Available online at www.sciencedirect.

com
Available online at www.sciencedirect.com

ScienceDirect
ScienceDirect
Available online at www.sciencedirect.com
Procedia Computer Science 00 (2024) 000–000 www.elsevier.com/locate/procedia
ScienceDirect
Procedia Computer Science 00 (2024) 000–000 www.elsevier.com/locate/procedia
Procedia Computer Science 242 (2024) 693–700

11th International Conference on Information Technology and Quantitative Management (ITQM 2024)
11th International Conference on Information Technology and Quantitative Management (ITQM 2024)
Text-Based Measure of ESG Risk Exposure
Text-Based Measure of ESG Risk Exposure
Fei Du a, Chunbing Bao a一, Qingchun Meng a, Yuqiao Hui a
Fei Du a, Chunbing Bao a一, Qingchun Meng a, Yuqiao Hui a
School of Management, Shandong University, Jinan 250100, China
a

a
School of Management, Shandong University, Jinan 250100, China

Abstract
Abstract
Against the backdrop of increasing instability and uncertainty in the world economy, evaluating corporate ESG (Environmental,
Social, and
Against Governance)
the backdrop risks has instability
of increasing become anand important toolinfor
uncertainty theenhancing resilience,
world economy, effectively
evaluating countering
corporate and mitigating
ESG (Environmental,
external risks, and achieving sustainable development. In order to assess the magnitude
Social, and Governance) risks has become an important tool for enhancing resilience, effectively countering and of ESG risk disclosure at themitigating
corporate
level, thisrisks,
external paperandfirst constructs
achieving dictionaries
sustainable for the three
development. Indimensions
order to assessof ESG using Word2vec
the magnitude of ESG model and a corpus
risk disclosure consisting
at the of
corporate
ESG reports
level, this paperandfirst
ESG-related
constructsnews. Secondly,
dictionaries for athefirm-level ESG riskofdisclosure
three dimensions ESG using measure
Word2vecis developed
model and using quarterly
a corpus earnings
consisting of
conference
ESG reportscalls and from 2017 tonews.
ESG-related 2024. Secondly,
Subsequently, the validity
a firm-level ESGof thedisclosure
risk measurement indicators
measure is verified
is developed usingthrough
quarterlya review of
earnings
human reading of phrase-level surface validity and excerpt snippets. Finally, the patterns of corporate
conference calls from 2017 to 2024. Subsequently, the validity of the measurement indicators is verified through a review of ESG risk disclosure are
demonstrated
human readingfrom both industry
of phrase-level and time
surface perspectives.
validity and excerpt These resultsFinally,
snippets. indicatethe
thatpatterns
text-based measurements
of corporate provide
ESG risk a reliable
disclosure are
quantification from
demonstrated of corporate-level
both industryESGand risk
timeexposure.
perspectives. By measuring
These resultscorporate ESG
indicate thatrisks, potential
text-based risks in environmental,
measurements social,
provide a reliable
and governanceofaspects
quantification can be timely
corporate-level ESG identified,
risk exposure. thus Bycontributing
measuringtocorporate
proactiveESG
risk management.
risks, potential risks in environmental, social,
© 2024 The Authors. Published by ELSEVIER B.V.
and governance aspects can be timely identified, thus contributing to proactive risk management.
© 2024
This The
is an Authors.
open accessPublished by Elsevier
article under B.V.
the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0)
© 2024 The Authors. Published by ELSEVIER B.V.
This is an open
Peer-review access
under article under
responsibility of the scientific
the CC BY-NC-ND license
committee (https://creativecommons.org/licenses/by-nc-nd/4.0)
of the 11th International Conference on Information Technology and
This is an
Peer-review open access
under article under
responsibility of the
the CC BY-NC-ND
scientific committeelicense
of (https://creativecommons.org/licenses/by-nc-nd/4.0)
the 11th International Conference on Information Technology and
Quantitative under
Peer-review Management
responsibility of the scientific committee of the 11th International Conference on Information Technology and
Quantitative
Keywords: Management
natural language processing; ESG risk; risk disclosures
Quantitative Management
Keywords: natural language processing; ESG risk; risk disclosures

1. Introduction
1. Introduction
In the 21st-century business environment, the significance of ESG (Environmental, Social, Governance) factors is
increasingly pronounced.
In the 21st-century Within
business corporations,
environment, the sustainability
significance ofisESG
no longer relegated Social,
(Environmental, to the Governance)
auxiliary concerns
factors of
is
corporate
increasinglysocial responsibility
pronounced. departments
Within but issustainability
corporations, instead a fundamental issuerelegated
is no longer at the level
to of
thechief executives
auxiliary of core
concerns of
business
corporateoperations [1].
social responsibility departments but is instead a fundamental issue at the level of chief executives of core
business operations [1].

* Corresponding author. Tel.: +86-531-88365157; fax: +86-531-88365157.

* Corresponding [email protected]
E-mail address:author. Tel.: +86-531-88365157; fax: +86-531-88365157.
E-mail address: [email protected]
1877-0509 © 2024 The Authors. Published by ELSEVIER B.V.
This is an open
1877-0509 access
© 2024 Thearticle under
Authors. the CC BY-NC-ND
Published by ELSEVIER license
B.V.(https://creativecommons.org/licenses/by-nc-nd/4.0)
Peer-review
This under
is an open responsibility
access of the
article under the scientific
CC BY-NC-NDcommittee of the
license 11th International Conference on Information Technology and Quantitative
(https://creativecommons.org/licenses/by-nc-nd/4.0)
Management
Peer-review under responsibility of the scientific committee of the 11th International Conference on Information Technology and Quantitative
Management
1877-0509 © 2024 The Authors. Published by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0)
Peer-review under responsibility of the scientific committee of the 11th International Conference on Information
Technology and Quantitative Management
10.1016/j.procs.2024.08.142
694 Fei Du et al. / Procedia Computer Science 242 (2024) 693–700
2 Fei Du et al. / Procedia Computer Science 00 (2024) 000–000

Existing research primarily focuses on investigating the relationship between ESG (Environmental, Social,
Governance) factors and investment performance, risk, and corporate performance. Pedersen, Lasse Heje elucidates
the potential costs and benefits of ESG-based investments, explaining how the adoption of ESG influences portfolio
selection and asset pricing [2]. Bang, Jeongseok analyzes the behavioral changes of three types of investors before
and after ESG conflict events occur, as well as investors' reactions to E/S/G controversies when examined separately
[3]. Lin, Shu-Ling employs ESG to predict bank bankruptcies [4]. Meng, Tiantian considers ESG factors in stock
prediction models [5]. Cui, Wei investigates the adjustment effects based on shareholder network and examines the
impact of media attention on the quality of corporate environmental, social, and governance disclosures [6]. Veltri,
Stefania studies whether incorporating ESG factors enhances the efficiency of utility companies using Data
Envelopment Analysis (DEA) models, and whether banks successfully optimize their portfolios by considering ESG
ratings when selecting utility companies [7]. Moskovics, Pedro explores the directional causality between market
structure (competition and concentration), ESG performance, and corporate efficiency under the Stochastic
Structure-Relations Planning (SSRP) model [8]. Teti, Emanuele examines the impact of post-merger environmental,
social, and governance scores on operational performance [9]. Yu, Xiaoling extracts ESG-related terms from
independent corporate social responsibility reports of Chinese A-share listed companies from 2009 to 2020 and
investigates the influence of ESG on annual stock returns and portfolio returns [10]. Additionally, some literature
studies ESG integration, demonstrating that integrating ESG standards enhances corporate sustainability
performance [11,12].
The majority of ESG literature focuses more on the ultimate ESG performance of companies (ESG rating data),
with less research on ESG risks and a lack of measurement for corporate ESG risks. Capelli, Paolo addresses the
regulatory requirements of incorporating sustainability-related risks into traditional financial risk measurement
standards. The new metric, VaRESG, considers the orthogonality of ESG standards with fundamental variables,
applying perturbation methods and the entropy function of ESG factors [13]. Ielasi, Federica employs entropy of
ESG scores to measure ESG risks [14].
In this paper, we propose a text mining-based approach to measure the extent of corporate disclosure of ESG risks.
The organization of this paper is as follows: Section 2 introduces the text mining of the corpus and the benefits of
using such text. Section 3 presents the proposed measure and descriptive statistical results. Section 4 validates the
results. Section 5 analyzes the characteristics of corporate ESG risk disclosure across different industries and time
periods. Finally, the concluding section of the paper provides research insights.

2. Earnings conference call data

We utilize textual data from quarterly earnings conference calls held by publicly listed companies to construct
metrics for assessing the degree of corporate ESG risk. Publicly traded companies are required to disclose their
earnings data and related financial statements every quarter. Alongside this disclosure, companies typically arrange a
conference call with stock analysts, during which the Chief Financial Officer or Chief Executive Officer first
provides an overview of various aspects of the company and then responds to 10 to 20 questions posed by the
analysts. These conference calls are centered around the topic of company profitability. Therefore, the discussion of
ESG content in earnings conference calls, compared to other forms of ESG disclosure by companies (such as
corporate ESG reports), provides a better perspective on the company's level of concern regarding ESG risks from
the standpoint of earnings and operations. Additionally, earnings conference calls have significant advantages in
terms of spontaneity and timeliness of content. Most conversations are unscripted, reducing the amount of pre-
prepared boilerplate language. While corporate ESG reports and regulatory filings such as 10-K forms are typically
released annually, earnings conference calls are held quarterly, providing more timely information.
The metrics utilize the entire content of the earnings conference calls, including management's presentation of the
company's status and the Q&A session with analysts. The data is sourced from The Motley Fool website, spanning
the years 2017 to 2024. The final sample comprises 14,568 observations from 1,700 companies listed on the New
York Stock Exchange (NYSE) and NASDAQ.
 Fei Du et al. / Procedia Computer Science 242 (2024) 693–700 695
Fei Du et al. / Procedia Computer Science 00 (2024) 000–000 3

3. Corporate ESG risk measurement

3.1. Construction of ESG Lexicon

Text analysis in ESG research is still in its nascent stage, and there is currently no comprehensive lexicon of
ESG-related vocabulary. The manual selection of phrases by researchers alone to form a corporate ESG lexicon is
susceptible to factors such as researchers' subjective perceptions and level of expertise, resulting in poor
completeness and comprehensiveness of the constructed lexicon. To build a comprehensive ESG vocabulary, we
utilize the word2vec algorithm to search for words similar to seed words in the corpus.
Google introduced Word2vec, a software tool for training word vectors, in 2013. Word2vec efficiently converts a
word into a vector form based on the given corpus using an optimized training model [15]. Compared with large
language models such as BERT GPT, it has a smaller computational complexity and relatively faster computational
speed, which can effectively solve this problem. First, we constructed a corpus consisting of 75 corporate ESG
reports and 100 ESG-related news articles. It is worth noting that in the future, real-time enterprise ESG reports and
ESG related news can be updated to enable the constructed ESG dictionary to adapt to the development and changes
of the times. Next, we used Word2vec to represent word pairs (composed of two words) in the corpus as vectors.
Then, for the "E," "S," and "G" dimensions of ESG, we selected "carbon emission," "social responsibility," and
"corporate governance" as seed words, respectively. Finally, we calculated the cosine similarity between seed words
and each word pair to characterize the similarity between words. Following the convention of text similarity in
large-scale text corpora, if the cosine similarity is greater than 0.3, the word pair is defined as similar to the seed
word pair. Ultimately, sets of 295, 397, and 466word pairs were constructed for the "E," "S," and "G" dimensions,
respectively, serving as dictionaries for each dimension.

3.2. Construction of corporate ESG risk measurement indicators

Using the constructed ESG lexicon and corporate business lexicon mentioned above, we can capture disclosure
statements from each company regarding ESG theme risks. These statements each contain potential risks that the
company faces in terms of corporate ESG risk.
We construct corporate ESG risk measurement indicators by calculating the frequency of ESG terms and risk
terms mentioned in the same sentence. Existing dictionaries are used for risk terms [16], and the length of the
earnings conference call text is scaled accordingly.

B i ,t
1
ESGrisk i , t 
Bi ,t
 (1[b 
b
]  1[ b , m  S ]) (1)

Where b=0,1, …, Bi,t represents the occurrence of word pairs in the earnings conference call text of company i in
the t-th quarter, and 1[·] denotes the indicator function. We employ the same method to construct integration
indicators for the environmental, social, and governance dimensions separately, denoted as Eriski,t, Sriski,t, and
Griski,t, respectively, and m represents the word pairs containing the risk lexicon. S denotes sentence units in the text.
Table 1 presents descriptive statistics of ESG risk measurement (for illustrative purposes, the measurement
indicators are multiplied by 103).

Table 1. Summary statistics.

Mean STD N
ESGriski,t 0.037 0.13 14568
Eriski,t 0.006 0.04 14568
Sriski,t 0.008 0.05 14568
Griski,t 0.024 0.12 14568
696 Fei Du et al. / Procedia Computer Science 242 (2024) 693–700
4 Fei Du et al. / Procedia Computer Science 00 (2024) 000–000

4. Validation

4.1. Surface validity of word pairs

Table 2, Table 3 and Table 4 present the top 25word pairs from each of the three dictionaries. These words
intuitively relate to their respective themes, indicating that using Word2vec yields effective results. Additionally, the
related word pairs also reflect the multidimensionality of the corresponding themes, capturing various aspects
related to ESG, including environmental protection, social issues, and governance. For example, among all the word
pairs from each of the three dictionaries, in the Environmental (E) dimension, word pairs capture not only terms
related to carbon reduction (e.g., "lower carbon" ,"carbon footprint", "emission reductions") but also terms related to
climate (e.g., "climate goals", "climate action"), water resources (e.g., "water conservation", "water efficiency"),
environmental protection (e.g., "vegetation management", "environmental issues"), and environmental policies (e.g.,
"Paris Agreement"). In the Social (S) dimension, terms such as "responsible sourcing" and "employee wellness" are
captured. In the Governance (G) dimension, terms such as "governance goals", "appointments remuneration", and
"independent auditor" are captured.Afterwards, experts will review and supplement the constructed ESG dictionary
to reduce the subjectivity of ESG risk capture.

Table 2. Dictionary of the E dimension.

Bigram Bigram Bigram Bigram Bigram

lower carbon reduce carbon air quality carbon neutrality Paris Agreement

carbon footprint global warming transition low-carbon pollution control methane emissions

water usage land use carbon neutral solar panels emissions intensity

fossil fuels reducing emissions environmental impacts ambition zero impacts biodiversity

water use climate goals alternative energy environmental footprint energy efficiency

Table 3. Dictionary of the S dimension.

Bigram Bigram Bigram Bigram Bigram

civil rights employees third communication channels develop skills employee resource
community relations corporate responsibility sustainable sourcing employee safety occupational health
employees contractors employee training individual rights worker health healthcare systems
responsible sourcing helping clients green supply value proposition communities live
cultural heritage employees across works council employee health employee wellbeing

Table 4. Dictionary of the G dimension.

Bigram Bigram Bigram Bigram Bigram
board matters delegating authority participation executive director executive market registrant
officer corporate evaluation disclosure concerning executive contained definitive related stockholder
governance goals committee audit responsibilities audit human capital independent auditor
preapproval policies executive compensation director corporate lobbying public Emerson meeting
supervision participation audit committee public responsibility finance professionals conducted evaluation

4.2. Excerpt review

Table 5 provides excerpts from the highest-scoring ESG risk disclosures of three companies, which are
statements identified by the algorithm as companies disclosing ESG risks. For instance, Emergent BioSolutions Inc
 Fei Du et al. / Procedia Computer Science 242 (2024) 693–700 697
Fei Du et al. / Procedia Computer Science 00 (2024) 000–000 5

(EBS), operating in the Biotechnology sector, raises concerns about public health threats, stating that the opioid
crisis has been a longstanding public health threat, claiming too many lives, and the pandemic has exacerbated the
situation; ONE Gas Inc (OGS) discusses actively the wastewater treatment plants and landfills owned by five cities
within its service area, exploring the possibility and risks of capturing methane emissions from these facilities for
transport on our system; Cerus Corporation (CERS) mentions the uncertainty of the impact of the 2019 coronavirus
disease on the global healthcare system and economy affecting revenues. These excerpted statements all specifically
address the ESG risk status of the respective companies.

Table 5. Snippets from top 3 companies

Firm Industry Time ESG Bigram Snippet
Finally, a third pillar of our 2024 strategy was to continue our focus
on public health threats, while diversifying our customer base
beyond the U. As you know, the opioid crisis has been a public
health threat, and it's claimed far too many lives and made even
Emergent
Q2 worse by the pandemic. Regardless of the outcome of the appellate
BioSolutions Biotechnology public health
2021 court's ruling, we continue to focus on the public health threat and
Inc (EBS)
our role as a provider of solutions to address the opioid epidemic.
We're on track with our 2024 strategy and Emergent is well
positioned to play a meaningful role in strengthening our national
public health threat preparedness.
We are also in active discussions with five city-owned wastewater
treatment plants and landfills within our service areas about the
methane
possibility of capturing methane emissions from those facilities for
emissions;
transportation on our system. We will continue development and
wastewater
ONE Gas Inc Q1 deployment of this technology throughout our territory, meaning that
Gas Utilities treatment;
(OGS) 2021 combined with our probabilistic risk ranking replacement program,
gas emissions;
we now have highly efficient technology deployed that positively
environmental
impacts capital planning and maintenance activities in our field
goals
operations, supporting both our safety goals and our environmental
goals.
healthcare
While uncertainty about the COVID-19 impact on the global
Cerus Healthcare systems;
Q2 healthcare systems and economy persist in the second half of 2020,
Corporation Equipment and global healthcare;
2020 we are confident in our ability to deliver on our guidance of $89
(CERS) Supplies global health;
million to $93 million in product revenue this year.
healthcare system

5. Discussion

5.1. Industry differences

We computed averages by industry (based on the industry dimension of the Global Industry Classification
Standard, GICS) and listed the top 10 ranked industries in the table.
The sector with the highest level of Environmental risk disclosure is Electric Utilities, followed by Gas Utilities
and Multi-Utilities. The results are shown in Table 6.
698 Fei Du et al. / Procedia Computer Science 242 (2024) 693–700
6 Fei Du et al. / Procedia Computer Science 00 (2024) 000–000

Table 6. Eriski,t
Industry Mean STD N

Electric Utilities 0.0555 0.1130 212

Gas Utilities 0.0550 0.1691 48
Multi-Utilities 0.0343 0.0732 96
Independent Power and Renewable 0.0337 0.0948 78
Water Utilities 0.0247 0.0757 49
Automobiles 0.0235 0.0723 72
Chemicals 0.0226 0.0613 246
Construction Materials 0.0215 0.0857 74
Oil, Gas and Consumable Fuels 0.0171 0.0611 658
Containers and Packaging 0.0159 0.0599 126

For Social risk disclosure, the highest-ranking sector is Diversified Financial Services, followed by Water
Utilities and Healthcare Providers and Services. The results are shown in Table 7.

Table 7. Sriski,t
Industry Mean STD N

Diversified Financial Services 0.2967 0.9337 14

Water Utilities 0.1802 0.2635 49
Healthcare Providers and Services 0.1405 0.2181 306
Health Care Technology 0.1270 0.2309 92
IT Services 0.0768 0.1773 459
Electric Utilities 0.0750 0.1165 212
Multi-Utilities 0.0690 0.1148 96
Metals and Mining 0.0679 0.1576 358
Building Products 0.0673 0.1188 143
Distributors 0.0666 0.1002 17

In terms of Governance risk disclosure, the highest-ranking sector is Multiline Retail, followed by Paper and
Forest Products and Wireless Telecommunication Services. The results are shown in Table 8.

Table 8. Griski,t
Industry Mean STD N

Multiline Retail 0.0659 0.0763 38

Paper and Forest Products 0.0626 0.1014 28
Wireless Telecommunication Serv 0.0427 0.0834 80
Health Care Technology 0.0407 0.1115 92
Thrifts and Mortgage Finance 0.0342 0.1245 179
Household Durables 0.0288 0.0625 205
Commercial Services and Supplier 0.0279 0.0637 224
Trading Companies and Distribution 0.0261 0.0600 165
Biotechnology 0.0256 0.0780 641
Banks 0.0229 0.1050 583
 Fei Du et al. / Procedia Computer Science 242 (2024) 693–700 699
Fei Du et al. / Procedia Computer Science 00 (2024) 000–000 7

5.2. Temporal changes

This paper calculates ESG risk disclosure values and plots them over time. With the gradual rise of ESG topics,
companies exhibited higher ESG disclosure values from 2019 to 2022, which aligns with the temporal trend of the
number of documents on ESG topics found in the Web of Science (WOS) search. In the fourth quarter of 2019, due
to the pandemic, companies reached the peak of ESG risk disclosure values. Subsequently, as the pandemic was
effectively controlled and business sector attention shifted, there was a significant decrease in ESG risk disclosure
values. The results are shown in Fig. 1.

Fig. 1. ESG risk

6. Conclusions

This paper introduces a novel text analysis method for measuring corporate ESG risk. By extracting content from
detailed discussions between management and stock analysts during quarterly earnings conference calls, this study
first develops an ESG corpus, utilizing word vector models to construct four intuitive and comprehensive
dictionaries of ESG-related vocabulary. Subsequently, this paper uses a combined bag-of-words approach with
lexicons of risk terms to obtain a corporate-level ESG risk assessment indicator. Furthermore, human review
confirms that it indeed captures the extent of ESG risk disclosure by companies. Finally, ESG risk magnitudes are
calculated for different industries and time periods.
This measurement has the following managerial implications. Firstly, it can serve as an explanatory variable for
empirical testing with other operational variables, providing new insights into addressing ESG risks or evaluating
the disclosure value of ESG risks. Current research primarily relies on ESG rating data measured by different
institutions, which suffer from issues such as differing indicator systems and limited accessibility. In contrast, this
measurement focuses on corporate ESG risks, utilizing quarterly earnings conference calls with a high update
frequency, thus providing practitioners and decision-makers with a precise and timely risk monitoring tool to
quantify where and to what extent ESG risks are occurring, enabling anticipation of the sustainability and resilience
of enterprises. Secondly, it allows for dynamic assessment and prediction of the sustainability capabilities of
different industries across industry and time dimensions.

Acknowledgements

This research was funded by the National Natural Science Foundation of China [grant numbers 72271144,
72134004]
700 Fei Du et al. / Procedia Computer Science 242 (2024) 693–700
8 Fei Du et al. / Procedia Computer Science 00 (2024) 000–000

References

[1] Liu, K. J. R. (2013) “Sustainable finance.” IEEE Signal Processing Magazine 30(6): 6–6.
[2] Pedersen, L. H., Fitzgibbons, S., and Pomorski, L. (2021) “Responsible investing: the ESG-efficient frontier.” Journal of Financial
Economics 142(2): 572–597.
[3] Bang, J., Ryu, D., and Yu, J. (2023) “ESG controversies and investor trading behavior in the Korean market.” Finance Research Letters 54:
103750.
[4] Lin, S.-L., and Jin, X. (2023) “Does ESG predict systemic banking crises? A computational economics model of early warning systems with
interpretable multi-variable LSTM based on mixture attention.” Mathematics 11(2): 410.
[5] Meng, T., Yahya, M. H., and Chai, J. (2022) “Deep learning model for stock excess return prediction based on nonlinear random matrix and
esg factor.” Mathematica Problems in Engineering 2022: 5239493.
[6] Cui, W., Chen, X., Xia, W., and Hu, Y. (2023) “Influence of media attention on the quality of environmental, social, and governance
information disclosure in enterprises: an adjustment effect based on the shareholder relationship network.” Sustainability 15(18): 13919.
[7] Veltri, S., Bruni, M. E., Iazzolino, G., Morea, D., and Baldissarro, G. (2023) “Do ESG factors improve utilities corporate efficiency and
reduce the risk perceived by credit lending institutions? An empirical analysis.” Utilities Policy 81:101520.
[8] Moskovics, P., Wanke, P., Tan, Y., and Gerged, A. M. (2024) “Market structure, ESG performance, and corporate efficiency: Insights from
Brazilian publicly traded companies.” Business Strategy and the Environment 33(2): 241–262.
[9] Teti, E., and Spiga, L. (2023) “The effect of environmental, social and governance score on operating performance after mergers and
acquisitions.” Business Strategy and the Environment 32(6): 3165–3177.
[10] Yu, X., Xiao, K., and Xu, T. (2023) “Does ESG profile depicted in CSR reports affect stock returns? Evidence from China.” Physical A-
Statical Mechanics and its Applications 627: 129118.
[11] Barbosa, A. de S., da Silva, M. C. B. C., da Silva, L. B., Morioka, S. N., and de Souza, V. F. (2023) “Integration of Environmental, Social,
and Governance (ESG) criteria: Their impacts on corporate sustainability performance.” Humanities & Social Sciences Communications
10(1): 410.
[12] Nosratabadi, S., Mosavi, A., Shamshirband, S., Zavadskas, E. K., Rakotonirainy, A., and Chau, K. W. (2019) “Sustainable Business Models:
A Review.” Sustainability 11(6): 1663.
[13] Capelli, P., Ielasi, F., and Russo, A. (2023) “Integrating ESG risks into value-at-risk.” Finance Research Letters 55: 103875.
[14] Ielasi, F., Capelli, P., and Russo, A. (2021) “Forecasting volatility by integrating financial risk with environmental, social, and governance
risk.” Corporate Social Responsibility and Environmental Management 28(5): 1483–1495.
[15] Church, K. W. (2017) “Emerging Trends Word2Vec.” Natural Language Engineering 23(1): 155–162.
[16] Hassan TA, Hollander S, Van Lent L, and Tahoun A (2019) “Firm-level political risk: Measurement and effects.” Quarterly Journal of
Economics 134(4): 2135–2202.