Zu Chongzhi Research Seminar

Date and Time (China standard time): Thursday, August 07, 1:30 pm – 2:30 pm

Location: IB 3106

Title: Risk disclosure premium in the cross-section of stock returns: A generative topic modeling approach

Speaker: Steve Yang

Abstract: We examine the pricing implications of financial risk disclosures by applying generative topic modeling to Section 1A of 10-K risk factor filings for S&P 1500 firms from 2010 to 2023. Using embeddings from FinBERT, GTE-Qwen, and SFR-Mistral, we quantify firm-level exposure to specific risk narratives and compute corresponding risk scores. These scores are used to construct narrative-based risk factors through a twostep empirical asset pricing framework. First, we test the return predictiveness of these factors using the Fama-French 5 factor model via long-short portfolio sorts. Second, we implement Fama-MacBeth cross-sectional regressions to evaluate their risk premia alongside the standard market, size, value, profitability, and investment factors. Our results reveal a robust risk premium in operational risk disclosure – which remains statistically significant even after applying the Bonferroni correction for multiple tests. In addition, several other disclosure factors, such as funding risk, intellectual property risk, and infrastructure risk, also show some level of significance. These findings suggest that firm-disclosed forward-looking risks, as identified through generative embedding model on firm risk disclosures, offer economically meaningful and statistically robust insights into cross-sectional return variation.

Bio: Steve Yang is an Associate Professor of the School of Business at Stevens Institute of Technology. He holds a Ph.D. in Systems and Information Engineering from University of Virginia with concentration on Financial Engineering. His research has been focused on understanding markets’ irrationality and its impact on trading, portfolio, risk management, and systemic risk using decision science tools such as Markov decision processes, reinforcement learning, and other artificial intelligence (AI) methods. His research has been funded by a variety of NGOs, government agencies and industry firms such as, NSF, SWIFT, IRRC, IAAER, CFTC, DoD, Accenture, Northrop Grumman, KPMG, etc.

Dr. Steve Yang is currently the Director of the Center for Research toward Advancing Financial Technologies (CRAFT – https://www.stevens.edu/school-business/craft-center/craft-center-leadership-and-affiliated-faculty) established by Stevens Institute of Technology and Rensselaer Polytechnic Institute, co-sponsored by the National Science Foundation (NSF). He has worked with several major federal financial regulators such as, the Securities and Exchange Commission (SEC), the Commodity Futures Trading Commission (CFTC) and U.S. Treasury in the capacity as either a research consultant or visiting scholar. He is a recognized scholar in Algorithmic trading and market stability. In 2020, he was selected as a visiting scholar in the Division of Economic and Risk Analysis at the SEC. He currently serves as an Associate Editor for European Journal of Finance. He also served as the Associate Editor for Expert Systems with Applications (ESWA) journal, the guest editor for Journal of Futures Markets, Quantitative Finance, and NSF review panellist in the past.