Modeling skewness with the linear stochastic plateau model to determine optimal nitrogen rates

Accurate modeling of skewness is needed to increase the actuarial fairness of crop insurance. We test Day's conjecture that crop yield skewness becomes negative as nitrogen rates increase and determine how well a linear response stochastic plateau (LRSP) production function matches the pattern of observed skewness using four long-term nitrogen experiments. Stillwater wheat is consistent with Day's conjecture, but the skewness for Lahoma and Altus wheat yields as well as Altus cotton yields are not. The LRSP assumes normal random effects and can explain only a small part of observed skewness, so a new LRSP with skew-normal random effects is introduced, which comes closer to explaining the observed skewness and should increase the accuracy of nitrogen rate recommendations. Negative skewness reduced optimal nitrogen rates and positive skewness increased optimal nitrogen rates.