This repository contains the agricultural input data of 12 crops used in the article analysis. The agricultural inputs used are (unit in parentheses): nitrogen (kg/ha) phosphorus (kg/ha) potassium (kg/ha) machinery (1000 metric horsepower cv) herbicides (original kg/ha, for analysis rescaled by dividing with 97.5th percentile) fungicides (original kg/ha, for analysis rescaled by dividing with 97.5th percentile) insecticides (original kg/ha, for analysis rescaled by dividing with 97.5th percentile) other pesticides (original kg/ha, for analysis rescaled by dividing with 97.5th percentile) soil organic carbon (t/ha) soil phosphorus (mg/kg( soil nitrogen (t/ha) non-mineral nitrogen (kg/ha) non-mineral phosphorus (kg/ha) irrigation (% of area under irrigation) agricultural workers (persons) This repository also includes the analysis results of yields after agricultural input shocks. This data is in rasters named e.g. wheat_phosphorus_shock_75.tif, meaning that this file contains the yield data (t/ha) for wheat after a 75% shock in phosphorus input. Rasters are provided for the following: yields after 25%, 50% or 75% nitrogen shock yields after 25%, 50% or 75% phosphorus shock yields after 25%, 50% or 75% potassium shock yields after 25%, 50% or 75% machinery shock yields after 25%, 50% or 75% pesticide shock yields after 25%, 50% or 75% shock in all fertilizers (fertilizer shock) yields after 25%, 50% or 75% shock in all inputs In addition, the observed crop yield (t/ha) used to construct the model is provided. Other yield rasters are: modelled baseline yield (t/ha), used to calculate the yield changes in shock scenarios modelled zero input yield (t/ha), calculated as a scenario with zero fertilizers, machinery and pesticide inputs Please see the article for references on the datasets.