Creating An Agent-Based Model To Establish Food Security In Malawi

Malawi’s agricultural sector is unable to establish food security for the majority of the population. Food security in Malawi is defined by access to maize, the staple food for over 96% of the population. Current rates of domestic maize production are too low to sustain the increasing population. Additionally crops are jeopardised by decreasing soil fertility and increasing occurrences of floods and drought brought about by climate change. Malawi’s agricultural sector is dominated by smallholder farmers. Increasing the yield of these smallholder farmers could alter the overall pattern of food production and food security. Increased yields creates households establishes food secure households who can generate income from the sale of surplus maize at market. A literature review identified fertiliser as a key solution to increasing maize yield for smallholder agricultural households. Using the paradigm of agent-based modelling, the Malawi Agent-Based Model (MW-ABM) was constructed to model the impact of fertiliser subsidy on food security, harvest and income of agricultural households. MW-ABM consists of a human-environment landscape, smallholder farmers are modelled as household agents while the agricultural fields are landscape patches. The integration of GIS data spatially-enabled the model; the spatial aspect is a vital component of agricultural activities. Five scenarios were simulated, each modelling a different blanket fertiliser subsidy programme lasting ten years. Anomalies in the results indicate that MW-ABM is still at the initial stage of construction; it currently poses more questions than it answers. The model is a starting point and further development should incorporate environmental factors and human decision-making. The complexity of this requires the modelling of social and cultural factors affecting human decision-making and risk assessment, and environmental factors of yield dynamics. An outline of data required for the next phase of model development is provided, illustrating the data intensive nature of agent-based modelling. The research questions whether agent-based modelling is an actually an appropriate method for examining food security.