An Empirical Analysis of Fertilizer Use in Pakistan

One of the stylized facts in developing countries is the wide variation in agricultural per-acre input and output levels across farm households. Here we focus on the variation in the level of chemical fertilizer applied, in particular to high-yielding wheat. The model used to explain the variation in fertilizer input across farms incorporates the process by which 'zeros' are generated, in particular differentiating between households who do not apply because of 'lack of access' and those who do not apply out of choice. We find that access is positively correlated with farm size but that, among users, per-acre fertilizer levels decrease with farm size. We interpret the latter by appealing to input choice under uncertainty. With this interpretation, our results suggest that, although larger farms have a higher level of non-stochastic wealth which, in the presence of decreasing absolute risk aversion, has the effect of increasing fertilizer intensity with farm size, the presence of increasing relative risk aversion means that the net relationship between farm size and fertilizer intensity is negative. This latter affect appears to dominate any tendancy for higher fertilizer intensity on larger farms arising from better knowledge or irrigation facilities.