Working Papers
Mahnaz Paydarzarnaghi
Mahnaz Paydarzarnaghi, David Rakowski, and Mahmut Yasar
Status: Submitted to the Journal of Business Finance & Accounting and under review.
This research is derived from Essay 2 of my dissertation. We explore how stock price reactions to Twitter (now known as X) posts are associated with the perceived credibility of social media users making the posts. We introduce new credibility metrics based on the sender and the content of Twitter posts. Less credible tweets influence prices through a transient and non-informational liquidity effect, while more credible tweets lead to a persistent information effect. Our results support the Elaboration Likelihood Model by demonstrating that the direct route of persuasion (represented by post credibility) is larger in magnitude and more persistent over time than the peripheral route of persuasion (represented by sender credibility).
Mahnaz Paydarzarnaghi and David Rakowski
This study examines the effects of bot-generated social media content on stock market behavior. We examine whether bots amplify or suppress the impact of social media on trading activity and returns. Using data from Twitter (now X), covering 2019 to 2022, we construct measures capturing the likelihood of bot activity in posts mentioning well-known stocks using cashtags. Our findings show that bot activity reduces the impact of social media on both trading volume and returns. Our results imply that investors have the ability to distinguish between genuine and automated activity on social media and that genuine activity is perceived as a more reliable source of financial information.
Herding Behaviors in Financial Markets: The Influence of Social Media on Market Perceptions
Mahnaz Paydarzarnaghi, Mahyar Vaghefi, David Rakowski, and Mahmut Yasar
Status: For submission to the Journal of the Association for Information Systems
This research is derived from Essay 1 of my dissertation. Based on the herding theory and the Emotion as Social Information model (EASI), we examine herding behavior in financial markets, especially Bitcoin. In this study, we focus on the StockTwits platform. The findings show that herding behavior significantly influences market perceptions, especially in volatile conditions, with unexpected interactions between crowd market trajectory and emotional content, offering new insights into online financial discussions.
Mahnaz Paydarzarnaghi, John David Diltz and Salil K Sarkar
Status: Revise and Resubmit (R&R) at Managerial Finance.
This study applies natural language processing (NLP) techniques to assess the degree of accounting conservatism in a large sample of 10-K filings over the period from 1999 through 2023. Following an extensive body of research focused on analysis of accounting conservatism using objective financial statement data, we explore the possibility that affective meaning (i.e., the subjective content of 10-K text) may provide insight into the degree of, and motivation for, conservative accounting practices. Our NLP measures of accounting conservatism, which employed the “log odds ratio informative Dirichlet prior,” generally align with conventional measures regarding conservatism, text uncertainty, and tone, but they differ regarding 10-K readability and analyst forecast accuracy. We conclude that NLP techniques represent a useful tool for the analysis of accounting conservatism exhibited in SEC filings.