Paper: arXiv 2508.18679
Authors: Zhi Chen, Zachary Feinstein, Ionut Florescu
Abstract
Environmental, Social, and Governance (ESG) factors aim to provide non-financial insights into corporations. In this study, we investigate whether we can extract relevant ESG variables to assess corporate risk, as measured by logarithmic volatility. We propose a novel Hierarchical Variable Selection (HVS) algorithm to identify a parsimonious set of variables from raw data that are most relevant to risk. HVS is specifically designed for ESG datasets characterized by a tree structure with significantly more variables than observations. Our findings demonstrate that HVS achieves significantly higher performance than models using pre-aggregated ESG scores. Furthermore, when compared with traditional variable selection methods, HVS achieves superior explanatory power using a more parsimonious set of ESG variables. We illustrate the methodology using company data from various sectors of the US economy.
Complexity vs Empirical Score
- Math Complexity: 6.5/10
- Empirical Rigor: 7.0/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: The paper employs advanced statistical techniques like the Box-Cox transformation and a novel hierarchical selection algorithm, demonstrating moderate-to-high mathematical density. It validates the methodology on real-world US corporate data, compares it against benchmarks (Lasso, PCA), and presents robust metrics like R-squared improvements, showing significant empirical rigor.
Research Flowchart
flowchart TD
A["Research Goal: Identify relevant ESG variables<br>to assess corporate risk (logarithmic volatility)"] --> B["Input: US Company Data<br>Raw ESG Variables with Tree Structure"]
B --> C["Hierarchical Variable Selection HVS Algorithm"]
C --> D{"Comparison & Evaluation"}
D --> E["Model 1: Pre-aggregated ESG Scores"]
D --> F["Model 2: Traditional Variable Selection"]
D --> G["Model 3: HVS Algorithm"]
E & F & G --> H["Outcomes"]
H --> I["HVS achieves superior explanatory power"]
H --> J["HVS uses a more parsimonious set of variables"]
H --> K["HVS significantly outperforms pre-aggregated scores"]
Evidence Card
What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card
Proposes
The paper proposes a Hierarchical Variable Selection (HVS) framework designed for ESG datasets with a hierarchical structure and many more variables than observations.
Source sentence
We propose a framework specifically designed for ESG datasets characterized by a hierarchical data structure and a significantly larger number of variables than observations.
Assumes 5
- Logarithmic transformation of volatility satisfies model assumptions and is used as the response variable. statistical
Source sentence
The results suggest that a logarithmic transformation (λ = 0) of volatility satisfies the model assumptions, as illustrated in Table 1.
- Missing values for boolean variables are set to zero and numerical variables with less than 80% coverage are removed. data
Source sentence
Many of the raw variables in the dataset contain missing values. We follow Cheng et al. (2014); Clarkson et al. (2008) by assigning 0 to missing values for boolean variables. We then remove numerical variables that have less than 80% data available.
- Observations with any remaining missing values after processing are excluded from the analysis. data
Source sentence
Finally, any company-year observation that still contains missing values after these steps is removed.
- The residuals of the regression are heteroskedastic and right‑skewed, requiring a transformation. statistical
Source sentence
Consistent with these economic phenomena, our diagnostic checks reveal that the model’s residuals are heteroskedastic and the distribution is right‑skewed.
Show 1 more
- The hierarchical structure of ESG data (tiers of raw variables, category scores, pillar scores, overall score) is appropriate for partitioning variables in the selection process. model
Source sentence
LSEG uses the category level (Tier 2 in Figure 1) in order to partition raw variables.
Data
- LSEG ESG Dataset alternative commercial
Source sentence
Taking the Energy sector as an example, the initial dataset has 617 variables and 695 company‑year data points.
Costs
Not applicable
Finds 5
- Raw ESG variables selected by HVS are significantly more relevant to financial risk than aggregated ESG scores. statistical
Source sentence
We show that raw variables selected by the proposed framework are significantly more relevant to financial risk than aggregated ESG scores.
- Using logarithmic volatility improves model fit, increasing R‑squared compared with regular volatility. statistical
Source sentence
Using logarithmic volatility also leads to an increase in the R-squared value of our framework, indicating improved model fit and explanatory power.
- HVS outperforms stepwise regression out‑of‑sample, showing lower MSE and stronger robustness. performance
Source sentence
These results indicate that the HVS framework reliably identifies relevant variables while keeping strong robustness on out-of-sample data.
- Selected raw ESG variables provide new information beyond traditional financial risk factors. predictive-accuracy
Source sentence
The selected raw ESG variables bring new information to traditional financial risk factors.
- Relevant ESG risk factors differ across sectors and between large‑ and small‑cap firms. other
Source sentence
different sectors have different relevance to ESG risk. Furthermore, our results reveal that the relevant ESG risk factors differ between large- and small-cap companies.
Checked by others
Supported by
- An Axiomatic Risk-Reward Framework for Sustainable Investing
Compare the findings
This paper: Raw ESG variables selected by HVS are significantly more relevant to financial risk than aggregated ESG scores.
That paper: Aggregated data show a limited relationship between ESG scores and financial performance (returns and volatility).
Code
No repository by the authors found in the paper.
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.
Cited References (90)
- ESG for all? The impact of ESG screening on return, risk, and diversification (2016)
- PG&E: The first climate-change bankruptcy, probably not the last (2019)
- Climate change 2014: mitigation of climate change (2015)
- Impacts of climate change on energy systems in global and regional scenarios (2020)
- SDGs and ESG disclosure regulation: is there an impact? Evidence from Top-50 world economies (2020)
- The impact of corporate reputation and reputation damaging events on financial performance: Empirical evidence from the literature (2015)
- The volkswagen scandal (2016)
- ESG Rating—Necessity for the Investor or the Company? (2021)
and 82 more
- ESG rating disagreement and stock returns (2021)
- Aggregate confusion: The divergence of ESG ratings (2022)
- Improving ESG scores with sustainability concepts (2022)
- On the validity of environmental performance metrics (2015)
- Do ratings of firms converge? Implications for strategy research (2014)
- Measuring the level and risk of corporate responsibility–An empirical comparison of different ESG rating approaches (2015)
- An analysis of the environmental information in international companies according to the new GRI standards (2018)
- CSR reporting as an important tool of CSR communication (2015)
- CSR reporting of companies on a global scale (2016)
- The effects of the EU non-financial reporting directive on corporate social responsibility (2022)
- Corporate social responsibility, Shariah-compliance, and earnings quality (2017)
- ESG ratings and stock performance during the COVID-19 crisis (2021)
- Does the ESG index affect stock return? Evidence from the Eurostoxx50 (2020)
- Foundations of ESG investing: How ESG affects equity valuation, risk, and performance (2019)
- Does it pay to treat employees well? International evidence on the value of employee-friendly culture (2018)
- Legitimacy theory: A story of reporting social and environmental matters within the Australian food and beverage industry (2007)
- Exposure, legitimacy, and social disclosure (1991)
- Integrating ESG in portfolio construction (2019)
- Integrating sustainability risks in asset management: the role of ESG exposures and ESG ratings (2020)
- ESG performance and stock price volatility in public health crisis: evidence from COVID-19 pandemic (2021)
- ESG and financial performance: aggregated evidence from more than 2000 empirical studies (2015)
- ESG and financial performance (2021)
- Carbon risk factor framework (2022)
- Can ESG indices improve the enterprises’ stock market performance?—An empirical study from China (2019)
- Environmental, social, governance risk and corporate sustainable growth nexus: Quantile regression approach (2021)
- Do ESG Ratings Reduce the Asymmetry Behavior in Volatility? (2022)
- Forecasting volatility by integrating financial risk with environmental, social, and governance risk (2021)
- ESG factors and risk-adjusted performance: a new quantitative model (2016)
- Is Corporate Social Responsibility investing a free lunch? The relationship between ESG, tail risk, and upside potential of stocks before and during the COVID-19 crisis (2022)
- Volatility modeling and dependence structure of esg and conventional investments (2022)
- Forecasting volatility persistence: Evidence from international stock markets (2023)
- Climate Risk, ESG Performance, and ESG Sentiment for US Commercial Banks (2022)
- A new quality factor: Finding alpha with ASSET4 ESG data (2010)
- ASSET4 Assetmaster professional reference guide (2011)
- Thomson Reuters ESG Scores (2017) — link
- Corporate social responsibility: A review of empirical research using Thomson Reuters Asset4 data (2022)
- ESG Factors and Firms’ Credit Risk (2024)
- Restoring trust in ESG investing through the adoption of just transition ethics (2024)
- Is 2024 past peak ESG? (2024) — [link](https://corpgov.law.harvard.edu/2024/09/16/is-2024-past-peak-esg/#:~:text=After%20years%20of%20ESG%20(and,%249%20billion%20outflow%20in%202023.)
- GLOBAL SUSTAINABLE INVESTMENT REVIEW 2022 (2022) — link
- The wages of social responsibility—where are they? A critical review of ESG investing (2015)
- ESG impact on performance of US S&P 500-listed firms (2020)
- ESG rating and market valuation of the firm: Sector approach (2022)
- ESG in China: A review of practice and research, and future research avenues (2023)
- Exploring the relationship of ESG score and firm value using cross-lagged panel analyses: Case of the Indian energy sector (2022)
- Board characteristics, social trust and ESG performance in the European banking sector (2023)
- The effect of financial materiality on ESG performance assessment (2021)
- Revisiting the relation between environmental performance and environmental disclosure: An empirical analysis (2008)
- Corporate social responsibility and access to finance (2014)
- The principle of parsimony (1981)
- Does ESG really matter—and why (2022)
- An Empirical Analysis on Performance Inconsistency among Environmental, Social and Governance Components of ESG Ratings (2024)
- A five-factor asset pricing model (2015)
- The volatility of a firm’s assets and the leverage effect (2016)
- A sparsity algorithm for finding optimal counterfactual explanations: Application to corporate credit rating (2023)
- Ridge Regression: Biased Estimation for Nonorthogonal Problems (1970) — link
- ESG in the financial industry: What matters for rating analysts? (2023)
- An overview of the European policies on ESG in the banking sector (2021)
- Can sustainable investment yield better financial returns: A comparative study of ESG indices and MSCI indices (2019)
- Application of deep neural networks to assess corporate credit rating (2020) — link
- Environmental, social and governance scores from Refinitiv (2022)
- ESG: Research progress and future prospects (2021)
- Corporate social and environmental reporting: a review of the literature and a longitudinal study of UK disclosure (1995)
- A systematic literature review on the determinants of sustainability reporting systems (2023)
- Take it with a pinch of salt—ESG rating of stocks and stock indices (2022)
- Less is more: Optimizing classification performance through feature selection in a very-high-resolution remote sensing object-based urban application (2018)
- On model selection consistency of Lasso (2006)
- LASSO and elastic net tend to over-select features (2023)
- ESG and volatility risk: International evidence (2023)
- Pricing transition risk with a jump-diffusion credit risk model: evidences from the CDS market — → 2303.12483
- The Impact of Controversial Events on Corporate Resilience: The Chain-Mediating Role of ESG and Value-at-Risk (2025)
- The distribution of realized stock return volatility (2001)
- Are ESG ratings informative to forecast idiosyncratic risk? (2025)
- Global pricing of carbon-transition risk (2023)
- Dissecting green returns (2022)
- Why did individual stocks become more volatile? (2006)
- Firm-specific, macroeconomic factors and stock price risk for Jordanian banks (2021)
- Idiosyncratic volatility and cash flow volatility: New evidence from S&P 500 (2018)
- Does the market value environmental performance? (2001)
- Firm-value effects of carbon emissions and carbon disclosures (2014)
- Five ways that ESG creates value (2019)
- Do higher-rated or enhancing ESG of firms enhance their long–term sustainability? Evidence from market returns in Korea (2020)
1 of 90 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.