Rainfall shocks and child health: the role of parental mental health

ABSTRACT

This study examines the impacts of rainfall shocks on child health in Vietnam. It uses Young Lives data matched with province level climate data covering the period 1970–2014. Existing literature demonstrates that shocks can impact on child health by reducing household income or through the incidence of disease. This paper identifies and confirms a third mechanism: shocks impacting on parents’ mental health which, in turn, reduce children's physical health. We find that one unit increase in parental mental health caused by rainfall shocks will increase the probability of a child being underweight by 0.976. Using an instrumental variable strategy, we can interpret these results as causal. We instrument parental mental health with a variable that captures whether the adult has been a victim of a crime. We also find that households that receive support, from both formal and informal channels, are less vulnerable to rainfall shocks, in terms of reducing negative health outcomes.

KEYWORDS:

• Rainfall shocks
• parental mental health
• child health
• Vietnam

Acknowledgements

The views expressed here are those of the authors. They are not necessarily those of Young Lives, the University of Oxford, DFID or other funders.

Disclosure statement

No potential conflict of interest was reported by the authors. The data used in this publication come from Young Lives, a 15-year study of the changing nature of childhood poverty http://www.younglives.org.uk.

Notes on contributors

Trong-Anh Trinh is a research officer at the school of Economics, Finance and Marketing, RMIT University. His areas of research include applied environmental economics, development economics and health economics.

Simon Feeny is a Professor at RMIT University, Australia. He has 20 years' experience as a development economist. Professor Feeny has been awarded more than US1.5 million in funding and has undertaken work for the Australian government's Department of Foreign Affairs and Trade, the United Nations, the South Pacific Forum Secretariat, the ASEAN Secretariat, Oxfam Australia, World Vision International, the Fred Hollows Foundation, Plan International and the Centre for Poverty Analysis in Sri Lanka. Professor Feeny has more than 80 academic publications and has produced over 25 reports for industry. He is currently an Associate Editor of the Journal of International Development.

Alberto Posso is a Professor of Economics at RMIT University, Australia. Professor Posso holds a PhD in Economics from the Australian National University (ANU) with specializations in labour economics, economic development, international trade and applied econometrics. Professor Posso has over 40 peer-reviewed publications and has authored reports for various governments and international organizations.

ORCID

Trong-Anh Trinh http://orcid.org/0000-0001-8102-7122

Simon Feeny http://orcid.org/0000-0002-5955-0045

Alberto Posso http://orcid.org/0000-0003-4226-7072

Notes

1 This study does not employ temperature to measure weather shocks due to the high correlation between temperature and rainfall. Also, rainfall conditions are closely linked to agricultural production through impacts on irrigation systems (see Trinh, 2018).

2 Most studies focus on impacts in developing countries since households in developed countries have greater resilience through free health care, and access to insurance and credit.

3 The Old cohort of Young Lives Project (about 1000 children) is not considered in this study due to missing information on parental and child health.

4 For more information on measurement process, see Fieldworker Instruction Handbook (available online at: https://www.younglives.org.uk/content/round-1-questionnaires).

5 Several studies employed the HAZ and WAZ directly in the model (Fichera & Savage, 2015; Humphries et al., 2017). In this study, we go one step further by using these indicators to identify stunting and being underweight. This is to prevent the problems of interpreting the sign and size of their coefficients. For example, while an increase in the weight-for-age z-score for a child at the lower end of the distribution would be viewed as favourable, this is not true for a child with an already high score at the upper end of the distribution.

6 In this study, we also use household expenditure as alternative proxy for income and find a similar result. The results can be obtained from the authors on request.

7 Information on parental mental health is only available in the first round of Young Lives.

8 Results are consistent when the impacts of droughts and floods are modelled in separate regressions. Results are available from the authors on request.

9 A Hausman test shows a preference for the fixed-effects model.

10 The critical value of 16.38 suggested by Stock and Yogo (2002) implies a rejection of the null hypothesis of weak instrument.

11 For each year during the period 1999–2010, Nguyen et al. (2013) compared birth weight information for 30 randomly selected children from the records of a district hospital in Vietnam with those obtained from the interviews with mothers. Differences were found to be very small. The overall means from the district hospital were 3.11 kg for boys and 3.05 kg for girls. The corresponding estimates from the mother reports were 3.14 and 3.06 kg.

12 Findings are consistent with support–drought interaction terms.

13 It should be noted that community health workers take an important role in improving child health. Therefore, it is likely that areas with more health workers will have lower probability of children being stunt and underweight. We test this hypothesis by using data of health workers at province level derived from the Statistical Yearbook of Vietnam. We run a simple regression model to see the impact of health workers on child health. The results are presented in Table A4 (Appendix) which confirms that children living in areas with more health workers have lower probability of being stunted and underweight.