Does Health Aid Reduce Infant and Child Mortality from Diarrhoea in Sub-Saharan Africa?

abstract

Achieving sustained improvements in health outcomes remains a challenge for sub-Saharan Africa, where diarrhoea remains a leading cause of death in children under the age of five. This paper examines the impact of foreign aid to the health sector on diarrhoea mortality in children under five in 47 sub-Saharan African countries, using panel data on the sectoral allocation of official development assistance in conjunction with country-level data on health outcomes. After controlling for fixed effects and the potential endogeneity of health aid, we find that increased health aid and increased public health expenditure are associated with lower diarrhoea mortality in children under five. In addition, health aid increases government spending on health, suggesting that the overall impact of health aid on diarrheal death rates could exceed the direct effect. Furthermore, increased access to improved sources of water and sanitation are important in reducing child mortality from diarrhoea.

Acknowledgements

We would like to acknowledge the constructive comments and suggestions of two anonymous reviewers. All data and code used in this analysis are available from the authors upon request.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. https://data.unicef.org/topic/child-health/diarrheal-disease/.

2. These figures are taken from the World Bank’s World Development Indicators (WDI) database.

3. Tests for the endogeneity of health aid indicate that an instrumental variable model is appropriate. The results from these tests can be provided to the reader on request.

4. Data from the WHO shows that the rate of increase in access to improved sources of water and sanitation is significantly below the rate of population increase in SSA.

5. Access to an improved water source refers to the percentage of the population using an improved drinking water source, defined by WHO/UNICEF as one that, by nature of its construction or through active intervention, is protected from outside contamination, in particular from contamination with faecal matter, such as piped water on premises, public standpipes, tube wells or boreholes, or protected dug wells.

6. Between 1990–2010 the UNDP education index is available for every fifth year. We interpolate to generate annual values for the intermediate years.

7. We test for under-identification, weak-identification, and over-identification of the instruments. The results of these tests are discussed in detail in the results section. The specifications meet the conditions for the relevance and validity of the instruments for aid when population access to improved water sources is used as a control in the specifications for all age groups.

8. The WHO uses different methods to estimate neonatal (1–27 days) and post-neonatal (1–59 months) causes of death. In countries with good quality vital registration (VR) data, the cause distributions are estimated directly from the VR data. The only data points in our sample for which this is true are the neonatal diarrhoea mortality rates for South Africa. For lower mortality countries with poor quality VR data, the distributions are predicted from a regression fit to data from countries with high-quality VR data. Access to water is a covariate in this regression. In our sample, only Cape Verde and Seychelles fall into this group, and we estimate separate regressions with and without these two countries as a robustness check. Our results do not change when these countries are excluded from the regression. For higher mortality countries with poor quality VR data (all other countries in our sample), the distributions are predicted from a regression fit to data assembled from studies of causes of death in high mortality countries. The covariates included in these regressions are female literacy, Gini coefficient, neonatal mortality rate, infant mortality rate, under five mortality rate, low birth weight, GNI per capita, antenatal care coverage, percentage of births with skilled birth attendance, general fertility rate, BGC vaccine coverage, PAB vaccine coverage, and indicator variables for world regions. To avoid spurious results, we do not include these variables in our regression models. For more information, see http://www.who.int/healthinfo/global_burden_disease/GlobalCOD_method_2000_2015.pdf?ua=1.

9. The implied elasticity is computed by multiplying the regression coefficient by the ratio of the mean of the dependent variable (that is, the average death rate for that age cohort) to the mean of the explanatory variable (for example, mean health aid).

10. The results for these regressions are available on request.

11. The Sargan-Hansen test statistic for this regression does not permit the rejection of the null; we provide the results from both the fixed effects and random effects estimations and find that the results are qualitatively similar.