Publication date: Available online 23 December 2019
Source: Finance Research Letters
Author(s): Simon Behrendt, Philipp Prange
Abstract
Google searches for stock tickers and company-specific Wikipedia searches may provide reasonable proxies for latent investor attention that are easily accessible for both researchers and practitioners. We draw upon Shannon transfer entropy, a model-free measure that detects any statistical dependence between time series and allows us to infer the dominant direction of the information transfer, to investigate if these different online search queries provide equivalent proxies for online investor attention. Our results show that this is not the case when considering information-theoretical arguments. Moreover, some of the detected bi-directional information transfer is nonlinear.