Determinants of Textual Transparency
Li focuses on factors that drive the transparency of textual financial reporting disclosures. Given the increased amount of textual financial disclosures in recent years, they will assess the availability, accuracy, and clarity of this information. They extend prior work by using bond-related disclosures and by singling out the effect of litigational concerns on the quality of textual information. Their key interest lies in exploring the cross-sectional variation of these textual disclosures and in understanding how textual disclosures are shaped by public equity and debt markets.
How do textual financial disclosures affect transparency, and how does this vary with the regulatory environment?
Texts in firms' reports should provide detailed information, complementing and explaining financial statements. However, stakeholders often complain that texts are too long and hard to understand in its entirety. Even an expert like Warren Buffett once said that: "Too often, I've been unable to decipher just what is being said or, worse yet, had to conclude that nothing was being said." It is also not clear what drives the disclosure representation of firm fundamentals in firms’ reports. Researchers suggest using textual analysis to study transparency. However, traditional measures are designed to capture linguistic characteristics irrespective of the specific economic fundamentals. As such, a text might be clearly written while only containing irrelevant information for a receiver.
We investigate whether and when companies meet stakeholders’ demands for transparent textual disclosure. In our view, firms are senders of private information about their economic fundamentals, and equity and bond investors are receivers of the signal. As such, firms may or may not provide transparent textual disclosure to meet investors' demands. Textual transparency in our project describes the availability of information in texts, i.e., a text contains information about firm fundamentals that are of interest to a receiver. It also describes the precision, reliability, and validity of texts, i.e., a text is consistent with the underlying economic fundamentals. Lastly, it describes the clarity of texts. We develop a new approach to measure textual transparency and examine its determinants in equity and bond markets. We are also interested in potential spillover effects among types of firm filings that address different audiences.
We develop a new approach to measure textual disclosure transparency using recent advances in computational linguistics (e.g., machine learning, multi-document summarization). We also investigate novel datasets providing insights into determinants of textual transparency.