Directional aspects of crop yield correlation: Implications for insurance rating and contract design

We estimate directional anisotropic models of linear correlation for crop yields. We find that directional distances (i.e., the degree to which yield correlation decays depending on direction) tend to differ from one another, with east to west distances being “longer” than north to south distances (i.e., correlation tends to decline slower in the east to west direction), at least for the major crops of corn and soybeans. Results for wheat are somewhat ambiguous. However, in all cases, average yield aggregates made up of east to west linked counties tend to be more accurate than is the case for north to south linkages. Implications for rating and contract design in the US area yield federal crop insurance program are offered.