Dynamic Crop Production Responses to Realizations of Weather and Production with Application in Jordan

Abstract

Uncertainty of output quantities and prices shape farming decisions across the globe. This axiom holds true in the semi-arid areas of Jordan too, where farmers must adapt throughout the calendar year to manage production risk due to the magnitude and timing of the onset of rains. In order to understand and model the response farming practiced by the Jordanian farmers where decisions are made in a dynamic setting with explicit time dimensions in a risky environment, a dynamic and stochastic modelling approach is required. Agricultural production decisions are shaped by the stochastic interactions of crop growth, weather, and financial outcomes. Moreover, crop production decision are made with due consideration of the livestock production decisions where both decisions are often made simultaneously - making a systems modelling approach a must. Incorporation of the biological responses with economic (dis)incentives, while intuitive, presents many challenges in modelling. Optimizing agents respond to current biological developments, with only past distributional knowledge (as opposed to perfect foresight into the future). The choice of the type and timing of production decisions is important for determining outcomes both in the short-run (e.g. yields, revenues, etc.), as well as in the long-run (e.g. soil moisture and soil organic matter). Producers choose which crops to plant, how much and when to apply inputs, whether to graze their animals on the crop/residue and when and what to harvest. These choices influence the profitability of production each year within a stochastic environment, and when incorporated in a single year model can easily be optimized. However, current agricultural production decisions also influence future profitability through their effect on soil properties, such as soil moisture or organic carbon, which are important for both the long term sustainability of the household, and avoiding environmental degradation. Optimization models analyzing intra-seasonal stochastic production and dynamic inter-seasonal resource management present a modeling challenge due to the often encountered curse of dimensionality, where model size increases exponentially with the number of time periods and stochastic events. In his paper, dynamic programming (DP) with a portfolio model is used to overcome the issues discussed above and develop an agricultural systems (AS) model. The AS model developed here values the production trade-offs of short- and long-run outcomes, within a stochastic choice model. This methodology allows for the dynamic testing and valuation of various production technologies, including conservation agriculture and crop varieties.