardised to mean 0 and SD 1.
4 Opposite is defined so that the variable takes into account only siblings ±4 years of the sample child. This was chosen so that when the cohort child enters kindergarten the sibling is past baby age where he/she would require a lot of parental attention, but also so that the sibling is not too far away in age such that interactions are rare. The effect remains significant as the age gap between siblings increases, but becomes smaller in magnitude.
5 Covariates are reported in detail in the notes of .
6 We emphasise that our identification assumption holds conditional on the total number of children in a family, because the sex composition of the first borns may have an impact on family size (cf. Ichino, Lindström, and Viviano Citation2014) and families may face a trade-off in terms of child quantity and quality (cf. Black, Devereux, and Salvanes Citation2005). Note, however, that our results remain virtually unchanged if we do not condition on family size or other family characteristics (see columns 1 and 4 of ).
7 Measured by six items: attentiveness, organization, flexibility, persistence in a task, eagerness to learn and independence in learning (Tourangeau et al. Citation2002).
8 Additional analyses, available upon request, show that all estimates of sibling effects are insignificant in higher grades, but these estimates are highly imprecise.