Corporate digital transformation, as a critical driver of high-quality economic and social development, has exerted an increasingly significant influence on operational efficiency, innovation vitality, and environmental, social, and governance (ESG) performance. A deeper investigation into the mechanisms through which corporate digital transformation influences ESG performance carries substantial theoretical value and practical implications. Based on annual report texts of A-share listed companies in Shanghai and Shenzhen between 2010 and 2023, this study employs a digital transformation performance index derived from large language models and an implementation intensity index constructed from digital patents, thereby enabling a precise assessment of firms’ overall digital transformation. The findings reveal that corporate digital transformation significantly enhances ESG performance, and this conclusion remains robust after addressing endogeneity concerns and conducting extensive robustness checks. Mechanism analysis indicates that digital transformation fosters ESG performance by increasing research and development (R&D) investment intensity and improving the efficiency of R&D transformation. Regarding heterogeneity, firms display diverse patterns in the impact of digital transformation on ESG performance. Notably, firms characterized by high digital homophily, severe financing constraints, and substantial resource redundancy demonstrate a distinctly stronger ESG-enhancing effect of digital transformation. This study offers feasible pathways and policy insights for advancing corporate digital transformation, enhancing ESG performance, and contributing to the realization of the “dual-carbon” goals.
In the context of the accelerated evolution of the global climate governance system and its deep integration with national sustainable development strategies, environmental, social, and governance (ESG) has evolved from a peripheral extension of corporate social responsibility into a formalized institutional framework for assessing the capacity for sustainable development of firms (Pu et al., 2025). As ESG principles have shifted from a regime of voluntary disclosure to mandatory compliance, their regulatory force and strategic significance have been substantially strengthened. In May 2024, the Council of the European Union formally approved the Corporate Sustainability Due Diligence Directive (CSDDD), thereby reinforcing corporate compliance obligations regarding human rights protection and environmental responsibility and signaling a more stringent phase for the global ESG regulatory framework. Based on this broader regulatory transformation, Chinese firms have accelerated the implementation of ESG practices. According to the China Listed Companies ESG Industry Report (2024), as of June 2024, 40.3% of A-share listed companies had disclosed ESG-related reports, an increase of more than five percentage points compared to the same period in the previous year, reflecting growing recognition of and engagement with ESG among Chinese enterprises.
However, considerable heterogeneity persists in corporate ESG performance. While certain leading firms have embedded ESG into their long-term strategic planning, thereby promoting green transformation and the co-creation of social value (Hu et al., 2025), many others are constrained by short-term performance pressures or insufficient internal incentives, resulting in symbolic ESG investment and, in some cases, greenwashing (Du & Lu, 2025; Mou et al., 2025). These structural imbalances in ESG practices undermine the substantive effectiveness of ESG governance and impede the realization of high-quality and sustainable corporate development (Gao et al., 2025; Palas et al., 2025). Accordingly, identifying mechanisms that can stimulate firms’ intrinsic motivation and reinforce their commitment to ESG engagement is of substantial theoretical and practical importance for fostering a distinctive and resilient ESG ecosystem.
In the context of innovation-driven development strategies and a new wave of technological revolution, digital transformation is widely recognized as a pivotal force in reshaping corporate development models. Currently, China’s traditional growth model is confronted with dual pressures arising from increasingly stringent environmental constraints and rising factor costs; under such circumstances, relying solely on external regulation or compliance mandates is unlikely to generate sustained and substantive improvements in corporate ESG performance. In contrast, digital transformation has the potential to fundamentally reshape resource allocation mechanisms of firms by integrating data elements into production decision-making, operational processes, and organizational management systems (Yang et al., 2026).
From an internal organizational perspective, the adoption of emerging digital technologies directly stimulates green technological innovation (Xu & Yin, 2025), enhances production and managerial efficiency (Zhao et al., 2023), and intensifies competition within supply networks (Bag et al., 2025). These effects collectively promote energy conservation and emissions reduction while simultaneously improving ESG performance. Moreover, enhanced information transparency and reduced information asymmetry between firms and their stakeholders (He et al., 2024) enable executives to more accurately evaluate project risks and expected returns, thereby strengthening corporate risk-taking capacity (Wang et al., 2024b). Consequently, firms become more inclined to actively fulfill ESG responsibilities in pursuit of reputational gains and positive market feedback (Chen & Zhang, 2024). Furthermore, systematic data collection and analytics render internal processes more transparent and controllable (Wanyan & Zhao, 2025), encouraging firms to undertake long-term value-creation initiatives conducive to sustainable development. These initiatives include improving employee welfare, reducing inefficient investment, enhancing internal control quality, adopting digitally integrated systems, strengthening dynamic learning capabilities, and improving digital literacy and leadership among executives (Cheng & Yang, 2026; Chen & Xie, 2025; Lizarelli et al., 2025; Qiao et al., 2025; Song et al., 2025; Zhang et al., 2025a).
From an external perspective, digital transformation also increases analyst coverage (Cen et al., 2023) and attracts greater external investment (Khalid et al., 2024), thereby alleviating the financial pressures associated with fulfilling ESG responsibilities and fostering a stronger corporate commitment to ESG-oriented development.
As ESG evolves from an externally imposed regulatory requirement into an endogenous mechanism of value creation, digital transformation has increasingly been recognized as a crucial internal determinant of corporate ESG performance.
However, whether digital transformation enhances ESG outcomes by influencing research and development (R&D) investment intensity and R&D conversion efficiency of firms is an empirical question that merits further examination.
Thus far, no consensus has been reached regarding the measurement of corporate digital transformation, and three primary approaches have been adopted in the literature. The first relies on textual analysis of the frequency of digital-related keywords in annual reports of firms (Han et al., 2023; Zareie et al., 2024). The second employs financial statement indicators, such as information technology investments and the proportion of intangible assets, as indirect proxies (Nucci et al., 2023). The third constructs categorical or graded indices based on firms’ adoption of digital systems, including ERP, cloud computing, and big data platforms (Amador & Silva, 2025). Although these approaches capture certain dimensions of digital transformation, they remain subject to notable limitations. The conceptualization of digital transformation varies considerably across studies, encompassing technological applications and organizational and strategic transformation, thereby weakening the comparability of empirical findings. Moreover, relying on a single data source often fails to capture both the strategic orientation and substantive implementation of digital transformation, which may lead to measurement bias, such as high textual emphasis without corresponding investment or strong technological adoption without explicit strategic commitment. Accordingly, the development of a comprehensive and robust measurement index that integrates both strategic representation and implementation intensity is a critical issue in the literature. To address this challenge, this study proposes, as its primary methodological innovation, the construction of a comprehensive digital transformation index.
From a mechanistic perspective, existing studies generally contend that digital transformation enhances corporate ESG performance through channels, such as improved information transparency, optimized resource allocation, and strengthened internal controls (Ding et al., 2024; Li et al., 2024b; Wang et al., 2025a). However, previous studies have mostly concentrated on estimating the aggregate effect of digital transformation on ESG outcomes (Guo & Pang, 2025; Wei et al., 2025; Zhao et al., 2025), while paying limited systematic attention to the firm’s internal innovation system as a central mediating mechanism. The existing literature has largely followed one of two approaches: some studies emphasized changes in R&D investment scale, conceptualizing digital transformation as an external driver of innovation expenditure (Xu et al., 2025); others focused on innovation outputs or technological performance, interpreting digital technologies as instruments that enhance organizational efficiency and learning capacity (Yu et al., 2024). However, innovation input and innovation efficiency are typically treated as separate dimensions rather than examined jointly in a unified analytical framework. Consequently, the dual impact of digital transformation on innovation resource allocation decisions and conversion efficiency of firms has not been systematically explored, obscuring the internal logic by which digital transformation may influence ESG performance through the complementary pathways of investment expansion and efficiency enhancement. Particularly, in an era where innovation-driven development has become a central engine of economic growth, this structural omission at the mechanistic level constrains a deeper understanding of how digital transformation empowers firms to achieve sustainable development. Accordingly, as its second methodological contribution, this study integrates investment expansion and efficiency enhancement into a unified analytical framework to address these limitations.
Furthermore, although a growing body of studies has begun to investigate heterogeneity in the economic consequences of digital transformation across firm characteristics, such as firm size, ownership structure, and industry affiliation (Guo et al., 2025; Liu et al., 2023). However, the boundary conditions under which digital transformation influences ESG performance remain insufficiently explored and lack a coherent theoretical framework. Existing studies have mostly operationalized heterogeneity along a single dimension, thereby overlooking the joint influence of firms’ external market environments and internal resource endowments (Clemente-Almendros et al., 2024; Xie et al., 2024). Moreover, relatively few studies have simultaneously incorporated both external competitive pressures and internal resource constraints to examine how the impact of digital transformation on ESG performance varies across different levels of market competition intensity, financing constraints, and resource allocation conditions (Nguyen et al., 2025; Zang & Wei, 2026). Consequently, the limited identification of boundary conditions constrains the policy relevance and practical applicability of existing findings. To overcome this shortcoming, this study advances a third contribution by providing a more systematic and integrated analysis of the relevant boundary conditions.
This study seeks to make incremental contributions in three key respects. First, leveraging annual report disclosures from listed companies and large language models, a digital transformation representation index is constructed and subsequently integrated with an implementation-intensity indicator derived from digital patent data to generate a comprehensive digital transformation index. This novel measurement approach resolves comparability issues associated with traditional metrics, often caused by ambiguous conceptual definitions and insufficient methodological rigor, and substantially improves the precision and reliability of firm-level digital transformation measurement. Second, the study advances an integrated analytical framework that examines how innovation-driven digital transformation enhances corporate ESG performance through two complementary channels: R&D investment intensity and R&D conversion efficiency. While R&D investment intensity reflects firms’ commitment of resources to innovation, R&D conversion efficiency captures the effectiveness with which such investments are transformed into productive outcomes. Moreover, this research design enriches the theoretical paradigm linking digital transformation and sustainable development and provides practical insights for advancing innovation-driven development strategies and achieving net-zero emission objectives by incorporating both dimensions into a unified framework. Third, from a multidimensional perspective, including market competition intensity, digital peer convergence, financing constraints, and resource slack, this study systematically examines the heterogeneous impact of digital transformation on ESG performance. In total, these findings offer actionable guidance for firms seeking to optimize resource allocation and enhance ESG effectiveness by identifying key boundary conditions in the Chinese institutional context.
Theoretical analysis and research hypothesesCorporate digital transformation and ESG performanceCorporate digital transformation reconstructs value networks through the effects of technology diffusion. Additionally, in a data-driven environment of resource reallocation and transparent governance, it drives ESG performance improvement through a coherent logic of “cost reduction, capability enhancement, and alignment strengthening.”
First, in terms of cost reduction, from the perspectives of transaction cost and information economics, digitalization significantly reduces information asymmetry and transaction costs associated with search, coordination, and supervision (Li et al., 2024a). It enhances process visibility and verifiability, thereby enabling environmental compliance, supply chain accountability, and internal controls at lower marginal costs, while also freeing up resources for energy-efficiency upgrades, safe production, and fulfilling social responsibilities (Tabaku et al., 2025).
Second, regarding capability enhancement, both the resource-based view and dynamic capabilities theory suggest that digital elements, such as data, algorithms, and platforms, can be restructured into reconfigurable organizational capabilities. This strengthens real-time perception in R&D and operations, facilitates rapid iteration and lean management, fosters the adoption of green technologies and clean production, improves workforce and quality management effectiveness, and bolsters governance improvements through enhanced risk identification and resilience.
Third, in terms of alignment enhancement, stakeholder and reputation capital theories highlight that digitalization strengthens real-time interactions and accountability between firms and employees, customers, suppliers, and regulatory bodies. It enhances the accessibility and reliability of information disclosure, optimizes incentive and constraint mechanisms (McWilliams & Siegel, 2001), curbs short-termism and opportunism (Lemieux et al., 2012), and solidifies the sustainability of ESG commitments. Through traceable disclosure and interactive channels, firms can promptly address stakeholder concerns, creating long-term reputational constraints and market-driven incentives (Li et al., 2020; Wang et al., 2023). Digitalized performance assessments align key indicators with ESG objectives, improving consistency between internal governance and external regulation, and fostering endogenous momentum for enhancing corporate ESG performance. Based on these insights, Hypothesis 1 is formulated:
Hypothesis 1 Digital transformation enhances the ESG performance of firms.
Digital transformation enhances information accessibility, process visualization, and data-driven decision-making, thereby reshaping firms’ evaluations of the marginal returns and risks of R&D and ultimately increasing R&D investment intensity as a critical form of “resource commitment.” On the one hand, digitalization reduces the costs of technology intelligence search, project comparison, and cross-departmental collaboration (Acemoglu & Autor, 2011), mitigates early-stage project uncertainty, enhances the accuracy of feasibility assessments and success expectations, and incentivizes firms to expand R&D budgets and sustain long-term investment (Altomonte et al., 2016). On the other hand, with a higher stock of digitalization, each additional unit of R&D produces greater marginal returns in demand insights, simulation-based iteration, and knowledge reuse, thereby reinforcing the “more investment–more return” incentive. Meanwhile, digital development generates refined budgeting and closed-loop performance systems, such as milestone tracking, process monitoring, and personal accountability, that effectively reduce internal coordination frictions, enhance funding transparency, and strengthen accountability (Wang et al., 2025b). These mechanisms increase the organizational feasibility of raising R&D intensity while mitigating resource crowding-out effects.
Based on this foundation, higher R&D investment intensity strengthens ESG performance across three dimensions: environment, society, and governance. Environmentally, sustained R&D provides stable resources and experimental platforms for clean processes, energy-efficient equipment, and emission-reduction technologies (Dachs et al., 2024), thereby lowering energy consumption and emission intensity. Socially, R&D-driven improvements in processes and quality enhance product safety, facilitate employee skill upgrading, and strengthen supply chain responsibility practices. In terms of governance, R&D projects impose stricter requirements on internal control, compliance, and disclosure, thereby promoting institutionalized processes and auditing standards while curbing short-termism (Zhang et al., 2025b). Based on these arguments, Hypothesis 2 is formulated:
Hypothesis 2 Corporate digital transformation enhances R&D investment intensity, thereby improving ESG performance.
Digital transformation significantly improves R&D conversion efficiency by strengthening the comprehensive management of knowledge acquisition, integration, and implementation. First, data platforms, document libraries, and knowledge graphs eliminate information silos, facilitating cross-departmental knowledge flow and solution accumulation (Zang et al., 2025). This reduces redundant experimentation and improves the first-pass success rate of the conversion process from R&D to pilot trials and, ultimately, to production. Second, simulation modeling, digital twins, and data-driven stage-gate management accelerate solution screening and iteration, effectively shortening the trial-and-error cycle. This allows inefficient paths to be eliminated early, concentrating resources on high-probability projects. Third, product lifecycle management (PLM) systems and manufacturing execution systems (MES), integrated with supply chain collaboration platforms, support parallel engineering. This enables real-time coordination with customers, suppliers, and research institutions, accelerating the transition from technology development to process formation and eventually, to mass production (Alharbi, 2024). Fourth, sensors and traceability systems provide process visualization, track product quality, reduce rework and scrap probabilities, and stabilize output quality. Meanwhile, digital intellectual property management strengthens the protection of R&D results, reducing procedural risks in patent applications, reviews, and conversions, thus enhancing the stability of outcome conversion (Wang et al., 2024c).
Based on improved conversion efficiency, the rapid introduction and large-scale application of green technologies effectively reduce energy consumption and emission intensity, minimize material waste, and lower the likelihood of environmental accidents (Baek & Lee, 2024). Moreover, through process optimization and enhanced quality consistency, employee and product safety improve, training outcomes and skill upgrades become more pronounced, and the implementation of supply chain responsibility practices becomes more feasible. Additionally, standardized conversion processes, data tracking, and real-time disclosure enhance the quality of internal controls and external oversight. These mechanisms facilitate the creation of data-driven accountability and incentive structures, strengthening the ability to transform technological advancements into societal benefits. Based on these insights, Hypothesis 3 is formulated:
Hypothesis 3 Corporate digital transformation enhances R&D conversion efficiency, thereby strengthening ESG performance.
The theoretical model is shown in Fig. 1.
Research designBenchmark regression model design
To empirically investigate the impact of corporate digital transformation on ESG performance, this study constructed the following benchmark regression model:
Specifically, ESGi,t represents the ESG performance of firm i in year t, EDTi,t represents the digital transformation of firm i in year t, and Xi,t includes a set of control variables that may influence the ESG performance of firms. The study further controls for individual fixed effects γi, year fixed effects δt, and industry-year fixed effects and ρi×δt. εi,t is the random error term. β0, β1, and β2 are the coefficients to be estimated, with β1 representing the effect of corporate digital transformation on ESG performance, which is the key parameter of this study.Variable construction and descriptive statistical analysisDependent variable
In this study, corporate ESG performance was measured using the comprehensive ESG score published by Bloomberg. The Bloomberg ESG score is derived from a standardized disclosure framework that quantitatively evaluates firms’ performance across ESG dimensions and aggregates these assessments into a composite index. Higher scores indicate stronger ESG performance.
The use of Bloomberg ESG data in this study was motivated by several considerations. First, Bloomberg ESG adopts a highly structured and standardized data-collection framework, using diverse sources, such as corporate annual reports, sustainability reports, official company websites, and independent investigations conducted by Bloomberg. This systematic approach mitigates measurement bias arising from heterogeneous disclosure practices and enhances comparability across firms and over time. Second, the Bloomberg ESG rating system features a clear and transparent indicator structure. In the composite score, each ESG pillar accounted for approximately 33% of the total weight. Each pillar was further subdivided into multiple themes and specific indicators, forming a hierarchical and multidimensional evaluation framework. For example, the social pillar includes themes such as community and customer engagement, diversity, ethics and compliance, health and safety, human capital, and supply chain management. Each theme accounted for approximately 5.55% of the total score and was further decomposed into several subdomains, thereby providing a comprehensive assessment of firms’ social responsibility practices.1 Third, Bloomberg ESG data have been widely utilized in international empirical studies (Avramov et al., 2022; Fang et al., 2023), demonstrating strong academic acceptance and data stability. These attributes make the dataset particularly well-suited for large-scale, long-term empirical studies.
From a methodological perspective, existing ESG evaluation systems can generally be categorized into three types: quantitative approaches based on raw disclosure data, qualitative assessments relying on expert judgment or survey instruments, and integrated rating models that synthesize information from multiple sources (Avramov et al., 2022; Wang et al., 2024a). Compared with qualitative methods that depend primarily on subjective evaluation, the Bloomberg ESG rating system is based on standardized quantitative indicators, thereby providing objective, replicable, and comparable across firms and over time. Moreover, relative to researcher-constructed indices, established third-party rating datasets offer broader coverage, stronger data consistency, and greater long-term availability. In addition, prior evidence indicates that Chinese listed companies exhibited relatively weak incentives to engage in systematic misreporting in ESG disclosures (Fang et al., 2023), further reinforcing the credibility of ESG ratings derived from publicly available data. Detailed information on the indicator composition and weighting scheme of the Bloomberg ESG score is presented in Appendix A.
Independent variableThe existing literature on measuring digital transformation primarily encompasses the following approaches: first, by developing a digital transformation questionnaire and measuring the level of corporate digital transformation based on questionnaire responses (Zahoor et al., 2023). Second, constructing a digital transformation dictionary and measuring corporate digital transformation levels based on the frequency of digital transformation-related keywords in annual reports. Scholars have different approaches to handling word frequency. Some argue that the total frequency exhibits a “right-skewed” distribution and requires logarithmic transformation (Jia et al., 2025), while others, accounting for the measurement error caused by text length, apply a ratio operation by multiplying the total word frequency by 100 (Wu et al., 2025). Additionally, some scholars measure the level of digital transformation using the ratio of total disclosure frequency to total word count in the corresponding annual report (Wu et al., 2023). Third, by measuring corporate digital transformation through the proportion of digital economy-related intangible assets disclosed in specific intangible asset items in the financial statement notes of listed companies (Qi et al., 2020).
Despite existing studies and practical efforts in academia on measuring corporate digital transformation, each of the aforementioned methods has notable limitations. While the questionnaire method relies on primary data, the quality of responses is often compromised, as respondents tend to provide answers that conform to social norms rather than reflecting their true views. Word frequency methods are relatively straightforward to apply, but they are subject to researchers’ biases, and issues, such as incomplete dictionary construction and confusion between substantive transformation and conceptual hype, are common. Furthermore, digital transformation indicators constructed using different dictionaries often lack comparability. The intangible asset proportion method encounters the problem of accounting measurement distortion, as a significant portion of soft digital investments is not included in the financial statements.
In the context of foregoing considerations, this study constructed a comprehensive index of corporate digital transformation from two complementary dimensions: strategic representation and substantive implementation. On the one hand, a digital transformation representation index was developed using large language models to capture firms’ strategic positioning and cognitive orientation toward digital transformation. On the other hand, an implementation-intensity measure was constructed based on firms’ digital patent data to reflect their tangible progress in digital technology adoption and related investment activities. Based on these two dimensions, an equal-weighting scheme was applied, assigning 50% weight to each component, and aggregating them into a comprehensive corporate digital transformation index.
Digital transformation performance indexBased on the method by Jin et al. (2024), this study used the annual reports of listed companies as textual data and applied large language models to measure the corporate digital transformation performance index. The detailed steps are as follows:
First, construct the textual database. The annual reports of A-share listed companies on the Shanghai and Shenzhen stock exchanges between 2010 and 2023 were gathered from company websites, the Cninfo database, and other relevant sources. The reports were then converted into text files, and the “Management Discussion and Analysis” (MD&A) section was extracted as the research data. Since the MD&A section covers company operations, future strategies, and risk disclosures, it is commonly used in existing studies as the basis for calculating the frequency or proportion of digital technology-related keywords. Some scholars also include the “Table of Contents, Definitions, and Major Risk Disclosures” section in their analysis, but after reviewing numerous reports, this study found that this section was often too brief to support a thorough investigation. Therefore, this section was excluded from the text analysis in this study. In total, 41,636 annual report texts from 2010 to 2023 were compiled for this study.
Second, construct the prediction corpus and the labeling corpus. The annual report texts were split by periods and semicolons to create the prediction corpus. Next, based on the study by Jin et al. (2024), the new digital technologies were categorized into six types: big data, artificial intelligence, mobile internet, cloud computing, the Internet of Things, and blockchain. Furthermore, using the digital technology keyword dictionary constructed by Jin et al. (2024), this study extracted sentences that contain keywords and those that do not from the prediction corpus. Moreover, due to the annual increase in the number of listed companies, directly randomizing sentence labeling would lead to a sample bias toward recent years. To balance the year distribution, sentences were grouped by year and randomly sampled without replacement, with 2000 sentences selected from each year, producing a labeling corpus of 28,000 sentences.
Third, manual annotation. Annotators reviewed every sentence of the sentence corpus and used the keyword dictionary to identify sentences that match the keywords. Based on the keyword matching results and considering the context and logic of each sentence, annotators assessed whether the sentence truly addressed digital transformation. If digital transformation was identified, they then classified the sentence into one of the new digital technology categories, ensuring that misclassification due to the mere presence of keywords was avoided. To ensure the reliability of the annotation results, this study used independent labeling by multiple annotators, followed by cross-review and consistency checks of the results. Specifically, the consistency coefficient between annotators was calculated, yielding a result of 0.83, which indicates a “high level of consistency” and suggests that the annotation quality was reliable. Ultimately, the sentences in the text database were categorized into seven labels, including six types of new digital technologies and non-digital transformation.
Fourth, train the large language models. The labeled sentence corpus was divided into training, testing, and validation sets with an 8:1:1 ratio. To enhance measurement accuracy, this study independently trained several mainstream large models, including BERT, ERINE, ChatGPT, and Qwen.2 After optimizing the model parameters and comparing the recall rate, F1-Score, and F.8-Score of each model, it was determined that the BERT model was more suitable for the classification task in this study.3
Fifth, construct the index. Using the trained model, predictions were performed on the sentences in the prediction corpus. If a firm was identified as applying a given technology, the corresponding indicator was assigned a value of 1; otherwise, it was assigned 0. The values of each firm’s technology indicators were then aggregated, thereby yielding the corporate digital transformation performance index, which took values from 0 to 6.
Digital transformation implementation intensityThis study followed the approach by Tao et al. (2025) and constructed an indicator of digital transformation implementation intensity based on digital patent data. First, invention patent data for China were retrieved from the incoPat database, and patents were identified and matched to listed companies according to the names of patent applicants. Next, the IPC information for digital technology fields provided in the Classification of Core Digital Economy Industries and International Patent Classification Correspondence Table (2023) was applied, and each invention patent’s IPC classification number was identified and matched accordingly. If a patent’s IPC classification number falls within the digital technology field, it is recognized as a digital patent. Finally, the total number of digital patents for each firm was aggregated, and the natural logarithm of one plus the number of digital patents was used as the measure of digital transformation implementation intensity.
Comprehensive digital transformation indexBased on the above and following the method by Li et al. (2022), this study assigned equal weights of 50% to the digital transformation performance index and the digital transformation implementation intensity, and integrated them to construct the comprehensive digital transformation index.
This innovative approach specifically addressed three critical challenges in the measurement of digital transformation. First, it alleviated the response bias inherent in questionnaire-based surveys. Large language models objectively captured the digital transformation performance of firms by semantically analyzing corporate annual reports, thereby avoiding distortions caused by socially desirable responses and restricted sample coverage. The integration of objective digital patent indicators further strengthened the robustness of quantitative measurement. Second, it addressed the strong subjectivity and conceptual ambiguity of word-frequency methods. Large models possess superior semantic recognition and contextual comprehension capabilities, enabling them to differentiate between substantive digital transformation and mere conceptual rhetoric, thereby effectively avoiding misclassifications arising from incomplete dictionaries or conceptual hype. Third, it addressed the accounting distortion inherent in the intangible asset proportion method. As authoritative public disclosures, annual reports more faithfully reflected digital strategies and practices of firms, while objective digital patent data effectively captured the intensity of digital technology transformation, thus avoiding the mismeasurement of soft investments under accounting conventions. Moreover, this method balances dynamic adaptability with structural accuracy by combining the two, thereby significantly enhancing the effectiveness of digital transformation measurement.
Mediating variablesTo investigate how digital transformation influences ESG performance at the firm level, this study examined two mediating dimensions: R&D investment intensity and R&D conversion efficiency, corresponding respectively to the questions of “how much is invested” and “how effectively it is converted.”
First, R&D investment intensity. Corporate R&D expenditure was treated as a core resource commitment in the process of digital transformation. To ensure comparability, 2010 was selected as the base year, and both “R&D expenditure” and “total assets” were deflated using the CPI to construct real-value indicators. Based on this, the ratio of deflated R&D expenditure to total assets was calculated as the measure of R&D investment intensity, and the ratio was further transformed using the inverse hyperbolic sine function to mitigate the impact of extreme values. This indicator captured the relative allocation of innovative resources after controlling for firm size, thereby reflecting the “input-side” changes associated with digital transformation.
Second, R&D conversion efficiency. Investment alone does not guarantee effectiveness; the key lies in the efficiency with which inputs are transformed into outputs. Following Wang and Huang (2007), this study employed the data envelopment analysis (DEA) method, using “R&D expenditure” and “number of R&D personnel” as input variables, and “revenue growth rate” and “gross profit margin” as output variables, to generate an R&D efficiency score for each firm. The score ranged from 0 to 1, with values closer to 1 indicating higher efficiency. This efficiency score served as a measure of R&D conversion efficiency, capturing a firm’s capability to transform innovative inputs into business performance in the context of digitalization, thereby reflecting the “efficiency-side” improvement.
Control variablesTo account for other firm-specific characteristics that may affect ESG performance, this study introduced a series of control variables, following the approach of previous studies (Zhang et al., 2025b). At the financial level, the control variables included return on equity, book-to-market ratio, Tobin’s Q, and the management expense ratio. At the governance level, CEO duality and board size were also included as control variables. The definitions of the relevant variables and their descriptive statistics are reported in Table 1.
Variable definitions and descriptive statistics.
The ESG rating data of listed companies were obtained from the Bloomberg database. Annual report texts were compiled from the Cninfo platform and official websites of listed companies. Digital patent data were collected and matched from the incoPat database, while additional financial information was sourced from the CSMAR database. This study selected A-share listed companies in Shanghai and Shenzhen from 2010 to 2023 as the research sample, based on three primary considerations. First, as the core component of China’s capital market, the A-share market aggregates leading enterprises across diverse sectors, ensuring broad industry coverage and representative firm characteristics. Second, listed companies are subject to dual disclosure requirements under the Securities Law and stock exchange regulations, and their financial data must be verified by third-party auditing institutions, thereby providing higher data completeness and reliability compared to non-listed firms. Third, the sample period spanned the complete policy cycles and technological diffusion stages of China’s digital economy, from the industrial upgrading initiatives introduced in the 12th Five-Year Plan in 2011, to the intensive release of strategic documents, such as the New Generation Artificial Intelligence Development Plan, in 2017, and to 2023, when the digital economy accounted for over 40% of China’s GDP. This evolution reflects the transition of digital transformation from a mere technological application to a fundamental restructuring of corporate strategy. This 14-year observation window captured the gradual trajectory of firms’ digital transformation while revealing the dynamic evolution of their relationship with ESG performance, thereby mitigating estimation bias arising from short-term policy fluctuations or economic cycles. To refine the dataset, the following procedures were applied: (1) excluded financial firms and those designated as ST, *ST, or PT; (2) excluded firms listed for less than one year and delisted or suspended firms; (3) removed observations with missing values; and (4) winsorized all continuous variables at the 1% level on both tails.
Baseline analysis and empirical resultsThe effect of corporate digital transformation on ESG performanceTable 2 presents the regression results examining the impact of corporate digital transformation on ESG performance. Column (1) reports results without control variables, whereas Column (2) incorporates the full set of control variables and absorbs year, firm, and industry-year fixed effects. The findings indicate that even under these stringent specifications, the coefficient of the key explanatory variable remained positive and statistically significant at the 1% level, thereby supporting Hypothesis 1, that corporate digital transformation exerted a significantly positive effect on ESG performance. Improvements in ESG performance had implications beyond regulatory compliance, being closely associated with capital market returns, financing conditions, and operational outcomes. At the managerial level, digital transformation enabled firms to establish value co-creation networks with employees, customers, supply-chain partners, and regulators, thereby strengthening process transparency and traceability. On the cost side, digital tools decreased search, monitoring, compliance, and coordination costs and freed organizational resources to support ESG-related practices. On the benefit side, enhanced ESG performance strengthened corporate reputation, alleviated financing frictions, improved investor preferences, and can generate long-term shareholder gains, resulting in valuation premia in capital markets. Therefore, from an input-output perspective, although digital transformation entails upfront investment, the resulting improvements in ESG performance and firm value can yield sustainable economic returns over time, implying positive net benefits and commercial viability for most firms.
Baseline regression results.
Note: ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. Robust standard errors clustered at the firm level are reported in parentheses. The same applies hereafter.
In the baseline regressions, relevant control variables were included, and fixed effects were applied to mitigate potential omitted variable bias. However, since ESG performance is influenced by multiple factors, the possibility of residual bias cannot be completely excluded. To further mitigate endogeneity concerns, this study employed instrumental variable (IV) estimation and propensity score matching (PSM) to verify the robustness of the baseline regression results.
IV approachFollowing Zhang et al. (2023), this study employed the natural logarithm of one plus the interaction between the number of post offices in the firm’s province in 1984 and the national information technology service revenue in the preceding year as the IV. Specifically, the number of post offices in 1984 captured the historical communication infrastructure and information flow capacity. A greater density of post offices indicates a more advanced information transmission network, which enabled firms in such regions to access diverse information, including advanced managerial practices and technological knowledge, thereby laying the groundwork for future digital transformation. National IT service revenue in the preceding year directly reflects industry development dynamics. Such growth fuels technological innovation and iteration, equipping firms with the essential tools for digitalization and automation. This ensures that the chosen IV satisfies the relevance condition. Moreover, the number of post offices, as a historical variable, is unlikely to directly affect firms’ contemporaneous ESG performance. Similarly, the previous year’s national IT service revenue reflects a macro-level shock rather than firm-specific conditions, thereby satisfying the exogeneity requirement.
The instrumental variable (IV) regression results are reported in Table 3. Columns (1) and (2) present the first- and second-stage results of the two-stage least squares (2SLS) estimation. The first-stage regression results indicate that the IV was significantly and positively correlated with the core explanatory variable at the 1% level, while the second-stage results confirmed that the positive effect of corporate digital transformation on ESG performance remained statistically significant at the 1% level, consistent with the baseline regression findings. Furthermore, the Kleibergen-Paap rk Wald F statistic equaled 28.495, exceeding the critical threshold for the 10% bias level, thereby eliminating concerns regarding weak instruments. Therefore, the chosen IV was valid.
PSM approachDue to the heterogeneous initial conditions across firms and their varying probabilities of adopting digital transformation, this study employed PSM to mitigate potential sample selection bias. First, ROE, B/M, Tobin’s Q, DUAL, BS, and AER, along with year and industry dummy variables, were included as covariates in a Logit model to estimate the propensity scores of firms undertaking digital transformation. Subsequently, three matching techniques, 1:1 nearest-neighbor, kernel, and caliper matching, were applied to construct the control sample. The regression results after matching are reported in Table 4. The findings were consistent across the three matching methods, and coefficients remained statistically significant at the 1% level, indicating that, after accounting for observable heterogeneity, the positive effect of digital transformation on ESG performance remained robust.
Results of the PSM approach.
To assess the robustness of the baseline regression results, two alternative measures of corporate ESG performance were employed.
First, following Fang et al. (2023), firms were ranked by their Bloomberg ESG scores and divided into ten groups, with values from 1 to 10 assigned to each group, where higher values indicate better ESG performance. The regressions were then re-estimated using these assigned values, and the results are reported in Column (1) of Table 5.
Results of robustness tests.
Second, to mitigate potential bias from relying on a single rating agency, the Huazheng annual ESG ratings were adopted as an alternative to Bloomberg ESG scores. The Huazheng ESG rating system consisted of three primary pillars (environment, society, and corporate governance), 16 secondary themes (e.g., climate change, social contribution, and business ethics), and 44 tertiary issues.4 The Huazheng ratings have been widely adopted in recent academic studies (Deng et al., 2023; Li & Cheng, 2024). Huazheng classified ESG performance into nine levels (AAA, AA, A, BBB, BB, B, CCC, CC, and C). In this study, AAA, AA, and A were coded as 3; BBB, BB, and B as 2; and CCC, CC, and C as 1. The regressions were then re-estimated, and the results are reported in Column (2) of Table 5. In both Columns (1) and (2), the regression coefficients remained significantly positive, thereby confirming the robustness of the study’s findings.
Alternative measurement of digital transformationFirst, following Seth and McGillivray (2018), the weights of the composite digital transformation index were adjusted by assigning 70% to the performance index and 30% to the implementation intensity, and vice versa, after which the regressions were re-estimated. The results are reported in Columns (3) and (4) of Table 5.
Second, following Chen et al. (2024), who used the word-frequency method to measure corporate digital transformation, this study constructed a dictionary of digital transformation keywords.5 The frequency of these keywords in the main text of annual reports was then calculated, and the natural logarithm of one plus the frequency was used as an alternative measure of the core explanatory variable. The regressions were subsequently re-estimated, with the results reported in Column (5) of Table 5. Across all specifications, the regression coefficients remained consistent with expectations, thereby confirming the robustness of the conclusions.
Employing more stringent controlsTo further alleviate potential omitted variable bias and enhance identification, two additional stringent controls were implemented on the baseline model. First, province-by-year fixed effects were incorporated, and second, the institutional investor ownership ratio was introduced as an additional control variable. The corresponding regression results are reported in Columns (6) and (7) of Table 5, showing that the coefficient of the core explanatory variable remained significantly positive, thereby reinforcing the robustness of the baseline regression findings.
Mechanism analysis: how corporate digital transformation affects ESG performanceBased on the preceding theoretical analysis, this study investigated the mechanisms through which corporate digital transformation influences ESG performance along two dimensions, R&D input intensity and R&D conversion efficiency. The empirical model is specified as follows:
Here, RDIIi,t denotes the R&D input intensity of firm i in year t, whereas RDCRi,t represents the R&D conversion efficiency of firm i in year t. The mechanism test results reported in Table 6 indicate that digital transformation significantly enhanced both R&D input intensity and R&D conversion efficiency. On the one hand, digital transformation enabled firms to more precisely identify market demand and potential growth opportunities, respond swiftly to technological changes, and consequently increase their R&D investment in developing new products, technologies, and services. Such investment helps in maintaining competitive advantage, fostering continuous innovation, and promoting sustainable development. The increase in R&D funding equipped firms with the capital and incentives to strengthen ESG governance mechanisms, thereby improving ESG performance. On the other hand, by leveraging advanced transformation mechanisms and platforms, firms can more efficiently convert R&D outcomes into tangible products or services and accelerate their commercialization. This acceleration in the transformation process enabled firms to realize quicker returns on R&D investment, improve conversion efficiency, and consequently enhance technological capabilities, market competitiveness, and ESG performance. Therefore, Hypotheses 2 and 3 were empirically supported.
Results of mechanism analysis.
Note: Owing to missing values in the mechanism variables, the number of observations in these regressions is lower.
Unlike the conventional practice of measuring industry concentration using administrative industry classifications, this study adopted a product-market definition approach based on textual similarity of firms’ “main business” disclosures, thereby delineating the extended product market at the firm-year level. Specifically, cosine similarity scores were computed between firms based on the “main business” sections of their annual reports. For each firm i in a given year, the 10 most similar firms (excluding itself) were identified as the potential competitor set. Using similarity-normalized weights and the market shares of competitor firms j, the extended concentration index (ExtHHI) was constructed, from which competition intensity was defined as 1 - ExtHHI.6 Subsequently, the sample was split into high- and low-competition groups based on the annual median of competition intensity, and subgroup regressions were estimated.
The results are presented in Columns (1) and (2) of Table 7. Across both high- and low-competition environments, digital transformation exerted a significant positive effect on ESG performance. From a mechanism perspective, more intense competition in the extended product market tends to amplify the marginal impact of digitalization: on the one hand, competitive pressure accelerates firms’ adoption of data platforms, process reengineering, and traceability systems into their supply chain and governance processes (Nguyen et al., 2025), thereby mitigating information frictions and compliance risks; on the other hand, digitalization enhances the measurement and disclosure quality of environmental performance and social responsibility (McWilliams & Siegel, 2001), increasing the benefits of external monitoring and stakeholder engagement, which collectively contribute to improvements in ESG performance.
Heterogeneity by market competition intensity and digital peer intensity.
Note: Due to missing values in certain measurements, the total number of observations in the subsamples is smaller than that in the full sample.
Corporate practices may diffuse across firms through channels, such as competitive effects, cooperative networks, organizational learning, and institutional isomorphism. To investigate whether the impact of digital transformation varies with firms’ exposure to digital peers, a digital peer exposure index was constructed to capture peer intensity. The construction procedure parallels that of market competition intensity: the “main business” texts were processed, cosine similarity scores were calculated between firms, and for each firm i in a given year, the ten most similar firms (excluding itself) were identified, with normalized similarity scores applied as weights. The weighted average of the lagged digitalization levels of peer firms was then used to construct the digital peer exposure index. Firms were divided into high- and low-peer groups based on the annual median of this index, and subgroup regressions were estumated.
As shown in Columns (3) and (4) of Table 7, digital transformation significantly improved ESG performance for firms with high digital peer intensity, whereas the effect was not significant among firms with low peer intensity. A possible explanation is that firms with higher peer intensity were more likely to establish mechanisms for information sharing and collaboration. Resource sharing and complementarity enabled them to access the latest ESG-related information and best practices, such as green supply chain management and environmental performance improvements, thereby jointly addressing environmental and social challenges (Cao et al., 2019). Moreover, higher digital peer intensity within an industry provided more benchmark firms for learning. These firms often exhibited superior ESG performance, serving as role models. Such demonstration effects incentivized other firms to pursue digital transformation and ESG initiatives more proactively, with the aim of reaching or exceeding the performance of benchmark firms (Zhao & Wang, 2024).
Heterogeneity by financing constraintsTo capture heterogeneity in financing constraints, following Hadlock and Pierce (2010), this study employed the SA index as a measure of financing constraints. In each sample year, firms were classified into high- and low-constraint groups according to the median of the SA index, and regressions were estimated separately for each group.
The results, presented in Columns (1) and (2) of Table 8, showed that digital transformation significantly improved ESG performance among highly constrained firms, whereas the effect was insignificant for firms with low financing constraints. A plausible explanation is that firms under tighter financing constraints relied more heavily on internal financing, and reducing operating costs served as an important means of expanding internal capital availability. Furthermore, through the adoption of advanced information technologies, automation, and artificial intelligence, digital transformation significantly enhanced production efficiency and operational management while reducing operating costs (Brynjolfsson et al., 2021). For highly constrained firms, these cost-reduction and efficiency-improving effects were especially pronounced. Moreover, digital transformation facilitated more precise resource allocation and enabled firms to direct limited funds toward ESG-related initiatives. This enhances environmental performance, strengthens social responsibility, and improves corporate governance (Liu et al., 2025), thereby contributing to an overall improvement in ESG performance.
Heterogeneity by financing constraints and resource redundancy.
Following the classification of organizational slack proposed by Bourgeois (1981) and Sharfman et al. (1988), this study categorized corporate slack resources into available, recoverable, and potential slack. Specifically, available slack is measured using the cash and quick ratio, recoverable slack is proxied by the selling and administrative expense ratio, and potential slack is measured using the debt-to-asset ratio and the interest coverage ratio. All indicators were standardized, and a composite redundancy index was constructed through principal component analysis (PCA). Firms were subsequently divided into high- and low-redundancy subsamples according to the annual median of the composite redundancy index, and heterogeneity tests were conducted.
The results, reported in Columns (3) and (4) of Table 8, showed that digital transformation significantly enhanced ESG performance in firms with high resource redundancy. This finding can be explained by the fact that both digital transformation and ESG improvements typically require substantial upfront investments and complementary governance arrangements. Firms with high resource redundancy possessed sufficient financial resources and organizational buffers to successfully implement integrated systems, such as data platforms, environmental monitoring tools, and compliance disclosure mechanisms, and are more likely to benefit from positive incentives from capital markets and supply chains (Sekimoto & Amran, 2025). In contrast, firms with low resource redundancy, constrained by financial and human capital limitations, tend to prioritize efficiency-oriented digitalization. Their structural investments in ESG are often delayed or fragmented, resulting in insignificant effects (He et al., 2024).
Conclusions and recommendationsAs a critical pathway for advancing sustainable development and reshaping the business ecosystem, corporate digital transformation served as a strategic lever in improving ESG performance. Based on data from Chinese A-share listed companies between 2010 and 2023, and employing large language models, this study systematically examined the impact of corporate digital transformation on ESG performance and its underlying mechanisms. The empirical results demonstrated that corporate digital transformation significantly enhances firms’ ESG performance. Mechanism analysis further revealed that digital transformation strengthened both R&D input intensity and R&D conversion efficiency, which in turn promoted ESG performance. The heterogeneity analysis suggested that the enabling effect of digital transformation differs across levels of market competition. Furthermore, firms characterized by high digital peer intensity, strong financing constraints, and abundant slack resources benefited more significantly from digital transformation in terms of ESG performance.
Based on the foregoing findings, the following policy and managerial recommendations were proposed. First, digital transformation should be closely integrated with innovation investment to reinforce an innovation-driven pathway for enhancing ESG performance. The empirical evidence indicates that digital transformation significantly improved ESG outcomes by increasing both R&D investment intensity and R&D conversion efficiency, highlighting a critical linkage between digitalization and innovation activities. Accordingly, firms should avoid restricting digital transformation to short-term efficiency gains and instead align digital strategies with the development of long-term innovation systems. Specifically, ESG-oriented R&D programs may be incorporated into digital transformation plans, integrating data analytics, intelligent manufacturing, and green technology development in a unified innovation framework. Performance evaluation mechanisms should also be designed to channel digital investments toward ESG-related innovation initiatives. At the policy level, governments can utilize instruments, such as R&D subsidies, tax incentives, and innovation funds, to encourage firms to allocate digital investments toward green technologies and sustainable development projects, thereby amplifying the ESG spillover effects of digital transformation.
Second, a coordinated digital-ESG ecosystem that balances competition-driven incentives with peer-based diffusion effects should be fostered. The empirical evidence demonstrated that digital transformation significantly enhanced ESG performance in both highly competitive and less competitive market environments, indicating that digitalization exerted a broadly applicable and robust driving force. Furthermore, the ESG-enhancing effect of digital transformation was significantly stronger among firms with a higher degree of digital peer convergence, whereas this effect was not statistically significant among firms with lower levels of peer similarity. These findings suggest that while competitive pressure provides fundamental incentives for transformation, peer learning and cluster-based diffusion mechanisms further amplify the ESG benefits of digital transformation. Accordingly, policymakers should safeguard a fair and competitive market environment while simultaneously promoting the cluster-based diffusion of digital transformation and ESG practices. On the one hand, efforts should be made to reduce institutional entry barriers and strengthen information transparency and regulatory oversight, thereby encouraging firms to advance digital and sustainable transformation under competitive pressure. On the other hand, industry-level digital collaboration platforms should be developed, benchmark cases illustrating successful digital–ESG integration should be disseminated, and cross-firm technical exchanges and joint training initiatives should be organized to facilitate the diffusion of best practices from leading firms to their peers. In particular, the establishment of “digital–ESG” demonstration clusters in industrial agglomerations and high-technology parks should be encouraged, so that peer effects can drive broader and systemic improvements.
Third, targeted financial and transformation support policies should be implemented for enterprises facing resource constraints. It is further found that digital transformation exerted a more pronounced positive effect on ESG performance among firms with higher financing constraints and those possessing greater resource slack, whereas such effects were not statistically significant among firms with lower financing constraints or limited resource slack. These findings underscore the pivotal role of firm-level resource endowments in shaping the outcomes of digital-enabled sustainable transformation. Accordingly, policy priorities should shift from broad-based subsidies toward more precisely targeted support mechanisms. For firms bound by severe financing constraints, the financial support system for digitalization and R&D investment should be strengthened through dedicated credit lines, R&D loan risk-compensation mechanisms, and intellectual property-backed financing, thereby lowering the financial barriers to digital and ESG investments. For firms with substantial resource slack, tax incentives and performance evaluation mechanisms should be designed to encourage the reallocation of idle resources toward long-term digitalization and ESG initiatives. Meanwhile, firms facing pronounced resource constraints should be guided to adopt highly cost-effective digital tools with rapid returns, promoting ESG improvements in a gradual and risk-mitigated manner.
Although this study advances the measurement of digital transformation and explores its underlying mechanisms, several limitations remain that warrant further examination in future studies. First, this study relied on A-share listed firms as the sample, making the conclusions more applicable to publicly traded entities with standardized disclosure and stronger external governance. The applicability to non-listed and small- and medium-sized enterprises is uncertain. Future studies could broaden the sample scope to assess policy effects across different types of firms. Second, this study primarily investigated the effects of digital transformation on ESG performance through R&D input intensity and R&D conversion efficiency, without systematically addressing other potential channels, such as organizational management optimization or supply chain integration. Future studies can incorporate management practice surveys or IoT monitoring data to conduct parallel tests with the identified mechanisms. Finally, ESG performance in this study was primarily measured using Bloomberg ESG scores, with Huazheng ESG ratings employed as an alternative for robustness checks. However, systematic differences in indicator frameworks and weighting schemes across rating agencies may still influence coefficient magnitudes and significance levels. Hence, future studies could incorporate multiple agency ratings to construct a composite index, thereby enhancing cross-framework comparability.
FundingThis study was supported by the Major Program of National Fund of Philosophy and Social Science of China [NO.24&ZD069] and the Jiangxi Provincial Graduate Innovation Special Fund [NO. YC2025-B102].
CRediT authorship contribution statementLu Guo: Writing – review & editing, Supervision, Funding acquisition, Conceptualization. Lei Wang: Writing – review & editing, Writing – original draft, Validation, Software, Methodology, Investigation, Formal analysis, Data curation. Wenxiu Zhao: Writing – review & editing, Validation.
All authors certify that they have no affiliations with or involvement in any organization or entity with any financial interest or non-financial interest in the subject matter or materials discussed in this manuscript.
We declare that this manuscript is original, has not been published before and is not currently being considered for publication elsewhere.
We confirm that the manuscript has been read and approved by all named authors and that there are no other persons who satisfied the criteria for authorship but are not listed. We further confirm that the order of authors listed in the manuscript has been approved by all of us. In addition, all authors’ affiliations have agreed to the submission.
We understand that the Corresponding author is the sole contact for the Editorial process. She is responsible for communicating with the other authors about progress, submissions of revisions and final approval of proofs.
For instance, the theme “Ecological and Biodiversity Impact” (one of the components of the “environment” pillar) is composed of five subfields: Biodiversity Policy, Number of Environmental Fines, Environmental Fines (Amount), Number of Significant Environmental Fines, and Amount of Significant Environmental Fines.
The versions of the major models used in this study are listed as follows: The BERT model version is BERT-base-Chinese, the ERINE model version is ernie-3.0-base-zh, the ChatGPT model version is gpt2, and the Qwen model version is qwen3.
Due to space constraints, the detailed performance of the major language models in terms of recall rate, F1-Score, and F.8-Score is provided in Appendix B.
The detailed composition and indicator weights of the Huazheng ESG ratings are presented in Appendix C.
articles
