题　目：Understanding the Demand Effects of Narratives: Method and Application to Brand News

嘉　宾：Yanqing Gui (桂彦青), Ph.D. Candidate, Cornell University

主持人：左思 助理教授 上海交通大学安泰经济与管理学院

时　间：2025年9月24日（周三）9:00-10:30

地　点：上海交通大学 徐汇校区 安泰经济与管理学院A303

内容简介:

News has wide reach among consumers, and a substantial share of these narratives focus on corporate brands. This paper develops a novel framework to quantify and interpret the causal effect of brand-related news on consumer demand. We combine five years of national newspaper coverage on Coca-Cola and Pepsi with Nielsen store-week brand sales data, leveraging cross-sectional variation in national news exposure to identify the causal effects of brand news. To estimate article-level effects, we model the causal effect as a nonparametric function of article embeddings, using a Gaussian Process prior that leverages textual similarity in the embedding space. To interpret these effects, we apply reinforcement fine-tuning to a reasoning-based large language model, using the posterior distribution of each article’s causal effect as the reward signal. The fine-tuned model recovers reasoning traces that best explain the posterior of news effects. We find that around 25% of brand-relevant articles have significant negative effects on demand, and that articles revealing corporate hypocrisy invariably reduce demand.  These findings underscore the economic consequences of such hypocrisy, given its visibility and influence through news. We also demonstrate how the model’s learned reasoning ability can support the generation of consumer-oriented public-relations responses to emerging news.

演讲人简介： 

Yanqing Gui is a Ph.D. candidate in Quantitative Marketing at the SC Johnson College of Business, Cornell University. Before joining Cornell, Yanqing earned a master’s degree in Statistics from the University of Chicago in 2020 and graduated with distinction from the University of Michigan in 2018, where he completed a double major in Mathematics and Statistics and a minor in Computer Science.

Yanqing’s research focuses on understanding the marketing implications of narratives through the lens of consumer outcomes. This intersection of causal inference and natural language processing poses exciting methodological and empirical challenges, especially as narratives are endogenous, unstructured, and often subtle in their implications. Yanqing take a flexible, question-driven approach to methodology, developing and applying tools from causal inference, Bayesian nonparametrics, and generative models to uncover how narratives shape consumer behavior. His work develops new approaches to understand the demand effects of brand news, evaluate the impact of large language models on consumer activism, and infer media exposure from search data—with the ultimate goal of uncovering the managerial implications of narratives for marketing, branding, and public relations.

欢迎广大师生参加!