Using satellite-observed geospatial inundation data to identify the impacts of floods on firm-level performance: The case of China during 2000–2009

Abstract

This paper compiles high-resolution geospatial inundation areas of China for the period 2000–2009 based on satellite imagery repositories filtered by the Global Flood Database (GFD). In parallel, we geocode a comprehensive firm-level dataset of China and combine these two sets of geospatial data to identify the set of inundated firms in each year of flood events, as well as the distances of all non-inundated firms from the inundated areas. Given the high-resolution inundation data, we adopt a generalized dynamic-panel specification to estimate dynamic and spatial spillover effects of floods on firm-level production activities (including outputs, capital and labor inputs, and productivities). We find negative and persistent effects of floods on firm-level performance measures, and negative but short-run spillover effects on non-inundated firms in nearby neighborhoods. In contrast, non-inundated firms located 6–12 km away from the inundated area expanded their production in the long run, suggesting reallocation of production activities/facilities away from the inundation area toward the outer rings of the neighborhood. We conduct various robustness checks and extended analyses, identify moderating/aggravating factors of inundation impacts, assess the aggregate effects at the economy-wide, province, and sector levels, and quantify the propagation of flood exposures via the input–output linkages.