During the last decades disasters have severely stricken Mexico in terms of economic and human losses. This article aims to measure the impact of hydro-meteorological phenomena and the response policies on the agricultural sector in Mexico following a production function approach that uses panel data models and spatial analysis. As a salient feature, we use an unbalanced panel dataset for 107 crops across 2,430 municipalities from 2003 to 2018. As expected, we find that a disaster declaration is related to a decrease in agricultural productivity, but when a municipality is stricken by one or two disasters over the same period, productivity can be recovered and even increased within the following agricultural year. This result is explained by the policies implemented by the federal government. Likewise, institutional variables like risk atlases are proven to be effective for farmers to plan the cropping season. In addition, we also provide evidence that disaster declarations in neighboring municipalities also diminish local yields.