Talking about uncertainty

Abstract

This empirical research seeks to demonstrate that we can decentralize, crowd source and aggregate signals of uncertainty coming from market agents and organizations, by using the Internet -and more specifically Twitter- as an information archive from which we extract the wisdom of the crowds concerning the state of uncertainty of a specific target system, like a country or a particular uncertainty source, in a given moment in time. We extract and aggregate these signals, constructing a set of specialized uncertainty indexes, by topic and/or by geographic-area. We model the dependence among these uncertainty indexes and other pre-existing uncertainty proxies and highlight differences in their reactiveness to real-world events that occurred in the year 2016, like the EU-referendum vote and the US presidential elections. Finally, we analyze and model the dynamics across time and space of uncertainty signals by geographic-area, to discriminate between, area specific feedback mechanisms, contagion among geographic-areas and international (multi-area) uncertainty shocks. Our results show that crowd-sourced uncertainty signals coming from Twitter may be fruitfully used to improve our understanding of uncertainty contagion and amplification mechanisms across geographic-areas and among market and non-market systems;