Quarterly Mark December 2023 - 90

ZHENGZHENG YANG
is an assistant vice
president of State
Street Corporation in
Hangzhou, China.
yangzz1985@gmail.
com
KEY FINDINGS
* This is the first time that the prompt-based learning approach is applied
on ESG text data analysis.
* The authors show that the prompt-based learning approach performs
better than a traditional pre-train and fine-tune framework.
LE ZHANG
is an assistant vice
president of State
Street Corporation in
Hangzhou, China.
lezhang112358@gmail
.com
XIAOYUN WANG
is a managing
director of State
Street Corporation in
Hangzhou, China.
xiaoyun99@gmail.com
YUBO MAI
is a PhD candidate at
Zhejiang University
in Hangzhou, China.
maiyb@zju.edu.cn
* The authors demonstrate that the prompt-based learning approach can
achieve outstanding performance with few training data.
ABSTRACT
Over the past decade, there is a surging trend to integrate environmental,
social, and governance (ESG) criteria into financial decision making. ESG
information extracted manually from text sources, such as company
statements, press releases, and regulatory disclosures, can be expensive
and inconsistent due to human interpretation. In this article, the authors
introduce the application of prompt-based learning, a cutting-edge
natural language processing (NLP) technology, to classify textual data into
ESG and non-ESG categories. In particular, the authors establish a promptbased
ESG classifier, using data from Refinitiv, and benchmark it against
a traditional pre-train and fine-tune classifier through statistical test. The
authors fine-tune the classifiers on various sizes of training data. The
experiment shows that the prompt-based learning approach outperforms
the traditional pre-train and fine-tune classifier and can generate
promising results when training data are limited.
E
nvironmental, social, and governance (ESG) criteria have been valued
increasingly by both retail and institutional investors in recent years. There
is a growing trend in the finance area to integrate ESG criteria into decision
making, which potentially impacts asset prices and company policies. As
summarized in Friede, Busch, and Bassen (2015), 90% of more than 2,000 studies
suggest a nonnegative relationship between ESG features and corporate financial
performance. Incorporating ESG information can be an effective approach to
reducing risk and improving performance. Hence, it is crucial to derive accurate
and dynamic ESG information based on specific applications, designed to extract
ESG-related information from raw data such as corporate social responsibility
(CSR) reports, investor calls, and regulatory disclosures. Such a process involves
analyzing a large scale of textual data and may rely heavily on a manual process
by a trained content analyst. The rapid development of natural language
processing (NLP) technology enables computers to understand human language
automatically and effectively. Employing NLP technology to extract ESG-related
information can be a promising approach, as it minimizes manual efforts and
maintains consistency when accessing raw data.
90 | ESG Text Classification: An Application of the Prompt-Based Learning Approace
Quarterly Mark December 2023

Table of Contents for the Digital Edition of Quarterly Mark December 2023
