Inflation, perception of economic uncertainty and COVID-19: Evidence from Central Bank communication

Abstract

Do extreme events have a significant effect about textual sentiment? The purpose of this article is to highlight the need to correct the estimation of indicators of economic uncertainty. The indicators were constructed from textual data about the perspective of extreme events. For this purpose, based on data extracted from the minutes of mee-tings of the Monetary Policy of eighteen Central Banks, we estimated two variables of perception of economic uncertainty: the first using only a traditional sentiment dictio-nary and the second incorporating terms associated with the extreme event (COVID- 19 Pandemic) in its word list. Initial results show that there is a significant effect of COVID-19 on the estimation of the perception of economic uncertainty; this effect acts as an accelerator that potentiates its impact. It was evident that incorporating conjunctural issues - be it local or global - is indispensable when performing sentiment analysis in texts during extreme events. Moreover, failing to take conjunctural issues into account throughout the estimation process can result in variables with biased in-formation.