Management Signals in Social Media on P2P Firms' Operation Health


Student thesis: Doctoral Thesis

View graph of relations


Related Research Unit(s)


Awarding Institution
Award date17 Jan 2022


Social media is an essential tool that individuals can use to share insights with others, and firms can use it to engage customers and investors. Numerous studies delineate the marketing effect of social media on firm performance. In this thesis, we argue that social media content, including user-generated content and firm-generated content, can also provide crucial information regarding firm management and survival.

To study this phenomenon, we use the peer-to-peer lending industry in China as an example for several reasons. First, the P2P lending industry in China has many private firms/platforms and has experienced a dramatic rise and fall in the last decade. It provides an appropriate opportunity for us to observe the dynamic performance of unlisted firms and identify their relationship with firms’ and investors’ social media activity. Second, investors are eager to assess P2P platforms’ operation health, but there is insufficient information for them to assess it. To this end, people would use social media content, including platforms’ disclosure and wisdom of crowds, as a channel to assess their management and operation health. Third, both P2P lending platforms and investors actively post content on social media. Platforms adopt social media as a major channel to approach customers, such as to share investment-related knowledge, publish campaigns, or interact with customers. Chinese P2P lending investors also tend to proactively share their opinions and experience about P2P lending on social media to reduce the information asymmetry among investors and between platforms and investors.

In the first study, we investigate whether FGC (firm-generated content) can reflect firms’ management. Using natural language techniques, we identify and extract two management signals (i.e., risk-mention and writing mistakes) in FGC. We choose the two signals since we think emphasizing risk to the public can reflect the level of platforms’ risk management and consensus of social responsibility, and writing mistakes can reveal platforms’ human resource management, including recruitment, training, and performance appraisal. Empirical results demonstrate that two signals do reflect platforms’ management conditions. This study is of value since it extends existing studies that mainly focus on FGC’s marketing effects by identifying the role of FGC as a reflection of management. This study contributes to social media literature and management literature by identifying the impact of firm management on firm social media wording, and our findings can also direct investors and the government to assess and monitor platform management conditions by observing firms’ social media postings.

In the second study, we investigate whether firms’ social media behavior is a good source of platform collapse risk indicators. More specifically, we uncover if platform operation health can be indicated by platforms’ having social media accounts, social media engagement, and wording. Empirical results demonstrate that P2P lending platforms with social media accounts are less likely to collapse. Moreover, firms’ social media engagement and wording are also indicative of platform collapse risk through different mechanisms. Firms’ social media engagement affects platform operation by positively influencing platform fundraising performance, which benefits firms’ operation health. Firms’ social media wording can indicate platform operation since it reflects firm management, which can significantly influence firms’ operations. As to the wording signaling of firm operation, we find that platforms that post articles with a higher level of risk-mention and fewer writing mistakes are more likely to continue running. According to topic modeling and predictive analysis, FGC topics, especially content related to management and marketing, can markedly increase the prediction power of platform survival. This study contributes to firm survival literature by expanding the range of potential indicators of firm collapse and develops a financial intermediary collapse prediction approach with topic modeling analysis, which contributes to the scant literature on the assessment of financial intermediary collapse risk.

The third study examines the relationship between investor opinion and the risk of platform collapse. We extract investor opinion from investors’ comments on platforms with natural language processing techniques. We subsequently measure two dimensions of investor opinion (i.e., valence and diversity). After constructing data into monthly-level panel data, we adopt a time-dependent Cox regression model to conduct platform survival analysis. Our findings show that if investors’ comments are more positive and concentrated, a platform will be less like to collapse, which presents the effectiveness of investor-generated content as an indicator of P2P lending platform health. This study strengthens the understanding of UGC’s effectiveness in reflecting firm operation health by identifying the relationship between investor opinion and firm survival, and supplements literature related to the firm survival by expanding the indicators of firm survival from firm-level characteristics or executive-level characteristics to investor opinion, which is a dynamic information source to predict firm success in the market.