Aggregate Temperature Measures and the Overestimation of the Impact of Global Warming on Crop Yield

Existing studies generally use ``aggregate'' temperature measures, such as mean temperature, degree-days, temperature bins, and piece-wise linear function within the growing season, to estimate the impact of global warming on crop yield. These temperature measures blend temperatures from different phenological stages of crop growth and thus implicitly assume that temperatures are additively substitutable within the growing season. However, this assumption contrasts with agronomic knowledge that crops are more sensitive to temperatures in certain phenological stages. Utilizing a unique site-level data on the detailed phenological stages of major crops in China, combined with crop production data and daily weather data, we develop an econometric model with stage-specific temperature measures. We then compare our estimates with models using traditional aggregate temperature measures, and find that adopting an aggregate temperature measure could overestimate the damage of global warming on crop yield up to two times that estimated using stage-specific temperature measures.