Direct interaction in digital interactive media and stock performance: Evidence from Panorama
Jinshui Huang,Jun Wang,Xiaoman Jin
DOI: https://doi.org/10.1371/journal.pone.0302448
IF: 3.7
2024-05-28
PLoS ONE
Abstract:Media information plays an essential role in the stock market. Recent financial research has verified that media information could shock stock price by influencing investors' expectation. Now, a new type of interactive media, called Digital Interactive Media (DIM), is popular in Chinese stock market and becomes the main channel for investors to understand listed companies. Unlike general news media or investor forums, DIM enables direct interaction between listed companies and investors. In the modern society where digital economy is booming, media information would largely affect investors' decisions. Therefore, it is urgent to use natural language processing (NLP) technology to deconstruct the massive questions and answers (Q&A) interactive information in DIM and extract valuable factors that affect stock prices and stock performances to explore the influence mechanism of digital interactive information on stock performances. This paper firstly uses web crawling technology to obtain approximately 110000 Q&A text information from the digital interactive platform ('Panoramic Network') from 2015 to 2021. Then we use big data text analysis technology and emotional quantification technology to extract valuable influencing factors from the massive text. A Multiple Linear Regression (MLR) model was created to explore specific influence mechanism of digital interactive information on stock price performance. The empirical results show that the emotions implicit in investors' questions do not significantly impact stock performance. However, the emotions and attitudes of the answers by listed companies can significantly affect corresponding stock prices, which indirectly confirms the Proximate Cause Effect of behavioral finance. This effect is particularly evident in the stock prices on the current trading day and the next trading day. In the Robustness Test, this paper replaces dependent variable and adds relevant control variables, and the conclusion remains valid. In the Endogeneity Test, this paper selects sample data before the launch of Panorama Network in 2014 as a comparison, and uses a Difference-in-Difference (DID) model to prove the significant impact of the launch of Panorama Network on Chinese stock market. In the Heterogeneity Test, the paper classifies the market value, region, and industry of listed companies and regressed the sub samples, once again confirming the reliability of the empirical conclusions. The results of Robustness Test, Endogeneity Test, and Heterogeneity Test conducted in this paper all support empirical conclusions.