Food for survival: diagnosing crop patterns to secure lower threshold food security levels in farm households of Burundi

Background: Burundi is one of the world's poorest countries, coming last in the Global Food Index (2013). Yet, a large majority of its population depends on agriculture. Most smallholder families do not produce enough to support their own families. Objective: To estimate the optimal crop mix and resources needed to provide the family with food containing sufficient energy, fat, and protein. Methods: This study uses mathematical programming to obtain the optimal crop mix that could maximize output given the constraints on production factor endowments and the need to feed the household. The model is calibrated with household-level data collected in 2010 in Ngozi Province in northern Burundi. Four models are developed, each representing a different farm type. The typology is based on 2007 data. Model predictions are compared with data collected during a revisit of the area in 2012. Results: By producing a smaller number of crops and concentrating on those in which they have a comparative advantage, and trading produce and input with other farms, large and medium-sized farms can improve their productivity and hire extra workers to supplement family labor. Predictions of crops to be planted coincided to a high degree with those that farmers planted 2 years after our survey on newly acquired plots. Conclusions: Despite land scarcity, it is still possible for households that own land to find optimal crop combinations that can meet their minimal food security requirements while generating a certain level of income. Nearly landless households would benefit from the increased off-farm employment opportunities. With only 0.05 ha of land per capita, the annotation Nearly Landless is used to highlight the limited access to land observed in this farm category