The purpose of this project is to investigate the extent, dynamics and consequences of automation in financial communication. The development of linguistic algorithms over the past decade has broadened the scope of automatic interpretation of textual news. We investigate whether this improves market efficiency with respect to the incorporation of information from text into prices. The potential downside of widespread automation is the marginalization of other sources of information, which by their nature cannot be automated, such as personal meetings between investors and company managers. We explore whether human interaction has indeed been marginalized by automation and whether this has decreased market efficiency with respect to the acquisition of information. To this end we first identify what it is that investors can learn from human interaction that is not conveyed by other sources.

Understanding the potential tradeoff between incorporation and acquisition of information and the net effect of automation on market efficiency has important implications for every investor. If markets are (close to) efficient and all (or most) information is instantly impounded into prices by automatic algorithms, then the best approach is to hold index funds, also known as passive investing. If on the other hand, sufficiently many pockets of information remain that can only be accessed by humans, then fundamental analysis of firms followed by active investing should still be worthwhile.

Project participants: Björn Hagströmer, Michał Dzielinski, Johan Graaf, Alex Wagner (University of Zürich)

Publications: In no (un)certain terms: Managerial style in communicating earnings news. Dzielinski, M., Wagner, A.F., Zeckhauser, R. Working paper, 2016.

Research grant: Human Interaction in Automated Financial Markets. 2 million SEK granted in 2016 by Handelsbankens Forskningsstiftelser. Project leader: Björn Hagströmer (SBS)