A stochastic dominance method for incorporating yield monitor data into the hybrid and variety decisions of Argentinean farmers

This paper presents a computer-based method that uses stochastic dominance techniques to combine yield monitor data and information from traditional hybrid and variety trials in guiding farm level seed choices. This choice is important since it affects farm profits and risks. The proposed method uses a non-parametric approach that takes into account information that is not easily summarized by statistical parameters, and would work with any type of distribution. The proposed method selects cultivars that have smaller probabilities of low yields, thus are more risk efficient than the ones selected by traditional methods. Therefore, the proposed method can potentially improve the choice of hybrids and varieties. An example of how this method can be applied is developed using data from the South-East region of the Buenos Aires province, Argentina. This country has more combine yield monitors than any other Latin American country and yield trials are conducted every year by farmers, farmer associations and research institutions. Therefore, the technology and data required to implement the proposed method are widely available. Advantages of the proposed method are discussed and presented graphically, and limitations are identified. The proposed method can be easily incorporated in computer software that automates the statistical procedures allowing its use by a broad sector of agriculture.