Researcher ORCID Identifier

https://orcid.org/0009-0007-4172-0812

Graduation Year

2026

Date of Submission

12-2025

Document Type

Open Access Senior Thesis

Degree Name

Bachelor of Arts

Department

Economics

Reader 1

Professor Benjamin Gillen

Rights Information

2025 Sofia K Weinstein

Abstract

This thesis examines how investors react to biotechnology firms’ clinical trial announcements and whether artificial intelligence (AI) can enhance our understanding of those market reactions. Using a representative sample of 95 firm-event observations between 2014 and 2025, the study measures cumulative abnormal returns (CARs) surrounding clinical readouts, regulatory decisions, and partnership disclosures. Event-study regressions test for heterogeneity across trial phase, therapeutic area, and textual sentiment derived from company press releases. Consistent with expectations, later-phase trials, particularly Phase II and III, produce stronger short-term abnormal returns. However, categorical variables such as phase or therapeutic area explain little of the total variation in market reactions (R² values remain below 0.15). Instead, qualitative tone extracted from press releases proves more informative: AI-based sentiment classification significantly predicts the direction and magnitude of CARs. Positive tones (“exceeded expectations”) correspond to CARs around +0.4%, while negative tones (“fell short”) are associated with small, directionally negative moves that are rarely statistically significant. These results suggest that market responses to biotech events are driven less by structural categories and more by the tone of how information is communicated. The findings support a hybrid interpretation of market efficiency: investors process technical information rapidly but rely heavily on linguistic cues to interpret complex outcomes. By integrating text-based machine learning with traditional event-study methods, this thesis demonstrates how large language models can improve empirical modeling of investor behavior in high-uncertainty industries.

Download

Share

COinS