Business Administration - Theses
Permanent URI for this collection
Search Results
1 - 1 of 1
-
ItemEssays on The Influence of The Media on Financial Markets( 2022)This thesis explores the influence of the media on stock markets through three channels: news outlets, social media, and the exposure of news outlets through social media. It includes three essays studying the characteristics of financial news and social finance and how these relate to stock markets. First essay presents a novel approach to sentiment analysis that combines topic modelling with the analysis of the deep structure of text from financial news headlines. Our approach provides two major contributions. First, it assesses the sentiment of text using tone-syntax patterns rather than a set of words. Second, it increases the accuracy of sentiment analysis by measuring tone across fine-grained topics in finance. We also integrate different machine learning algorithms trained by abnormal returns with our approach to examine whether and to what extent machine learning improves sentiment models in finance. Second Essay studies whether networks have the danger of creating disintermediation, whereby investors bypass professional financial advice and seek counsel in a social network. To discover what determines influence in social finance, we examine the structural and functional connectivity of the largest social finance network, StockTwits. According to theory, network activity can be predicted from network structure. In social finance networks, however, activity can bypass the network structure through top-down or bottom-up content curation. We also investigate how Message content and sentiment appear to matter as much as location and status for influence. The third essay presents an empirical setting to study the attention biases of investors, which result in different interpretations of a given piece of public information (i.e., disagreement among investors). To capture attention biases, it studies the introductions and adoptions (i.e., posting and resharing) of a large set of media news in the largest digital social network for investment: StockTwits. To capture investors’ attention biases, we classify the news stories according to the reliability and political orientation of publishers, and the topic and sentiment of the news stories.