Stereotypes and Global Capital Flows

Emanuele Colonnelli, Joseph L. Gidwitz Professor of Finance and Entrepreneurship
Suproteem Sarkar, Assistant Professor of Finance

This project studies how stereotypes about developing countries shape foreign investment and economic growth. We construct quantitative measures of stereotypes using text embeddings derived from influential economic and financial reports. We track how the semantic representation of countries evolves over time and across contexts, capturing shifts in perceived risk, quality, and opportunity. We validate these measures by linking them to future investment flows and real economic outcomes, allowing us to identify when beliefs reflect distortions rather than fundamentals. To understand the drivers of these stereotypes, we apply modern machine learning tools, including sparse autoencoders, to uncover latent features and events associated with biased narratives. By leveraging algorithmic bias as a lens into human belief formation, the project offers a novel framework to quantify how misperceptions affect capital allocation and development.