This work is co-authored by Mistry Malcolm, De Cian Enrica and Wing Ian Sue. This work is co-authored by Mistry Malcolm, De Cian Enrica and Wing Ian Sue.

A rapidly growing literature employs historical observations or pseudo-data generated by Global Gridded Crop Models (GGCMs) to empirically estimate reduced-form crop yield responses to meteorology. The resulting fitted response surfaces, when forced by Earth System Model (ESM) simulations of future climate, function as computationally tractable statistical emulators of climatic shocks to crop productivity that can be coupled with Integrated Assessment Models (IAMs) to evaluate the broader energy and economic  implications of the agricultural climate change impacts.

We develop a statistical emulator of the yields of four major cereal crops (maize, rice, wheat, and soybeans), over 1972-2099 under two climate change scenarios (Representative Concentration Pathways 4.5 and 8.5), using a combination of six GGCMs and one ESM, under rainfed cultivation regimes. We characterize the reduced-form response functions to temperature and precipitation, and assess their stationarity across time, crop suitability zone and different models. We demonstrate how adaptation plays a contrasting role across GGCMs in reducing the potential negative percentage yield shocks on crops in future. Our simple and flexible statistical emulator holds considerable potential as a diagnostic methodology to elucidate uncertainties in the processes simulated by GGCMs, and to support the development of climate impact inter-comparison exercises within the integrated assessment modelling community.