Sentiment Analysis on Climate Change for Sustainable Investment
Lists
Only you see this
Abstract
This research involves examining whether sentiment on climate change can be accounted as a systematic risk factor within sustainable finance. Tweets related to climate change from 2014-2022 are collected via the Twitter API. RoBERTa is then deployed to label the collected tweets according to sentiment polarity. Thereafter, DistilBERT is utilised to classify tweets according to their different climate change topics Aggregated, Impact, Politics & Policy, Mitigation, Rootcause. Using these sentiment and topic labels, monthly sentiment scores are aggregated from the frequency of sentiment polarity (number of positive, and negative tweets), and segmented according to the topics that the tweets are classified under. On the other hand, we collect the monthly returns of various green ETFs, alongside ETFs that contain stocks from other sectors (technology, oil and gas), as well as 2 baskets of stocks (sustainable and heavy-polluting). We run Fama-French regressions by conducting time series regression of monthly returns on the traditional 3 factors, with specific addition of sentiment scores.
BibTeX
@inproceedings{ong2024sentiment,
author = {Keane Ong and Rui Mao and Deeksha Varshney and Ranjan Satapathy and Erik Cambria and Gianmarco Mengaldo},
title = {{Sentiment Analysis on Climate Change for Sustainable Investment}},
booktitle = {Workshops @ ICDM},
year = {2024},
doi = {10.1109/icdmw65004.2024.00067},
}Loading related content...