Economic Geography of Political Animosity: Analyzing 2 Billion Tweets, 2013-2023
Fri, September 6, 2:00 to 3:30pm, Pennsylvania Convention Center (PCC), 106AAbstract
Advanced democracies around the world are witnessing increasing political animosity. Voters report intense negative sentiments towards citizens with different political opinions or political entities in general. This trend is not uniquely American, with similar patterns observed in multiple European and Asian democracies, such as France, Spain, and Korea. What explains this rising emotional temperature of politics in advanced industrial democracies? I propose a novel measure of political animosity and explain it focusing on economic disparities.
To better study the relationship, I first propose a novel measure of political animosity, using the universe of geo-located tweets spanning a decade and covering three advanced democracies across regions (the United States, France, and South Korea). I use cutting-edge deep learning language models to measure political animosity in 2 billion tweets. The breath of this data enables me to analyze long-term changes and to test the generalizability of my theory of political animosity across diverse institutional and cultural contexts. This measure is also fine-grained in time (up to a second) and geography (exact longitude and latitude), which allows me to merge it with detailed economic indicators available at granular geographic and temporal units.
Then, using this measure, I explain how geographic economic disparities and populist politicians have interacted to generate rising political animosity during the past decade. Economic disparities between local communities have widened, due to deindustrialization, automation, and globalization disproportionately harming certain geographic areas. I propose three hypotheses: (1) Economically declining places have developed resentment and high political animosity, even before the salient electoral success of populist candidates in races for top office: Trump in 2016, Le Pen in 2017, and Yoon in 2022. (2) The success of these candidates increased political animosity both in economically worse-off and better-off areas, but especially in economically better-off communities geographically insulated from changes in declining areas. (3) Before the rise of these populist leaders, citizens of worse-off communities directed their animosity at more diffuse targets (e.g. all politicians, the government), while afterwards, their animosity became more focused towards politicians opposing populists or groups blamed by populists for their economic decline.
This work contributes to the recent scholarship on affective polarization by offering a new cross-temporal and cross-national measure of political animosity based on real-time political behavior. This adds further context and contrasts with previous approaches focusing on survey evidence or social media data around specific political events. Further studies can utilize this fine-grained measure to study the roles of different factors in shaping political animosity.
It also contributes to the literature on populism, which frequently cites “anger, resentment, indignation” as theoretical mechanisms behind the rise of populism. I directly measure these over time and examine which local communities have expressed higher levels of anger and to whom this has been directed. This work also differs from previous literature on populism in that it studies animosity among both populist and anti-populist voters. I highlight dynamic changes in animosity within these groups and their interaction across the span of a decade and how this has shaped the overall emotional temperature of politics.
Lastly, this study contributes to the comparative political economy literature, by highlighting that common changes in economic structure across countries can produce similar trends in political domains, despite institutional and cultural differences.