Abstract:
We study the effect of terrorist attacks on the migration discourse in the European Parliament (EP). First, using an LLM model, we develop an original dataset on sentiments of all parliamentary speeches concerning migration for 2009-2019, building on a novel dataset by Sylvester et al. (2023). Second, following Brodeur (2018), we employ a causal identification strategy based on quasi-natural randomization in the success or failure of terrorist attacks. We find that while a successful terrorist attack does not change the overall migration sentiment, it has heterogeneous effects conditional on the political position of the speaker: left-wing and, to a lesser extent, centrist politicians become more favorable toward migration after successful attacks, while the rightwing politicians become more negative. Politicians of different ideologies adjusting migration-related sentiment in a direction aligned with their pre-existing partisan positions indicate an increasing polarization among policymakers as a direct consequence of terrorism.