Information Disclosure in the Context of Combating Climate Change: Evidence from the Chinese Natural Gas Industry

Abstract

Natural gas (NG) is a key transitional energy source for clean energy transition. Against the backdrop of a grim climate change situation, the sustainable development of the Chinese NG industry is emphasized. Climate change disclosure (CCD) has become an important way for corporations to fulfill their social responsibility and demonstrate their capacity for sustainable development. In order to understand the current status of CCD in the Chinese NG industry and to improve the deficiencies, this paper assesses the quality of CCD in the Chinese NG industry. Climate change information is not fully covered by the existing quality evaluation systems. This study establishes a highly applicable system for evaluating the quality of CCD based on the theory pillar perspective. It includes the following five dimensions: completeness, balance, reliability, comparability, and understandability. This study evaluates the quality of CCD of 58 NG corporations using content analysis and quality evaluation index methods, incorporating Skip-Gram and CRITIC models. The evaluation results indicate that the quality of climate reports in the Chinese NG industry has shown general improvement over time; however, inconsistencies remain, making comparisons challenging. There are differences in the level of quality of CCD in the Chinese NG industry. Policy incentives with clear guidance and regional economic development conditions have a notable impact on the quality of CCD. For Chinese NG corporations themselves, disclosing climate change information related to risk management is the focus of narrowing the reporting gap. The CCD quality evaluation system constructed in this paper provides a theoretical reference for all industries to accurately promote disclosure quality. It also provides practical guidelines for corporations to identify weak links in CCD.

1. Introduction

Human activities have had far more profound and extreme adverse effects on the global climate than anticipated [1]. Stakeholders are increasingly inclined to criticize corporations that emit excessive amounts of greenhouse gases [2]. Corporations are both the primary contributors to greenhouse gas emissions and victims of the increasing frequency and intensity of extreme weather events [3]. Climate change poses both physical and transitional risks to businesses [4]. The development of strategies to manage climate-related risks and the disclosure of both the risks and strategies for risk management have become crucial components of corporate social responsibility [5].

COP29 explicitly emphasizes the role of information support in enhancing transparency on climate change, focusing on the importance of corporate CCD in achieving net-zero emissions [6]. International organizations and investment institutions have established a series of voluntary and consistent sustainable disclosure guidelines, creating essential transparency conditions to mitigate information asymmetry [7]. CCD becomes an effective way for stakeholders to understand that corporations are undertaking actions to address climate change [8,9]. Corporations utilize textual reports as the primary medium for the disclosure of information, and the reporting characteristics directly influence users’ decisions and market responses [10].

The global CCD is currently in a transitional phase from voluntary to mandatory reporting. The autonomy nature of content selection and the absence of unified disclosure standards have led to substantial variations in the quality of corporate CCD [11]. A subset of corporations opportunistically employ climate reporting as a tool for social legitimacy [12,13]. Corporations use narratives centered on decarbonization and greenhouse gas reduction commitments, while lacking substantive quantitative evidence of their impacts [14]. This practice exacerbates information asymmetry while facilitating “greenwashing” behaviors, thereby significantly undermining stakeholders’ trust in corporate practices and disclosure quality [15]. The transparency and completeness of CCD constitute key dimensions in curbing “greenwashing” behavior [16,17]. These two aspects, respectively, ensure the usability of CCD through normative practice [18] and content comprehensiveness (substantive value of information and scope of information coverage) [19]. Therefore, it is necessary to conduct quality assessments of CCD to enhance disclosure quality and regain stakeholder trust.

Energy has been prioritized in climate action plans [20]. The fossil fuel industry is facing increasing pressure from governments, climate change-concerned investors, and the public to decarbonize [21]. NG, as one of the representatives of clean energy, emits approximately 45% less carbon compared to coal for the same calorific value [22]. The US is the world’s largest producer and exporter of NG, mainly consumed in the power industry, with a high degree of marketization. Europe mainly imports NG, and the price is susceptible to geopolitics, supply relations, and other factors. The diversification of supply sources has resulted in a well-developed and flexible pipeline network. China boasts a coal-dominated energy consumption structure, while large-scale energy storage technology for renewable energy has yet to achieve breakthroughs [23]. “Coal-to-gas” transitions have emerged as a practical choice for ensuring energy security and facilitating the transformation of the energy mix. As an important part of the Chinese “carbon neutral” process, the “Action Plan for Carbon Dioxide Peaking Before 2030” clearly proposes guiding natural gas consumption in an orderly manner [24]. The structure of China’s energy consumption has led to the expectation that NG will play an active role in transitioning to a net-zero energy system.

However, the NG industry also faces challenges posed by the frequent occurrence of extreme weather events. Extreme weather can lead to risks such as the failure of NG extraction, transportation, and storage, as well as damage to infrastructure (including pipelines, gas storage facilities, filling stations, etc.). These risks not only jeopardize supply stability but may also lead to safety incidents. To avoid the cessation of the NG business under pressure to address climate change in the future, its ability to enforce CCD has begun to gain traction. The state enhances the level of corporate information disclosure by guiding public participation in environmental affairs [25] and collaborating with multiple departments to jointly issue policies [26]. The State Power Investment Corporation and other organizations jointly release the “ESG Disclosure Guidelines for Energy Corporations” and “ESG Evaluation Guidelines for Energy Corporations”. These promise to fill the gap in China’s information disclosure standard development in the energy sector [27]. China is in the initial stage of developing CCD, facing issues such as large differences in disclosure quality, incomparable disclosure information, and the randomness of disclosure [28]. Consequently, an assessment of the quality of CCD within China’s NG industry can shed light on the prevalent challenges in current disclosure practices. Based on the results of the evaluation, corporations improve CCD in a targeted manner to meet the information needs of stakeholders.

The literature focuses on the quality influencing factors of CCD. From the perspective of external factors, the recognition and enforcement of mandatory directives are the most direct and effective factors affecting the quality of CCD [29,30]. Among third-party stakeholders, environmental organizations and auditing institutions can enhance the level of CCD [31,32]. Giannarakis et al. propose that government ownership and the independent verification of environmental data determine the level of CCD [8]. García-Sánchez et al. indicate that business transparency related to climate change is explained by climate governance and the detailed coverage of corporations by financial analysts [33]. Even-Tov et al. found that after expanding government procurement opportunities, corporations with higher exposure to government contracts significantly increased their climate disclosures [34]. Assaf et al. argue that the quality of climate change-related information improves during periods of political uncertainty, providing moral legitimacy and building trust among all stakeholders [35]. Linares-Rodríguez et al. argue that geographic regional differences, including regional planning and economic levels, can have an impact on the quality of corporate CCD [36]. From an internal perspective, Shvarts and He et al. found that state-owned enterprises (SOEs) and large listed companies exhibit higher-quality CCD, with SOEs outperforming large listed firms [37]. Girma et al. argue that SOEs are subject to more restrictions from national regulations and political burdens [38]. Lee et al. observed that board independence and diversity positively contribute to CCD [39]. Mou et al. argued that firm size, dual listing, and board size have a positive impact on the quality of CCD, whereas institutional ownership is negatively correlated with disclosure quality [40]. Kouloukoui et al. suggested that firm size or whether a corporate originates from a developed country does not necessarily explain the level of climate risk information disclosed [41]. Bui et al. demonstrated that governance is positively correlated with CCD levels, and integrating climate governance into corporate governance mechanisms enhances carbon-related disclosure [42]. Dilling et al. found that the presence of a Chief Sustainability Officer and the adoption of climate transition plans are positively and significantly associated with climate governance disclosure [43]. Haque et al. highlighted that the lack of proactive stakeholder engagement, coupled with an excessive focus on financial performance and shareholder value, negatively affects disclosure levels [44]. Cho et al. argue that the use of climate information reporting as a symbolic tool to gain organizational legitimacy significantly reduces the improvement of substantive information for climate governance [45]. Tyagi et al. argued that corporations’ voluntary reporting of non-material climate change information and the use of non-quantitative descriptions adversely impact information quality [46]. The existing empirical analysis literature mainly uses a sample of international corporations and focuses on carbon information, a core component of climate change information. Different countries differ significantly in terms of institutional environment, stage of economic development, industry characteristics, and corporate characteristics. Empirical findings based on foreign samples may lack generalizability across country scenarios. The Chinese policy-driven market environment and the concentrated industrial landscape of state-owned enterprises may mark a unique driving path for improving CCD quality. However, China remains in the nascent stage of CCD, and data availability constrains large-scale empirical investigations.

China has been slow to respond to CCD, and research on disclosure has mainly focused on carbon and environmental information. There has been no assessment of the quality of CCD in the Chinese NG industry. The existing assessment systems have their own focuses and fail to cover all aspects of CCD. In order to construct a scientific and rigorous quality evaluation system, this study is based on the theoretical foundation of CCD and systematically analyzes the quality characteristic elements under each theoretical framework. Through theoretical integration and complementation, we construct a multi-dimensional evaluation perspective. A specified keyword extraction model is constructed by incorporating the Skip-Gram model to expand the lexical meaning, thus realizing the deep semantic parsing of unstructured text data. The content analysis method is used to analyze the disclosure text in a structured way and quantify the indicators of each dimension. Based on the CRITIC method, the Climate Change Disclosure Quality Index (CCDQI) is constructed to measure the level of CCD. Based on the results of the CCDQI, we analyze the causes of the discrepancies and propose effective suggestions. This study contributes to CCD in three ways. First, it proposes a new evaluation index system for measuring the level of corporate CCD, incorporating the theoretical pillars of CCD, which complements the research on the quality of CCD. These principles include completeness, balance, reliability, comparability, and understandability. Compared with previous studies, this evaluation indicator incorporates climate change information from multiple mainstream international disclosure guidelines and is highly applicable. Second, it expands the scope of research on the CCD quality in China, analyzes the Chinese NG industry for the first time, and explores the factors affecting CCDQI levels. Instead of relying on theoretical speculation, this study analyzes corporations’ actual disclosure reports. Third, it encourages corporations’ near-zero emissions to provide theoretical references. The scientific quality evaluation system constructed in this paper provides an operational reference framework for corporations to optimize CCD. It also helps report preparers to self-check the disclosure quality.

The structure of the remaining sections of this paper is as follows. Section 2 summarizes the evaluation dimensions and evaluation methods of the CCD quality evaluation framework. Section 3 describes the data sources and specific dimension measures of this paper. Section 4 analyzes the results of the CCD quality evaluation in the Chinese NG industry and discusses the quality influencing factors. Section 5 discusses the suggestions for improving CCD in the Chinese NG industry. Conclusions are drawn in Section 6.

2. Methods

2.1. Theoretical Foundation

Based on the research on information disclosure theory [7,19,47,48,49,50,51], this mainly involves the theories summarized in Table S1. This study analyzes the meaning of information disclosure theory and its performance in CCD, and it extracts the information focus of CCD quality evaluation. The information focuses, viewed from different theoretical perspectives, are both different and intrinsically related. It is difficult for a single theory to comprehensively cover the multiple characteristics of corporate climate change information disclosure. This study constructs a multidimensional comprehensive evaluation framework by systematically integrating relevant theoretical foundations. Specifically, this study analyzes the information focus of each theory and identifies its complementary relationship. It refines and summarizes the key quality features through theoretical cross-validation. Eventually, we form the following five core evaluation dimensions, which provide a theoretical foundation for CCD quality evaluation research:

(1)

Completeness, relevance, comprehensiveness, validity, and prospectiveness.

(2)

Balance.

(3)

Reliability, verifiability, and authenticity.

(4)

Comparability and consistency.

(5)

Understandability and substantiveness.

2.2. CRITIC Method

The CRITIC method determines the weights of indicators by evaluating both the contrast intensity and the interdependence among them. Unlike subjective approaches such as the AHP, the CRITIC method relies entirely on objective data, thereby eliminating subjective bias [52]. In comparison to the entropy method and standard deviation method, CRITIC accounts for the correlation between indicators, offering a more robust assessment [53]. In comparison to the PCA method, CRITIC preserves the original data for weight allocation, ensuring no loss of information due to dimensionality reduction [54].

Data processing. To mitigate the influence of different magnitudes on the evaluation results, it is essential to conduct dimensionless processing on each indicator. Positive indicators follow Formula (1), negative indicators follow Formula (2).

$${\mathrm{X}}_{\mathrm{ab}}={\displaystyle \frac{{\mathrm{X}}_{\mathrm{b}}-{\mathrm{X}}_{\min }}{{\mathrm{X}}_{\max }-{\mathrm{X}}_{\min }}}$$

(1)

$${\mathrm{X}}_{\mathrm{ab}}={\displaystyle \frac{{\mathrm{X}}_{\max }-{\mathrm{X}}_{\mathrm{b}}}{{\mathrm{X}}_{\max }-{\mathrm{X}}_{\min }}}$$

(2)

Variability reflects the degree of dispersion of each indicator across different evaluation scenarios. The greater the variability, the more significant the indicator’s contribution to distinguishing between different scenarios, and thus, a higher weight is assigned. Conflict reflects the degree of linear dependence between indicators. The higher the correlation between two indicators, the more likely they provide redundant information, and consequently, a lower weight is assigned. The calculation process is as follows.

Step 1. Original indicator data matrix. The original dataset includes n samples evaluated across p key indicators. Where ${\mathrm{X}}_{\mathrm{ab}}$ represents the value of the b-th evaluation indicator for the a-th sample.

$${\mathrm{X}}_{\mathrm{ab}}=\left(\begin{array}{ccc}{\mathrm{X}}_{11}& \dots & {\mathrm{X}}_{1\mathrm{p}}\\ ⋮& \ddots & ⋮\\ {\mathrm{X}}_{\mathrm{n}1}& \dots & {\mathrm{X}}_{\mathrm{np}}\end{array}\right)$$

(3)

Step 2. Indicator variability. Using the standard deviation to represent the fluctuation in values within each indicator. ${\mathrm{S}}_{\mathrm{b}}$ represents the standard deviation of the b-th indicator.

$${\mathrm{X}}_{\mathrm{b}}={\displaystyle \frac{1}{\mathrm{n}}}{{\displaystyle \sum }}_{\mathrm{a}=1}^{\mathrm{n}}{\mathrm{X}}_{\mathrm{ab}}$$

(4)

$${\mathrm{S}}_{\mathrm{b}}=\sqrt{{\displaystyle \frac{{{\displaystyle \sum }}_{\mathrm{a}=1}^{\mathrm{n}}\left({\mathrm{X}}_{\mathrm{ab}}-{\mathrm{X}}_{\mathrm{b}} \right)²}{\mathrm{n}-1}}}$$

(5)

Step 3. Indicator conflict. The correlation coefficient is employed to express the interdependence between indicators. In addition, ${\mathrm{r}}_{\mathrm{ab}}$ is the correlation coefficient between evaluation indicators a and b.

$${\mathrm{R}}_{\mathrm{b}}={\displaystyle \sum }_{\mathrm{a}=1}^{\mathrm{p}}\left(1-{\mathrm{r}}_{\mathrm{ab}}\right)$$

(6)

Step 4. Information content. The larger the ${\mathrm{C}}_{\mathrm{b}}$ is, the greater the role of the b-th evaluation indicator within the entire evaluation index system, thereby warranting the allocation of a higher weight to it.

$${\mathrm{C}}_{\mathrm{b}}={\mathrm{S}}_{\mathrm{b}}\times {\mathrm{R}}_{\mathrm{b}}$$

(7)

Step 5. Objective weight. The objective weight W_b for the b-th indicator.

$${\mathrm{W}}_{\mathrm{b}}={\displaystyle \frac{{\mathrm{C}}_{\mathrm{b}}}{{{\displaystyle \sum }}_{\mathrm{b}=1}^{\mathrm{p}}{\mathrm{C}}_{\mathrm{b}}}}$$

(8)

2.3. Keyword Extraction Method

2.3.1. Text Preprocessing

The three-step operation flow of text preprocessing is shown in Figure 1, which is completed in Python 3.12.

2.3.2. Designated Keyword Thesaurus Construction

Chinese can use a variety of words with similar meanings to express the same concept. It is necessary to augment the initial keyword thesaurus by incorporating synonymous terms.

Word2vec is a word vector model based on a shallow neural network, primarily comprising the following two models: CBOW and Skip-Gram. The core principle is that words occurring in similar contexts tend to possess similar meanings. The Skip-Gram model predicts the surrounding context words based on a given target word. In this study, the Skip-Gram model is trained using the Gensim library in Python.

The Skip-Gram model is structurally divided into the following three layers: input layer, hidden layer, and output layer (Figure 2). Each node in the input layer corresponds to a target word, represented by a one-hot encoded vector. The center word vector is mapped to the hidden layer through a weight matrix. The output of the hidden layer is mapped to the vocabulary through another weight matrix. The probability distribution of each context word is computed by the softmax function. If we want to obtain two words around the target word w, the output layer is w(t − 2), w(t − 1), w(t + 1), w(t + 2). The calculation process is as follows.

Word vector generation. Assuming that the target word is indexed by i in the vocabulary list V = $\left[{\mathrm{w}}_1{,\mathrm{w}}_2,\dots {,\mathrm{w}}_{\mathrm{k}}\right]$, ${\mathrm{w}}_{\mathrm{i}}$ can be expressed as follows:

$${\mathrm{w}}_{\mathrm{f}}=\left\{\begin{array}{c}1, if f=i\\ 0,\mathrm{otherwise}\end{array}\right.$$

(9)

The objective of the Skip-Gram model is to maximize the conditional probability of the context words ${\mathrm{w}}_{\mathrm{t}+\mathrm{j}}$ given the target word $w_t$. The word vectors are updated using the gradient descent optimization algorithm.

$$\mathrm{Maximize}{\displaystyle \frac{1}{\mathrm{T}}}{\displaystyle \sum }_{\mathrm{t}=1}^{\mathrm{T}}{\displaystyle \sum }_{-\mathrm{c}\le \mathrm{j}\le \mathrm{c},\mathrm{j}\ne 0}\log \mathsf{{\rm P}}\left({\mathrm{w}}_{\mathrm{t}+\mathrm{j}}{|\mathrm{w}}_{\mathrm{t}}\right)$$

(10)

T is the total number of words in the corpus; c is the size of the context window.

The softmax function transforms the raw output logits of the model into probability values, ensuring that the sum of all probabilities equals 1.

$$\mathsf{{\rm P}}\left({\mathrm{w}}_{\mathrm{t}+\mathrm{j}}{|\mathrm{w}}_{\mathrm{t}}\right)={\displaystyle \frac{\exp \left({\mathrm{V}}_{{\mathrm{w}}_{\mathrm{t}+\mathrm{j}}}^{\mathrm{T}}{\mathrm{V}}_{{\mathrm{w}}_{\mathrm{t}}}\right)}{{{\displaystyle \sum }}_{\mathrm{k}=1}^{\mathrm{V}}\exp \left({\mathrm{V}}_{{\mathrm{w}}_{\mathrm{k}}}^{\mathrm{T}}{\mathrm{V}}_{{\mathrm{w}}_{\mathrm{t}}}\right)}}$$

(11)

${\mathrm{V}}_{{\mathrm{w}}_{\mathrm{t}}}$ represents the word vector of the target word; ${\mathrm{V}}_{{\mathrm{w}}_{\mathrm{t}+\mathrm{j}}}$ denotes the word vector of the context word; and ${\mathrm{V}}_{{\mathrm{w}}_{\mathrm{k}}}$ is the word vector for all words in the vocabulary list.

Since the denominator of the softmax function requires summation over the entire vocabulary, the Skip-Gram model typically employs negative sampling as an optimization technique.

$$\log \mathsf{\sigma }\left({\mathrm{V}}_{{\mathrm{w}}_{\mathrm{t}+\mathrm{j}}}^{\mathrm{T}}{\mathrm{V}}_{{\mathrm{w}}_{\mathrm{t}}}\right)+{\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{k}}{\mathsf{{\rm E}}}_{{\mathrm{w}}_{\mathrm{i}}}~{\mathsf{{\rm P}}}_{\mathrm{n}}\left(\mathrm{w}\right)\left[\log \mathsf{\sigma }\left(-{\mathrm{V}}_{{\mathrm{w}}_{\mathrm{i}}}^{\mathrm{T}}{\mathrm{V}}_{{\mathrm{w}}_{\mathrm{t}}}\right)\right]$$

(12)

k is the number of negatively sampled samples; ${\mathsf{{\rm P}}}_{\mathrm{n}}\left(\mathrm{w}\right)$ is the distribution of negatively sampled samples (usually 3/4th power of word frequency).

The “most_similar” method is used to retrieve words that are most similar to a given word based on cosine similarity between word vectors. To optimize the results, it is necessary to set an appropriate vocabulary size, adjust the expansion threshold, and perform manual validation and filtering.

$$\mathrm{similarity}\left({\mathrm{w}}_1{|\mathrm{w}}_2\right)={\displaystyle \frac{{\mathrm{v}}_{{\mathrm{w}}_1}·{\mathrm{v}}_{{\mathrm{w}}_2}}{||{\mathrm{v}}_{{\mathrm{w}}_1}||||{\mathrm{v}}_{{\mathrm{w}}_2}||}}$$

(13)

2.3.3. Word Frequency Statistics

Step 1. Perform word frequency statistics on the preprocessed text.

Utilize the “collections. Counter” to conduct all word frequency statistics on the preprocessed text from Section 2.3.1, generating a word frequency distribution dictionary, denoted as word_freq.

Step 2. Extract specified keyword frequencies.

Apply “Counter” to further extract the specified keyword word frequency and filter the specified keyword word frequency using dictionary derivation as follows: {keyword: word_freq [keyword] for keyword in keywords if keyword in word_freq}. Generate the dictionary keyword_freq for the specified keyword and its word frequency.

3. Constructing a Climate Change Disclosure Quality Evaluation System

3.1. Data Source

China is in the early stages of developing its CCD. As industry head corporations, listed companies exhibit high sensitivity to policies and regulatory requirements and possess the financial capacity to implement the related practices. The three major stock exchanges in China (Hong Kong, Shenzhen, and Shanghai) host the primary listed companies in the NG industry, covering the entire industrial chain from upstream exploration and exploitation, midstream storage and transportation, to downstream direct marketing and distribution. These stock exchanges have clear requirements for information disclosure. If listed companies engage in CCD, they are required to publish the relevant reports on their official websites.

This study aligns the industrial chain modules of the Chinese NG sector with the industry classification guidelines of the stock exchange. Based on the industry code to which the industry chain modules belong, a covered corporation search was conducted on the stock exchange website and a total of 95 corporations were obtained. Pipe China, as a state-owned enterprise, represents a core initiative of the Chinese “pipeline-transport separation” reform. It is responsible for the nationwide operation of natural gas pipeline infrastructure.

This paper downloads CCD-related reports from the official websites of corporations, 58 of which have released ESG, CSR, or sustainability reports individually. The starting time for corporations to carry out CCD is inconsistent. Sample corporations have been disclosing relevant information since 2007. In 2021, China issued the “Reform Plan for the System of the Law-based Disclosure of Environmental Information”, which requires key corporations to disclose environmental information on a mandatory basis. The number of corporate disclosures increased significantly, and the sample size of the study was enlarged (Figure 3). The post-2021 time period can cover the latest trend and avoid the lag of earlier data. Second, the mainstream sustainability disclosure standards selected under the completeness dimension of this paper have been updated or new standards have been issued after 2020. The content of earlier reports may not be able to meet the current disclosure requirements and reflect the current stage of the implementation of the disclosure standards. Therefore, this paper selects 2021–2023 as the study period. Finally, 147 research reports were obtained, including 61 ESG reports, 69 CSR reports, and 17 sustainability reports.

3.2. Climate Change Disclosure Quality Index System

• Establishment of an evaluation index system

Based on the five quality dimensions delineated in Section 2.1, and taking into account the quality characteristics of each dimension, the evaluation indicators for each dimension are defined. The scoring strategy for the evaluation indicators is also specified (Figure 4).

IFRS s2, as the first comprehensive international standard for CCD, covers information on governance, strategy, risk management, and metrics and targets. Keyword categorization under the completeness dimension follows these four themes.

• Governance: How organizations manage climate-related risks and opportunities.
• Strategy: The implications of climate-related risks and opportunities for an organization’s business, strategic, and financial planning.
• Risk management: The process by which the organization identifies, assesses, and manages climate-related risks.
• Metrics and targets: The metrics and targets the organization uses to manage climate-related risks and opportunities.

Since the Skip-Gram model training data are based on large-scale corpus sampling, this paper uses the pre-trained corpus as the initialization value. The data sources include the Chinese Wikipedia corpus, the text of corporate ESG, CSR, sustainability reports, and the relevant national government documents, such as Responding to Climate Change: China’s Policies and Actions. The final keywords in

Table 1. Integrity dimension keywords.

Themes	Specific Keywords	Name
Governance	Environmental Policy/Environmental Management Policy	EP
	Board Oversight, Supervision, Governance, Management/Management Oversight, Supervision, Governance, Management	BS
	Governance Structure	GS
	Management Responsibility	MR
	Employee Rewards/Employee Incentives/Employee Benefits/Employee Development	EMR
Strategy	Climate Risks/Climate Opportunities	CR
	Strategy and Decision Making/Transition Plans/Development Plans/Business Models	SD
	Value Chain Management/Supply Chain Management	SCM
	Climate Scenarios/Climate Adaptation/Climate Change Recovery/Climate Change Transformation	CS
Risk Management	Climate Risk Identification	CRI
	Climate Risk Assessment	CRA
	Climate Risk Monitoring	CRM
	Climate Risk Control	CRC
	Climate Risk Management	CRN
	Climate Risk Impacts	CRP
Metrics and Targets	Emission Reduction Actions/Measures	ERA
	Energy Consumption/Water Use/Land Use/Waste Management	RC
	Scope 1 (i)	S1
	Scope 2 (ii)	S2
	Scope 3 (iii)	S3
	Low Carbon Products	LCP
	Low Carbon Investments	LCI
	Carbon Trading/Carbon Pricing/Carbon Credits	CT
	Biodiversity/Ecological Impacts	EI
	GHG Emission Targets/GHG Reduction Targets/Carbon Emission Targets/Carbon Reduction Targets	GHG
	Emissions Performance/Environmental Performance	ER

Note: Scope 1, Scope 2, and Scope 3 are delineated by the World Resources Institute (WRI) and World Business Council for Sustainable Development.

are obtained.

• Measurement of the response of the reference texts

Specific dimension scores are obtained according to the evaluation criteria, and the data are summed after dimensionalization. Because the CCD in the Chinese NG industry is not comparable (see Tables S2 and S3), weights are calculated only for completeness, balance, reliability, and understandability (

Table 2. Results of CRITIC weighting.

Item	Indicator Variability	Indicator Conflict	Amount of Information	Weighting
Completeness	0.194	3.307	0.641	20.64%
Balance	0.278	2.557	0.710	22.87%
Reliability	0.366	2.547	0.931	29.97%
Understandability	0.343	2.399	0.824	26.52%

). The weights are obtained using CRITIC’s objective assignment method. The following formula is used to obtain the CCDQI:

CCDQI_i,t = Completeness_i,t × 20.64% + Balance_i,t × 22.87% + Reliability_i,t × 29.97% + Understandability_i,t × 26.52%

(14)

CCDQI∈[0,1]

The CCDQI i,t is the score of corporation i in year t.

Spearman’s correlation coefficient is widely used to verify the validity of data. This is because it possesses the advantages of nonparametric properties, the ability to measure monotonic relationships, as well as being easy to understand and interpret [56]. Spearman’s correlation coefficient is used to verify the validity of the CCDQI on the final results of each dimension. The calculation process is as follows.

$$\mathsf{\rho }=1-{\displaystyle \frac{6\sum {\mathrm{di}}^2}{\mathrm{n}\left({\mathrm{n}}^2-1\right)}}$$

(15)

$\mathrm{di}$ is the absolute value of the difference between the rank (ordinal) of the two variables X and Y in the i-th pair of observations.

$\mathrm{n}$ is the total number of observations.

The results show a significant positive correlation between the CCDQI and completeness, balance, reliability, and understandability, affirming the validity of the CCDQI (

Table 3. Spearman correlation coefficient.

	CCDQI	Completeness	Balance	Reliability	Understandability
CCDQI	1
Completeness	0.737 **	1
Balance	0.724 **	0.522 **	1
Reliability	0.627 **	0.282 **	0.226 **	1
Understandability	0.810 **	0.621 **	0.439 **	0.301 **	1

Note: ** p < 0.01.

). The database contains qualitative and quantitative data. The qualitative content contains the complete CCD, which provides a reference for companies’ disclosure on environmental issues. The quantitative information contains the results of the analysis of the dimensions and comprehensive scores of the sample corporations. This facilitates the analysis of the quality level of disclosure and the development and evolution of the level of information disclosure.

4. Results and Discussion

4.1. CCD Analysis

4.1.1. Status of CCD in the NG Industry in China

When analyzing the use of report preparation guidelines for sample corporations, most explicitly chose multiple reference guidelines. Ten corporations use a single guideline, and six corporations do not specify or follow the preparation guidelines for three years. China’s climate disclosure guidelines have been issued extensively. These guidelines involve 31 types, accounting for approximately 68% of the total, with a relatively low overall utilization rate. The c and f are the most commonly used standards. The h, k, and x preparation guidelines are expected to become widely available by 2023 (see Table S4).

Corporations can refer to or follow these guidelines. The number of corporations that strictly indicate metrics will increase from 38% in 2021 to 55% in 2023. Most corporations use c, f, or a combination of the two to indicate the topics related to their disclosure. In response to the fragmentation of climate disclosure standards, the Ministry of Finance of the People’s Republic of China proposed building on the ISSB standard and establishing a national unified sustainability disclosure standard system by 2030 [57].

4.1.2. CCDQI Differences in the Quality Level of Each Corporation

The Chinese NG industry CCDQI results are shown in Table S5. The quality of CCD in the Chinese NG industry has shown a consistent upward trend over the years. The number of high-score data points has been increasing each year. This indicates that Chinese NG corporations disclosed more or higher-quality climate change information during 2021–2023. There are CCDQI-based differences in the quality level of each corporation. The data points are centrally distributed in the range of 0.5 and below, with the overall quality level being low (Figure 5).

The increase is mainly due to two reasons. In terms of the external environment, substantial Chinese progress in the construction of an information disclosure system has exerted significant external pressure on CCD in the NG industry (

Table 4. Policy developments related to CCD.

Year	Policy	Practice
2003 [59]	“Announcement on Enterprise Environmental Information Disclosure”	Initial explorations in environmental information disclosure by corporations.
2017 [60]	“Cooperation Agreement on Jointly Carrying Out Environmental Information Disclosure for Listed Companies”	Promoting the establishment of a mandatory environmental information left system for listed companies.
2017 [61]	“CSRC Announcement [2017] No. 17”	Encourage corporations to take the initiative to disclose their efforts to actively fulfill their social responsibilities, taking into account the characteristics of their industries.
2021 [62]	“Reform Plan for the System of the Law-based Disclosure of Environmental Information”	Promote the mandatory disclosure of environmental information by corporations and improve the working norms for the participation of third-party organizations.
2021 [63]	“Measures for the Administration of the Law-based Disclosure of Environmental Information by Enterprises”	Clear disclosure of information relating to the climate change response of projects for which financing is provided.
2021 [64]	“Format Standards for the Law-based Disclosure of Environmental Information by Enterprises”	Refinement of the content and format of the legal disclosure of corporate environmental information.
2022 [65]	“Work Guidelines for the Investor Relations Management of Listed Companies”	Encouraging listed companies to disclose ESG information.

). Analyzing the internal motivations, the positive disclosure of non-financial performance information is of strategic importance and creates an incentive for voluntary disclosure in the industry [58]. The research data show that a CCDQI score below 0.5 indicates a low level of corporate response to the CCD quality framework, which is mainly characterized by missing information or the disclosure of non-substantive content. This phenomenon reflects that some corporations use climate information reporting mainly as a symbolic tool to gain organizational legitimacy, aiming to build a good image through compliance disclosure, rather than being truly committed to substantive improvements in climate governance.

Geographic Area Differences

The quality of CCD in all regions of China shows a significant trend of improvement, but there are still significant differences between regions. In 2021–2023, Hong Kong continued to maintain the leading level, with the quality of information disclosure scores of 0.4869, 0.6715, and 0.7041, respectively, followed by Beijing, with the scores of 0.4685, 0.4819, and 0.5719 in the same period (Figure 6). Heterogeneity is mainly attributed to two key factors. First, there are significant differences in socio-economic characteristics across regions. Second, differences in the strength of local governments in the formulation and implementation of policy documents related to CCD (

Table 5. Evolution of government policy in Hong Kong, China.

Year	Policy
2013	Voluntary disclosure recommendations for listed corporations, “Environmental, Social and Governance Reporting Guidelines”
2018	“Green Finance Strategy Framework” to strengthen the disclosure of climate-related information by listed corporations
2019	Implement the principle of “no disclosure without justification”
2021	Release the “Guidance on Climate-related Disclosures”, benchmarking against the international standard TCFD, and emphasizing the need for corporations to raise their focus on climate change-related risks
2023	Propose to amend the “Environmental, Social and Governance Reporting Guidelines” to enhance the mandatory nature of CCD and transparency

Source: https://www.hkex.com.hk/?sc_lang=en (accessed on 10 February 2025).

). Hong Kong, as an international financial center, has a highly internationalized capital market, a well-established third-party professional consulting service system, and a strong market-driven mechanism due to the high level of stakeholder interest in corporate climate change performance, which provides institutional safeguards and a market foundation for maintaining a highly quality level of disclosure.

Industry Chain Differences

Chinese upstream and midstream corporations in the NG industry exhibit higher levels of CCD compared to downstream corporations, with downstream corporations growing the most (

Table 6. The CCDQI scores of sample corporations in the industrial chain, 2021–2023.

	Upstream	Midstream	Downstream
2021	0.3092	0.3069	0.2864
2022	0.3575	0.3476	0.3333
2023	0.4023	0.4245	0.4143

). The primary reason for the higher CCDQI score among upstream and midstream corporations is the strong monopoly in the industry. This monopoly is particularly evident in the exploration and exploitation, storage, and transportation sectors, which are predominantly SOEs. The bureaucratic structure of SOEs enhances their level of environmental information disclosure. But at the same time, they are subject to more constraints from national regulations and political burdens. When external pressures intensify, the impact on SOEs becomes more pronounced. The State-Owned Assets Supervision and Administration Commission of the State Council has issued the “Work Plan for Improving the Quality of Listed Companies Controlled by Central Enterprises” [66]. Due to their unique attributes and advantages, SOEs are driven by the combined forces of state regulation and are responsible for promoting CCD development. There are more private corporations in the lower reaches, and the proportion of Hong Kong corporations is larger than that of the upper and middle reaches. The government issues the “Opinions of the CPC Central Committee and the State Council on Promoting the Development and Growth of the Private Economy” [67]. Private corporations are aware of the development of social trends and importance of stakeholder preferences. This structural difference reflects the differentiated development paths and driving mechanisms in the CCD practices of corporations under different ownership.

Completeness Dimension Differences

The governance dimension had the highest score for the quality of CCD, while the risk management dimension had a significantly lower score than the other dimensions (Figure 7). This difference can be explained by both policy-driven and real-world conditions. First, disclosure at the governance level is usually driven by explicit policy regulations and regulatory requirements. The Chinese government attaches great importance to the issue of climate change, proposing the strategic goal of “carbon peaking and carbon neutrality” and introducing the “1 + N” policy system, which provides a clear direction for enterprises to disclose information. Secondly, the governance level is usually directly promoted by the top management of the corporation. The board of directors and management are more inclined to respond to the national policy guidance and external expectations in their strategic priorities. At the same time, disclosures at the governance level usually involve more structured content such as governance structure and management responsibilities. Corporations are able to carry out disclosure work relatively easily based on the existing governance framework. In contrast, low compliance at the risk management level is mainly due to practical considerations. First, risk management is highly complex and requires a high level of technical skill, such as scenario analysis methods and quantitative risk modeling for climate change risk analysis. These tasks require a high degree of data quantification, and there is a lack of corresponding development tools and unified methodological support. On the other hand, the calculation standards and regulatory system for risk management have not yet been perfected, which makes it difficult for enterprises to grasp the degree of disclosure in actual operation. In addition, risk management requires substantial financial and resource investments, yet the majority of corporations have not prioritized it in their CCD practices. This has resulted in a limited number of practical case studies, leaving organizations with insufficient mature and referenceable experiences in this domain.

4.2. Climate Change Disclosure—Environmental Topics

As in

Table 7. CCD under the completeness dimension—environmental issues.

Framework	Indicators	Disclosure
Governance	EP	(1) Description of environment-related policies formulated by the corporation in accordance with the state * (2) Description of environment-related policies formulated by the corporation on its own *
	BS	(1) Description of the work of the board of directors/governance in terms of responsibility checking, periodic review, and progress follow-up of work related to environmental management, climate governance, ESG, etc. * (2) The Board of Directors/Governance takes a position on significant resolutions related to environmental management, climate governance, and ESG
	GS	(1) Description of roles and responsibilities of each management level in environmental management, climate governance, ESG management * (2) Description of governance framework: ESG Committee, Environmental Work Leadership Group, etc. *
	MR	(1) Responding to Stakeholders’ Demands on Environmental Protection, Climate Governance, etc. * (2) Identify significant issues in the environmental field (3) Participation in local infrastructure environmental governance (4) Participate in developing industry standards related to the environment
	EMR	(1) Employee compensation linked to environmental indicators (2) Employee training and promotion of green thinking, climate change, ESG, etc.
Strategy	CR	(1) Identification of the organization’s short-, medium-, and long-term climate risks (physical and transformation risks) and response measures (2) Opportunities arising from climate risks faced by the organization (3) Impacts of climate risk on the organization—financial impacts
	SD	(1) Organizational low-carbon transition business development layout/path planning (2) Climate strategy/ESG strategy description (3) Environmental protection strategy cooperation—universities, corporations, and governments *
	SCM	(1) Assessment of suppliers’ environmental performance and environmental management qualification system (2) Preparation of environmental standards for each link in the supply chain of corporations
	CS	(1) Strategic resilience of the organization under different climate scenarios
Risk Management	CRI	(1) Carbon footprint study (2) ESG climate risk
	CRA	(1) Climate risk modeling
	CRM	-
	CRC	-
	CRN	(1) Climate change risk management team (2) Description of environmental/climate risk management system—ISO
	CRP	(1) Improvement and evaluation of environmental/climate risk management systems
Metrics and Targets	ERA	(1) Specific description of actions/measures taken to reduce emissions * (2) Collaboration with universities, corporations, government, etc.
	RC	(1) Indicators related to energy consumption, water, land use, and waste management * (2) Description of specific practice cases * (3) Zero-carbon parks
	S1	(1) Quantification of Scope 1 GHG emissions * (2) Accounting standards used
	S2	(1) Quantification of Scope 2 GHG emissions * (2) Accounting standards used
	S3	(1) Quantitative Scope 3 GHG emissions (2) Accounting standards used
	GHG	(1) Quantification of the target value of each GHG emission/reduction set by corporations
	LCI	(1) Amount of the organization’s investment in environmental protection and governance (2) Description of the invested environmental protection projects and their effectiveness
	CT	(1) Descriptive representation of the organization’s participation in the indicator (2) Quantifying the amount of transactions (3) Green power consumption trading, carbon sinks, carbon inclusion, and blue bonds
	EI	(1) Actions and measures taken * (2) Description of relevant cases * (3) Impact assessment—environmental impact assessment of projects, assessment of potential impacts on biodiversity (4) Publication of Biodiversity Conservation Report
	LCP	(1) Patents and technologies (2) Cooperation with universities and other organizations
	ER	(1) Expression of importance

Note: * indicates greater disclosure; - indicates no disclosure.

, at the governance level, corporations tend to disclose information related to EP, BS, GS, and MR. Regarding BS information, corporations disclose a more superficial description of the board’s work in the environmental field, with less mention of specific practices and their effectiveness, which are highly homogenized. In MR information, corporations emphasize communication with stakeholders, actively identify major issues in the environmental field using tools such as the materiality matrix, and pay more attention to this aspect. Corporations’ data at the governance level are less quantitative, with little data on top management’s responsibilities related to environmental protection and climate governance. EMR rarely mentions employee compensation linked to environmental indicators. There is a lack of data on the environmental performance standards of corporations.

At the strategic level, corporations are willing to identify climate change risks and opportunities. They rarely specify short-, medium-, and long-term risks or the integration of climate opportunities into the corporation’s strategic development. To select the climate scenarios, three representative concentration pathways (RCP2.6, RCP4.5, and RCP8.5) issued by the Intergovernmental Panel on Climate Change (IPCC) are used. No further actions or quantification are performed.

At the risk management levels, the Chinese NG industry has a low level of quantification and sophistication, with a limited number of corporations disclosing information, involved in only CRI, CRA, CRN, and CRP. None of the corporations describe processes for identifying and assessing climate change risks. Few corporations use climate risk modeling (such as the Acute Climate Risk Optimization Demand Forecasting Model), ESG risk classification identification, and carbon footprint studies as complementary tools. Few corporations have separate Climate Risk Management teams. These work solely on climate risk management and conduct International Organization for Standardization (ISO) assessments and certifications.

At the metric and target levels, S1 and S2, the RC, ERA, and EI are relatively well disclosed. However, communication between corporate GHG settings and CT aspects is poor. In S3, only three corporations are included, China Gas, Sinopec Kantons, and Xinao, mainly for employee travel commuting and paper waste disposal. Specifically analyzing the GHG aspect, the number of corporations involved in net-zero or carbon-neutral targets accounted for less than 20%. The Chinese NG industry lacks motivation and self-confidence in its decarbonization targets. However, there is a lack of scientific targets based on SBTi. Eleven corporations have different timeframes for net-zero targets, focusing on 2050 and 2060, which aligns with China’s “2060 Carbon Neutral” policy objective. While six corporations have set a 2030 target, which is in line with China’s “2030 Carbon Peak” policy objective, two corporations did not set specific time targets. Corporations that have set net-zero or carbon-neutral targets lack specific target information, including the scope and timeframe of their GHG emissions. These corporations do not consider the importance of supply chain decarbonization. The effectiveness of the disclosure of decarbonization targets has yet to be confirmed.

In general, by quantifying and disclosing governance effectiveness; identifying climate risks and opportunities; and integrating these into corporate strategic development outcomes, carbon reduction targets, and Scope 3 emissions, corporations can significantly address the shortfall in CCD expectations. Furthermore, the disclosed information should provide feedback to corporations, for instance, by illustrating the financial impacts of climate change risks and opportunities, and by demonstrating how carbon reduction statistics influence the decarbonization of supply chains. Currently, corporations tend to demonstrate CCD practices and achievements, while ignoring the role of providing a basis for decision making. As a result, the disclosed information fails to provide detailed data and indicators, thereby hindering the comparison and evaluation of indicators.

CCD by Chinese corporations is still dominated by qualitative descriptions. The Chinese language itself is characterized by semantic richness and words with multiple meanings. It is often difficult to accurately capture and quantify the substantive information in the qualitative textual disclosures of Chinese corporations if a single international mainstream sustainability disclosure standard is mechanically applied. The evaluation model constructed in this study effectively solves this problem by introducing text mining technology and semantic analysis algorithms, integrating multiple international mainstream sustainability disclosure standards. The empirical analysis shows that the level of disclosure quality in the Chinese NG industry is mainly influenced by the government’s policy orientation. Other factors that significantly affect disclosure quality in the international sample have limited explanatory power in this research sample. Therefore, improving the quality of CCD in different countries and industries requires a targeted analysis, including institutional environment differences, industry characteristic differences, and cultural and linguistic differences.

The limitations of this study provide opportunities for further research. A significant limitation of this study is its small sample size. The lack of data limits the scope of the content analysis, resulting in the incomplete coverage of the analyzed dimensions. Consequently, these dimensions may fail to fully reflect all the information disclosed by the sample corporations. The inclusion of scoring aspects in the analyzed dimensions inevitably introduces a certain degree of subjectivity.

5. Suggested Improvements

The development of the CCDQI indicates an increase in both the quantity and quality of information disclosed by corporations, but the coverage of corporate disclosure topics is insufficient. Completeness of disclosure is significantly better at the governance level than at the risk management level, a phenomenon that highlights the shortcomings of corporations in building capacity for climate change risk management. Further content analysis reveals a clear lack of corporate attention to key information on potential climate change and a low motivation to decarbonize. Most corporations still demonstrate low compliance with CCD policies and standards, with unclear data sources and calculation standards, highlighting reliability issues. Corporations are concerned about their social image and reputation, and they are therefore more willing to showcase their practices and results to the outside world. While qualitative descriptions are relatively complete, quantitative data are insufficient, greatly reducing comprehensibility and making it difficult to assess comparability. The causes of the above problems are analyzed, and suggestions for improvement are provided as follows:

(1) The distinction between CSR and ESG concepts is unclear. The reports disclosed by corporations integrate the contents of both CSR and ESG theories. The disclosure focus is fragmented and interspersed, leading to indicators being disclosed incompletely. When indicators within the disclosure focus are not clearly distinguishable, common language is used to describe them. This results in neither type of reports that can achieve the desired results. Therefore, corporations should strictly differentiate the definitions, contents, objectives, focuses, and preparation and disclosure requirements of different types of reports. It is also crucial to enhance the understanding and training of CSR and ESG concepts within each corporation to optimize the report preparation process and improve the quality level.

(2) Lack of unified disclosure standards and corresponding regulatory measures. When disclosing climate change information, corporations face multiple optional disclosure standards. Different corporations choose different combinations of standards, leading to varying measurement methods and disclosure characteristics. CCD follows a voluntary principle, and there are no mandatory requirements for corporations to adhere to the standards. When reports cannot be compared, doubts arise regarding the reliability of the information. Therefore, the government should develop nationally harmonized and operational CCD guidelines. This should clarify the content, format, and quality requirements of the disclosure to avoid the duplication of disclosure content. Quantitative indicators should provide detailed calculation methods and units of measurement for indicators. Second, government departments and regulators should establish a multilevel and comprehensive supervision system to check and assess the disclosure situation regularly. The government should link the results of the evaluations of the quality of CCD to corporate credit ratings. A corresponding penalty mechanism to increase the cost of non-compliance should be established for corporations that fail to disclose as required or disclose inaccurate information. The government and regulators should also promote a third-party verification system and formulate a verification standard for CCD applicable to China. While strictly regulating the verification process, a qualification system should be implemented for third-party verification organizations. Government departments should provide tax incentives and financial subsidies to corporations that actively disclose climate change information. CCD can also be used as an important evaluation indicator for corporations to obtain financing such as green credit and green bonds.

(3) Inadequate capacity to disclose potential climate change information indicators. The quantification of potential climate change indicators is demanding. Corporations are reluctant to carry out substantive work based on the cost of disclosure and the delayed effect of reporting feedback. Corporations lack the technology to measure potential CCD indicators and are unaware of the need for disclosure. Therefore, for potential climate change information, the government’s implementation process should include a buffer period. Additionally, the government should establish optional disclosure items and implement transitional measures, allowing corporations to use methods that match their capabilities and resources. Second, corporations should have the awareness of improving information disclosure independently. Corporations should keep abreast of the relevant policies and regulations issued by the state and developments at home and abroad. They should stablish their own CCD management system to build a foundation for subsequent in-depth work. At the same time, they should actively focus on the reduction in emissions in the industrial chain, build corporation communication bridges, and form data interoperability with upstream and downstream corporations.

6. Conclusions

Based on the theoretical framework of CCD and incorporating mainstream international sustainability disclosure standards, this study develops a robust and highly applicable quality assessment system for CCD. In this study, a quantitative assessment of corporate CCD was conducted using a content analysis and quality evaluation index method, leading to the development of the CCDQI. The results indicate that the quantity and quality of climate change information disclosure by enterprises have shown an upward trend. But the differences between individual corporations and regions are more significant, reflecting the unevenness of the level of information disclosure. The external policy environment has an important impact on the quality of corporate CCD. Policies with clear guidance requirements can more effectively exert institutional pressure on corporations. In addition, the level of regional economic development is positively related to the quality of CCD. From the perspective of the corporations themselves, the disclosure of climate change information related to risk management is a key link in narrowing the reporting gap.

According to the evaluation results, the reasons for the poor quality of CCD in the Chinese NG industry are analyzed, so as to put forward relevant recommendations for effectiveness and give a set of simplified and complete information disclosure processes. To support national decarbonization goals, high ratings of climate disclosure help direct financial flows to low-carbon projects. Customer-based environmental disclosures incentivize suppliers to reduce emissions and promote supply chain-wide carbon neutrality, thereby better coordinating the relationship between the economy, society, and the environment. Future studies could extend this framework to investigate the evolution of CCD quality in other sectors and countries and focus on unlisted companies. To improve the CCDQI, comparable measurement methods at the national and industrial scales can also be explored. The identification of climate change risks and opportunities should also be considered to strengthen climate action and CCD quality in a cross-cutting manner. Corporations’ responses to climate change are dynamic and it may take years to detect these changes. These can be built upon in the future to continue to expand the time span of the study.