Jeremy Foltz, Ursula Aldana, and Paul Laris (2011). Prepared for the NBER African Successes Conference, Zanzibar, Tanzania, August 3-5.

Using micro-level panel data from farmers, this study argues that Mali's impressive growth in maize production can be attributed to improved fertilizer usage rather than to changes in seed varieties or management practices.

Mali's experience with maize over the past 50 years is one of African agriculture's success stories.  Since its independence in 1960,Malian maize production has increased by a factor of ten.  Both land area under cultivation and yields have increased.  Maize production has grown to a point where it is now on par with the traditional Sahelian crops of millet and sorghum.  Though they are currently low by world standards at 2 tons per hectare, Malian maize yields have grown faster than those in Senegal, Mauritania, and Niger yet not as fast as those in Burkina Faso. 

Much of the economics literature shows that there are high returns to fertilizer in Africa, yet many farmers do not use it.  While agricultural productivity growth stagnated for several decades, recent evidence suggests that has increased over the past decade.  Some scholars argue that fertilizer usage has been a key determinant of this recent growth, and that maize has the potential to catalyze a green revolution.  However, this is not a topic with a clear answer.  Several studies have shown that the returns to fertilizer usage and maize adoption are heterogeneous across various regions.

In this paper, the authors pose the question of, “What has pushed the great expansion of maize production in Mali?” (2).  In particular, is this expansion due to inputs such as fertilizer or changes in seeds and management?  They address this question with a new dataset of maize production in Sikasso, a southern region of Mali, over the past 20 years.  Collected by the Malian agricultural research center Institut d'Economie Rurale, the dataset began with annual observations on 149 farmers in 1988.  Due to sample attrition, there were only 84 farmers in the sample by 2008, which does pose a problem for the study.  This panel data is remarkably detailed, including daily rainfall, fertilizer by type and crop, tractor and plow usage, and some data on labor.  This dataset was supplemented by interviews on farmer decision making.

Using fixed effects estimation, the authors find that fertilizer has heterogenous effects on Malian maize yields.  It has the greatest impact for households with a high percentage of literate members.  Additionally, yield growth seems to have been principally driven by fertilizer adoption rather than seed and management improvements.  The authors comment that “This is not to say that farmers did not adopt new technologies, but rather the maize revolution came as a sequential adoption process in which farmers adopted parts of a package in succession: seed first, appropriate levels of organic and chemical fertilizer later” (23). 

The heterogeneity outlined by this study is important—if it were ignored, econometric estimation would wrongly attribute the growth in Malian maize yields to seed and management changes and not to fertilizer.  The authors note that “Such an underestimate of the returns to fertilizer could seriously call into question public policies such as the Malian government's current program to subsidize fertilizer” (23). 

An additional interesting point that this micro-level panel data unearths is that farmers make rational and sophisticated decisions about fertilizer based on their soil conditions.  The authors report that “farmers seem to respond to reduced fertility of their soils, as happens with cotton cultivation, with increased applications of fertilizer.  This suggests a sophistication in African farmer knowledge that goes beyond that commonly suggested in the economics literature.  Malian farmers are making both temporally and dynamically rational decisions about the fertility of their soils” (23).