Conference Agenda

I study how equity analysts form subjective beliefs about firms' earnings using their own written text from over 1.1 million equity research reports. Using large language models, I identify the topics discussed by analysts and represent topic-level information using textual embeddings. I introduce a two-stage procedure that builds on the classic Coibion-Gorodnichenko regression to uncover analysts' over- and underreaction to specific information. Using this new procedure, I find that the overreaction in long-term forecasts is concentrated in qualitative, intangible topics rather than quantitative, statistical ones. Revisions driven by qualitative information in long-term earnings forecasts strongly predict future stock returns. From linguistic expression, I find suggestive evidence that overconfidence is an important driver of the overreaction to qualitative information.