Abstract

This paper studies how discriminatory fake news arises and spatially diffuses. We focus on India at the onset of the COVID-19 pandemic: on March 30, a Muslim convention (the Tablighi Jamaat) in New Delhi became publicly recognized as a COVID hotspot, and the next day, fake news on Muslims intentionally spreading the virus spiked. Using Twitter data, we build a comprehensive novel dataset of georeferenced tweets to identify anti-Muslim fake news. We find, in cross-sectional and difference-in-difference settings, that discriminatory fake news became much more widespread after March 30 (1) in New Delhi, (2) in districts closer to New Delhi, and (3) in districts with higher social media interactions with New Delhi. Further, we investigate whether deeply rooted historical factors may have also played a role in the diffusion of anti-Muslim fake news: we show that, after March 30, discriminatory fake news was more common in districts historically exposed to attacks by Muslim groups.