An investigation of the impact of electronic communication on financial markets

Abstract

This paper uses the popular social networking site Twitter to test recent theories on the importance of communication through networks. I create a unique data set that precisely measures the network size and credibility of each contributor to the site. I test the theory of DeMarzo et al (2003) that an agent's influence is a function of his credibility and the size of his network, as well as the Colla and Mele (2010) theory of correlated trading among social groups. I show that influence is a function of network credibility and network size. I also find that there is positively correlated trading among agents in a social network and negatively correlated trading with agents outside of the network. Finally, using portfolios sorted on level of attention, I find support for the Merton (1987) theory of investor recognition which predicts that low attention stocks should exhibit positive risk adjusted returns.