Debt Dictionaries

Abstract

Using the debt and equity response to the release of textual information from earnings calls, we demonstrate that stock and bond investors for the same firm interpret different aspects of firm information as value-relevant. These differences in interpretation-captured by distinct functions mapping text to returns-cannot be explained by differences in security payoffs. If anything, the importance of investor-specific approaches to information processing increases as debt and equity payoffs converge. In our strongest test, we examine junior bonds with equitylike payoffs (positive convexity with respect to firm value) and find that investors continue to interpret information through a creditor's lens. Using a combination of structural topic modeling and large language models, we show that debt investors systematically emphasize news about crises, economic challenges, liquidity, and operational issues, while equity investors focus more on growth opportunities, technology, and innovation.

Keywords: Cross-asset returns, Natural Language Processing, Machine Learning, Market Segmentation