|Title||Drought risk and vulnerability of Moroccan dryland wheat farmers|
This dissertation is comprised of three essays on drought risk, coping, and vulnerability. The first essay develops two methods for assessing household vulnerability: one measures ex ante vulnerability based on estimated income distributions, and the other measures ex post vulnerability based on household asset positions and reported responses in a severe drought year. Unlike most vulnerability measures, neither relies on repeated cross-sections of consumption data, which are difficult to obtain. Both measures are used to assess the vulnerability of a sample of rural Moroccan households, making for a unique, side-by-side comparison of an ex ante and ex post vulnerability measure. Households deemed vulnerable according to both measures are then compared to their resilient counterparts. Many drought shock measures treat risk management as a static decision. The second essay demonstrates how households can manage risk through intra-seasonal crop management choices and quantifies the significance of such dynamic risk management. A dynamic optimization model of wheat production is calibrated using survey data on yields and costs, as well as rainfall data from Morocco. The model is then used to simulate drought income shocks, first under the assumption that farmers manage their crops statically, and then under the assumption of dynamic, intra-seasonal management. Comparison of the two measures shows that ignoring dynamic risk management behavior significantly overstates drought shocks. Crop rotation, the strategic sequencing of crops, may drive the on-farm crop diversity observed in many areas throughout the developing world. The third essay quantifies the intertemporal productivity benefits rotation effects) that form the basis of crop rotations, using uncontrolled field data from Moroccan farms. Where rotations are practiced, ignoring rotation effects can lead to erroneous conclusions about other potential cropping motives such as avoiding risk, easing input constraints, and being self sufficient. Quantifying rotation effects is the first step in disentangling these many motives. A novel maximum likelihood specification that accounts for both sample selection and heteroskedasticity is developed. Though estimated rotation effects are mostly insignificant and, at times, counterintuitive, the exercise highlights the non-trivial gains from modeling both sample selection and heteroskedastic errors, and presents a simple method for doing so.
|Favorite||ADD TO FAVORITE|