Assessing the sensitivity of climate change impacts in agriculture to the climate–economy functional form

Due to anthropogenic global warming, the intersection of climate and the economy has garnered considerable attention in recent decades. The growing economics literature in this area is chiefly empirical and showcases a swath of statistical modelling approaches seeking to estimate the effect of temperature and precipitation on a variety of outcomes. This paper evaluates the consequences of alternative modelling choices in estimating the average and distributional impacts of weather and climate change on broad agricultural outcomes. I first develop a long-run dataset of county-level weather and climate metrics, including hourly temperature measurements, to assess differences in estimated marginal effects as well as non-marginal damages using climate projections. Model specifications including flexible, nonlinear terms that allow the marginal effects of temperature to vary across space reveal more intuitive results and tend to perform best in terms of model selection criteria. Notably, when aggregating county-level projected damages to discern an average annual impact for the entire United States, findings are consistent across models. However, methodologies like degree days and temperature binning-predominant in contemporary literature-manifest geographically conflicting estimates. These results have important implications for targeted policy intervention and equity considerations in benefit-cost analysis.