Stature, Skills and Adult Life Outcomes: Evidence from Indonesia

ABSTRACT

We investigate the effect of height on earnings, occupational choices and a subjective measure of wellbeing among Indonesian men. We explore the extent to which height captures the effects of endowments set before entry in the labour market. Physical and cognitive skills, co-determined with stature early in life, do not explain much of the height earnings premium directly. Yet, human capital more broadly, including cognition, educational attainment and other factors related to childhood conditions, explains around half of the height premium and does so through occupational sorting. Indeed, taller workers tend to have more education, and educated workers tend to work in more lucrative occupations that require brain and social skills, not brawn. The unexplained share of the height earnings premium may reflect a specific role of stature on social interaction, labour market advantages or self-assurance. We also find a height premium in happiness, half of which simply accounts for the educational and earnings advantages of taller workers.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. For rich countries, see for instance Lundborg et al. (2014) and Lindqvist (2012) on Sweden; Böckerman, Johansson, Kiiskinen, and Heliövaara (2010) and Böckerman and Vianiomäki (2013) on Finland; Hübler (2009) on Germany; Case and Paxson (2008a) and Persico et al. (2004) on the United States and the UK.

2. Numerous essays in anthropometric history and cliometric research, exploring the secular changes in living standards, have suggested using biological measures like height as complements to conventional indicators of wellbeing (see the surveys by Steckel, 1995, 2009).

3. This is well-documented in rich countries (see Case & Paxson, 2011). Yet, height may be an even better indicator in developing countries because of greater variability in early life conditions (see LaFave & Thomas, 2013; Strauss & Thomas, 1998; Vogl, 2014).

4. While Sohn (2015) focuses on earnings only, and does not consider the role of education through occupational sorting that characterise the ‘explained’ height premium, his study completes ours by testing for the nonlinearity of the premium and showing results for women.

5. This may explain implausibly large IV estimates found in previous studies (for example, Schultz, 2002, 2003a, 2003b). Alderman et al. (1996) use several school quality indicators as well as parental education to instrument for endogenous cognitive skills, highlighting the challenges of this approach. See the discussion in Sohn (2015).

6. This is different from Vogl (2014) who uses the Raven’s Colored Progressive Matrices (CPM) questions. The latter evaluate an individual’s ability to recognise patterns through identification of the missing elements that best match the incomplete patterns. The Raven test is available in the IFLS for individuals aged 7–24 only.

7. This is indeed a short test leading to a possibly noisy measure of cognitive abilities. For Indonesia, LaFave & Thomas (2013) nonetheless show that the variance in word recall scores explains 40 per cent of the variance in Raven scores. They also show very similar correlations between both measures of intelligence and height.

8. Very similar patterns are obtained with other specification including the log of height or a quadratic form. Linearity is convenient for interpreting how the height effect varies with the addition of codetermined variables like cognitive skills.

9. A one standard deviation increase (a 25% increase from the mean) in the word recall test score is related to a 24 per cent (17%) earnings gain.

10. In the more recent figures from the World Bank, mean annual earnings were $PPP 765 in the informal sector compared to $PPP 2382 for entrepreneurs, $PPP 2036 for civil servants and $PPP 1100 for formal private sector employees.

11. A difference, obviously, is the nature of occupational types between rich and poorer countries. Given the high rate of informality and the key role of this sector in poverty analysis, we have constructed our occupation categorisation along the formal/informal divide rather than trying to match more detailed occupation types according to Western standards as done in Vogl (2014).

12. Note that to avoid spurious correlation that may arise from potential correlation between the average height of the reference group and its wealth, we also include the mean income of all men in the reference group as an additional control.