Sie können Operatoren mit Ihrer Suchanfrage kombinieren, um diese noch präziser einzugrenzen. Klicken Sie auf den Suchoperator, um eine Erklärung seiner Funktionsweise anzuzeigen.
Findet Dokumente, in denen beide Begriffe in beliebiger Reihenfolge innerhalb von maximal n Worten zueinander stehen. Empfehlung: Wählen Sie zwischen 15 und 30 als maximale Wortanzahl (z.B. NEAR(hybrid, antrieb, 20)).
Findet Dokumente, in denen der Begriff in Wortvarianten vorkommt, wobei diese VOR, HINTER oder VOR und HINTER dem Suchbegriff anschließen können (z.B., leichtbau*, *leichtbau, *leichtbau*).
Der Artikel untersucht die wachsende Bedeutung von Umwelt-, Sozial- und Governance-Prinzipien (Environmental, Social, and Governance, ESG) in Unternehmensstrategien, die von den Forderungen des Marktes und der Investoren nach Nachhaltigkeit angetrieben werden. Sie unterstreicht die zunehmende Zahl von ESG-Vorschriften weltweit und die entscheidende Rolle der Nachhaltigkeitsberichterstattung bei der Einhaltung von Unternehmensvorschriften. Die Studie unterscheidet zwischen Unternehmen, die freiwillig ESG-Praktiken anwenden, und solchen, die diese nur unter regulatorischem Druck einhalten und verwendet ESG-Werte als Messgröße für ihre Nachhaltigkeitsverpflichtung. Er analysiert die Entwicklung der ESG-Werte im letzten Jahrzehnt und identifiziert Unternehmen, die ihre ESG-Bemühungen proaktiv betrieben haben, verglichen mit Unternehmen, die ein langsameres Tempo verfolgt haben. Der Artikel untersucht auch die Auswirkungen der ESG-Integration auf die Finanzleistung, indem er das Fama-Modell und das französische Fünf-Faktor-Modell zur Bewertung des Risikos und der Überschussrenditen heranzieht. Sie zeigt signifikante Unterschiede im Risikoprofil zwischen ESG-Pionieren und ESG-Verfolgern, wobei Pioniere stabilere und konzentriertere Risikoprofile aufweisen. Die Studie unterstreicht die Bedeutung proaktiver Nachhaltigkeitsbemühungen bei der Gestaltung der finanziellen Widerstandskraft und Marktpositionierung eines Unternehmens und liefert wertvolle Erkenntnisse für Investoren und Interessengruppen, die versuchen, die finanziellen Auswirkungen der ESG-Integration zu verstehen.
KI-Generiert
Diese Zusammenfassung des Fachinhalts wurde mit Hilfe von KI generiert.
Abstract
Transitioning to sustainable business models has become a critical challenge for companies globally. While some firms proactively align their strategies with the UN Sustainable Development Goals, others adopt sustainability measures primarily in response to regulatory pressures. This study investigates whether financial markets recognize and value differing approaches to sustainability by clustering listed companies based on the evolution of their ESG scores. Focusing on companies within the S&P Index, we identify two distinct clusters: ESG Pioneers, firms with consistently high ESG scores since 2013, and ESG Chasers, firms that have significantly improved their ESG performance over time. To assess the financial implications of these strategies, we analyze excess returns across the clusters, using the Fama-French framework. Our findings highlight diverse corporate approaches to sustainability, shedding light on the costs and benefits of sustainable business practices in a competitive environment.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
1 Introduction
The market and investors push companies to integrate the principles of the Sustainable Development Goals (SDGs) into their business strategies, driving the transformation toward a sustainable economy. At the moment, there are 2400 Environmental, Social, and Governance (ESG) regulations worldwide, and sustainability reporting is becoming a crucial part of corporate compliance. Companies aiming to stay ahead nowadays, have to understand and comply with these regulations to become leaders in sustainability, setting the standard for corporate responsibility, and aligning with investor and consumer expectations. How do investors and consumers differentiate between companies that voluntarily adopted ESG principles ahead of regulatory mandates and those that aligned only when compelled to remain competitive? Does the voluntary nature of adopting corporate responsible business model influence market perception and, consequently, the financial performance of these companies compared to the late adopters? Using the ESG score as a measure of the company’s sustainability commitment, we analyze their evolution over the last decade, to distinguish between companies who have been ESG proactive with respect to those who have followed a slower pace.
Incorporating ESG criteria into the business process is an ever-evolving concept, shaped by regulation changes, investor preferences, and the search for accurate and reliable evaluation metrics. The European Union, in particular, has been quite active in introducing comprehensive ESG regulations that have far-reaching implications for businesses operating within its jurisdiction. The Corporate Sustainability Reporting Directive (CSRD) effective since 2023, is only one of the several EU directives, it requires a broader range of large companies and non-listed small and medium enterprises (SMEs) to disclose social and environmental information (European Commission 2023). In the US, ESG and non-financial reporting regulations are evolving but lack a unified federal framework. The Securities and Exchange Commission (SEC) has been expanding its focus on ESG disclosures, particularly around climate risks and human capital management, requiring publicly traded companies to disclose material ESG-related information (Securities and Exchange Commission 2024). While not mandatory, industry-specific guidelines from the Sustainability Accounting Standards Board (SASB) are widely adopted by companies to meet investor demands for financially relevant ESG data (International Financial Reporting Standards. International Sustainability Standards Board (2021), Sustainability Accounting Standards Boards (2022)). The absence of standardized regulations presents challenges but also opportunities for companies to lead in transparency, risk management, and competitive advantage. With growing stakeholder pressure and a global emphasis on sustainability, the US is expected to soon introduce more defined ESG reporting requirements.
Anzeige
The relationship between sustainability and financial performance remains blurry, partly due to business sector differences and the wide variety of available sustainability metrics, of which ESG ratings are the most popular. ESG scores, albeit have faced criticism regarding their reliability (see Berg et al. (2022)) and temporal inconsistency, are considered as essential tools for developing investment strategies, implementing effective risk management practices, and meeting disclosure requirements for investors and financial institutions. In its 2023 report, LSEG asserts that its methodologies ensure the consistency of ESG scores, as debated by Berg et al. (2021) (LSEG 2023a). These refinements are intended to align the ratings with evolving standards and improve their utility for ESG-focused strategies.
In this paper, we analyze changes in a company’s ESG score over time to account for its efforts to enhance sustainability. ESG scores often cluster around medium-to-high values, making it challenging to differentiate companies genuinely committed to sustainability from those taking minimal actions to meet standards (D’Ecclesia et al. 2024). Companies that voluntarily implement sustainability initiatives independent of regulatory mandates demonstrate a deeper commitment to these issues. These firms should be analyzed separately from those that merely comply with regulatory requirements. Furthermore, industries at the forefront of the sustainability revolution, such as energy and heavy manufacturing, face greater challenges in improving ESG ratings compared to sectors like finance or utilities, as substantial resource investments are required for meaningful progress (Battiston et al. 2017). By examining the dynamics of ESG scores, we better evaluate a company’s actual commitment to sustainability and categorize firms based on their levels of dedication and effort. In other words, we find who is truly committed and who is just compliant.
We use the evolution of companies’ approaches to sustainability over time to study how the market prices their sustainable choices. We focus on the US market, where companies retain discretion in adopting ESG practices, unlike the more regulated European context. Our analysis is based on a sample of US-listed companies in the S&P 500 index over the period 2013–2022. To capture patterns in ESG score evolution, we employ a hierarchical time-series clustering method, Dynamic Time Warping (DTW), to group companies according to similar ESG score changes rather than static ESG values. We analyze the market characteristics of each cluster to uncover potential differences arising from varying levels of alignment with the SDGs over time. Using the Fama and French five-factor model we investigate if the excess returns of companies belonging to separate clusters are affected by different risk factors.
The main results of the empirical analysis are as follows. Despite various shocks and events in recent geopolitical and market history, such as the COVID-19 outbreak and the Russian-Ukrainian conflict, that has temporarily distracted the economy from the collective push toward sustainability, companies are now more aligned with the UN SDGs than in the early 2000s: overall the average ESG score shows an upward trend. Examining the dynamics of ESG scores over time proves particularly effective in differentiating between voluntary and mandatory sustainable choices among firms. Using the DTW clustering method, we identify two main clusters: ESG Pioneers (71 companies), which exhibit consistently high ESG scores throughout the entire period, and ESG Chasers (235 companies), which significantly increased their ESG scores over time. We find significant differences in the risk exposure of ESG Pioneers and ESG Chasers, which vary across market conditions and economic sectors. While median risk factor sensitivities appear similar, their distributions reveal asymmetries, particularly for the size (SMB) and value (HML) factors. ESG Pioneers show a more stable and concentrated risk profile, whereas ESG Chasers display higher variability, suggesting a less predictable financial structure. Sector-specific dynamics further shape these differences. Defensive sectors like Utilities and Consumer Non-Cyclicals exhibit lower market exposure and negative sensitivity to size and value factors, reflecting their stability and limited reliance on external financing. In contrast, Basic Materials firms, central to the sustainability transition, show greater divergence between clusters. ESG Pioneers in this sector, having made earlier and larger investments in sustainability, display heightened sensitivity to market fluctuations but are positioned to benefit more from ESG-driven growth trends.
Anzeige
The remainder of the paper is organized as follows. Section 2 provides the review of the literature; in Sect. 3 the dynamic clustering methodology and the Fama and French model are recalled; Sect. 4 presents the dataset and the main results, while Sect. 5 concludes.
2 Literature review
Determining whether responsible investments are also profitable is compounded by factors such as the analysis timeframe, sector-specific discrepancies, local sustainability policies, and the abundance of available ESG metrics. In this section, we present the key findings from the literature on this topic.
The financial performance of socially responsible investments (SRI) remains a topic of debate, with some studies reporting under-performance or minimal differences compared to traditional benchmarks (see D’Amato et al. (2022) and references therein). Under certain conditions, ESG investing can enhance risk management and yield returns comparable to, or even exceeding, those from traditional investments (Boffo and Patalano 2020; Zhou et al. 2022; Dumitrescu et al. 2023). Companies with robust ESG practices often benefit from lower capital costs, reduced volatility, and fewer instances of corruption and fraud (Fulton et al. 2012; Lins et al. 2017; Chen et al. 2023; Wang et al. 2023). High ESG scores are generally associated with lower systemic risk, suggesting that strong ESG performance contributes positively to financial stability (Bax et al. 2023; Erhart 2022; Eskantar et al. 2024). For larger firms, ESG factors are positively correlated with profitability and credit ratings (Kim and Li 2021; Michalski and Low 2024), whereas the impact on smaller firms tends to be less significant (Chen et al. 2023). This implies that financial benefits entailed by well-conducted ESG practices possibly depend on company size, business sectors, and context (Fan et al. 2024). In portfolio optimization, the inclusion of ESG constraints lowers the efficient frontier and reduces the maximum attainable Sharpe ratio, highlighting a trade-off between responsible investing and financial returns (Pedersen et al. 2019). Lin et al. (2019) find a positive relationship between corporate financial performance (CFP) and corporate social responsibility (CSR), yet they caution that improving CSR entails costs, which may not always translate into superior CFP. Further evidence supports the trade-off hypothesis, indicating that socially responsible actions can constrain shareholder wealth and profitability. For instance, D’Amato et al. (2023) show through machine learning that companies often need to fundamentally adapt their business models to align with ESG criteria in order to achieve enhanced profitability. However, the lack of clear and consistent criteria for distinguishing ESG-aligned companies contributes to confusion in evaluating their financial performance (Berg et al. 2021; Berg et al. 2022). Disagreement among ESG ratings further complicates the landscape, as Brandon et al. (2019) find that greater divergence in ratings is associated with higher stock returns, potentially reflecting a risk premium tied to environmental disagreements.
Integrating ESG factors into asset pricing models offers mixed evidence on their role as risk drivers, leaving the evidence of ESG as a systematic risk driver as still inconclusive. Some studies sow that ESG captures new dimensions of systematic risk linked to sustainability transitions (Hubel and Scholz 2020; Díaz et al. 2021). For instance, Jin (2018) extend the Fama-French five-factor model by including an ESG-related factor, finding it significantly priced in the US market. However, Naffa and Fain (2022) find opposite results and show that ESG does not represent an additional risk factor. Cornell (2021) argue that while high ESG ratings may lower the cost of capital due to investor preferences, they might also lead to lower expected returns, reflecting the social desirability of ESG investments. Similarly, Sassen et al. (2016) find that stronger CSR reduces firm risk, particularly idiosyncratic and total risk, with the social dimension driving the effect. Environmental performance exhibits industry-specific impacts, while governance appears unrelated to risk. These findings suggest that ESG’s impact on risk may vary across dimensions and contexts. Further complexity arises from methodological differences in constructing ESG factors. Lioui and Tarelli (2022) highlight substantial variability in factor performance depending on data sources and approaches, noting that media attention significantly influences ESG pricing. Finally, Pastor et al. (2022) focus on the green component of ESG, showing that green assets tend to have lower expected returns due to their utility for investors and their hedging properties against climate risks, but they may outperform during positive shifts in investor sentiment toward sustainability.
While extensive research has explored clustering based on CSR or ESG performance (Iamandi et al. 2019; Vilas et al. 2022; Gonzaga et al. 2024), little attention has been given to the dynamic evolution of ESG scores over time or how these scores respond to shifting regulatory frameworks. The existing literature predominantly examines the relationship between corporate financial performance and sustainability measures but often neglects whether the market differentiates between voluntary sustainability efforts and mere compliance with mandatory regulations. Some studies investigate the impact of mandatory ESG disclosure on corporate sustainability worldwide (Aghamolla and An 2023; Chung et al. 2024; Krueger et al. 2024), yet, to the best of our knowledge, no work explicitly evaluates how such disclosures influence financial performance. This gap is critical. We believe that the trajectory of the historical ESG score can reveal the firm’s genuine commitment to sustainability. Companies that embraced sustainability during times of limited attention and became leaders in the field may be perceived, and valued differently, by the market with respect to those that adopted sustainable practices only in response to regulatory pressure. This generates different financial performance: In an era where nearly all firms achieve high ESG scores, distinguishing genuine innovators from regulatory conformists provides additional insights. Examining the evolution of ESG practices can help us determine whether the market rewards true pioneers and evaluate the risks associated with sustainability. This approach not only sheds light on how sustainability efforts are perceived but also explores the broader implications for corporate risk management and financial performance. This paper aims to fill this gap. Using the ESG score dynamics to classify the company’s commitment toward sustainability and analyzing the drivers of its financial outcomes, we reveal whether the market recognizes and rewards authentic sustainability leaders.
3 Methodology
We employ a hierarchical time-series clustering approach to group companies which show similar ESG scores dynamics. We utilize the traditional Fama and French five-factor model to analyze the excess returns of companies with similar sustainability commitments in order to capture different exposures to the various risk factors. In this section, we describe the models and techniques used for our analysis.
3.1 Time series clustering
Time series clustering is typically a challenging task due to the complexity of the data and the difficulty in defining an appropriate similarity measure between time series (see, e.g., Rajabi et al. (2020) for a review of the literature and data clustering methods). Most standard algorithms are effective when a suitable similarity measure for sequential data is specified. However, determining a similarity measure for sequential data is more complex than for traditional data because the order of elements in the sequences must be considered. To compare similarity measures involves addressing issues such as sample uniformity, measurement scale, and time series length. Various similarity measures exist, Dynamic Time Warping (DTW) has become quite common and it is often used to compare two or more time series with different lags, peaks, and slopes by disregarding any shifts and variations in speed (Sakoe and Chiba 1978).
Let \(A=\left\{ a_1, a_2,\ldots, a_i,\ldots, a_T\right\} \) and \(B=\left\{ b_1, b_2,\ldots, b_j,\ldots, b_T\right\} \) be two time series, and let d be a distance between each pair of the series. To achieve the global similarity D(A, B), we determine the distance between A and B according to the DTW method recursively minimizing Eq. (1):
Using the hierarchical clustering approach, we construct a hierarchy of groups that defines an ordered sequence of groupings, where each level of the hierarchy represents a separate set of data into disjoint clusters of observations (Hastie et al. 2009). We compare four hierarchical methods characterized by the linkage criteria (or intergroup dissimilarity): average, complete, Ward.D (Ward 1963), and Ward.D2 (Murtagh and Legendre 2014). The linkage criterion specifies the metric to merge the clusters and obtain the final cluster containing all the observations. In Table 1, we list and describe the linkage criteria used in our analyses, where \(C_1\) and \(C_2\) are two clusters, \(|C_1|\) is the number of observations belonging to cluster \(C_1\), \(|C_2|\) is the number of observations belonging to cluster \(C_2\), and \(d(a_i,b_j)\) the distance between the observations \(a_i\) and \(b_j\).
Number of clusters selection. A suitable clustering method should ensure cluster compactness and cluster separation from each other. To identify the optimal number of clusters we use four Clustering Validity Indexes (CVIs) chosen among the most well known indexes developed in the literature (see Arbelaitz et al. (2013) for a comprehensive overview). The four CVIs are reported in Table 2: (i) Silhouette Index (Sil) (Rousseeuw 1987) evaluates cluster cohesion and separation by measuring how similar a data point is to its own cluster compared to others. A higher score (range:[−1,+1]) indicates better-defined clusters; (ii) Dunn Index (D) (Bezdek and Pal 1988) assesses cluster compactness and separation by comparing the minimum distance between points in different clusters to the maximum distance within the same cluster. Higher values indicate well-separated and compact clusters; (iii) Condition of Order Preservation Index (COP) (Gurrutxaga et al. 2010) measures how well the clustering preserves the original order of data instances. Lower scores indicate greater consistency with the original data order; (iv) Calinski-Harabasz Index (CH) (Calinski and Harabasz 1974) evaluates cluster separation and compactness by comparing the ratio of between-cluster variance to within-cluster variance. Higher scores indicate better separation and internal compactness.
Table 2
Clustering validity indexes
CVI
Description
Objective function
Objective
Sil
Measures how similar an object is to its own cluster
\(\frac{b - a}{\max (a, b)}\)
Maximize
D
Measures the compactness and separation of clusters
Ratio of between-cluster dispersion to within-cluster dispersion
\(\frac{B(k)}{W(k)} \times \frac{N - k}{k - 1}\)
Maximize
3.2 Fama-French five-factor model
The Fama-French five-factor model (FF5F) (Fama and French 2015), quantifies risk and estimates expected equity excess returns considering five market factors derived from the difference between the returns of opposing portfolios. Let \(R_{i,t}\) be the return on stock i at time t and \(R_{f,t}\) the risk-free return for the same period. The FF5F model is designed to capture the relation between the average excess return of security i, the stock market excess return, and four additional factors:
where \(MKT_t\) is the stock market excess return over the risk-free rate of the value-weighted market portfolio (the risk premium), \(SMB_t\) (Small Minus Big, the size premium) is the return on a diversified portfolio of small stocks minus the return on a diversified portfolio of big stocks, \(HML_t\) (High Minus Low, the value premium) is the difference between the returns on diversified portfolios of high and low (B/M) stocks, \(RMW_t\) (Robust Minus Weak, the profitability premium) is the difference between the returns of the diversified portfolios of stocks with robust and weak profitability, and \(CMA_t\) (Conservative Minus Aggressive, the investment strategy premium) is the difference between the returns on diversified portfolios of the stocks of low (conservative) and high (aggressive) investment firms; \(\theta _i = \{b_i, s_i, h_i, r_i,c_i\}\) and \(a_i\) are OLS parameters and, if \(\theta _i\) captures all variation in the expected returns, the model intercept, \(a_i\), is zero for all securities in the sample; and \(e_{it}\) are the zero-mean residuals of the model with constant variance \(\sigma _{e_i}^2\). The model consists of an extension of the Capital Asset Pricing Model (CAPM), which considers the sole risk premium as a risk driver for the excess returns.
4 Empirical analysis
We collect daily prices for all the stocks constituent the S&P 500 index from 2013 to 2022, together with the relative yearly ESG scores (including joiners and leavers). Data are collected from the LSEG database (LSEG 2023a). After cleaning the data by removing companies lacking ESG scores, the dataset consists of 306 companies distributed over the ten different Thomson Reuters Business Classifications (TRBC) Economic Sectors.
Fig. 1
Boxplots of the ESG score distribution (2013–2022)
The distribution of the ESG score among companies is described using a boxplot (Fig. 1). Median ESG scores increase over time while variability decreases. As shown in Table 3, the range of the ESG score drastically shrinks over the ten years, with only a few companies having an ESG score lower than 50 in 2022 (considered outliers). In 2013, the minimum ESG score was 5.64, while in 2022 it rose to 22.85. Similarly, the standard deviation dropped from 19.07 to 11.90. The dynamics of the ESG scores for the US companies witness an increasing attention to sustainability. Some companies exhibited a high ESG score from the early years of the analysis, while others began to increase their values after 2014–2015. For illustration purposes, Fig. 2 shows the ESG score dynamics of three companies: United Health Group (UNH, Healthcare TRBC sector), Conagra (CAG, Consumer Non-Cyclicals TRBC sector), and Freeport McMoRan Copper and Gold Inc. (FCX, Basic Materials TRBC sector). We observe that UNH made significant strides toward a sustainable business model during 2013–2022, with its ESG score rising from 10 to 80 in nine years. In contrast, the other two companies exhibit relatively stable and high ESG scores, with little change from 2013 to 2022.
4.1 Dynamic cluster analysis
We assume the ESG score measures the company’s level of commitment to sustainability; an increase in the ESG score is interpreted as an effort by the company to achieve better alignment with the SDGs. Conversely, a decrease in the score indicates either a deterioration in the company’s sustainability efforts or, more likely, that the company has remained stagnant. To identify clusters of companies with similar attitude toward sustainability, we use a DTW clustering technique. This method allows us to group companies based on the ESG score yearly changes: we assume that companies which show similar ESG score changes are driven by a common attitude toward sustainability.
Searching for the optimal hierarchical clustering of companies, we compare all possible combinations of clusters using various CVIs (see Table 2), allowing the number of clusters to vary within a reasonable range (2–5) and applying the four linkage criteria listed in Table 1. We conduct the cluster analysis using the TSclust R package (Montero and Vilar 2014), which offers several alternative procedures, and the DTW R package developed by Giorgino (2009). We identify two clusters using the Silhouette and CH Indexes, these CVIs determine the best linkage criterion three out of four linkage criteria (Complete, Ward.D, and Ward.D2), identifying Ward.D as the best linkage criterion taking into account the cluster sizes with the aim to have a fair distribution of the companies between the two clusters. (Fig. 3) We validate the optimal number of clusters using the dendrograms shown in Fig. 4.
Fig. 3
ESG score CVI’s comparison. The number of clusters varies between 2 and 5
We group the 306 companies into two clusters: Cluster 1 (\(C_1\)) containing 71 companies, and Cluster 2 with 235 companies (\(C_2\)). The descriptive statistics of the ESG scores of companies in the two clusters are reported in Table 4. Cluster 1 has an average ESG score of 68 with a standard deviation of 15, while Cluster 2 has a lower mean of 57 and a higher standard deviation of 17. The dynamics of the average ESG score in the two clusters differ dramatically as shown in Fig. 5, companies in \(C_1\) exhibit a steady level of ESG scores, high values (68) in 2013, and almost the same value (67) in 2022; while the average ESG score for companies in \(C_2\) shows a steady positive trend, starting at 45 in 2013 and reaching 68 in 2022. This clustering indicates that US companies approach the transition to sustainability in two distinct ways. One group, which we call the “ESG Pioneers” (\(C_1\)), started to be committed as early as 2013 and maintained high ESG scores over the entire period. The other group, the “ESG Chasers” (\(C_2\)), has moved slowly and improved its commitment toward sustainability, reflecting the overall trend of rising ESG scores in recent years.
The contrasting ESG trajectories of the two clusters are far more evident in Fig. 6. The left panel, representing the ESG Pioneers, shows distributions consistently centered around high ESG scores, with only a slight and irregular reduction in the left tails over time. This highlights how companies in this cluster have maintained strong ESG commitments with minimal variation. In contrast, the right panel, representing the Chasers, reveals a clear upward shift in ESG scores. In the early years (2013–2016), their scores were predominantly low, but over time, the distribution moves toward higher values while also narrowing, gradually resembling the Pioneers. Two key trends emerge within the Chasers: (i) a narrowing range of ESG scores, as the distribution gradually loses its platykurtic shape, and (ii) an overall shift toward higher values, reflecting increased ESG commitment. By the final years of the sample, the Chasers’ distribution closely mirrors that of the Pioneers, underscoring a structural shift in corporate ESG behavior.
Fig. 5
Average ESG score for each cluster over the period 2013–2022
A closer look at how ESG scores are distributed within each cluster is possible thanks to the classification of ESG ratings provided by LSEG. They form different rating classes based on the size of the ESG score, precisely they identify the following classes: ‘A’ = (ESG Score 75–100) excellent ESG performance and high transparency in reporting ESG data; ‘B’ = (ESG Score 50–75) denotes good ESG performance and above-average transparency in reporting ESG data; ‘C’ = (ESG Score 25–50) indicates satisfactory ESG performance and moderate transparency in reporting ESG data; ‘D’ = (ESG Score 0–25) reflects poor ESG performance and insufficient transparency in reporting ESG data (LSEG 2023b). Using this classification we report in Fig. 7 the distribution within the two clusters. In Cluster 1, companies are predominantly rated A and B, with B ratings prevailing each year except for 2017. The distribution among classes A and B remains relatively stable, suggesting that companies in \(C_1\) have kept their ESG score quite stable over time, showing a sustainable business as usual model. Companies belonging to Cluster 2 appear quite active with the number of A-rated companies increasing over time, while the number of C-rated and D-rated companies decreases, with D-rated companies disappearing in 2022. This trend highlights that companies in Cluster 2 are constantly improving their sustainability and seeing their ESG ratings adjusted accordingly.
Fig. 7
Distribution of ESG score according to the LSEG Refinitv classification (classes D:0–25, C:25–50, B: 50–75, and A: 75–100) (2013–2022)
The distribution of companies by TRBC sector in the two clusters does not differ much as reported in Fig. 8. We cannot assume a difference in the ESG score patterns due to the business sectors, the only exceptions is found for the Utilities and Consumer Non-Cyclicals sectors. Utilities companies represents 11% in Cluster 1 and only 5% in Cluster 2, while the Consumer Non-Cyclicals companies accounts for 11% in Cluster 1 against 6% in Cluster 2. Companies in these sectors are not correlated with the economic cycle, so their financial performance is relatively stable, and they tend to outperform the market during economic downturns. Additionally, Utility companies have been among those pushed by regulators to increase the sustainability of their business since the early 2000s. They belong to the Climate Policy Relevant Sectors (CPRS) identified by Battiston et al. (2017) and are significant contributors to achieving SDGs (D’Amore et al. 2024). According to a recent survey conducted by Price Waterhouse Coopers International Limited (PwC), Utility companies have increased spending on ESG-related initiatives over the last 10 years, with nearly half reporting a 25% or more increase in investments (PwC 2023). This clearly explains the imbalance between the two clusters in terms of Utility companies.
4.2 The financial performance of ESG Pioneers and ESG Chasers
In this section, we assess whether companies that have pursued sustainability in different ways over time exhibit distinct exposures to key market risk factors. Sharp fluctuations in ESG ratings indicate that some firms are actively transitioning toward more sustainable business models, while others have maintained a stable ESG profile over time. By analyzing these differences through an asset pricing lens, we aim to quantify potential variations in risk exposure linked to different approaches to sustainability.
To investigate potential discrepancies, we first examine the risk-return profiles of ESG Chasers and Pioneers using the Capital Asset Pricing Model (CAPM). While CAPM provides a foundational framework for understanding market-related risk, its single-factor structure may overlook other systematic risk drivers. To address this limitation, we extend our analysis by incorporating the FF5F model, which accounts for additional sources of risk beyond the market premium alone. In particular, we analyze how each of the factors outlined in Sect. 3.2 influences the expected returns of the two groups. To isolate the contribution of each factor while controlling for market exposure, we estimate a series of bivariate models that pair the market factor (MKT) with each of the remaining four factors (SMB, HML, RMW, and CMA) individually. This layered approach allows us to assess whether differences in sustainability pathways translate into distinct exposures to systematic risk, shedding light on the financial implications of varying ESG adoption strategies.
We retrieve the daily FF5F from the French Data Library.1 We compute each stock’s daily excess returns (2577 observations) using ten years of daily quotes for the 306 S&P components. To capture the time-varying exposure to each FF factor, we estimate each of the six models using a yearly rolling window (w = 252 days). This results in 2325 model estimates for each company i spanning from \(w+1\) to T. We examine the estimated coefficients based on their magnitude and statistical significance, considering only those that are significantly different from zero at the 99% confidence level. To ensure robust inference and mitigate potential biases arising from heteroskedasticity and autocorrelation in return time series, we use heteroskedasticity and autocorrelation consistent (HAC) standard errors when testing the statistical significance of OLS parameters (Newey and West 1986).
We label companies according to the cluster analysis results and estimate the coefficients \(\theta _{i,m,t}=\{b_{i,m,t}, s_{i,m,t}, h_{i,m,t}, r_{i,m,t}, c_{i,m,t} \}\) at time t within each group. Then, we compute the median of each company, \(\theta _{i,m}\), considering only statistically significant coefficients.2 For example, the median coefficients \(b_{i,m}\) for cluster m, \(C_{m}\), where \(m = \{c, p\}\) represents the ESG Chasers and Pioneers, respectively, is obtained by computing the median of the \(N^{*}_{m}\) MKT statistically significant coefficients estimated at each time t for the stocks in cluster \(C_m\). By doing this, we obtain a vector \(b_{i,m}\), of size \(N_m\) with \(N = N_c+ N_p\), which contains the median values of every company \(i \in C_m\). To summarize the results, we compute the average median coefficient \(\theta _m = \{b_m, s_m,h_m,r_m,c_m\}\) for each cluster as:
In Table 5, we report the average of the median coefficients, the proportion of statistically significant coefficients over the total estimated, and the average adjusted-\(R^2\), obtained by computing the average of the adjusted-\(R^2\) of every model estimated. We report the output for the CAPM, the four bivariate models obtained as combinations of the market factor with one of the other FF factors, and the FF5F model. The CAPM average coefficient (\(b_{i,m}\)) for the ESG Pioneers (ESG Chasers) is equal to 0.987 (0.974), which is statistically significant in 97% (98%) of the cases. The coefficient \(b_i\) estimated in the four two-factor models is statistically significant in 98–99% of cases for both clusters, with an average median value that is nearly identical (spanning from 0.957 to 1.025). In contrast, the other coefficients estimated by these models (i.e., \(s_i\), \(h_i\), \(r_i\), and \(c_i\)) are statistically significant in a much smaller percentage of cases but often diverge within the two clusters, highlighting differences entailed by the different approach to the sustainability. The average coefficient \(s_p\) (\(s_c\)) for the ESG Pioneers (ESG Chasers) is 0.123 (0.167) and statistically significant in 45% (41%) of cases, while \(h_p\) (\(h_c\)) is 0.376 (0.237) with a higher level of significance (55–57%). Similarly, \(r_p\) (\(r_c\)) is 0.376 (0.231) and statistically significant in 48% (43%) of cases, while \(c_p\) (\(c_c\)) is 0.708 (0.433) and significant in 52% (48%) of cases. Thus, ESG Pioneers are significantly more exposed to the profitability factor (RMW) than ESG Chasers and to the investment component (CMA), indicating an overperformance of the ESG Pioneers in periods when the market rewards aggressive investment companies. This is likely due to the nature of the ESG Pioneers, who aggressively seek innovative products in the market, moving ahead of others as they did when enhancing their sustainability business models.
The full model exhibits the best goodness of fit (average adjusted \(R^2 = 0.38-0.39\)). We test the FF hypothesis of a null intercept in the model, which ensures the correct specification of the model, using the Gibbons, Ross, and Shanken (GRS) test (Gibbons et al. 1989), conducted across the total number of estimated models (\(T - w\)) and for every company i. The results consistently reject the null hypothesis of a statistically significant model intercept. The FF5F estimates for the coefficient \(b_i\) are statistically significant in 99% of the cases for both clusters, with an average median of 1.025 for the ESG Pioneers and 0.985 for the ESG Chasers. The other coefficients do not differ significantly between the two clusters, with the percentage of statistically significant cases varying between 31 and 47%. The only coefficient that shows a large difference is \(c_i\): its average value is higher for the ESG Pioneers (0.509, 38%) compared to the ESG Chasers (0.255, 35%), which is consistent with the previous findings.
Table 5
Average of the median coefficients for the CAPM, the two-factor models, and the FF5F model, with the percentage of statistically significant coefficients in brackets (Jan 2014–Dec 2022)
Model
ESG Pioneers
ESG Chasers
\(b_{i,p}\)
\(s_{i,p}\)
\(h_{i,p}\)
\(r_{i,p}\)
\(c_{i,p}\)
adj-\(R^2\)
\(b_{i,c}\)
\(s_{i,c}\)
\(h_{i,c}\)
\(r_{i,c}\)
\(c_{i,c}\)
adj-\(R^2\)
CAPM
0.987
0.33
0.974
0.34
(97%)
(98%)
MKT + SMB
0.969
0.123
0.35
0.957
0.167
0.35
(98%)
(45%)
(98%)
(41%)
MKT + HML
1.004
0.376
0.34
0.985
0.237
0.36
(98%)
(55%)
(98%)
(57%)
MKT + RMW
0.999
0.376
0.35
0.978
0.231
0.36
(98%)
(48%)
(99%)
(43%)
MKT + CMA
1.025
0.708
0.36
0.995
0.433
0.36
(98%)
(52%)
(98%)
(48%)
FF5F
1.025
0.042
0.197
0.19
0.509
0.38
0.985
0.052
0.111
0.114
0.255
0.39
(99%)
(32%)
(47%)
(41%)
(38%)
(99%)
(31%)
(46%)
(37%)
(35%)
The average adjusted \(R^2\) (adj-\(R^2\)) is calculated over the 2325 estimated models
The descriptive statistics of the median estimated coefficients for the FF5F model are presented in Table 6, along with boxplots detailing graphically the distribution of coefficients in the two clusters. Despite the median values of the \(b_i\), \(r_i\), and \(c_i\) parameters exhibiting similar distributions, albeit differing in the range between the two groups, the distributions of the \(s_i\) and \(h_i\) parameters are markedly distinct. Specifically, the \(s_{i,c}\) distribution shows a pronounced negative skewness, whereas the \(s_{i,p}\) exhibits a strong positive skewness. The opposite pattern emerges for the \(h_i\) parameter, where the ESG Chasers’ distribution is highly positively skewed, while that of the Pioneers is notably negatively skewed. These findings suggest that the pursuit of sustainability leads to significantly different outcomes in terms of risk exposure, but such effects are asymmetric across different risk factors.
We find further evidence of discrepancies between the groups by analyzing the statistics and boxplots for the CAPM and two-factor models in Fig. 7. Isolating the effect of each risk driver highlights that while the median MKT parameters remain fairly similar across clusters, several key discrepancies emerge in the distribution of the other parameters. First, the comparison of \(s_{i,c}\) and \(s_{i,p}\) confirms the findings from the FF5F model, with the Chasers (Pioneers) displaying a pronounced negative (positive) skewness. Also, the discrepancies in the \(c_i\) parameter distributions persist, albeit to a lesser extent, with the Chasers exhibiting a significantly wider range of values compared to the Pioneers. Third, both the \(r_i\) and \(c_i\) parameter distributions for the Pioneers appear much more concentrated around the median, as evidenced by the boxplots, with few extreme values classified as outliers. In contrast, the distributions for the Chasers are largely more platykurtic and significantly negatively skewed. These results confirm and stress how the approach to sustainability influences a company’s exposure to systemic risk factors. Companies that anticipated the integration of ESG principles into their strategy (Pioneers) tend to exhibit more stable and concentrated risk factor exposures, suggesting a more predictable and resilient financial structure. Conversely, firms that adopt sustainability reactively or opportunistically (Chasers) face a more erratic and asymmetrical risk profile, potentially exposing them to greater uncertainty. Thus, the pathway to sustainability seems to be not merely an ethical or regulatory choice but a strategic decision that fundamentally reshapes a firm’s financial and risk positioning in the market.
Table 6
Summary statistics and boxplots of the median FF5F model coefficients estimated separated by cluster (ESG Pioneers, in purple, and ESG Chasers, in green) (Jan 2014–Dec 2022)
Table 7
Summary statistics and boxplots of the median coefficients estimated in the four 2F models and the CAPM model (first row) separated by cluster (ESG Pioneers, in purple, and ESG Chasers, in green) (Jan 2014–Dec 2022)
We analyze the evolution of risk exposures for the two clusters over time to assess whether their sensitivities to market factors change as their sustainability profiles evolve. By examining the time-series dynamics of the FF5Fs for each group, we aim to determine whether shifts in sustainability alignment correspond to changes in systematic risk exposure. In addition, we investigate whether, as ESG Chasers converge toward similar sustainability values as Pioneers, their risk exposures also become more alike. So, for each cluster, we compute \(\theta _{m,t} = \{ b_{m,t}, s_{m,t}, h_{m,t}, r_{m,t}, c_{m,t} \}\) as the elementwise average of the statistically significant coefficients within the cluster for each time t:
with \(N^{*}_{m,t} \le N_m\) the number of statistically significant coefficients for each coefficient in t and \(\alpha =0.01\).3
We report the dynamics of the average FF5F model coefficients over time by cluster in Fig. 9. Their time series are non-stationary, specifically integrated of order one (\(\mathcal {I}(1)\)), and exhibit multiple structural breaks. We perform a cointegration test using the methodology in (Engle and Granger 1991) to assess whether the two series share a common stochastic trend and report the results in Tables 8 and 9 in the Appendix. The two CAPM MKT parameter series are not cointegrated, suggesting that they are influenced by distinct sources of randomness, possibly shaped by the sustainability commitment, which modifies the company’s exposure to market risk. The values of the other FF parameters fluctuate over time, are not stationary, and present spikes and collapses that suggest a certain dependence on the market health and the attention to sustainability. The dynamics of the Market factor (panel a), the profitability factor-RMW, (panel d), and the investment factor-CMA (panel e) differ significantly between the two clusters, whereas the Size and B/M Factors exhibit very similar dynamics over time. However, the cointegration test rejects the null of the absence of cointegration exclusively for the MKT and the CMA parameters.
The dynamics of the model coefficients over time are reported in Fig. 10. Among the two-factor models, the average SMB coefficients (\(b_{m,t}\)) show similar patterns across the two clusters, whereas the coefficients for the other three factors differ during specific subperiods. Consistent with the CAPM parameter, the time series of the average two-factor coefficients exhibit non-stationary behavior within each cluster. The cointegration test rules out any form of cointegration among the pairs of average coefficients, except among the MKT parameters of the bivariate model specification that includes the RMW factor.
Fig. 9
Average coefficients computed as in Eq. (4) in the FF5F models (Jan 2014–Dec 2022). ESG Pioneers (ESG Chasers) are represented in purple (green)
Average coefficients computed as in Eq. (4) in the 2F models and CAPM (panel a) (Jan 2014–Dec 2022). ESG Pioneers (ESG Chasers) are represented in purple (green)
In this section, we examine how sustainability improvements affect the financial performance of companies across TRBC economic sectors. Since firms within the same sector share similar business models and risk exposures, the impact of sustainability is expected to vary across industries. Moreover, sustainability receives uneven attention across sectors, with some facing greater scrutiny and regulatory pressure than others. To explore these differences, we analyze the excess returns of companies within each cluster, assuming that similar business models exhibit specific sensitivities to various risk factors. We estimate coefficients for the CAPM model, the four two-factor models, and the FF5F model to assess how sustainability influences systematic risk and return dynamics across industries.
The median coefficients, along with the percentage of statistically significant parameters, are reported in the Appendix (Tables 10 through 19). The market coefficient, \(b_i\), is statistically significant in nearly 100% of cases and is close to one for both ESG Chasers and ESG Pioneers in most sectors, with exceptions in Basic Materials, Technology, and Utilities. In the Basic Materials sector, ESG Pioneers exhibit significantly higher market betas than ESG Chasers (1.324 vs. 1.070 in the FF5F model). This suggests that firms with an early commitment to sustainability tend to be more exposed to market risk. A possible explanation is that these companies have undertaken substantial upfront investment costs to transition toward sustainable operations, potentially increasing their financial leverage and market exposure. For instance, DOW, a major producer of sealants and silicone, has been actively reducing emissions since 2015, while Newmont Mining, a leading gold mining company, already had a high ESG score in 2015. These firms likely internalized sustainability-related costs earlier, but as these efforts paid off, they may have become more competitive and growth-oriented, leading to a higher market beta. A similar pattern is observed in the Technology sector, where ESG Pioneers have an average beta of \(b_m = 1.189\), compared to 1.041 for ESG Chasers. This could reflect the sector’s reliance on innovation and growth, where early adoption of sustainability practices might have correlated with a more aggressive investment strategy and greater exposure to market fluctuations. The higher betas of ESG Pioneers suggest that these firms, by embedding sustainability into their core business models earlier, may have positioned themselves in higher-growth segments of the industry, potentially amplifying their market sensitivity. Conversely, firms in the Utilities sector display much lower sensitivity to the market factor. The \(b_i\) coefficients are statistically significant in 86–97% of cases, but their values remain well below one. For instance, in the CAPM model, the market beta is 0.474 for ESG Pioneers and 0.437 for ESG Chasers. This aligns with the sector’s traditionally defensive nature, as utility firms operate in regulated environments with stable cash flows, making them less sensitive to broader market fluctuations. The relatively small difference between ESG Pioneers and ESG Chasers suggests that sustainability transitions in this sector do not substantially alter market risk exposure, possibly because regulatory incentives and long-term infrastructure investments smooth out financial volatility. Overall, these findings highlight how the financial implications of sustainability initiatives are sector-dependent. While in cyclical and growth-oriented industries (e.g., Basic Materials and Technology), early sustainability adopters tend to exhibit higher systematic risk, in more defensive sectors (e.g., Utilities), the transition to sustainability appears to have a more muted effect on market exposure.
The sensitivity of different sectors to the remaining risk factors varies significantly. In the Basic Materials sector, firms exhibit different exposure to both the value (HML) and investment (CMA) factors. For Pioneers, the median \(h_i\) is 0.765 compared to 0.561 for Chasers in the two-factor model, while the average median investment factor, \(c_i\), is substantially higher for Pioneers (1.299 vs. 0.933). This suggests that early adopters of sustainability in capital-intensive industries tend to follow more aggressive investment policies, likely driven by the need to fund technological innovation and regulatory compliance. These firms may also be more growth-oriented, reducing their exposure to the value factor. A key driver of these differences is the regulatory framework in high-emission industries, where stringent environmental policies set minimum sustainability thresholds for market participation. Firms that transitioned earlier toward sustainable business models may have gained competitive advantages but at the cost of greater sensitivity to investment-related risks. Their higher CMA coefficients indicate that sustainability-driven investments play a central role in their strategic positioning, differentiating them from firms that adopted sustainability practices later. Conversely, Utility sector firms exhibit a negative sensitivity to both the size (SMB) and value (HML) factors among Chasers, while only the size factor is negative for Pioneers. This asymmetry likely reflects the different sustainability approaches adopted by the two groups. Pioneers, having maintained high sustainability standards from the outset, may have implemented long-term investment strategies that emphasize efficiency and innovation, thereby reducing their exposure to the value factor. In contrast, Chasers, which transitioned to sustainable practices more recently, may still retain characteristics of traditional utility firms, such as capital-intensive structures and lower growth expectations, leading to a stronger negative loading on both factors.
Focusing on the FF5F model outputs, within Consumer Non-Cyclicals, the investment factor (CMA) emerges as the second most significant, with 51% of coefficients being statistically significant for Pioneers and 59% for Chasers. The average CMA coefficient is \(c_p = 0.857\) for Pioneers and 0.885 for Chasers, indicating that investment intensity plays a comparable role across both groups and that firms in this sector, regardless of their sustainability approach, rely on stable capital allocation strategies, likely due to lower sensitivity to market cycles and a focus on long-term operational efficiency. In contrast, Consumer Cyclicals show a much weaker relationship with investment, with CMA coefficients of \(c_p = 0.182\) for Pioneers and \(-0.234\) for Chasers, with statistical significance below 30%. This sign reversal suggests that sustainability influences investment behavior differently across clusters. For Pioneers, the positive coefficient indicates a more stable investment approach, as these firms integrate sustainability into long-term capital planning, reducing exposure to cyclical fluctuations. Conversely, the negative coefficient for Chasers suggests a more pro-cyclical investment strategy, where sustainability commitments are more reactive and linked to short-term financial conditions. This discrepancy further stresses how early ESG adoption fosters investment stability, while late adoption leaves firms more exposed to cyclical pressures. Instead, profitability (RMW) appears to be the dominant risk factor in this sector, with \(r_p = 0.879\) for Pioneers and \(r_c = 0.674\) for Chasers. The stronger RMW coefficient for Pioneers suggests that firms with a long-standing commitment to sustainability benefit from higher operating efficiency and better margins, possibly due to stronger brand positioning and cost advantages from sustainable practices.
To sum up, the empirical results highlight that the financial impact of sustainability is not uniform but shaped by sector-specific characteristics. While both investment intensity and profitability influence returns, their relative importance varies depending on industry dynamics. Also, the approach to sustainability itself plays a crucial role: Pioneers, with early and consistent ESG integration, exhibit more stable financial patterns, whereas Chasers, adapting sustainability practices later, display higher sensitivity to sector-specific risks. This highlights the need to account for both industry structure and sustainability timing when evaluating ESG-driven financial performance.
5 Conclusion
Over the past decade, transitioning to a sustainable business model has posed a significant challenge for most companies. Firms worldwide strive to enhance their sustainability while maintaining competitiveness. However, while some companies voluntarily align their strategies with the UN Sustainable Development Goals (SDGs), others comply with regulations only when mandated by authorities. This raises a critical question: has the market, increasingly seeking sustainable stocks, rewarded these differences in corporate approaches?
In this paper, we cluster companies within the S&P 500 Index based on the evolution of their ESG scores over the years, which serve as a proxy for sustainability performance. We identify two distinct clusters of companies, each representing a different approach to improving sustainability. The first group consists of companies that reported high ESG scores at the start of our analysis in 2013 and have maintained relatively stable scores over time. We refer to these firms as “ESG Pioneers”. The second group includes companies that have significantly increased their ESG scores throughout the period, reflecting a more recent commitment to sustainability. We call these “ESG Chasers”. These findings highlight diverse strategies in the corporate world regarding sustainability, with some companies being committed and others just compliant.
Using an Asset Pricing framework, specifically the Fama and French five-factor model, we assess whether different sustainability commitments influence firms’ profitability and risk exposure. While market exposure remains similar across groups, significant differences emerge in other risk factors. The distributions of size (\(s_i\)) and value (\(h_i\)) factors show opposite skewness patterns between ESG Pioneers and Chasers, indicating asymmetries in risk sensitivity. Chasers exhibit wider dispersion in investment (\(c_i\)) and profitability (\(r_i\)) factor exposures, suggesting greater uncertainty in their financial structure. These results highlight that a firm’s approach to sustainability shapes its systemic risk profile. Pioneers, having integrated ESG principles early display more stable and predictable risk exposures, reinforcing financial resilience. In contrast, Chasers, adapting reactively, face more volatile and asymmetric risk dynamics. This underscores that sustainability is not just a compliance choice but a strategic decision with tangible implications for financial stability and risk positioning.
The impact of sustainability on firms’ risk exposure and excess returns varies significantly across economic sectors, underscoring the need for a sector-specific approach in financial analysis. Defensive sectors like Utilities and Consumer Non-Cyclicals exhibit lower market exposure and negative sensitivity to size and value factors, reflecting their stability and weaker dependence on economic cycles. In contrast, Basic Materials firms, central to the sustainability transition, show a stark divergence between ESG Pioneers and Chasers in market sensitivity. Pioneers, having undergone early and often expensive business model shifts, display higher exposure to market fluctuations, positioning them to benefit more from ESG-driven market trends. These findings suggest that sustainability commitments reshape firms’ financial risk profiles differently across industries. Companies in stable sectors face lower systemic risk, while those in transformative industries, such as Basic Materials, experience amplified market sensitivity due to the capital-intensive nature of their transition. As such, financial assessments of ESG integration must account for sector-specific risk-return dynamics to avoid oversimplified conclusions and bias due to a rough aggregation of the results.
In aiming to build a sustainable economy, the way managers take initiative does count. The timing and the strength of actions to target the SDGs affect market performance as investors seek sustainable and profitable investments. Our analysis shows that early and strong commitments to sustainability matter. ESG Pioneers show high sensitivity to market and investment risk drivers, having adopted sustainable practices sooner. In contrast, ESG Chasers, still adapting their business models, show different risk exposures. Proactive sustainability efforts meet regulatory and ethical standards, improve market performance, and attract investors. Managers should act quickly and decisively toward sustainability to meet market expectations and gain long-term advantages. Further studies will help identify which ESG component influences the market perception of sustainability most and whether these effects are consistent across countries or vary due to local policies. This will provide insights into how different regulations and cultural factors impact the market’s response to sustainability efforts.
Declarations
Conflict of interest
The authors have no relevant financial or non-financial interests to disclose.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
We run cointegration tests to search for long-run relationships between each couple of time series. The presence of cointegration implies that, despite short-term deviations, the series move together in the long run, indicating a similar pattern throughout the years. In particular, we run the Engle-Granger test (Engle and Granger 1991), based on a two-step residual-based approach. The null hypothesis states that the residuals from a static regression are non-stationary (i.e., no cointegration), while the alternative hypothesis suggests that they are stationary, indicating a cointegrating relationship. Empirical results are reported in Table 8 (Table 9).
Table 8
Cointegration tests for each specification (No trend, linear trend, and quadratic trend) between the parameters of the Pioneers and Chasers FF5F models (Jan 2014–Dec 2022)
Parameter
NoTrend
LinearTrend
QuadTrend
MKT
−2.524
0.962
−0.190
SMB
−4.259***
0.398
−0.270
HML
−3.645***
0.659
−2.167
RMW
−3.603***
−1.129
1.477
CMA
−2.304
−1.185
1.239
\(^{*}p<0.1\); \(^{**}p<0.05\); \(^{***}p<0.01\)
Table 9
Cointegration tests for each specification (No trend, linear trend, and quadratic trend) between the parameters of the Pioneers and Chasers models (Jan 2014–Dec 2022)
Parameter
NoTrend
LinearTrend
QuadTrend
CAPM
−2.072
0.747
0.737
MKT (SMB)
−2.464
0.779
0.452
MKT (HML)
−2.538
1.28
−0.358
MKT (RMW)
−3.191***
1.025
0.148
MKT (CMA)
−2.374
1.054
0.250
SMB
−2.409
−0.004
1.221
HML
−2.139
−0.515
−0.611
RMW
−1.544
−0.456
−0.030
CMA
−1.942
−0.728
0.187
The first row reports the test on the CAPM parameter, while the subsequent rows present the tests on the two-factor models, which combine the RMRF factor with one of the other risk factors
The following Tables report the output of the analysis conducted separated by business sector, according to the LSEG classification (TRBC) (Tables 10, 11, 12, 13, 14, 15, 16, 17 and 18).
Table 10
Model comparison within the basic material TRBC sector (Jan 2014–Dec 2022)
Basic Materials
ESG Pioneers
ESG Chasers
Model
\(b_{i,p}\)
\(s_{i,p}\)
\(h_{i,p}\)
\(r_{i,p}\)
\(c_{i,p}\)
adj-\(R^2\)
\(b_{i,c}\)
\(s_{i,c}\)
\(h_{i,c}\)
\(r_{i,c}\)
\(c_{i,c}\)
adj-\(R^2\)
CAPM
1.294
0.33
1.032
0.35
(100%)
(97%)
\(MKT+SMB\)
1.257
0.876
0.36
1.008
0.537
0.36
(100%)
(50%)
(96%)
(35%)
\(MKT+HML\)
1.344
0.765
0.38
1.052
0.561
0.39
(100%)
(71%)
(97%)
(55%)
\(MKT+RMW\)
1.269
0.368
0.34
1.048
0.473
0.36
(100%)
(36%)
(96%)
(35%)
\(MKT+CMA\)
1.376
1.299
0.37
1.080
0.933
0.37
(100%)
(69%)
(97%)
(50%)
FF5F
1.324
0.874
0.835
\(-\)0.368
1.072
0.40
1.070
0.408
0.512
0.021
0.713
0.41
(100%)
(36%)
(52%)
(26%)
(32%)
(97%)
(28%)
(39%)
(27%)
(29%)
The adj-\(R^2\) column contains the average adj-\(R^2\) computed over time
Table 11
Model comparison within the Consumer Cyclicals TRBC sector (Jan 2014–Dec 2022)
Consumer Cyclicals
ESG Pioneers
ESG Chasers
Model
\(b_{i,p}\)
\(s_{i,p}\)
\(h_{i,p}\)
\(r_{i,p}\)
\(c_{i,p}\)
adj-\(R^2\)
\(b_{i,c}\)
\(s_{i,c}\)
\(h_{i,c}\)
\(r_{i,c}\)
\(c_{i,c}\)
adj-\(R^2\)
CAPM
1.003
0.31
1.038
0.30
(100%)
(99%)
\(MKT+SMB\)
0.971
0.632
0.32
1.006
0.586
0.32
(99%)
(44%)
(99%)
(42%)
\(MKT+HML\)
0.997
0.476
0.33
1.039
0.356
0.33
(100%)
(36%)
(100%)
(48%)
\(MKT+RMW\)
1.042
0.737
0.32
1.062
0.656
0.31
(100%)
(51%)
(99%)
(42%)
\(MKT+CMA\)
1.018
0.759
0.32
1.048
0.639
0.31
(100%)
(31%)
(99%)
(37%)
FF5F
1.031
0.612
0.203
0.879
0.182
0.35
1.051
0.529
\(-\)0.038
0.674
\(-\)0.234
0.36
(100%)
(40%)
(30%)
(50%)
(26%)
(99%)
(37%)
(36%)
(41%)
(28%)
The adj-\(R^2\) column contains the average adj-\(R^2\) computed over time
Table 12
Model comparison within the Consumer Non-Cyclicals TRBC sector (Jan 2014–Dec 2022)
Cons Non-Cyclicals
ESG Pioneers
ESG Chasers
Model
\(b_{i,p}\)
\(s_{i,p}\)
\(h_{i,p}\)
\(r_{i,p}\)
\(c_{i,p}\)
adj-\(R^2\)
\(b_{i,c}\)
\(s_{i,c}\)
\(h_{i,c}\)
\(r_{i,c}\)
\(c_{i,c}\)
adj-\(R^2\)
CAPM
0.784
0.31
0.624
0.24
(96%)
(94%)
\(MKT+SMB\)
0.802
\(-\)0.339
0.33
0.662
\(-\)0.422
0.27
(96%)
(50%)
(95%)
(56%)
\(MKT+HML\)
0.795
0.359
0.35
0.649
0.277
0.27
(98%)
(50%)
(95%)
(47%)
\(MKT+RMW\)
0.814
0.582
0.34
0.707
0.672
0.28
(98%)
(61%)
(96%)
(71%)
\(MKT+CMA\)
0.827
0.723
0.34
0.697
0.791
0.27
(97%)
(63%)
(96%)
(61%)
FF5F
0.857
\(-\)0.276
\(-\)0.059
0.384
0.857
0.40
0.753
\(-\)0.316
\(-\)0.268
0.506
0.885
0.34
(98%)
(35%)
(44%)
(45%)
(51%)
(97%)
(38%)
(40%)
(56%)
(59%)
The adj-\(R^2\) column contains the average adj-\(R^2\) computed over time
Table 13
Model comparison within the Energy TRBC sector (Jan 2014–Dec 2022)
Energy
ESG Pioneers
ESG Chasers
Model
\(b_{i,p}\)
\(s_{i,p}\)
\(h_{i,p}\)
\(r_{i,p}\)
\(c_{i,p}\)
adj-\(R^2\)
\(b_{i,c}\)
\(s_{i,c}\)
\(h_{i,c}\)
\(r_{i,c}\)
\(c_{i,c}\)
adj-\(R^2\)
CAPM
1.089
0.26
1.151
0.26
(100%)
(100%)
\(MKT+SMB\)
1.037
0.788
0.28
1.094
1.033
0.28
(98%)
(39%)
(99%)
(38%)
\(MKT+HML\)
1.181
0.981
0.38
1.208
1.058
0.36
(100%)
(95%)
(100%)
(84%)
\(MKT+RMW\)
1.052
\(-\)1.017
0.29
1.056
\(-\)0.964
0.29
(98%)
(48%)
(100%)
(46%)
\(MKT+CMA\)
1.229
1.800
0.36
1.268
1.663
0.34
(100%)
(79%)
(100%)
(71%)
FF5F
1.131
\(-\)0.061
0.88
\(-\)0.894
0.865
0.45
1.174
0.346
0.974
\(-\)0.991
0.628
0.44
(100%)
(20%)
(75%)
(63%)
(49%)
(100%)
(15%)
(65%)
(57%)
(50%)
The adj-\(R^2\) column contains the average adj-\(R^2\) computed over time
Table 14
Model comparison within the Financials TRBC sector (Jan 2014–Dec 2022)
Financials
ESG Pioneers
ESG Chasers
Model
\(b_{i,p}\)
\(s_{i,p}\)
\(h_{i,p}\)
\(r_{i,p}\)
\(c_{i,p}\)
adj-\(R^2\)
\(b_{i,c}\)
\(s_{i,c}\)
\(h_{i,c}\)
\(r_{i,c}\)
\(c_{i,c}\)
adj-\(R^2\)
CAPM
0.934
0.42
1.008
0.44
(100%)
(100%)
\(MKT+SMB\)
0.917
0.285
0.44
0.976
0.343
0.46
(100%)
(47%)
(100%)
(41%)
\(MKT+HML\)
1.028
0.726
0.54
1.089
0.742
0.56
(100%)
(86%)
(100%)
(84%)
\(MKT+RMW\)
0.948
0.371
0.44
1.001
0.247
0.46
(100%)
(45%)
(100%)
(42%)
\(MKT+CMA\)
1.017
0.768
0.45
1.072
0.782
0.47
(100%)
(59%)
(100%)
(55%)
FF5F
0.991
\(-\)0.011
0.998
\(-\)0.375
\(-\)0.632
0.57
1.039
\(-\)0.158
0.954
\(-\)0.544
\(-\)0.521
0.58
(100%)
(23%)
(81%)
(42%)
(43%)
(100%)
(26%)
(78%)
(35%)
(44%)
The adj-\(R^2\) column contains the average adj-\(R^2\) computed over time
Table 15
Model comparison within the Healthcare TRBC sector (Jan 2014–Dec 2022)
Healthcare
ESG Pioneers
ESG Chasers
Model
\(b_{i,p}\)
\(s_{i,p}\)
\(h_{i,p}\)
\(r_{i,p}\)
\(c_{i,p}\)
adj-\(R^2\)
\(b_{i,c}\)
\(s_{i,c}\)
\(h_{i,c}\)
\(r_{i,c}\)
\(c_{i,c}\)
adj-\(R^2\)
CAPM
0.775
0.29
0.871
0.32
(98%)
(100%)
\(MKT+SMB\)
0.797
\(-\)0.318
0.30
0.880
\(-\)0.313
0.33
(99%)
(47%)
(100%)
(30%)
\(MKT+HML\)
0.782
\(-\)0.283
0.31
0.868
\(-\)0.434
0.35
(99%)
(49%)
(100%)
(60%)
\(MKT+RMW\)
0.793
0.399
0.30
0.884
\(-\)0.411
0.33
(99%)
(36%)
(100%)
(34%)
\(MKT+CMA\)
0.805
0.678
0.30
0.878
\(-\)0.560
0.33
(99%)
(43%)
(100%)
(40%)
FF5F
0.813
\(-\)0.162
\(-\)0.294
0.008
0.816
0.34
0.897
\(-\)0.051
\(-\)0.489
\(-\)0.316
0.556
0.36
(99%)
(36%)
(45%)
(31%)
(41%)
(99%)
(23%)
(48%)
(25%)
(24%)
The adj-\(R^2\) column contains the average adj-\(R^2\) computed over time
Table 16
Model comparison within the Industrials TRBC sector (Jan 2014–Dec 2022)
Industrials
ESG Pioneers
ESG Chasers
Model
\(b_{i,p}\)
\(s_{i,p}\)
\(h_{i,p}\)
\(r_{i,p}\)
\(c_{i,p}\)
adj-\(R^2\)
\(b_{i,c}\)
\(s_{i,c}\)
\(h_{i,c}\)
\(r_{i,c}\)
\(c_{i,c}\)
adj-\(R^2\)
CAPM
1.066
0.38
0.980
0.40
(100%)
(100%)
\(MKT+SMB\)
1.044
0.502
0.42
0.965
0.367
0.43
(100%)
(34%)
(100%)
(37%)
\(MKT+HML\)
1.075
0.470
0.991
0.393
(100%)
(54%)
(100%)
(49%)
\(MKT+RMW\)
1.089
0.649
0.39
1.013
0.550
0.41
(100%)
(44%)
(100%)
(44%)
\(MKT+CMA\)
1.104
0.814
0.40
1.006
0.715
0.41
(100%)
(47%)
(100%)
(47%)
FF5F
1.097
0.516
0.421
0.618
0.438
0.44
1.021
0.362
0.234
0.550
0.554
0.45
(100%)
(29%)
(39%)
(37%)
(29%)
(100%)
(33%)
(34%)
(37%)
(25%)
The adj-\(R^2\) column contains the average adj-\(R^2\) computed over time
Table 17
Model comparison within the Real Estate TRBC sector (Jan 2014–Dec 2022)
Real Estate
ESG Pioneers
ESG Chasers
Model
\(b_{i,p}\)
\(s_{i,p}\)
\(h_{i,p}\)
\(r_{i,p}\)
\(c_{i,p}\)
adj-\(R^2\)
\(b_{i,c}\)
\(s_{i,c}\)
\(h_{i,c}\)
\(r_{i,c}\)
\(c_{i,c}\)
adj-\(R^2\)
CAPM
0.870
0.36
0.669
0.23
(100%)
(95%)
\(MKT+SMB\)
0.875
\(-\)0.255
0.37
0.689
\(-\)0.347
0.26
(100%)
(39%)
(96%)
(50%)
\(MKT+HML\)
0.908
0.301
0.37
0.692
\(-\)0.297
0.27
(100%)
(27%)
(95%)
(50%)
\(MKT+RMW\)
0.936
0.532
0.37
0.723
0.643
0.25
(100%)
(41%)
(97%)
(47%)
\(MKT+CMA\)
0.929
0.737
0.37
0.717
0.642
0.25
(100%)
(44%)
(95%)
(34%)
FF5F
0.995
0.313
\(-\)0.287
0.497
0.876
0.40
0.739
\(-\)0.083
\(-\)0.252
0.373
0.510
0.31
(100%)
(25%)
(39%)
(30%)
(45%)
(98%)
(38%)
(62%)
(37%)
(45%)
The adj-\(R^2\) column contains the average adj-\(R^2\) computed over time
Table 18
Model comparison within the Technology TRBC sector (Jan 2014–Dec 2022)
Technology
ESG Pioneers
ESG Chasers
Model
\(b_{i,p}\)
\(s_{i,p}\)
\(h_{i,p}\)
\(r_{i,p}\)
\(c_{i,p}\)
adj-\(R^2\)
\(b_{i,c}\)
\(s_{i,c}\)
\(h_{i,c}\)
\(r_{i,c}\)
\(c_{i,c}\)
adj-\(R^2\)
CAPM
1.193
0.43
1.053
0.36
(100%)
(100%)
\(MKT+SMB\)
1.207
\(-\)0.376
0.44
1.054
\(-\)0.182
0.37
(100%)
(33%)
(100%)
(37%)
\(MKT+HML\)
1.196
\(-\)0.551
0.46
1.046
\(-\)0.457
0.39
(100%)
(49%)
(100%)
(50%)
\(MKT+RMW\)
1.201
0.318
0.44
1.060
\(-\)0.307
0.37
(100%)
(39%)
(100%)
(31%)
\(MKT+CMA\)
1.155
\(-\)0.809
0.45
1.028
\(-\)0.812
0.38
(100%)
(52%)
(100%)
(47%)
FF5F
1.189
\(-\)0.107
\(-\)0.521
0.250
\(-\)0.369
0.48
1.041
0.129
\(-\)0.504
\(-\)0.161
\(-\)0.713
0.41
(100%)
(23%)
(39%)
(28%)
(29%)
(100%)
(27%)
(32%)
(25%)
(26%)
The adj-\(R^2\) column contains the average adj-\(R^2\) computed over time
Table 19
Model comparison within the Utilities TRBC sector (Jan 2014–Dec 2022)
Utilities
ESG Pioneers
ESG Chasers
Model
\(b_{i,p}\)
\(s_{i,p}\)
\(h_{i,p}\)
\(r_{i,p}\)
\(c_{i,p}\)
adj-\(R^2\)
\(b_{i,c}\)
\(s_{i,c}\)
\(h_{i,c}\)
\(r_{i,c}\)
\(c_{i,c}\)
adj-\(R^2\)
CAPM
0.474
0.15
0.437
0.14
(82%)
(79%)
\(MKT+SMB\)
0.506
\(-\)0.448
0.18
0.471
\(-\)0.446
0.17
(91%)
(74%)
(89%)
(75%)
\(MKT+HML\)
0.524
0.204
0.17
0.469
\(-\)0.190
0.16
(85%)
(47%)
(81%)
(40%)
\(MKT+RMW\)
0.556
0.642
0.19
0.514
0.630
0.18
(90%)
(69%)
(88%)
(77%)
\(MKT+CMA\)
0.548
0.717
0.18
0.483
0.714
0.17
(88%)
(55%)
(86%)
(56%)
FF5F
0.608
\(-\)0.386
\(-\)0.042
0.343
0.813
0.24
0.580
\(-\)0.411
\(-\)0.257
0.324
0.838
0.24
(97%)
(49%)
(44%)
(47%)
(53%)
(95%)
(51%)
(42%)
(60%)
(57%)
The adj-\(R^2\) column contains the average adj-\(R^2\) computed over time
We use the median instead of the mean to reduce potential biases caused by heterogeneity within each cluster, ensuring a more robust comparison of factor coefficients over time.
In this case, we compute the average instead of the median because the number of statistically significant parameters at certain time points t is small, which can lead to large daily fluctuations in the median of the parameter estimates.
Aghamolla, C., An, B.J.: Mandatory vs. voluntary ESG disclosure, efficiency, and real effects. Nanyang Business School Research Paper (22–41) (2023)
Arbelaitz, O., Gurrutxaga, I., Muguerza, J., et al.: An extensive comparative study of cluster validity indices. Pattern Recogn. 46(1), 243–256 (2013)CrossRef
Battiston, S., Mandel, A., Monasterolo, I., Schutze, F., Visentin, G.: A climate stress-test of the financial system. Nat. Clim. Chang. 7(4), 283–288 (2017)CrossRef
Bax, K., Bonaccolto, G., Paterlini, S.: Do lower environmental, social, and governance (ESG) rated companies have higher systemic impact? Empirical evidence from Europe and the United States. Corp. Soc. Responsib. Environ. Manag. 30(3), 1406–1420 (2023)CrossRef
Berg, F., Fabisik, K., Sautner, Z.: Is history repeating itself? The (un) predictable past of ESG ratings. SSRN Electron. J. 66, 1–59 (2021)
Berg, F., Koelbel, J.F., Rigobon, R.: Aggregate confusion: the divergence of ESG ratings. Rev. Finance 26(6), 1315–1344 (2022)CrossRef
Bezdek, J.C., Pal, N.R.: Some new indexes of cluster validity. IEEE Trans. Syst., Man, Cybern. Part B (Cybern.) 28, 301–315 (1988)CrossRef
Boffo, R., Patalano, R.: ESG Investing: Practices. Progress and Challenges, OECD Paris (2020)
Brandon, R., Krueger, P., Schmidt, P.: ESG rating disagreement and stock returns. Financ. Anal. J. 77, 104–127 (2019)CrossRef
Calinski, R.B., Harabasz, J.: A dendrite method for cluster analysis. Commun. Stat. 3(1), 1–27 (1974)
Chen, S., Song, Y., Gao, P.: Environmental, social, and governance (ESG) performance and financial outcomes: analyzing the impact of ESG on financial performance. J. Environ. Manage. 345, 118829 (2023)CrossRef
Chen, Y., Li, T., Zeng, Q., Zhu, B.: Effect of ESG performance on the cost of equity capital: evidence from China. Int. Rev. Econ. Finance 83, 348–364 (2023)CrossRef
Chung, R., Bayne, L., Birt, J.L.: Determinants of ESG disclosure among listed firms under voluntary and mandatory ESG disclosure regimes in Hong Kong. J. Appl. Acc. Res. 25(4), 812–836 (2024)
D’Amore, G., Testa, M., Lepore, L.: How is the utilities sector contributing to building a sustainable future? A systematic literature review of sustainability practices. Sustainability 16(1), 374 (2024)CrossRef
D’Ecclesia, R.L., Levantesi, S., Stefanelli, K.: Measuring business impacts on the sustainability of European-listed firms. Socioecon. Plann. Sci. 96, 102078 (2024)CrossRef
Díaz, V., Ibrushi, D., Zhao, J.: Reconsidering systematic factors during the Covid-19 pandemic—the rising importance of ESG. Financ. Res. Lett. 38, 101870 (2021)CrossRef
Dumitrescu, A., Jarvinen, J., Zakriya, M.: Hidden gem or fool’s gold: can passive ESG ETFs outperform the benchmarks? Int. Rev. Financ. Anal. 86, 102540 (2023)CrossRef
Engle, R., Granger, C.: Long-run economic relationships: Readings in cointegration. Oxford University Press, Oxford (1991)CrossRef
Erhart, S.: Take it with a pinch of salt-ESG rating of stocks and stock indices. Int. Rev. Financ. Anal. 83, 102308 (2022)CrossRef
Eskantar, M., Zopounidis, C., Doumpos, M., Galariotis, E., Guesmi, K.: Navigating ESG complexity: an in-depth analysis of sustainability criteria, frameworks, and impact assessment. Int. Rev. Financ. Anal. 95, 103380 (2024)CrossRef
Fama, E.F., French, K.R.: A five-factor asset pricing model. J. Financ. Econ. 116, 1–22 (2015)CrossRef
Fan, K.Y., Shen, J., Hui, E.C., Cheng, L.T.: ESG components and equity returns: evidence from real estate investment trusts. Int. Rev. Financ. Anal. 96, 103716 (2024)CrossRef
Fulton, M., Kahn, B., Sharples, C.: Sustainable investing: Establishing long-term value and performance. Available at SSRN 2222740 (2012)
Gibbons, M.R., Ross, S.A., Shanken, J.: A test of the efficiency of a given portfolio. Econom. J. Econom. Soc.: 57, 1121–1152 (1989)
Giorgino, T.: Computing and visualizing dynamic time warping alignments in R: the dtw Package. J. Stat. Softw. 31(7), 1–24 (2009)CrossRef
Gonzaga, B.R., Klotzle, M.C., Brugni, T.V., Rakos, I.S., Cioca, I.C., Barbu, C.M., Cucerzan, T.: The ESG patterns of emerging-market companies: are there differences in their sustainable behavior after COVID-19? Sustainability 16(2), 676 (2024)CrossRef
Gurrutxaga, I., Albisua, I., Arbelaitz, O., et al.: SEP/COP: an efficient method to find the best partition in hierarchical clustering based on a new cluster validity index. Pattern Recogn. 43(10), 3364–3373 (2010)CrossRef
Hastie, T., Tibshirani, R., Friedman, J.H., Friedman, J.H.: The elements of statistical learning: data mining, inference, and prediction, vol. 2, pp. 1–758. Springer, New York (2009)
Hubel, B., Scholz, H.: Integrating sustainability risks in asset management: the role of ESG exposures and ESG ratings. J. Asset Manag. 21, 52–69 (2020)CrossRef
Iamandi, I.E., Constantin, L.G., Munteanu, S.M., Cernat-Gruici, B.: Mapping the ESG behavior of European companies. A holistic Kohonen approach. Sustainability 11(12), 3276 (2019)CrossRef
Jin, I.: Is ESG a systematic risk factor for US equity mutual funds? J. Sustain. Finance Invest. 8(1), 72–93 (2018)CrossRef
Kim, S., Li, Z.: Understanding the impact of ESG practices in corporate finance. Sustainability 13(7), 3746 (2021)CrossRef
Krueger, P., Sautner, Z., Tang, D.Y., Zhong, R.: The effects of mandatory ESG disclosure around the world. J. Account. Res. 62(5), 1795–1847 (2024)CrossRef
Lin, W.L., Law, S.H., Ho, J.A., Sambasivan, M.: The causality direction of the corporate social responsibility-corporate financial performance nexus: application of panel vector autoregression approach. North Am. J. Econ. Finance 48, 401–418 (2019)CrossRef
Lins, K.V., Servaes, H., Tamayo, A.: Social capital, trust, and firm performance: the value of corporate social responsibility during the financial crisis. J. Financ. 72(4), 1785–1824 (2017)CrossRef
Lioui, A., Tarelli, A.: Chasing the ESG factor. J. Banking Finance 139, 106498 (2022)CrossRef
Rajabi, A., Eskandari, M., Ghadi, M.J., Li, L., Zhang, J., Siano, P.: A comparative study of clustering techniques for electrical load pattern segmentation. Renew. Sustain. Energy Rev. 120, 109628 (2020)CrossRef
Rousseeuw, P.J.: Silhouettes: a graphical aid to the interpretation and validation of cluster analysis. J. Comput. Appl. Math. 20, 53–65 (1987)CrossRef
Sakoe, H., Chiba, S.: Dynamic programming algorithm optimization for spoken word recognition. IEEE Trans. Acoust. Speech Signal Process. 26(1), 43–49 (1978)CrossRef
Sassen, R., Hinze, A.K., Hardeck, I.: Impact of ESG factors on firm risk in Europe. J. Bus. Econ. 86, 867–904 (2016)
Vilas, P., Andreu, L., Sarto, J.L.: Cluster analysis to validate the sustainability label of stock indices: an analysis of the inclusion and exclusion processes in terms of size and ESG ratings. J. Clean. Prod. 330, 129862 (2022)CrossRef
Wang, L., Ji, Y., Ni, Z.: Spillover of stock price crash risk: do environmental, social and governance (ESG) matter? Int. Rev. Financ. Anal. 89, 102768 (2023)CrossRef
Ward, J.H., Jr.: Hierarchical grouping to optimize an objective function. J. Am. Stat. Assoc. 58, 236–244 (1963)CrossRef
Zhou, G., Liu, L., Luo, S.: Sustainable development, ESG performance, and company market value: mediating effect of financial performance. Bus. Strateg. Environ. 31(7), 3371–3387 (2022)CrossRef
