Abstract

We analyze district-level seasonal yields of paddy, wheat, and millets across India from 2010 - 2019 to identify yield hotspots - districts with consistently high or rising yields - and coldspots - districts showing significant yield declines. Using spatial clustering and regression analysis, we link yield performance to seasonal weather, irrigation coverage, fertilizer use, and farmers' access to agricultural information through call centers. Coldspot districts produce roughly half the yields of hotspot areas, largely due to lower input use and limited access to crop-specific information. Hotspots are concentrated in northern and southern India, while coldspots or low-performing clusters prevail in the western, eastern, and northeastern regions. Average yield gaps between hotspot and coldspot districts are 2.0 t per ha for paddy, 2.4 t per ha for wheat, and 1.0 t per ha for millets. Regression results show that a 1% increase in irrigation coverage, farmers' queries to call centers, and 1 t per ha of fertilizer use are associated with yield gains of 7, 6, and 0.5 kg per ha, respectively. An illustrative comparison with Darbhanga district in Bihar State of eastern India suggests that closing the hotspot-coldspot yield gap could raise household revenues by up to 148% for paddy and 79% for wheat.