How information technologies shape investor sentiment: A web-based investor sentiment index

摘要

This paper proposes a new investor sentiment indicator that combines the use of principal component analysis with web searches. This proposal provides economic meaning to the underlying variables, a sound factor structure, and reduces the noise regarding to web searches, when compared to standard search-based sentiment indicators. In fact, our indicator not only confirms the relevance of sentiment for future assets performance and provides greater predictive capacities than standard formulations, but also generates new insights in terms of globalization of investor sentiment and the role that information flows and technology play on that process. Moreover, it challenges some general beliefs present in the literature of sentiment such as the prevalence of a local bias, the greater impact of sentiment in developed markets or the fact that institutional investors are not sensitive to sentiment. Finally, an investment strategy is implemented showing how a sentiment-based investment rule generates above-market returns.