Near-real time analysis of war and economic activity during Russia's invasion of Ukraine

This paper introduces near-real time data on Russia's full-scale invasion of Ukraine, and uses these data to investigate the short-term impact of occupation and violence on local economic activity. The data project - VIINA (Violent Incident Information from News Articles) - parses news reports from Ukrainian and Russian media, georeferences them, and classifies them into standard event categories (e.g. artillery shelling) through machine learning. As we show, VIINA is more geographically comprehensive and more thoroughly documented than other open-source event databases on Ukraine, and is the only such effort to track territorial control. We illustrate applications for research on political economy, by utilizing remote sensing data on luminosity and vegetation as proxies for urban economic activity and agricultural land use. We find that economic activity declined most in urban areas that neither side fully controlled, and in places most exposed to artillery shelling. Areas under Russian occupation, however, were more insulated from these negative shocks. Although not causal, these findings are in line with the view that war brings not only economic devastation, but also vast geographic inequalities that are almost immediately observable from space.