Do input subsidies crowd in or crowd out other soil fertility management practices? Evidence from Zambia

It is recognized that inorganic fertilizer, as is commonly distributed in large-­‐scale input subsidy programs, must be used along with soil fertility management (SFM) practices in order to maximize its efficacy. We use nationally representative data with 8,839 household observations to assess the impact of the Zambian input subsidy program on the use of five SFM practices: (i) manure and/or compost application, (ii) soil erosion preventative measures, (iii) minimum tillage, (iv) rotations between cereals and legumes, and (v) leaving land fallow. We estimate at the household level the effect of subsidized fertilizer on probability of adoption of each practice using a maximum likelihood probit model and the effect on number of hectares under each practice with a maximum likelihood Tobit model. The endogeneity of fertilizer distribution is tested and controlled for using the control function approach. We find a small but positive statistically significant crowding in effect of receiving subsidized fertilizer on all SFM practices except for fallow land, where we report a statistically significant crowding out effect of larger magnitude than estimated for the other practices (a decrease in hectares equal to 11.3% of the unconditional mean hectares of fallow land per household).