Food for thought: the birth-order effect and resource allocation in Indonesia

ABSTRACT

Despite the large literature linking birth order to socio-economic outcomes, research uncovering the mechanisms that drive these effects has been rare. We construct a unique measure of nutritional variety to examine whether differential resource allocation among siblings explains the birth-order effect on cognitive ability. We document birth-order differences among children and young adults, showing that second and third borns perform worse on cognitive assessments compared to their firstborn siblings. We find further evidence suggesting that second- and third-born children frequently eat a smaller variety of food than their first-born counterparts. However, this differentiation in food allocation does not appear to drive the differences in cognitive ability in our sample. To our knowledge, this article is the first both to confirm the presence of a birth-order effect on cognition in a developing country and to test empirically whether a difference in food allocation contributes to this effect.

KEYWORDS:

• Birth order
• cognition
• food allocation

JEL CLASSIFICATION:

• J13
• D13

Acknowledgements

We are grateful to an anonymous referee as well as seminar participants at the 2014 Western Economics Association’s Annual Conference and 2015 Southern Economics Association’s Annual Conference for helpful comments and feedback.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ Educational attainment decreases as birth order increases in the United States and Europe (Black, Devereux, and Salvanes 2005; Booth and Kee 2009, Katarevic and Mechoulan 2006), whereas studies set in developing countries typically find the opposite effect, in that younger siblings have higher average levels of schooling compared to their older siblings (Ejrnaes and Portner 2004; Emerson and Souza 2008; Park and Chung 2012).

² The spatial assessment questions are similar in format to those used in typical tests of spatial intelligence. The actual assessment instrument can be downloaded at http://www.rand.org/labor/FLS/IFLS/ifls4.html.

³ Although standardized, the sample average assessment scores reported for the full sample in Table 1 do not equal zero. These averages are not zero because we performed our standardization based on the entire sample of individuals who took the cognitive and math assessments. However, the descriptive statistics reported in Table 1 for assessment scores are given for the fixed-effects sample only. Thus, it appears that individuals in our fixed-effects sample (who all have siblings taking the assessments as well) score slightly higher on average than individuals in the full data sample. Regardless, the fixed-effects regression results reported in Table 2 are robust to standardizing assessment scores based on either the full sample or the fixed-effects sample only.

⁴ The empirical results of Table 4 column 1 serve to further minimize this concern. If firstborns are being allocated food more frequently simply because they are older, then our results should indicate decreasing frequency of food allocation by order of birth. However, while second- and third-born children do appear to have lower food allocation than firstborns, fourth and higher borns do not. Thus, it is unlikely that frequency of food allocation is simply being driven by age.

⁵ Approximately 30 per cent of families in our dataset report no variation in food allocation among family members.

⁶ Sibling composition variables include number of older brothers, number of younger brothers, number of older sisters, and number of younger sisters.

⁷ Household assets include the total value of the home, other homes besides the principal residence, non-farm land, livestock, poultry, karas plants, vehicles, appliances, savings, receivables, jewellery, furniture, and other household assets.

⁸ Our regressions also show that male children tend to score higher on assessments than females. However, in ancillary regressions not shown, the birth-order effect does not vary according to sex.

⁹ The fixed-effects results from Table 2 are robust to limiting the sample to that used in OLS regressions. In fact, using fixed-effects on the smaller sample actually doubles the magnitude of the birth-order effect for second-born children. OLS results from Table 2 are also robust to the limited fixed-effects sample. All results available upon request.

¹⁰ In our sample, the median household reports a total asset value of 24,800,000 rupiah.

¹¹ The magnitude and significance of all birth-order differences in Table 4 become slightly larger when the sample is limited to children under 25 (since individuals older than 24 years old did not participate in the cognitive assessments).

¹² In regressions not shown, interaction terms with sex and birth order indicate that while male children receive more food overall, birth order differences themselves do not appear to vary according to sex.

¹³ The sample in column 5 is very small (1495 observations) because it is limited to those individuals who took the cognitive assessments in 2000 only. Many individuals took the assessments in both 2000 and 2007, and we restrict those individuals to the 2007 sample only (the latest year). However, when we estimate a regression using the full sample of all individuals who took cognitive assessments in 2000, regardless of whether they took the tests again in 2007, we obtain similar results to those reported for the small sample in column 5.