Department of Agricultural, Food, and Resource Economics Feed the Future Innovation Lab for Food Security Policy

William J. Burke, Emmanuel Frossard, Stephen Kabwe, and Thomas S. Jayne. 2016. Understanding Fertilizer Effectiveness and Adoption on Maize in Zambia. Feed the Future Innovation Lab for Food Security Policy Research Paper 30. East Lansing: Michigan State University

EXECUTIVE SUMMARY
As populations continue to rise and land becomes scarcer in Africa’s rural areas, there is increasing urgency for farmers to adopt land management practices that sustainably raise land and labor productivity. Considerable effort has focused on promoting inorganic fertilizers, but it is increasingly recognized that smallholder farmers’ demand for fertilizer can be depressed by soil conditions that reduce crop response to and the profitability of fertilizer use. This article quantifies the impacts of soil characteristics on maize response to fertilizer in Zambia using a nationally representative sample of 1,453 fields. In addition to economic and farm management surveys, composite soil samples were collected and analyzed for several characteristics at the Zambia Agricultural Research Institute. Soil’s role in agricultural production and fertilizer efficiency is more nuanced than most economic literature has acknowledged. We believe ours is the first model in economic literature that simultaneously allows for the effects of multiple soil characteristics. We estimate critical threshold effects on yield response to fertilizer to be between pH levels of 5.4 and 5.6, soil organic matter levels of 1.2-1.4%, and find significant soil texture―and cation exchange―related thresholds. Depending on these soil characteristics, average maize yield response estimates range from insignificant (0) to 5.7 maize kg per fertilizer kg. We estimate fertilizer use on maize is not profitable at commercial prices for the majority of Zambian farmers (under current practices). Even ignoring transfer costs, about 80% of fertilized maize fields still have an estimated average value-cost-ratio for fertilizer less than one at commercial prices. To the best of our knowledge, the flexibility of our model and data with this scope of geography and content are novel contributions to the literature.