Overview and details of the sessions of this conference. Please select a date or location to show only sessions at that day or location. Please select a single session for detailed view (with abstracts and downloads if available).
Please note that all times are shown in the time zone of the conference. The current conference time is: 30th Sept 2023, 07:45:44am CEST
Anthony Cookson1, Runjing Lu2, William Mullins3, Marina Niessner4
1University of Colorado at Boulder; 2University of Alberta; 3University of California, San Diego; 4University of Pennsylvania
Discussant: Isabella Wolfskeil (Federal Reserve Board of Governors)
We examine social media attention and sentiment from three major platforms: Twitter, StockTwits, and Seeking Alpha. We find that attention is highly correlated across platforms, but sentiment is not: its first principal component explains little more variation than purely idiosyncratic sentiment. We attribute differences across platforms to differences in users (e.g., professionals vs. novices) and differences in platform design (e.g., character limits in posts). We also find that sentiment and attention are both positively related to retail trading imbalance, but contain different return-relevant information. Sentiment-induced retail trading imbalance predicts positive next-day returns, in contrast to attention-induced retail trading imbalance, which predicts strongly negative next-day returns. These results highlight the importance of distinguishing between social media sentiment and attention, and suggest caution when studying the social signal through the lens of a single platform.
Information Waves and Firm Investment
Feng Chi
Cornell University, United States of America
Discussant: Olivier Dessaint (INSEAD)
This paper measures the impact of information quality on the success of firms' investment decisions. Firms in the retail and restaurant industries rely on the demographic information provided by the U.S. decennial census. Over the course of a decade, information from the decennial census snapshot likely deviates from the evolving market condition, thereby making the data less relevant. I find that on average, outdated census information increases establishment failure rate by 1.6% per year, translating into a 16% increase over a 10 year gap. The effects are stronger for geographic areas that experience large changes in demographics, for industries that rely on precise information in small trade areas, and for independent retailers that lack alternative sources of demographic information.
