Agricultural correlates of aggregate nutritional outcomes in Malawi: District-level rank analysis

Abstract

Obtaining empirical evidence on the importance of agriculture in determining nutritional status is difficult, as there are few data sets that contain information on both nutrition outcomes and the agricultural livelihoods of survey respondents. Here we take advantage of the fact that the Demographic and Health Survey fielded in Malawi in 2010, which collected detailed information on nutrition outcomes, used comparable survey strata to a household consumption and expenditure survey in 2010–2011 that collected extensive information on agricultural production. While the samples for the two surveys involved different households, the results of both are representative at district level. Using a non-parametric rank correlation approach, we use the district-level results from the surveys to examine whether there are associations between five aggregate nutritional outcomes and characteristics of agricultural production in 27 districts of Malawi. The analysis is then extended to explore other non-agricultural determinants of nutritional status. Limited associations between aggregate agricultural production and aggregate nutritional outcomes were found, suggesting that the direct relationship between the two is relatively weak in Malawi. Moreover, the sign of many of the associations runs contrary to expectations: in some cases, strengthened agricultural livelihoods are negatively related to nutritional outcomes. When the analysis is extended to consider non-agricultural determinants of nutritional status, slightly stronger associations are found – aggregated measures of women’s empowerment, the availability of food and micronutrients and measures of household wealth are positively related to aggregate nutrition outcomes. Nonetheless, this analysis should be treated as exploratory. It is presented to stimulate more detailed examination of the association in Malawi between nutrition outcomes and agriculture using individual- and household-level data.