Abstract
Health and education are known to be highly correlated, but the mechanisms behind the relationship are not well understood. In particular, there is sparse evidence on whether adolescent health may influence educational attainment. Using a large registry dataset of twins, including comprehensive information on health status at the age of 18 and later educational attainment, we investigate whether health predicts final education within monozygotic (identical) twin pairs. We find no evidence of this and conclude that health in adolescence may not have an influence on the level of schooling. Instead, raw correlations between adolescent health and schooling appear to be driven by genes and twin-pair-specific environmental factors.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
Birth weight divided by gestation.
A few studies, including Smith (2009) and Currie et al. (2010), have examined the role of childhood health using sibling comparisons, thus not fully accounting for potential genetic confounders. As will be shown, accounting for genetic confounders is important for the interpretation of the relationship.
This has previously been pointed out by Miller et al. (2005).
Although we use the term “adolescent” throughout, it should be noted that health is a process from birth up to the current age and that our measure will reflect a sum of health shocks up to the age of 18.
Up until 1972, individuals were usually called to undergo the military enlistment test at age 19, whereas in later years most individuals underwent the enlistment test at age 18.
A number of studies suggest that childhood disease burden has strong effects on the growth of a person (e.g., Bozzoli et al. 2009), and a number of recent studies have used indicators of height or stunted growth to proxy for health early in life (e.g., Case and Paxson 2008; Bhalotra and Rawlings 2011). Bozzoli et al. (2009) suggest that height is related to rates of postneonatal mortality due to pneumonia and possibly congenital anomalies and intestinal disease. They find little evidence that height would be related to mortality rates for other conditions. In poorer countries, stunting has also been related to diarrhea (Martorell et al. 1975; Lutter et al. 1989).
With the exception of conditions of the sensory organs; low hearing acuity and low visual acuity are instead studied separately.
BMI, body mass index, is calculated as (weight in kilograms)/(height in meters)2.
See Carlsson et al. (2015) for a further explanation of the subtests.
To be specific, we use information on grades and results from a math test described by Pettersson (1993) and restrict attention to children with Swedish-born parents, yielding 7399 individuals. We standardize test scores based on subtest 1a and 2 (as these were taken by all children), sum these, and standardize again. Finally, the average test score is then calculated for each grade in general and special math. Each individual in our data is then assigned this average standardized test score corresponding to their grade, whether in general math or in special math.
Our approach has the advantage that twin pairs with missing data will not contribute to identification of health’s effect on schooling. However, in order to take into account the interrelations between regression variables and the occurances of missings, one can also use multiple imputation (Rubin 1986). Implementing this analysis, using the STATA command mi input chained makes little difference to our results. Results are also similar if dropping individuals with missing information on any variable.
More specifically, individuals are excluded if global health is missing. It is not possible to determine whether there is missing information on specific health conditions. However, for individuals reported to suffer from specific health conditions, data on global health is always available. Missing data on global health usually reflects that the data from the enlistment process has not been electronically recorded, which is the case for all conscriptions that took place in 1985 and for parts of several other years.
Descriptive statistics for DZ twins are reported in Appendix Table 7. There are many significant, albeit small, differences in average characteristics when comparing the full male population sample, the MZ twin sample, and the DZ twin sample. For example, both MZ and DZ twins tend to have slightly better global health than the overall male population at age 18. We have run regressions where we re-weight the MZ and DZ twin samples to match the frequencies of global health observed in our full sample; however, this makes little difference to the results.
An effect of parental education on child’s health and abilities (as measured at military enlistment in Sweden) was shown in Lundborg et al. (2014a).
This is most likely because the NSA de facto raised their thresholds and evaluated individuals’ health less favorably in more recent years. Accounting for this by the use of year of birth dummies increases the effect of global health in absolute value from −0.46 to −0.67 (an increase by 46 %). Adding controls for parental education then reduces this effect by 24 %, which reduces by another 6 % when also controlling for family income.
It is possible to run separate OLS regressions on twin pairs differing in global health (thus contributing to the FE estimate) and on those not differing (thus not contributing to the FE estimate). Doing this, with controls, we find an effect of −0.34 in the former sample and −0.58 in the latter. It is not surprising that the latter estimate is larger, as it relies entirely on between-pair comparisons, where healthier twin pairs are also likely to have other advantageous characteristics compared to less healthy ones. Still, it is noteworthy that the former estimate is substantial and strongly significant, as opposed to the result obtained using fixed effects.
We can also control for different subscores of cognitive ability. The significant effect turns out to be driven by logical skills and synonym understanding, but whether we use the overall score or the subscores makes virtually no difference for the coefficient on health.
The strong positive relationships between respiratory conditions and education in OLS models without controls may be explained by the fact that smoking used to be more common among individuals with higher levels of educational attainment, a fact that has previously been shown for the USA (Sander 1995). Using the Swedish Level of Living Survey from 1968, we regressed a smoking dummy on a dummy indicating “more than basic education” (no controls) and obtained a coefficient of 0.06; t = 4.88. With this positive relationship, in combination with a positive intergenerational transmission of education and the fact that children to smokers are more likely to get respiratory problems (e.g., Cook and Strachan 1999), we would indeed expect these results.
While using the 5th percentile to define “low physical capacity,” “weak handgrip strength,” “short,” “low visual acuity,” and “low hearing acuity” is somewhat arbitrary, we have experimented with using the 25th percentile, and this did not change our conclusions. Moreover, since the correlations between different health measures are not too strong, results are quite similar when including them one by one rather than simultaneously. See Appendix Tables 11 and 12 for these results.
We can also account separately for either cognitive or noncognitive ability. These two are positively correlated, and controlling for either of them renders the effect of health small and insignificant.
References
Adler NE, Ostrove JM (1999) Socioeconomic status and health: what we know and what we don’t. Ann N Y Acad Sci 896:3–15
Adler NE, Boyce TM, Chesney MA, Cohen S, Folkman S, Kahn RL, Syme SL (1994) Socioeconomic status and health. The challenge of the gradient. Am Psychol 49:15–24
Almond D, Mazumder B (2013) Fetal origins and parental responses. Ann Rev Econ 5:37–56
Ashenfelter O, Krueger AB (1994) Estimates of the economic return to schooling from a new sample of twins. Am Econ Rev 84:1157–73
Ashenfelter O, Rouse C (1998) Income, schooling and ability: evidence from a new sample of identical twins. Q J Econ 113:253–84
Behrman JR, Rosenzweig MR (2004) Returns to birth weight. Rev Econ Stat 86:586–601
Bhalotra S, Rawlings S (2011) Intergenerational persistence of health in developing countries: the penalty of gender inequality? J Public Econ 95:286–99
Black SE, Devereux PJ, Salvanes KG (2007) From the cradle to the labor market? The effect of birth weight on adult outcomes. Q J Econ 122:409–39
Bonjour D, Cherkas L, Haskel J, Hawkes D, Spector T (2003) Returns to education: evidence from UK twins. Am Econ Rev 93:1799–1812
Booth ML, Owen L, Bauman A, Gore CJ (1996) Relationship between a 14-days recall measure of leisure-time physical activity and a submaximal test of physical work capacity in a population sample of Australian adults. Res Q Exerc Sport 67:221–7
Bound J, Solon G (1999) Double trouble: on the value of twins-based estimation of the return to schooling. Econ Educ Rev 18:169–82
Bozzoli CG, Deaton AS, Quintana-Domeque C (2009) Adult height and childhood disease. Demography 46:647–9
Card D (1999) The causal effect of education on earnings. In: Handbook of labor economics, Vol. 3A, ed. Ashenfelter O, Card D. Amsterdam, NL: North Holland
Carlsson M, Dahl GB, Öckert B, Rooth D-O (2015) The effect of schooling on cognitive skills. Rev Econ Stat 97:533–47
Case A, Paxson C (2008) Height, health, and cognitive function at older ages. Am Econ Rev 98:463–7
Chiteji N (2010) Time preference, noncognitive skills and well being across the life course: do noncognitive skills encourage healthy behavior. Am Econ Rev Pap Proc 100:200–4
Conti G, Heckman J, Urzua S (2010) The education-health gradient. Am Econ Rev Pap Proc 100:234–8
Conti G, Heckman J, Yi J, Zhang J (2015) Early health shocks, intra-household resource allocation and child outcomes. Econ J 125:F347–71
Cook DG, Strachan DP (1999) Summary of effects of parental smoking on the respiratory health of children and implications for further research. Thorax 54:357–66
Cunha F, Heckman JJ, Lochner L (2006) Interpreting the evidence on life-cycle skill formation. In: Handbook of the economics of education, Vol 1, Hanushek E, Welch F (eds). North-Holland, Amsterdam
Currie J (2009) Healthy, wealthy, and wise: socioeconomic status, poor health in childhood, and human capital development. J Econ Lit 47:87–122
Currie J, Hyson R (1999) Is the impact of health shocks cushioned by socioeconomic status? The case of low birth weight. Am Econ Rev 89:245–50
Currie J, Stabile M, Manivong P, Roos LL (2010) Child health and young adult outcomes. J Hum Resour 45:517–48
Cutler DM, Lleras-Muney A (2006) Education and health: evaluating theories and evidence. NBER working paper, No 12352
Deaton A, Paxson C (2001) Mortality, education, income, and inequality among American cohorts. In: Wise DA (ed) Themes in the economics of aging. University of Chicago Press, Chicago
Figlio DN, Guryan J, Karbownik K, Roth J (2013) The effects of poor neonatal health on children’s cognitive development. Am Econ Rev 104:3821–55
Gale C, Martyn C, Cooper C, Sayer A (2007) Grip strength, body composition and mortality. Int J Epidemiol 36:228–35
Griliches Z (1979) Sibling models and data in economics: beginning of a survey. J Polit Econ 87:37–64
Isacsson G (2004) Estimating the economic return to educational levels using data on twins. J Appl Econ 19:99–119
Lichtenstein P, de Faire U, Floderus B, Svartengren M, Svedberg P, Pedersen NL (2002) The Swedish twin registry: a unique resource for clinical, epidemiological and genetic studies. J Intern Med 252:184–205
Lindqvist E, Vestman R (2011) The labor market returns to cognitive and noncognitive ability: evidence from the Swedish enlistment. Am Econ J: App 3:101–28
Lundborg P, Nilsson A, Rooth D-O (2014a) Parental education and offspring outcomes: evidence from the swedish compulsory school reform. Am Econ J: App 6:253–78
Lundborg P, Nilsson A, Rooth D-O (2014b) Adolescent health and adult labor market outcomes. J Health Econ 37:25–40
Lutter CK, Mora JO, Habicht J-P, Rasmussen KM, Robson DS, Sellers SG, Super CM, Guillermo H (1989) Nutritional supplementation: effects on child stunting because of diarrhea. Am J Clin Nutr 50:1–8
Marmot M, Wilkinson RG (2006) Social determinants of health. Oxford University Press, Oxford
Martorell R, Yarborough C, Lechtig A, Habicht J-P (1975) Diarrheal diseases and growth retardation in preschool Guatemalan children. Am J Phys Anthropol 43:341–6
Mendolia S, Walker I (2014) The effect of non-cognitive traits on health behaviours in adolescence. Health Econ 23:1146–58
Metter E, Talbot L, Schrager M, Conwit R (2002) Skeletal muscle strength as a predictor of all-cause mortality in healthy men. J Gerontol A Biol Sci Med Sci 57:B359–65
Miller P, Mulvey C, Martin N (1995) What do twin studies reveal about the economic returns to education? A comparison of Australian and U.S. findings. Am Econ Rev 85:586–99
Miller P, Mulvey C, Martin N (2005) Birth weight and schooling and earnings: estimates from a sample of twins. Econ Lett 86:387–92
Pettersson A (1993) Matematik Åk 9 Den nationella Utvärderingen av Grundskolan Våren 1992. Huvudrapport, Skolverkets Rapport Nr. 15
Rubin DB (1986) Multiple imputation for nonresponse in surveys. Wiley, New York
Sander W (1995) Schooling and smoking. Econ Educ Rev 13:23–33
Sandewall Ö, Cesarini D, Johannesson M (2014) The co-twin methodology and returns to schooling—testing a critical assumption. Labour Econ 26:1–10
Smith JP (1999) Healthy bodies and thick wallets: the dual relation between health and economic status. J Econ Perspect 13:145–66
Smith JP (2009) The impact of childhood health on adult labor market outcomes. Rev Econ Stat 91:478–89
Worme JD, Lickteig JA, Reynolds RD, Deuster PA (1991) Consumption of a dehydrated ration for 31 days at moderate altitudes: energy intakes and physical performace. J Am Diet Assoc 91:1543–9
Acknowledgments
Research grants from the Centre for Economic Demography and from the Jan Wallander and Tom Hedelius Foundation are gratefully acknowledged. We thank two anonymous referees for their help and guidance.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Funding
This study was funded by the Centre for Economic Demography and the Jan Wallander and Tom Hedelius Foundation.
Conflict of Interest
The authors declare that they have no conflict of interest.
Additional information
Responsible editor: Erdal Tekin
Appendix
Appendix
Rights and permissions
About this article
Cite this article
Lundborg, P., Nilsson, A. & Rooth, DO. The health-schooling relationship: evidence from Swedish twins. J Popul Econ 29, 1191–1215 (2016). https://doi.org/10.1007/s00148-016-0598-8
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00148-016-0598-8