Maternal education, child health and nutrition — evidence from China’s compulsory education law

ABSTRACT

The effect of maternal education on child health in China is studied by exploiting a change in the Compulsory Education Law from 1986. Data from four waves (2004–2011) of the China Health and Nutrition Survey are used. Variations across cohorts and provinces induced by the timing of the reform are used as instrumental variables to account for the endogeneity of education. Results show that mothers’ educational attainment has significant and sizable positive effects on their children’s health, as measured by height, weight and BMI Z-scores. Maternal education is associated with improved food structure (increased consumption of meat, egg, and dairy products) and healthier nutrient intake (increased share of calorie obtained from fat and proteins) of their offspring. Mothers who were exposed to the education reform are more knowledgeable about healthy diets and healthy behaviours than those not exposed and hence had fewer years of schooling.

KEYWORDS:

• Maternal education
• child health and nutrition
• educational reform
• China

JEL CLASSIFICATION:

• I12
• I21
• I38
• O12

Acknowledgments

We thank UNC Carolina Population Center for providing the data.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

¹ In some regions, children enrol in school at 7 years old.

² In fact, the educational reform was not strictly enforced after the enactment. Nine-year compulsory education was not provided free of charge in rural China until 2006 and in urban China until 2008 (Xiao, Li, and Zhao 2017). This applied to the first six years as well as to the last three (junior high school) years.

³ The reform was implemented at different paces, not only across provinces but also across cities and counties within provinces (Liu 2007). However, detailed information about this is not available, and so, provincial-level effective year, as shown in Table 1, is used in the subsequent analysis.

⁴ Due to the low proportion of within-variation of health measurements, we cannot apply fixed effects model to address endogeneity concerns.

⁵ Height and weight measured by a physician are less subject to measurement error than self-report indicators (Lindeboom, Llena-Nozal, and van der Klaauw 2009).

⁶ Weight-for-age reference data are not available beyond age 10 because this indicator does not distinguish between height and body mass in an age period where many children are experiencing the pubertal growth spurt and may appear as having excess weight when in fact they are just tall (de Onis 2006). Hence, in our sample there are considerably fewer observations on children’s weight.

⁷ Household-level per capita income was winsorized in the analysis: that is, the lowest and highest percentile values were set to the values of 1th and 99th percentiles, respectively.

⁸ We cannot estimate the effects of father’s schooling on child health and nutrition outcomes due to the absence of valid instruments. The preference for sons implies males have higher educational attainment before the reform. Hence, the effects of the compulsory education law on educational attainment are smaller for males than their female counterparts.

⁹ Table A2 reports the results by using regular OLS method without considering endogeneity problems. In general, all OLS estimates are smaller than IV estimates, displayed in Table 3.

¹⁰ As maternal education and household income are positively correlated, we experimented with excluding household per capita income from the baseline model. Estimates of the effect of maternal education changed only marginally as a result of the exclusion. Thus, the impact of maternal education is independent of household income. Results are available upon request.

¹¹ This can be compared to previous studies on Chinese data. Chen and Li (2009) found that an additional year of maternal education increases HAZ by 0.091 standard deviations for children aged 0 to 4 years. Cui, Liu, and Zhao (2019) report OLS estimates suggesting significant associations between maternal education and a higher HAZ (0.058) of adolescents aged 10 to 19 years. For another developing country, Turkey, Güneş (2015) also found sizable impacts of an educational reform increasing mothers’ years of schooling.

¹² The first-stage results also indicate that the educational reform increased the probability of completing junior high school. However, for WAZ, the F-statistic scores are below 10, indicating that the instrumental variable is weak.

¹³ It should be noted, that the construction and application of the growth references in Zong and Li (2013) are based exclusively on data on urban Chinese children.

¹⁴ Although the use of the covariance matrix of disturbances improve the efficiency of the 3SLS estimators, 2SLS estimates would be consistent.

¹⁵ Naturally, improved educational level of mothers can also have had other impacts, which in turn can have had a positive effect on their children’s health. Examples are: they are older when they give birth and give birth to fewer, and they are more likely to work off-farm and learn new ways to raise their children, including an increased focus on healthy diet and behaviours.

Additional information

Funding

The work was supported by the National Natural Science Foundation of China [72003091]; Special Project of Ecological Civilization Construction Research in Humanities and Social Sciences of Nanjing Forestry University [A2021YB13]; Project of Philosophy and Social Science in Colleges and Universities of Jiangsu Province [2022SJYB0158].