Powering Up Child Growth: The Impact of Electricity Outages on Children’s Anthropometric Outcomes in the Kyrgyz Republic

ABSTRACT

Using household-level panel data and a coarsened exact matching procedure, this paper investigates the relationship between frequent electricity outages and the anthropometric outcomes of children. Specifically, we study the differences in the anthropometric outcomes of children aged under 5 in the Kyrgyz Republic. We find that children living in households with frequent outages have z-scores of height-for-age (HAZ) that are 0.37 units lower compared to the children living in similar households but without frequent outages. We also document that HAZ values of girls are more sensitive toward frequent outages, which indicates a possible presence of gender bias among Kyrgyz households.

KEYWORDS:

• Anthropometric outcomes
• child health
• electricity outages
• Kyrgyz republic
• transition economies

JEL CLASSIFICATION:

• I12
• I14
• J13
• P36
• Q53
• Q41

Acknowledgments

The financial support was provided by the Grant Agency of Charles University (grant number 327421). This paper is part of a project that has received funding from the European Union’s Horizon 2020 (GEOCEP) research and innovation program under the Marie Skłodowska-Curie grant agreement No 870245. We also thank the participants of the 40th International Energy Workshop (IEW) 2022, the 12th Biennial Conference of the Czech Economic Society (CES) in Prague 2022, the 23rd Annual Conference on Finance and Accounting (ACFA) in Prague, 2022, the 23rd Annual Conference Environmental Economics, Policy, and International Relations VSE-UK 2021, 22nd International Conference on Environmental Economics, Policy and International Environmental Relations VSE-UK 2020, the Sustainability Environmental Economics and Dynamics Studies (SEEDS) Annual Workshop 2020, the 1^st Online IAEE Conference 2021, the ArmEA Annual Meetings 2021, the International Autumn School on “Climate Policy and Energy System Transformation” in Freiberg 2021, 6th AIEE Energy Symposium on Current and Future Challenges to Energy Security 2021, and the CERCIS 2021 annual workshop at the University of Ferrara for many helpful comments. Responsibility for any errors remains with the authors.

Disclosure statement

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

Notes

1. Leroy et al. (2015) also argue that using HAZ is inappropriate to measure the changes in linear growth over time because they are constructed using standard deviations from cross-sectional data (population level catch up in height). Instead, they argue that using height-for-age differences (HAD) is more appropriate. In this paper, we are not interested in the changes in linear growth of the population over time. Instead, we are concerned about the possible effects of the frequent outages on the prevalence of children aged 5 and below to have a linear growth retardation or to be stunted which is defined universally by HAZ.

2. https://www.climatestotravel.com/climate/kyrgyzstan

3. https://www.climatestotravel.com/climate/kyrgyzstan.

4. According to the World Bank (2017) less than 1/5 of the population has access to central heating

5. Source: State Agency for Regulation of the Fuel and Energy under the Ministry of Energy and Industry of the Kyrgyz Republic http://regultek.gov.kg.

6. There are no weights assigned to individuals and households.

7. About 80.6% of initial sample of households that participate in the survey in 2010 also participated in the following three years that we use for the purpose of our analysis (2011–2013).

8. This continuity is crucial for capturing the time-invariant heterogeneity of households and children, which is essential for the HAZ and WAZ analysis of children below 60 months of age. Using data with wide gaps between waves is not feasible since most of the children under 60 months in 2013 will be above this age limit in 2016. Furthermore, none of the children present in 2013 will be in the 2019 wave.

9. We use global child growth standards for infants and children up to the age of 5 years introduced by the World Health Organization (WHO) in (2006). Please refer to the Introduction section for justification of using the anthropometric scores as well-being indicators.

10. As a robustness check (see, Appendix) we also perform all the analysis using a cutoff indicating the households that experience outages once a month and more (one cutoff below the one used in the main analysis), and a cutoff indicating the households that experience outages once a week and more (one cutoff above the one used in the main analysis). The effect of the outages is smaller in size when cutoff also includes the households that experience outages less often, indicating that the effect is driven mostly by the households experiencing outages more than once a month. In case when the cutoff includes only the households that experience outages more often (once a week and more) the effect is mainly statistically insignificant due to the low number of households that experience outages that often. Please, see Tables A1-A6 for more details.

11. The Hausman specification test rejects the Random effects model in favor of Fixed effects model at all conventional levels of significance.

12. In developing countries, height-for-age Z-scores have a nonlinear relationship with age, with older children having lower Z-scores than younger children, as nutritional and other deficits accumulate with age (Martorell and Habicht 1986). For further analyses we control for age in our fixed effect regression.

13. It should be noted that the absolute values of the L1 statistics mean less than comparisons between the matching solutions. In this sense, the L1 statistics works for imbalance as R-squared works for the model fit.

14. Some research suggests that the height of an adult is determined mostly in the age interval up to 24 months (or earlier), and is difficult to change thereafter (see for instance, Stein et al. 2010). We present regression results in the Appendix (see, Table A7) assessing the effect of frequent outages on the HAZ and WAZ indicators for the children exposed to the outages between 24 months and below (about 6.6% of the sample). The effect of frequent outages, in this case, becomes statistically insignificant (the effect of frequent outages exhibits statistical significance only in the case of HAZ indicator for the unmatched (full) sample).

15. We test for the difference between the two coefficients via Wald-test. The difference is statistically significant at 1%.

Additional information

Funding

This work was supported by the Grantová Agentura, Univerzita Karlova [327421]; Horizon 2020 Framework Programme.