Title: Conservative and timely ESG-compliant investment universe screening using textual sentiment analysis
Authors: Samuel Borms - Universite de Neuchatel (Switzerland) [presenting]
David Ardia - HEC Montréal (Canada)
Kris Boudt - Vrije Universiteit Brussel and VU Amsterdam (Belgium)
Andres Algaba - Vrije Universiteit Brussel (Belgium)
Abstract: Combining textual sentiment analysis and econometric techniques, we improve the construction of an ESG best-of-class investment universe by more timely detection of companies and countries likely to succumb to a sustainability downgrade. Our approach consists of several layers. First, we score the sentiment in the relevant news articles employing various ESG-specific lexicons. Second, we aggregate the textual sentiment scores and remove the noise through application of the Kalman filter. Third, we use predictive modelling and forecast combination over multiple horizons to assess the likelihood of downgrading for every investable stock. Fourth, we define decision rules to construct and update both the sustainable investment universe and a blacklist. In an empirical portfolio application, we test for the added value of our more accurate and timelier sustainability signals.