Visiting Graduate Student at UCSD- Rady School of Management

Our research leverages large language models (LLMs) to analyze text data from Reddit and StockTwits, exploring differences in investing styles between Republican and Democrat retail investors. Building on Cookson et al. (2020), we will use advanced LLM techniques to classify political affiliations based on language patterns and analyze retail investor sentiment from 2010 onwards. Cookson et al. revealed partisan Republicans’ relative optimism about the stock market during the pandemic, showing specific biases in stock preferences. We expand this by incorporating broader social media data, capturing a wider spectrum of sentiment and political discourse.

Literature on partisanship and belief formation (Milner and Judkins, 2004; Gaines et al., 2007) and studies by Gentzkow et al. (2019) and Kaplan et al. (2019) highlight political identity’s impact on economic perceptions. Our study examines how partisan-driven narratives influence retail investors, particularly during uncertainty, using frameworks like motivated reasoning (Brunnermeier and Parker, 2005; Benabou, 2015) and uncertainty identity theory (Hogg, 2007).

By analyzing sentiment in user-generated content, we aim to identify belief divergences between Republicans and Democrats, providing insights into the intersection of political identity and market behavior. This research contributes to behavioral finance and political economy, offering practical insights for policymakers and financial analysts.

Faculty Supervisor:

Mark Kamstra

Student:

Partner:

University of California, San Diego

Discipline:

Business

Sector:

Education

University:

York University

Program: