Transatlantic Emotions: an emotional perspective on alliance burden sharing

Emotion is incredibly important in international relations because it conveys a lot of information in diplomatic exchanges. However, emotion in international relations has not been extensively studied, despite recent progress in scholarship. In this article, emotion is examined in the context of security cooperation, using NATO as example. The research question is: how do emotions in U.S. presidential speeches affect NATO allies defense spending behavior in the post Cold War period? I scraped 8,137 speeches from the American Presidency Project and wrote a Python program that extracts and quantifies 6 different emotions from these speeches using DistilBERT. I developed a hypothesis for each of the emotions and tested them using a country-year level panel dataset. Based on these hypotheses, anger, fear, and sadness are more significant in influencing NATO allies defense spending behavior than others, but this conclusion is also limited by the capabilities and capacities of the large language model I chose in this paper. Despite this, this paper still make several contributions to the literature, especially the specific hypotheses about each emotion.