Research Professional at University of Chicago Booth School of Business
I’m a research professional in the Accounting group at the University of Chicago Booth School of Business. My primary interest lies in exploring the transformative potential of large language models (LLMs) to assist investors in processing information and driving real change in corporations.
At Booth, I have been fortunate to be mentored by Prof. Valeri Nikolaev and Prof. Maximilian Muhn. I'd also like to give a special shout-out to brilliant PhD student Alex Kim, an amazing team member. Together, we developed effective AI tools for public use and explored how Generative AI and Large Language Models (LLMs) can help investors make well-informed decisions. Our work aims to promote more efficient and equitable market outcomes.
I'm also grateful to have worked with Prof. Christina Brown and Prof. Rachel Glennerster in development economics at the Becker Friedman Institute, where we used randomized controlled trials (RCTs) to examine how asymmetric information shapes labor and education market imperfections.
Before joining Booth, I earned my Master’s degree in Computational Social Science with a focus on Data Science and Economics, and my Bachelor’s degree in Economics from the Institute for Economic and Social Research (IESR) at Jinan University.
Joint with Alex Kim, Maximilian Muhn and Valeri Nikolaev
Abstract: We introduce a novel approach to learn fundamental information that investors respond to within lengthy textual disclosures. Our attention-based model is considerably more effective at explaining stock market reactions to disclosed information than attention-free models. We find that topics such as special items (losses) and debt obligations receive the highest attention among investors, whereas internal control disclosures are the least important. Building on this technology, we show that the regulatory interventions (modernization of Regulation S-K) increased investors' attention to the textual content. We also show that firms strategically position fundamental information within MD&A to influence investor focus. Overall, our findings underscore the value of attention-based analysis of corporate communications and open new avenues for future work.
Under Review by Journal of Economic Behavior & Organization
Joint with Shuaizhang Feng, Yujie Han and Binglan Wu
Presented at "7th IESR-GLO Joint Workshop on Aging Societies 2024" and the "Future of Cities: Migrant Children Education Policy Seminar and Principals' Forum 2023."
Abstract: This paper uses a large longitudinal dataset to study the predictive power of children’s time preferences on their future school outcomes, accounting simultaneously for the influences of cognitive (IQ) and noncognitive skills (the Big Five). We show that children’s time preferences significantly predict future cognitive outcomes but not behavioral outcomes when both IQ and the Big Five factors are controlled for. In terms of their respective incremental predictive power, IQ and conscientiousness are the single most important predictors of cognitive and behavioral outcomes, respectively, while time preferences only have minimal predictive power.
On this website, we provide twelve important topics about S&P 500 firms produced with generative AI by analyzing quarterly earnings call transcripts. The topics range from financial performance to value drivers and also include an overall summary of earnings calls. We update this website frequently so that it covers the most recent company insights.
Related papers:
Kim, A., Muhn, M., & Nikolaev, V. (2024). Financial statement analysis with large language models.
Kim, A., Muhn, M., & Nikolaev, V. (2023). From transcripts to insights: Uncovering corporate risks using generative ai.
This interactive app leverages GPT's capabilities to assist with financial statement analysis, focusing on processing 10-Ks and 10-Qs step-by-step while incorporating narrative context. Designed for ChatGPT Plus users, it demonstrates the potential of LLMs in financial analysis but requires careful verification of accuracy due to possible retrieval errors.
Related paper:
Kim, A., Muhn, M., & Nikolaev, V. (2024). Financial statement analysis with large language models.
[1] Dissecting News Engagement on Social Media: The Role of Novelty and Polarization in Information Diffusion
Joint with Ken Moon, Senthil Veeraraghavan and Jiding Zhang
Presented at INFORMS 2023