Effects of the Fertilizer Subsidy Program on Fertilizer Use, Farm Productivity and Crop Sales in Mali

Mali’s most recent phase of fertilizer subsidies began during the global food crisis of 2008/09, but there is little evidence-based information concerning its effects. To generate information of potential use to policymakers in Mali, we implemented a survey to a random sample of 2400 extended farm family households in two major agroecological zones of Mali—the Delta du Niger and the Plateau de Koutiala. In this paper, we test the effects of the fertilizer subsidy on total fertilizer applied, yield, target cropincome and quantity of all crops sold. We find that subsidized fertilizer accounts for most of thetotal fertilizer applied by farmers, suggesting that in some instances it is displacing demand forcommercial fertilizer. Average fertilizer use rates in kgs appear to be below the recommended quantities for all target crops, despite subsidy receipts. In future research, we intend to verify these findings converting units to nitrogen nutrient kgs, which standardizes across fertilizer types and permits a more exact comparison. We compare regression results across several econometric approaches to improve their reliability. Each econometric approach provides evidence that considering all crops combined, the fertilizer subsidy has a positive effect on total fertilizer applied per ha, yields, and crop revenues of target crop, as well as on quantities of all crops sold. However, important differences are observable among crops. On average, subsidy effects on millet and sorghum outcome variable were weak or not statistically significant. Average subsidy effects on all outcome variables were strong for rice. Average subsidy effects were strong on maize yields, but not revenues or sales of other crops. For cotton, the subsidy only allowed an increase in the mean quantities of fertilizers used without improving productivity or other outcomes. The dose-response estimation suggests efficiency intervals in which the fertilizer subsidy has a positive marginal effect on fertilizer use, productivity and crop sales. These also vary from one crop to another, but are estimated only for rice, maize and cotton given that mean effects are not significant for sorghum and millet. We find no positive marginal effect of subsidized fertilizer on yields below 65 kg/ha for rice and 87 kg/ha for maize. The graphs also show peaks at high levels of subsidized fertilizer for both crops, with declining marginal returns after that point. For rice, marginal effects on rice revenues have a similar shape to that of the yield effect, and effects on quantities of all crops sold are strong through much of the range of subsidized fertilizer applied in the data. This last result is observable also for maize at higher levels of the subsidy, suggesting some spillovers from rice and maize to non-target crops. No positive effect on cotton yields, cotton revenues or quantities of all crops sold is discernible regardless of the level of subsidization. The fertilizer subsidy in Mali is currently designed to target particular crops and enhance their productivity. We conclude that the design could be made more efficient by either reconsidering target crops or targeting the subsidy according to different criteria. We consider that applying the subsidy to cotton represents a deadweight loss—that is, a public expenditure that leads to no discernible supply shift. This last finding could be season dependent, or result from factors we could not measure in this analysis—such as cottonseed quality.