Abstract
The paper studies the effect of crop output value and livestock ownership on the nutrition of children, adolescents and adults in agricultural households. Using anthropometric data to measure nutritional status, this paper finds that both crop values and large livestock ownership have positive and significant effects on the nutrition of children under age 10. The effects persist after controlling for household socioeconomic status. Higher crop values and ownership of livestock are linked to better long-term indicators of nutrition (height-for-age) among the youngest children and better short-term indicators (BMI-for-age and weight-for-age) among older children. The effects also vary between boys and girls.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Acknowledgements
The completion of the paper would not have been possible without the invaluable guidance of Paul Winters. I am also thankful for the comments provided by participants in the Agriculture and Nutrition Workshop that took place at the World Bank on 7 June 2013 and for the critical comments of two anonymous reviewers. The remaining errors are all mine.
Notes
1. The Tropical Livestock Unit (TLU) is equivalent to 250 kg live weight and it is used to standardise live animals by species mean live weight (disregarding differences in breed, sex, age and health). We use the following conversion factors: cattle (0.70), sheep and goats (0.10), donkey and horses (0.50), pigs (0.25) and poultry and rabbits (0.01).
2. The use of household level fixed effects in estimating the nutrition equations further restricts the analysis to household with at least two children (adults) in each age group.
3. For children under 24 months, height is measured in a recumbent position, while for older children height is measured in standing up position. For simplicity, throughout the paper we refer to the indicator as height-for-age.
4. The anthropometric indicators for children ages 0 to 5 are calculated using the zscore06 command in STATA which incorporates WHO 2006 child growth standard. The anthropometric indicators for children ages 5 to 20 are calculated using WHO STATA code.
5. There were issues with the birth and interview date in the data we use. We had to make some strong assumptions in order to calculate preschoolers’ age in months, which is needed for the calculations of the anthropometric indicators. The indicators are sensitive to the age of children, especially at very low ages. However, comparisons of the sample means of the anthropometric indicators to the means of the same indicators obtained from other sources paint a similar picture and reduce measurement error concerns.
6. Household fixed effects are based on the household identifiers as of the first year of the survey, that is households that split between the survey waves are still treated as one in the sense that they share a common background. It is likely that for such households the unobserved household factors that may be correlated with the error term are similar
7. Variables with limited within-household variation are not precisely estimated when household fixed effects are included. Such variables are the gender and education of the household head, and parents’ education. These are excluded from the model, but their exclusion does not change significantly the estimates of the explanatory variables of interest (namely, agricultural production).
8. We impute the height of mother and father who are not present at the time of the survey (mostly because they live outside the household or are dead) and include a control for whether they were present at the time of the survey.
9. However, with or without controls for SES, the coefficients on the agricultural variables of interest are very similar (see Table A2, Table A5, Table A6 and Table A7 in the Online Appendix).
10. The main results for the link between agricultural production and nutrition hold regardless of whether we keep pure pastoralists or not. Table A4 in the Online Appendix reports the results for the full set of anthropometric indicators for all children under five excluding pure pastoralists (the variable ‘No crops’ is therefore dropped).