The influence of weather extremes on the spatial correlation of corn yields

Annual production shocks at the farm-level are driven by year-to-year weather variability. While identifying drivers of these shocks is important and well-researched, little attention has been paid to the extent to which these shocks aggregate up to the regional or national level. Here, we provide a method for simultaneously modeling the mean, variance, and spatial correlation of crop yields in the presence of evolving technology. Our approach allows one to condition spatial correlations on variables of interest—such as weather—in a straightforward manner. An application to state-level Iowa and Illinois corn yields provides evidence that spatial correlations roughly double in both good and bad weather years relative to normal years. Furthermore, we consider several functional forms for conditioning spatial correlations on weather and find that less flexible relationships generate misleading results as they vastly underestimate the degree of correlation in bad weather years. These findings have important implications for the climate change, food price volatility, crop insurance, and yield modeling literatures.