Individual Submission Summary
Share...

Direct link:

Party Lines or Voter Preferences? Explaining Political Realignment

Fri, September 6, 10:00 to 11:30am, Marriott Philadelphia Downtown, 308

Abstract

While low education used to reliably predict right-wing affiliation in the 1970s, this tendency has today completely reversed. What has triggered this realignment along education lines in the US? Is it that political parties have shifted ideologically or is it voter preferences that have changed? This paper quantifies the extent to which this political realignment can be attributed to changes of national party platforms, party discipline, voter demographics, and citizens' preferences, respectively. Using campaign websites of candidates for the United States House of Representatives and applying an unsupervised probabilistic topic model, I recover the position of each candidate on a series of key topics for congressional elections between 2000 and 2020. I find that the distance between the average position of the two main US parties has considerably widened during that period, in particular on cultural questions. I use fine-grained election results at the precinct level from 2000 onwards to show that the relationship between the precinct share of college graduates and the Democratic vote exhibits a U-shape but has flattened over time. Leveraging exogenous variation in candidate policy propositions, I estimate a model of heterogeneous demand for policy positions. I show that educated voters' ideal points are more liberal than the median voter, in particular on cultural questions (e.g., reproductive rights, climate change), and that this demand has barely changed since the early 2000s. I also show that the weights attributed to cultural topics have increased but not along educational lines. I then estimate a model of supply where each candidate tries to maximize their vote share but has to comply with the position of the party leadership. I show that party discipline has increased over time, especially on cultural questions. Lastly, I am able to study counterfactual scenarios and show that most of the political realignment along educational lines over that period can be attributed to changing national party positions.

Author