The "Ricardian" technique for projecting climate change impacts on agriculture has generated an unusual amount of critical attention. Ricardian climate sensitivities are typically viewed as the necessary and static result of exclusively local economic and biophysical conditions. In this paper, six spatial econometric models are estimated to explore how human-environment relationships associated with climate sensitivities have varied over space and time in the Great Plains, 1969-1992. Results indicate that spatial effects, such as extra-local communication processes and proximity to and regulation of Ogallala irrigation water, are important influences on climate sensitivities. Projected climate change impacts also vary significantly with the scale, location, and time of analysis. Under a hypothetical climate change, at the county scale, land values would decline (by up to one-third) in the western counties, but increase (by up to one-half) in the eastern counties. In some cases, the projected impacts for a given county change algebraic sign or order of magnitude during the study period. At the regional scale, impacts are significantly higher in the early years (a projected increase of about 5 percent of regional land values, similar to$7 billion [1992$]) than in the later years (an increase of about one-half of one percent, similar to$0.7 billion). These results suggest that, suitably modified, the Ricardian framework can be used constructively to explore subtle yet important social dimensions of dynamic climate risk, and that on balance the Great Plains system of agricultural production, despite a heterogeneous picture of projected impacts, appears to be increasingly vulnerable to the effects of climate change.
