Improving Agronomic Structure in Econometric Models of Climate Change Impacts

Economists are relying on agronomic concepts to construct weather or climate independent variables and improve the reliability and efficiency of econometric models of climate change impact on U.S. agriculture. The use of cumulative heat measures in agronomy (growing degree-days), has recently served as a basis for the introduction of plurimonthly calendar heat variables in these models. However, season-long weather conditions seem at odds with conventional agronomic wisdom that emphasizes crucial differences in crop stage sensitivity to environmental stress. In this paper I show that weather variables matched to key corn development stages provide an enhanced and more stable fit than their calendar counterparts. More importantly, the proposed season-disaggregated framework yields very different implications for adaptation than its calendar counterparts as it indicates that most of the projected yield damages are accounted during the flowering period, a relatively short period in the crop cycle. This should open the door to more advanced yield models that account for additional possibilities of adaptation and thus provide a more nuanced outlook on the potential impacts of climate change on crop yields.