Corporate Sustainable Growth, Carbon Performance, and Voluntary Carbon Information Disclosure: New Panel Data Evidence for Chinese Listed Companies

Abstract

At present, to comply with carbon reduction commitments, China has only required energy-intensive enterprises to report their carbon information to regulators, aside from mandatory public disclosures. Although some enterprises have been disclosing their own carbon information voluntarily by means of corporate reports in order to shape their green image, their carbon information disclosures (CID) still need to be improved. This study attempts to systematically investigate links between corporate growth, carbon emission (CEP) or reduction performance (CRP) in two forms (intensity and amount), and CID in industries with different carbon intensities on the basis of stakeholder theory as well as legitimacy theory. This study took Chinese companies listed on the main board market from 2009 to 2021 as samples. The Arellano–Bover/Blundell–Bond dynamic panel data model was used for regression analysis. Results showed that sustainable growth enhanced the promotion effect of CEP (both in terms of intensity and amount) on CID in carbon-intensive industries, while sustainable growth enhanced the inhibition effect of CEP measured in terms of amount, rather than intensity, on CID in low-carbon industries. This revealed that CEP, not CRP, had a significant influence on CID, and uncovered the influence mechanism between carbon performance and CID from the perspective of sustainable enterprise growth. The carbon information disclosure of high-carbon industries is closely and positively related to carbon performance, indicating that the interaction between high-carbon industries and capital markets will be more affected by the mitigation of carbon information asymmetry. Further, circulating A-shares are moderators for better CID in both carbon-intensive industries and low-carbon industries, which fits the expectation of stakeholder theory as well as legitimacy theory. Additionally, measurement habits or preference for carbon emissions performance (in the form of amount or intensity) in different industries should be brought to the forefront to enhance investors’ confidence in CID. This study has certain guiding value for the formulation of CID standards and contributes to the process of mandatory CID.

1. Introduction

As climate change has become more and more serious, the goal of controlling global temperatures within 2 °C (compared with the pre-industrial age) promotes close and substantial cooperation within the international community to curb greenhouse gas emissions. Studies have shown that compensation policies and carbon market mechanisms contribute to carbon emission reduction [1,2]. For example, the number of participating companies has doubled, and the emission load of big emitters was adjusted down to half of the initial criterion determined in Beijing in 2016 [3]. The status quo of enterprise carbon emissions as well as carbon emission reduction efforts will be more and more important for the carbon predominance of enterprises in the era of carbon trading. Carbon emission reduction efforts and carbon performance are inseparable from internal development, such as enterprises’ sustainable growth, and so on. Enterprises’ sustainable growth may interact with the external environment when it comes to potential financing and information asymmetry from the perspective of stakeholders.

For carbon predominance, enterprises, the main body of greenhouse gas emissions, must attach importance to carbon information disclosure (presentation behavior) in order to maintain a green image. Carbon information disclosure is an important means for demonstrating the integrity, detail, and credibility of key carbon emissions data, including that of scope 1 and 2 emissions and carbon intensity, as well as data related to emission mitigation measures, to the government, the public, and investors [4,5]. Since investors and creditors’ demand for high-quality carbon information is increasing in green finance, China’s mainland bourses and the China Securities Regulatory Commission (CSRC) encourage enterprises to disclose carbon emission data. The content and format of Guideline No. 2—Corporate Information Disclosure of Publicly Issued Securities has been revised to ensure the indicative value of disclosure [6]. The Hong Kong Stock Exchange also increased the requirements for environmental, social, and governance (ESG) reports, issued via several pieces of guidance on climate disclosure reporting since 2016 [7]. Eighty-two listed companies in Hong Kong provided 224 carbon footprint reports between 2011–2017 [8].

However, enterprises tend to actively disclose favorable information related to carbon emission reduction in mainland China [9,10,11,12,13,14]. Since CID is not mandatory, and rules for CID are not well-established, relevant supervision and initiatives have played a limited role in Chinese companies [15]. After the government proposed its goals for carbon peak and carbon neutrality, the coming transition period before mandatory CID is implemented will be key for investigating the carbon behavior of Chinese companies from the perspective of the companies. In particular, stakeholders seem more concerned about the CIDs of enterprises from high-carbon industries than those of low-carbon industries. However, carbon predominance still exists in low-carbon industries in terms of intra-industry or inter-industry production. Considering stakeholders’ concerns about corporate sustainable growth from the perspective of stakeholder theory, we took corporate sustainable growth as the internal driver for carbon information disclosure.

Therefore, these research questions are proposed: 1. Does carbon emission performance have different effects on carbon information disclosure in carbon-intensive industries compared to low-carbon industries? 2. Do the two aspects of carbon emission performance, measured via carbon emission and carbon emission reduction, affect carbon information disclosure in different directions? 3. Does corporate sustainable growth strengthen the relationship between carbon emission performance and carbon information disclosure?

This study collected carbon information from 2009–2021 social responsibility reports and ESG reports from A-share firms listed on the main board market. A set of corporate CID items was identified in the study. As the value and urgency of CIDs are obviously different in carbon-intensive industries versus low-carbon industries, we investigated whether the effects of two kinds of carbon performance, namely carbon emission performance and carbon emission reduction performance, on CID are consistent. As an important aspect of enterprise development, the sustainable growth of enterprises may drive carbon performance. This study further explores whether sustainable development has a moderating effect on the relationship between carbon emission performance or carbon emission reduction performance and CID. The Arellano–Bover/Blundell–Bond (A–B/B–B) estimation of dynamic panel data statistical method was adopted.

Investigating the voluntary CID of Chinese listed companies broadens the influencing factors (corporate sustainable growth) of the behavior of enterprises from industries with different carbon intensity characteristics and has not only enriched stakeholder theory and legitimacy theory, but has also been helpful in promoting the formulation of mandatory carbon information disclosure for the near future. In practice, carbon emission performance other than carbon emission reduction performance, measured in terms of both intensity and amount, of carbon-intensive industries should be emphasized continuously by government oversight as well as enterprises. While carbon emission performance measured via amount of carbon emissions has a significantly negative effect on CID in low-carbon industries, carbon information disclosure in high-carbon industries is closely and positively related to carbon performance, indicating that the interaction between high-carbon industries and capital markets will be more affected by the mitigation of carbon information asymmetry. Therefore, more refined and instructive measures of carbon performance should be proposed for industries with different carbon intensities, considering the opposite influences of carbon performance on CID in different industries. Further, tradable shares will help industries to promote their CIDs, which can be explained by legitimacy theory.

2. Literature Review and Hypothesis Development

2.1. Literature Review

Environmental performance, CSR measures (reputation ranking) and environmental disclosures, and even early investment in environmental responsibility positively affect stock prices and make enterprises appealing to investors and analysts [16,17,18,19,20]. ESG performance improves corporate credit ratings and is related negatively to risk-taking [21,22]. Further, environmental information disclosure (EID) has reduced corporate risk and promoted firm value through investor perceptions [23,24,25,26,27,28,29,30,31]. However, benefits from ESG disclosure seem to be more pronounced in less-developed districts [32,33]. This might be one factor that damages enterprises’ enthusiasm for ESG disclosure.

From the perspective of financial performance, the positive effect of EGS, corporate environmental responsibility, or corporate sustainability performance and EID on firms’ financial performance is also constantly significant [19,34,35,36,37,38,39,40]. Various and different connections between CSR and financial indicators exist in various samples [41,42]. Further, EID and mandatory CSR disclosure are positively correlated with enterprise innovation [43,44]. Both environmental performance and EID are related to corporate stock market performance, financial performance, and firm value, indicating that significant correlations between environmental performance and EID exist and supporting voluntary disclosure theory [21,28,45,46,47,48].

Especially as the global climate change situation has become aggravated, carbon performance as a component of environmental performance has become increasingly important. Similarly, carbon performance has been found to be correlated with firm value and financial performance in some high-carbon industries besides agriculture and the new energy industry [2,14,49,50,51,52,53,54,55,56,57]. However, Zhang also found that positive relationships between carbon and financial performance exist only when carbon performance is beyond a certain threshold [58].

The improvement of carbon performance is influenced by enterprise energy choice, abatement investment, and internal carbon management, including internal carbon pricing, corporate climate targets, establishing corporate GHG emission inventories, and so on [22,59,60]. From the perspective of corporate governance, managerial discretion has a positive effect on low-carbon performance [61,62]. These influencing factors, along with control tools and the adoption of a Carbon Management Strategy, also have a certain impact on carbon management and CID (for scope 1 and 2 emissions) [63,64,65,66,67,68].

Similar to carbon performance, CID is also significantly related to financial performance, (financial) risk, and value [50,69,70,71,72,73,74,75,76]. Investors are willing to acknowledge enterprises with better performance in GHG emissions and transparency regardless of lower returns [16]. Consistency in corporate carbon performance and CID also improve enterprise value in manufacturing industries [55]. However, carbon-intensive industries are inclined to avoid voluntary CID [77]. Research on CID also includes normative studies (methodological diversity and integrity of content) to maintain consistency and to reconcile the usefulness of climate change-related data [78,79,80,81].

From correlations between green performance and corresponding information disclosure [82] to the bridges between carbon performance and CID, such as enterprise value, financial performance, internal management (carbon management), investment, and low-carbon awareness among managers, these associations provide a strong basis for investing in improved carbon performance and CID [8,74,83,84,85,86]. However, investigations into the carbon performance and CID of enterprises are still under way because of uncertainty [87,88], which may indicate that industrial characteristics, enterprise characteristics such as growth, stock liquidity, and ownership concentration, as well as specific measurements for CID, also need to be investigated.

The connections between sustainable growth and carbon performance [89], environmental performance in step with sustainable growth [90,91,92], harmony between carbon performance and environmental performance, environmental performance, and CID, and thus the connections between all of these things and profitability, risk-taking, and innovation and sustainable growth [93,94,95], as well as CID, corporate value improved by carbon performance or CID, environmental policy uncertainty, and sustainable growth (or carbon performance or financial aspects) and disclosure [96], provide some new possibilities for understanding the relationship between sustainable growth and CID. Sustainable growth is of great value in reducing the ambiguity of the relationship between CEP (or CRP) and CID.

To sum up, studies on carbon performance and CID are generally one-sided, and there are not abundant studies investigating enterprises’ sustainable growth. Research objectives usually focus on excellent enterprises such as the Standard & Poor’s 500 and do not investigate enterprises with poor financial performance or those from low-carbon industries. Additionally, carbon performance is often single-faceted or carbon performance measurements are not uniform, which can be seen by referring to the explanatory variables in this paper. CID measurement methods should also be clarified [87]. From the perspective of voluntary CID, this study calculated the number of CID items and the CID score on the basis of CID items collected from corporate reports including ESG reports. Additionally, the relationships between performance, sustainable growth, and CID in carbon-intensive and low-carbon industries was systematically studied.

2.2. Hypothesis Development

Signaling theory and stakeholder theory predict that stakeholders concerned about climate change favor companies with more or valuable CIDs. Voluntary disclosure or CSR disclosure is also compatible with legitimacy theories [97,98]. Although heavy-polluting companies tend to garner negative reactions from investors, they are more inclined to disclose environmental information and their environmental information disclosure reduces stock price crash risk [99,100]. Enterprises in carbon-intensive industries exposed to regulatory climate risks have received more attention from regulators and should therefore disclose more environmental information [101,102,103,104,105,106,107,108]. Belkhir et al. found that changes in CID and carbon performance were going in the same direction [109]. Carbon performance positively affects carbon disclosure in companies in the industrial sector, consumer staples sector, consumer discretionary sector, and so on, according to CDP data [110].

Specifically, stakeholder theory explained why growth capacity strengthens the positive relationship between CEP and CID in carbon-intensive industries and legitimacy theory explained the positive relationship between CEP and CID observed via the ratio of tradable stocks from the perspective of the capital market in carbon-intensive industries. In terms of legitimacy theory, enterprises are subject to fewer disclosure rules. From the perspective of stakeholder theory, namely stakeholder’s expectations and preference for growth as well as reduction of information asymmetry, sustainable growth may have a positive impact on carbon disclosure by improving carbon performance. See Figure 1.

Further, enterprises in high-carbon industries have more information to disclose in terms of carbon emission measurements, carbon management methods, and carbon emission reduction actions. Since enterprises classified as reporting enterprises for carbon emissions or big emitters in mainland China must report their emissions data, carbon quota, and other carbon information on a specific platform [111,112], it is easier for them to prepare carbon information reports that may not require a lot of extra investment or cost. To explain the positive impact on environmental performance and environmental information disclosure in addition to CEP, CRP, and CID for specific industries, hypothesis 1 is proposed.

H1. 

Corporate carbon emission performance has a positive impact on CID in carbon-intensive industries.

Although carbon-intensive enterprises emit more carbon dioxide, they also have greater carbon reduction potential. In addition, CRP is a representative of an enterprise’s investment and innovation capacity, as well as its low-carbon business transfer plan. Disclosure and carbon-reduction efforts, such as the use of renewable energy, as well as carbon management, which is associated with stakeholder pressure, benefit CEP and CRP [59,87,113,114,115,116]. In summary, the theoretical expectation of stakeholders and the two intermediate variables of environmental information disclosure and carbon emission performance reflect the possible relationship between CRP and CID. Therefore, hypothesis 2 is proposed.

H2. 

Corporate carbon emission reduction performance has a positive impact on CID in carbon-intensive industries.

Companies with varied carbon intensities produced CIDS of significantly different quality [12,117]. Symbolic disclosure is more common in less intensive sectors [118] and can be explained by low pressure from the media and regulations, from an equilibrium perspective. Some firms remain silent and do not disclose environmental performance when they face diverse audiences [119], and studies show that carbon emission is positively correlated with CID [120]. In terms of legitimacy theory, enterprises are subject to fewer disclosure rules in low carbon industries. Therefore, hypotheses 3 and 4 are proposed.

H3. 

Corporate carbon emission performance has a negative influence on CID in industries with low carbon intensity.

H4. 

Corporate carbon emission reduction performance has a negative influence on CID in industries with low carbon intensity.

Higher sustainable ventures in growth will perform better in industries inclined to higher returns [121]. Models for growth potential and emissions were explored [90]. Green growth could improve corporate sustainability performance [91], and profitability and growth relationships are mediated via investment in China’s manufacturing firms [95]. An association between R&D volatility or high innovation efficiency and firm growth exists [93,94], and EPU, economic policy uncertainty, is bound up with sustainable financial growth and carbon emission intensity as well as corporate risk-taking [122,123,124]. From the perspective of stakeholder theory, namely stakeholder’s expectations and preference for growth, as well as for reductions in information asymmetry, sustainable growth may have a positive impact on carbon disclosure by promoting innovation, reducing risk-taking, or improving carbon performance. Therefore, combined with H1–H4, the following hypotheses are proposed.

H5. 

The sustainable growth of enterprises enhances the promotion effect of carbon emission performance on corporate CID in carbon-intensive industries.

H6. 

The sustainable growth of enterprises enhances the promotion effect of carbon emission reduction performance on corporate CID in carbon-intensive industries.

Growth rates in China’s manufacturing industries negatively influence improvements in sustainable performance in production [89], and most of the manufacturing sector is low carbon intensity. That is, growth rates in low-carbon industries may negatively affect carbon performance. Moreover, a firm’s growth rate is an important condition for disclosure [96]. Based on hypotheses 3 and 4, hypotheses 7 and 8 are proposed.

H7. 

The sustainable growth of enterprises strengthens the inhibition effect of carbon emission performance on CID in low-carbon industries.

H8. 

The sustainable development of enterprises strengthens the inhibition effect of carbon emission reduction performance on CID in low-carbon industries.

3. Research Design

3.1. Variables

The dependent variable is CID. CID refers to public disclosure of monetary, quantitative, or qualitative descriptions related to carbon emission, which is helpful for stakeholders in recognizing carbon performance. The quality of carbon information disclosure can be measured by the number of CID items. This study collects carbon information items from corporate social responsibility reports, sustainability reports, and environmental, social, and governance reports. In total, there were 62 items including carbon emissions, carbon emission reductions, identification of climate-related risks, emission reduction targets, carbon asset management, low-carbon transfer plans, low-carbon issues in the value chain, green and low-carbon office measures, and so on. Different types of carbon information items were distinguished and higher scores were assigned to the carbon information disclosure items that contain monetary or quantitative information. Those are the Scores of CID. More specifically, three points, two points, and one point were assigned for each monetary, quantitative, and qualitative carbon information disclosed, respectively. Number of CID items and Scores of CID (for the robustness test) were used as indicators to reflect the quality of CIDs in this study.

The explanatory variable of this study is carbon performance. Carbon performance refers to carbon emissions (reduction) or carbon intensity when applying low-carbon technologies, energy restructuring, relevant projects, carbon trading, and so on [125]. Carbon intensity is the ratio of carbon emissions (or emission reduction) and enterprise output value (or income). See Figure 2. Carbon emissions from Scope 1 and Scope 2, rather than Scope 3, are used in order to maintain consistency with the scope of carbon emissions disclosed by most enterprises and widely used disclosure guidelines regarding the difficulty of quantifying carbon emissions and reduction in Scope 3 emissions [109,126]. In detail, Scope 1 refers to the direct emissions generated by directly controlled or owned emission sources, including combustion, etc., and Scope 2 refers to indirect emissions, including the indirect emissions generated by power purchased for the enterprise’s own use, such as steam, heating, cool air, etc. Scope 3 refers to indirect emissions other than those covered by Scope 2, including all emissions that may be produced upstream or downstream. Rather than the rate of change in carbon emissions, CRP refers to measurements based on carbon emission reduction from technology improvements, clean energy generation, etc.

Good governance, including managerial ownership, is positively related to pollution disclosure, supporting legitimacy theory [127,128]. Additionally, cross listing is preferable to other types of listings in performance and fine environmental disclosure to mitigate the liability of foreignness [28,47,129]. This paper takes equity concentration to represent an aspect of corporate governance, then adds solvency, capital utilization as well as stock liquidity considering the financing method, financing demand, and information requirements [28,65,110], as shown in

Table 1. Variable descriptions.

Variables	Proxy Variables	Symbolics	Explanations
CID	Number of CID items	CarboDisc	Nominal variables
Carbon Performance (CPerf)	Carbon Emission Intensity Performance	CPerf_E	Numerical variable
	Carbon Reduction Intensity Performance	CPerf_R
Sustainable development	Sustainable growth rate	SusDeve
Size	Asset	Asset
	Liabilities	Liab
Profitability (Profit)	Operating income	Sales
	Return on total assets	ROA
Solvency	Quick ratio	QuiRa
Capital utilization	Fixed asset ratio	FiRa
	Total asset turnover ratio	ToTuRa
Equity concentration (EquityConcen)	Shareholding ratio (SR) of the first largest shareholder	ECSR1
	SR of top 5 shareholders	ECSR5
	SR of top 10 shareholders	ECSR10
	Number of shareholders	EC11
Stock liquidity (StockLiqu)	Shares that can be traded freely in RMB in the A-share market in Mainland China (Proportion of tradable A-shares)	TrdaSP
	Shares listed in Hong Kong Stock Exchange (Proportion of tradable H-shares)	TrdhSP

.

3.2. Empirical Models

Considering that there may be some continuity between corporate carbon information disclosure and previous disclosures, the study set regression models with a lag period on explained variables of four. In order to solve the endogeneity problem, the Arellano–Bover/Blundell–Bond dynamic panel data model was used for regression analysis, and lags on the endogenous variables and the predetermined variables were used as instrumental variables. Considering data availability and matrix problems, the first-order, second-order, third-order, and fourth-order lag terms on the predetermined variables and endogenous variables were used as instrumental variables. If there are data matrix problems or high autocorrelation problems and collinearity problems, the number of lag terms on the dependent variables and the number of instrumental variables should be reduced accordingly.

The command line “xtset industry and year” was set in the statistical software, STATA, before the application of the Arellano–Bover/Blundell–Bond dynamic panel data model. The fixed effects of time and industry were automatically considered. As for the control of individual fixed effects, effective instrumental variables were included, namely the lag on the explanatory variables and the differential lag term. The Sargan test of overidentifying restrictions was used to verify the effectiveness of the instrumental variables to control for individual fixed effect.

Model 1:

$$\begin{array}{cc}\hfill \mathit{CarboDis}{c}_{i,t}=& \alpha +{\alpha }_1\mathit{CarboDis}{c}_{i,t-1}+{\alpha }_2\mathit{CarboDis}{c}_{i,t-2}\hfill \\ \hfill & +{\alpha }_3\mathit{CarboDis}{c}_{i,t-3}+{\alpha }_4\mathit{CarboDis}{c}_{i,t-4}+{\alpha }_5\mathit{CPer}{f}_{i,t}\hfill \\ \hfill & +{\alpha }_6\mathit{Siz}{e}_{i,t}+{\alpha }_7\mathit{Profi}{t}_{i,t}+{\alpha }_8\mathit{QuiR}{a}_{i,t}+{\alpha }_9\mathit{FiR}{a}_{i,t}\hfill \\ \hfill & +{\alpha }_{10}\mathit{ToTuR}{a}_{i,t}+{\alpha }_{11}\mathit{EquityConce}{n}_{i,t}+{\alpha }_{12}\mathit{StockLiqu}{u}_{i,t}\hfill \\ \hfill & +\epsilon \hfill \end{array}$$

Model 2:

$$\begin{array}{cc}\hfill \mathit{CarboDis}{c}_{i,t}=& \alpha +{\alpha }_1\mathit{CarboDis}{c}_{i,t-1}+{\alpha }_2\mathit{CarboDis}{c}_{i,t-2}\hfill \\ \hfill & +{\alpha }_3\mathit{CarboDis}{c}_{i,t-3}\hfill \\ \hfill & +{\alpha }_4\mathit{CarboDis}{c}_{i,t-4}\hfill \\ \hfill & +{\alpha }_5{\mathit{CPer}{f}_E}_{i,t}+{\alpha }_6\mathit{SusDev}{e}_{i,t}+{\alpha }_7\mathit{CPer}{f}_{i,t}\ast \mathit{SusDev}{e}_{i,t}\hfill \\ \hfill & +{\alpha }_8\mathit{Siz}{e}_{i,t}+{\alpha }_9\mathit{Profi}{t}_{i,t}+{\alpha }_{10}\mathit{QuiR}{a}_{i,t}+{\alpha }_{11}\mathit{FiR}{a}_{i,t}\hfill \\ \hfill & +{\alpha }_{12}\mathit{ToTuR}{a}_{i,t}+{\alpha }_{13}\mathit{EquityConce}{n}_{i,t}+{\alpha }_{14}\mathit{StockLiq}{u}_{i,t}\hfill \\ \hfill & +\epsilon \hfill \end{array}$$

3.3. Sample Selection and Data Sources

Considering the specific platform access limits (only the government and companies themselves) on the corporate carbon data of China’s big emitters and the lack of disclosure of China’s listed companies by CDP [111,112,130,131], at present, CSR reports and ESG reports are still the main media and authoritative bases by which the public and investors can acquaint themselves with the carbon performance of enterprises in China. We selected enterprises listed on the main board market in China from 2009–2021 which published social responsibility reports and ESG reports as samples. Using these parameters, enterprise quantity reached 1229. This study also collected more than 3000 social responsibility reports and ESG reports from 37 industries. In this study, reports that disclosed corporate carbon emission or carbon emission reduction were selected as samples. If reports only contained non-numerical carbon information or carbon data that could not represent the overall carbon performance of an enterprise, they were not included in the sample. For example, if only the benefits of saving coal were disclosed, but corresponding carbon emission reductions were not disclosed, this was taken to indicate that the enterprise did not consider carbon data to be key to disclosure and we did not take them as samples. There were 481 valid observations in total. The valid number of enterprises was 91. The disclosure of CEP and CRP among China’s A-share companies listed on the main board market is still very low.

Among the samples, carbon-intensive enterprises accounted for 37.36%, as shown in

Table 2. Sample distribution by industry.

CI Code	Sector	Frequency	Percent (%)	Cum. (%)
HCI1	Electricity, heat production, and supply industry	14	15.38	15.38
HCI2	Non-metallic products industry	6	6.59	21.98
HCI3	Coal mining and washing	4	4.40	26.37
HCI4	Chemical raw materials and chemical products manufacturing industry	4	4.40	30.77
HCI5	Smelting and pressing of non-ferrous metals	2	2.20	35.16
HCI6	Crude petroleum and natural gas industry	2	2.20	36.26
HCI7	Other high-carbon industries	2	2.20	47.25
LCI1	Manufacturing of computers, communications, and other electronic equipment	9	9.89	57.14
LCI2	Financial industry	9	9.89	62.64
LCI3	Manufacturers of general machinery	5	5.49	68.13
LCI4	Electrical machinery and equipment manufacturing	5	5.49	73.63
LCI5	Special equipment manufacturing	5	5.49	78.02
LCI6	Gas production and supply	4	4.40	82.42
LCI7	Pharmaceutical manufacturing industry	4	4.40	85.71
LCI8	Motor-car industry	3	3.30	89.01
LCI9	Non-ferrous metals mining and dressing	3	3.30	91.21
LCI10	Wine, beverage, and refined tea manufacturing	2	2.20	93.41
LCI11	Water production and supply	2	2.20	94.51
LCI12	Other low-carbon industries	6	6.59	100.00
	Total	91	100.00

HCI and LCI represent carbon-intensive industries and low-carbon industries, respectively.

. Electricity and thermal power production and supply enterprises accounted for the highest proportion (15.38%). The percentage of computer, communications, and other electronics manufacturers in low-carbon industries was 9.89%.

See Figure 3. Observations of CEP disclosures exceeded those of observations of CRP disclosures since 2017. Observations from carbon-intensive industries and low-carbon industries were 202 and 279, respectively.

4. Empirical Results

4.1. Descriptive Statistics

STATA software was adopted to analyze statistics.

Table 3. Descriptive analysis.

Variable	Obs.	Mean	Std. Dev.	Min	Max
CarboDisc	481	8.366	5.668	0	31
Carbon Emission	250	11.474	101.297	0	1147.752
Carbon Emission Reduction	231	6.852	10.588	0	46.049
SusDeve	481	0.077	0.105	−0.674	0.806
Asset (billion yuan)	481	736.348	342.010	1.819	30,253.981
Liab (billion yuan)	481	637.052	3133.085	0.501	27,639.859
Sales (billion yuan)	481	87.465	281.439	0.367	2966.193
ROA	481	0.043	0.049	−0.200	0.281
QuiRa	481	1.024	0.726	0.085	7.958
FiRa	481	0.303	0.211	0.001	0.876
ToTuRa	481	0.596	0.444	0.023	2.561
ECSR1	481	0.422	0.173	0.078	0.990
ECSR5	481	0.662	0.174	0.232	1.005
ECSR10	481	0.700	0.162	0.278	1.012
ECSR11	481	11.436	1.246	0.693	13.860
TrdaSP	481	0.997	2.989	0	66.274
TrdhSP	481	0.195	1.542	0	33.726

reports the key descriptive data of the sample. The mean of CID items was 8.366. Carbon emission and carbon emission reduction were 11.474 and 6.852, respectively.

In terms of industries, petroleum processing, coking, and nuclear fuel processing, a carbon-intensive industry, disclosed 17 items, which is the most. Additionally, other low-carbon industries such as the metal products industry, food manufacturing, and wine, beverages, and refined tea manufacturing also disclosed more than others. The mean of CID items reported by the non-metallic mineral products industry was 4.632, which was the least. By year, the mean of CID items was the largest in 2021, as shown in

Table 4. Panels for sector and year.

	HC1 (N = 95)	HC2 (N = 38)	HC3 (N = 19)	HC4 (N = 22)	LC1 (N = 53)	LC2 (N = 53)	LC3 (N = 21)	LC4 (N = 21)	LC5 (N = 14)	LC6 (N = 24)
Panel A
CarboDisc	6.168	4.632	11.368	4.909	7.528	11.736	9.286	7.857	7.929	6.250
CPerf_E (ton/thousand yuan)	13.674	9.419	3.120	2.382	0.076	0.010	0.088	0.080	0.132	573.485
CPerf_R (ton/thousand yuan)	18.065	0.837	0.264	10.614	0.229	1.173	0.222	0.034	-	6.197
SusDeve	0.068	0.129	0.056	0.066	0.05	0.071	0.054	0.192	0.075	0.063
Asset	1220.712	549.391	3681.01	123.046	351.015	58,190.44	1164.516	1191.888	191.081	145.122
Liab	800.624	180.274	1397.538	35.097	214.262	53,069.42	786.961	814.699	102.845	80.441
Sales	303.801	360.107	1852.668	101.313	238.466	1631.334	521.308	1061.16	120.287	77.554
ROA	0.032	0.077	0.057	0.071	0.032	0.013	0.018	0.072	0.032	0.042
QuiRa	0.673	1.119	1.070	1.837	1.045	1.473	0.939	0.895	1.253	0.680
FiRa	0.556	0.447	0.387	0.333	0.213	0.005	0.132	0.11	0.181	0.390
ToTuRa	0.319	0.547	0.563	0.864	0.846	0.042	0.527	1.075	0.68	0.595
ECSR1	0.517	0.364	0.623	0.456	0.367	0.38	0.466	0.248	0.353	0.434
ECSR5	0.702	0.659	0.874	0.581	0.548	0.715	0.656	0.532	0.649	0.778
ECSR10	0.739	0.689	0.885	0.626	0.595	0.776	0.679	0.585	0.681	0.796
ECSR11	11.143	11.055	11.816	10.884	10.888	12.229	11.808	11.508	10.515	10.910
TrdaSP	0.954	0.834	0.785	1	0.928	0.669	0.796	4.008	0.758	0.971
TrdhSP	0.034	0.148	0.215	0	0.034	0.331	0.12	1.697	0.242	0.029
Panel B
Variable	2021 (N = 70)	2020 (N = 63)	2019 (N = 61)	2018 (N = 52)	2017 (N = 63)	2016 (N = 44)	2015 (N = 37)	2014 (N = 30)	2013 (N = 20)	2012 (N = 18)
CarboDisc	14.086	10.81	9.213	7.904	7.048	6.114	5	5.5	5.6	4.833
CPerf_E (ton/thousand yuan)	1.744	2.249	2.680	3.730	31.509	65.276	0.869	1.338	1.873	2.868
CPerf_R (ton/thousand yuan)	5.638	6.556	6.911	9.168	6.793	10.464	6.128	5.452	6.530	5.255
SusDeve	0.086	0.07	0.074	0.088	0.081	0.055	0.053	0.071	0.091	0.064
Asset	13,248.55	13,628.16	12,707.62	4607.983	7286.227	2314.871	2270.546	2050.349	640.768	655.088
Liab	11,615.18	12,005	11,171.77	3790.145	6310.844	1860.731	1845.725	1657.593	322.311	334.161
Sales	1350.614	1186.695	1283.177	1128.498	932.083	389.369	323.302	334.065	305.654	286.357
ROA	0.054	0.042	0.043	0.045	0.042	0.038	0.03	0.037	0.043	0.038
QuiRa	1.07	1.012	1.087	1.18	1.058	1.079	0.842	1.172	0.762	0.756
FiRa	0.262	0.265	0.269	0.279	0.272	0.333	0.393	0.344	0.391	0.404
ToTuRa	0.679	0.568	0.618	0.643	0.56	0.537	0.489	0.55	0.6	0.557
ECSR1	0.398	0.407	0.408	0.411	0.401	0.419	0.45	0.454	0.477	0.486
ECSR5	0.659	0.687	0.682	0.679	0.661	0.632	0.644	0.639	0.643	0.656
ECSR10	0.697	0.729	0.721	0.721	0.706	0.675	0.677	0.665	0.668	0.692
ECSR11	11.797	11.57	11.492	11.47	11.542	11.43	11.472	10.876	11.219	11.285
TrdaSP	0.836	0.825	0.824	0.857	1.874	0.91	0.921	0.896	0.923	0.937
TrdhSP	0.154	0.174	0.175	0.141	0.69	0.086	0.065	0.05	0.056	0.033

, and CIDs show an increasing trend year on year.

This study constructed 62 items from the CID items on the basis of company standards for greenhouse gas emission agreement found in the Global Reporting Initiative, Guidelines for Writing Environmental Reports, and CDP questionnaire. Observations around five items were dense. See Figure 4. The fluctuation range of the CID items was 5.668 during 2009–2021.

4.2. Univariate Analysis

First, the maximum and minimum normalization method was used to process CarboDisc and CPerf_E (contrary indicator), while CPerf_R, Sales, Asset, Liab and EC 11 were processed via logarithm. CID had no significant correlation with carbon emission reduction or carbon emissions. Then, the data were winsorized. All coefficients are Pearson correlation coefficients. CID is significantly positively correlated with operating income, assets, liability, turnover, and equity concentration. CEP is significantly positively correlated with sustainable development, while CRP is significantly positively correlated with FiRa and equity concentration and negatively correlated with TrdaSP, as shown in

Table 5. Correlations between variables.

Variables	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)	(13)	(14)	(15)	(16)	(17)
(1) CarboDisc	1.000
(2) CPerf_E	0.083	1.000
	(0.191)
(3) CPerf_R	−0.089 (0.175)	-	1.000
(4) SusDeve	−0.079	0.131 **	−0.005	1.000
	(0.211)	(0.039)	(0.939)
(5) Sales	0.433 ***	0.025	−0.109 *	−0.027	1.000
	(0.000)	(0.695)	(0.098)	(0.665)
(6) ROA	−0.005	−0.036	−0.044	0.617 ***	−0.011	1.000
	(0.943)	(0.572)	(0.506)	(0.000)	(0.862)
(7) Asset	0.278 ***	0.096	−0.054	0.031	0.308 ***	−0.166 ***	1.000
	(0.000)	(0.131)	(0.413)	(0.628)	(0.000)	(0.009)
(8) Liab	0.266 ***	0.098	−0.055	0.033	0.284 ***	−0.169 ***	1.000 ***	1.000
	(0.000)	(0.121)	(0.404)	(0.606)	(0.000)	(0.007)	(0.000)
(9) FiRa	−0.056	−0.579 ***	0.603 *	−0.126 **	0.065	0.173 ***	−0.260 ***	−0.265 ***	1.000
	(0.376)	(0.000)	(0.000)	(0.047)	(0.304)	(0.006)	(0.000)	(0.000)
(10) QuiRa	−0.060	0.117 *	−0.171 *	0.145 **	−0.131 **	0.437 ***	−0.007	−0.009	−0.308 ***	1.000
	(0.344)	(0.066)	(0.009)	(0.022)	(0.039)	(0.000)	(0.915)	(0.891)	(0.000)
(11) ToTuRa	0.185 ***	0.110 *	−0.460 *	0.241 ***	0.162 **	0.420 ***	−0.288 ***	−0.292 ***	0.238 ***	−0.150 **	1.000
	(0.003)	(0.081)	(0.000)	(0.000)	(0.010)	(0.000)	(0.000)	(0.000)	(0.000)	(0.018)
(12) EC 1	0.188 ***	0.091	0.415 *	0.024	0.175 ***	0.065	0.114 *	0.110 *	0.020	0.012	0.219 ***	1.000
	(0.003)	(0.149)	(0.000)	(0.704)	(0.005)	(0.304)	(0.073)	(0.084)	(0.758)	(0.848)	(0.000)
(13) EC 5	0.211 ***	−0.035	0.226 *	−0.037	0.223 ***	−0.063	0.224 ***	0.219 ***	0.028	−0.025	0.057	0.746 ***	1.000
	(0.001)	(0.580)	(0.000)	(0.555)	(0.000)	(0.324)	(0.000)	(0.001)	(0.661)	(0.689)	(0.372)	(0.000)
(14) EC 10	0.223 ***	−0.040	0.230 *	−0.041	0.227 ***	−0.085	0.248 ***	0.243 ***	0.009	−0.006	0.013	0.695 *	0.983 ***	1.000
	(0.000)	(0.525)	(0.000)	(0.523)	(0.000)	(0.181)	(0.000)	(0.000)	(0.886)	(0.921)	(0.844)	(0.000)	(0.000)
(15) EC 11	0.179 ***	−0.070	−0.023	−0.148 **	0.241 ***	−0.156 **	0.207 ***	0.201 ***	−0.012	−0.059	−0.308 ***	−0.437 ***	−0.273 ***	−0.255 ***	1.000
	(0.005)	(0.268)	(0.728)	(0.019)	(0.000)	(0.014)	(0.001)	(0.001)	(0.850)	(0.356)	(0.000)	(0.000)	(0.000)	(0.000)
(16) TrdaSP	−0.042	0.021	0.125 *	0.171 ***	−0.023	0.088	−0.039	−0.039	−0.013	−0.031	0.128 **	−0.022	0.005	0.006	−0.040	1.000
	(0.508)	(0.741)	(0.057)	(0.007)	(0.713)	(0.166)	(0.535)	(0.543)	(0.832)	(0.627)	(0.043)	(0.723)	(0.934)	(0.919)	(0.527)
(17) TrdhSP	−0.010	0.023	−0.176 *	0.175 ***	0.002	0.065	0.028	0.029	−0.034	−0.042	0.105 *	−0.013	0.062	0.063	−0.038	0.992 ***	1.000
	(0.880)	(0.715)	(0.007)	(0.005)	(0.980)	(0.309)	(0.657)	(0.651)	(0.592)	(0.509)	(0.096)	(0.836)	(0.331)	(0.318)	(0.549)	(0.000)

*** p < 0.01, ** p < 0.05, * p < 0.1.

.

4.3. Multivariate Analysis

Considering that there may be some continuity between indicators and their previous value, the study set regression models with a lag period on the variables. In order to solve the endogeneity problem, the Arellano–Bover/Blundell–Bond (A–B/B–B) estimation of dynamic panel data was used for regression analysis, and lags on endogenous variables and the predetermined variables were used as instrumental variables. The model includes four lags on the dependent variables. Equity concentration and tradable shares are pre-determined variables, and the endogenous variables are carbon performance, operating income, assets, liabilities, ROA, solvency, FiRa, and ToTuRa.

For hypothesis 1, results show that the Wald chi2(11) = 3.06 × 10⁶, Prob > chi2 = 0.0000, model is well established. The Arellano–Bond test for zero autocorrelation in first-differenced errors showed z = 1.498 (p = 0.1341) for Order 2, meaning there is no autocorrelation. The instrumental variables are shown in

Table 6. Instrumental variables.

	Instruments for Differenced Equation	Instruments for Level Equation
GMM-type	L (2/5). CarboDisc L (1/4).EC1 L (1/4).EC5 L (1/4).EC10 L (1/4).EC11 L (1/4). TrdhSP L(2/4). CPerf L (2/4). FiRa L(2/4).QuiRa L(2/4).ToTuRa L(2/4).Sales L(2/4).Asset L(2/4).Liab L(2/4).ROA	LD. CarboDisc D.EC1 D.EC5 D.EC10 D.EC11 D.TrdhSP LD. CPerf LD. FiRa LD.QuiRa LD.ToTuRa LD.Sales LD.Asset LD.Liab LD.ROA
Standard	D. CPerf D.FiRa D.QuiRa D.ToTuRa D.Sales D.Asset D.Liab D.ROA D.EC1 D.EC5 D.EC10 D.EC11 D.TrdhSP	_cons

. The Sargan test of overidentifying restrictions showed chi2(44) = 18.877, p = 0.9997. The overidentifying restrictions were valid. The coefficient of CPerf_E was 15.601, which is significant at the level of 0.01. See

Table 7. Regression results.

	Model 1 (HC)	Model 1 (LC)	Model 1 (HC)	Model 1 (LC)
	CarboDisc_-CPerf_E		CarboDisc_-CPerf_R
L	0.328 * (1.85)	0.141 (0.54)	0.555 ** (1.98)	0.581 ** (2.30)
L2	0.300 ** (2.20)	0.432 *** (4.09)	−0.327 ** (−2.08)	-
L3	-	0.001 (0.01)	0.239 (1.29)	-
L4	-	-	0.034 (0.18)	-
CPerf	15.601 *** (2.80)	−0.314 (−1.44)	−0.044 (−1.03)	0.001 (0.01)
FiRa	0.001 (0.34)	−0.001 (−0.43)	−0.002 *** (−3.39)	0.001 (0.82)
QuiRa	−0.005 (−0.15)	−0.014 (−0.24)	−0.028 (−1.37)	−0.029 (−0.85)
ToTuRa	0.098 (0.86)	0.064 (1.06)	0.026 (0.46)	0.006 (0.08)
Asset	0 *** (4.03)	0 ** (2.06)	0 (1.51)	0 *** (6.44)
Liab	0 *** (−2.62)	0 ** (−2.09)	0 (−1.6)	0 *** (−6.59)
ROA	0.002 (0.65)	−0.006 *** (−3.33)	0.008 *** (2.71)	−0.003 (−0.91)
Sales	0 (−0.36)	0 ** (2.14)	0 (−0.55)	0 (0.44)
EC1	−1.268 *** (−7.81)	0.145 (0.79)	−0.066 (−1.05)	0.273 ** (1.98)
EC5	0.155 (0.21)	0.651 (0.90)	−0.205 (−0.34)	1.001 *** (2.71)
EC10	0.638 (0.81)	−0.763 (−0.83)	0.335 (0.54)	−1.011 ** (−2.40)
EC11	0 (−0.01)	−0.039 (−0.72)	0.053 (1.50)	−0.043 ** (−2.06)
TrdaSP	0.008 *** (5.00)	-	0.008 (0.98)	0.003 * (1.91)
TrdhSP	-	0 (−0.01)	0.002 (0.19)	0.020 *** (5.13)
Constant	−16.114 *** (−2.72)	0.874 (1.20)	−1.251 (−1.35)	0.167 (0.60)
Number of obs	36	54	45	44
Number of instruments	60	76	76	71
Wald chi2	3.06 × 106 ***	600.815 ***	3819.721 ***	712,832.305 ***

*** p < 0.01, ** p < 0.05, * p < 0.1 HC and LC represent carbon-intensive industries and low-carbon industries, respectively.

. The better the carbon emission intensity performance of enterprises in carbon-intensive industries, the higher was the CID. Thus, hypothesis 1 was verified. However, hypothesis 2, hypothesis 3, and hypothesis 4 were not as predicted.

The coefficient of interaction between carbon emission intensity performance and the sustainable growth of enterprises in carbon-intensive industries was 84.989, as shown in

Table 8. Regression results of the moderating effect.

	Model 2 (HC)	Model 2 (LC)	Model 2 (HC)	Model 2 (LC)
	CarboDisc_-CPerf_E		CarboDisc_-CPerf_R
L	0.320 ** (2.13)	0.153 (0.58)	0.590 ** (2.04)	0.525 ** (2.51)
L2	0.259 * (1.86)	0.386 *** (3.89)	−0.270 (−1.35)	-
L3	-	−0.008 (−0.09)	0.353 * (1.86)	-
L4	-	-	−0.025 (−0.13)	-
CPerf	12.501 ** (2.22)	0.010 (0.04)	−0.156 ** (−2.08)	−0.061 (−1.18)
SusDeve	−2.391 *** (−4.72)	−0.265 (−0.71)	−0.257 (−1.61)	−0.040 (−0.54)
CPerf*SusDeve	84.989 *** (4.94)	0.337 (0.32)	−0.850 * (−1.91)	1.614 *** (3.85)
FiRa	−0.001 (−0.8)	0 (0.14)	−0.002 ** (−2.52)	0.002 (1.58)
QuiRa	−0.008 (−0.28)	−0.006 (−0.10)	−0.012 (−0.46)	−0.030 (−1.17)
ToTuRa	0.097 (1.04)	0.064 (0.94)	−0.056 (−0.73)	−0.007 (−0.1)
Asset	0 *** (4.47)	0 ** (2.11)	0 (1.44)	0 *** (4.36)
Liab	0 *** (−3.88)	0 ** (−2.14)	0 (−0.90)	0 *** (−4.52)
ROA	0 (−0.04)	0.002 (0.24)	0.016 *** (2.59)	0 (−0.08)
Sales	0 (0.59)	0 ** (1.96)	0 (−0.65)	0 *** (3.13)
EC1	−1.102 *** (−6.46)	0.161 (0.85)	−0.099 (−1.59)	0.430 *** (2.8)
EC5	−0.108 (−0.22)	0.845 (1.32)	−0.041 (−0.06)	0.449 (1.32)
EC10	0.846 (1.56)	−0.942 (−1.12)	0.058 (0.08)	−0.585 (−1.52)
EC11	0.003 (0.11)	−0.020 (−0.39)	0.064 * (1.86)	−0.047 * (−1.92)
TrdaSP	0.008 *** (5.94)	-	0.012 (1.24)	0.004 ** (2.39)
TrdhSP	-	0 (0.29)	0.004 (0.42)	0.020 *** (6.27)
Constant	−12.554 ** (−2.14)	0.456 (0.69)	−1.513 (−1.43)	0.139 (0.52)
Number of obs	36	54	45	44
Number of instruments	60	80	76	71
Wald chi2	459.273 ***	2606.275 ***	3170.564 ***	5,213,984.419 ***

*** p < 0.01, ** p < 0.05, * p < 0.1.

. Sustainable growth enhances the promotion of carbon emission intensity performance to the CID of enterprises in carbon-intensive industries. Hypothesis 5 was verified. However, sustainable growth enhances the inhibitory effect of carbon emission reduction intensity performance on the CID of carbon-intensive industries. Hypothesis 6 was not verified. In low-carbon industries, the coefficient of CPerf*SusDeve was not significant. Hypothesis 7 was not verified. The coefficient of CPerf*SusDeve was 1.614, which is significant at the 0.01 level, but CRP was still insignificant (compared with the results of the model with no interaction term), the results were not clear, and hypothesis 8 was not verified.

4.4. Robustness Checks

In this study, the robustness test was conducted by replacing the dependent variable with another proxy indicator. That is, the CID items were replaced by the CID score (CarboDisc_). Categories of carbon information items were distinguished, and CID items containing monetary or quantitative information were given higher scores, that is, each item of monetary, quantitative, and qualitative carbon information disclosed was awarded three points, two points, and one point, respectively. The carbon information items disclosed by enterprises were thus assigned and summed according to the corresponding coefficients to calculate the CID score.

Results showed that carbon emission intensity performance has a significantly positive effect on the CID score in carbon-intensive industries, as shown in

Table 9. Regression results of carbon performance in the form of carbon intensity and CID scores in carbon-intensive industries.

	Model 1 (HC)	Model 2 (HC)
	CarboDisc_-CPerf_E
L	0.481 *** (2.74)	0.425 *** (3.33)
L2	0.234 ** (2.50)	0.228 ** (2.04)
L3	-	-
L4	-	-
CPerf	14.917 *** (3.59)	11.463 ** (2.43)
SusDeve	-	−2.507 *** (−3.49)
CPerf*SusDeve	-	90.315 *** (3.72)
FiRa	0.001 (0.80)	−0.001 (−0.32)
QuiRa	−0.005 (−0.18)	−0.009 (−0.32)
ToTuRa	−0.041 (−0.43)	0.048 (0.62)
Asset	0 *** (−3.53)	0 *** (4.31)
Liab	0 *** (−3.37)	0 *** (−5.01)
ROA	0.002 (0.79)	−0.001 (−0.20)
Sales	0 (0.61)	0 * (1.86)
EC1	−1.381 *** (−8.59)	−1.225 ***−(8.79)
EC5	0.063 (0.08)	−0.209 (−0.38)
EC10	0.838 (1.06)	1.069 * (−1.79)
EC11	−0.014 (−0.53)	−0.008 (−0.33)
TrdaSP	0.008 *** (6.39)	0.009 *** (8.89)
TrdhSP	-	-
Constant	−15.374 *** (−3.51)	−11.447 ** (−2.41)
Number of obs	36	36
Number of instruments	60	60
Wald chi2	109,175.061 ***	15,939.747 ***

*** p < 0.01, ** p < 0.05, * p < 0.1.

. Sustainable growth enhances the promotion of carbon emission intensity performance to the CID of enterprises in carbon-intensive industries. Hypothesis 1 and hypothesis 5 were verified, while others have not been verified, which is consistent with the previous results. Other results can be seen in

Table A1. Regression results of carbon performance in form of intensity and CID scores.

	Model 1 (LC)	Model 1 (HC)	Model 2 (LC)	Model 2 (HC)
	CarboDisc_-CPerf_E	CarboDisc_-CPerf_R	CarboDisc-CPerf_E	CarboDisc-CPerf_R
L	0.111 (0.44)	0.067 (0.29)	0.088 (0.37)	0.091 (0.42)
L2	0.566 *** (6.51)	−0.247 *** (−4.54)	0.534 *** (6.59)	−0.231 *** (−3.36)
L3	0.003 (0.02)	0.403 (1.59)	0 (−0.00)	0.484 * (1.85)
L4	-	0.247 (1.57)	-	0.17 (1.01)
CPerf	−0.146 (−0.56)	00	−0.074 (−0.30)	−0.087 (−1.44)
SusDeve	-	-	−0.209 (−0.60)	−0.122 (−0.96)
CPerf*SusDeve	-	-	−0.779 (−0.63)	−0.700 * (−1.76)
FiRa	0.001 (0.14)	−0.002 *** (−4.01)	0.001 (0.31)	−0.002 *** (−2.64)
QuiRa	−0.012 (−0.22)	−0.032 *** (−3.09)	0.004 (0.07)	−0.026 ** (−2.13)
ToTuRa	0.037 (0.46)	0.038 (0.96)	0.044 (0.58)	−0.003 (−0.05)
Asset	0 * (1.72)	0 (−0.27)	0 * (1.79)	0 (−0.50)
Liab	0 * (−1.74)	0 (0.20)	0 * (−1.82)	0 (0.53)
ROA	−0.004 * (−1.81)	0.006 *** (2.90)	−0.002 (−0.20)	0.01 ** (2.43)
Sales	0 * (1.94)	0 * (1.76)	0 ** (2.06)	0 * (1.95)
EC1	−0.031 (−0.19)	−0.02 (−0.34)	0.003 (0.02)	−0.038 (−0.57)
EC5	0.338 (0.45)	−0.467 ** (−2.25)	0.235 (0.37)	−0.467 * (−1.88)
EC10	−0.389 (−0.41)	0.511 *** (2.81)	−0.213 (−0.25)	0.422 * (1.75)
EC11	−0.061 (−0.97)	0.04 ** (2.31)	−0.053 (−0.88)	0.054 *** (2.82)
TrdaSP	-omited due to collinearity	−0.005 (−0.69)	-omited due to collinearity	−0.002 (−0.25)
TrdhSP	0 (−0.24)	−0.009 (−1.32)	0 (0.11)	−0.008 (−0.99)
Constant	0.963 (1.25)	0.172 (0.21)	0.822 (1.12)	−0.152 (−0.17)
Mean (dep var)	0.478	0.133	0.478	0.133
SD (dep var)	0.189	0.104	0.189	0.104
Number of obs	54	45	54	45
Number of instruments	76	76	80	76
Wald chi2	816.183 ***	1296.134 ***	2171.390 ***	32,420.282 ***
AB test for ZA	Order (2) = 1.333, p = 0.182	Order (2) = −0.876, p = 0.381	Order (2) = 1.427, p = 0.154	Order (2) = −1.069, p = 0.285
Sargan test of OVERID	chi2 (59) = 32.118, p = 0.9983	chi2 (57) = 26.671, p = 0.9998	chi2 (61) = 27.894, p = 0.9999	chi2 (55) = 23.376, p = 0.9999

*** p < 0.01, ** p < 0.05, * p < 0.1.

.

4.5. Further Investigation

The results show that relationship between carbon emission intensity performance and CID in carbon intensive industries meets expectations. The uncertainty about the relationship between carbon performance and the CID of enterprises in low-carbon industries may be due to the small amount of carbon emission and carbon emission reduction, which does not draw enough attention from the enterprises. In order to deeply analyze the effect of carbon performance on CID in carbon-intensive industries and low-carbon industries, this paper further adopts the substitution variable method, that is, carbon performance is replaced by carbon emission amount performance and carbon emission reduction amount performance, and then attempts further regression testing.

Further tests and corresponding robustness tests (see

Table 10. Regression results of carbon performance in form of carbon amount and CID.

	Model 1 (HC)	Model 1 (LC)	Model 2 (HC)	Model 2 (LC)	Model 2 (LC)
	CarboDisc_-CPerf_EA				CarboDisc_-CPerf_RA
L	0.347 ** (2.15)	0.012 (0.05)	0.373 *** (2.59)	−0.021 (−0.09)	0.616 *** (3.25)
L2	0.358 ** (2.53)	0.472 *** (4.19)	0.286 * (1.91)	0.502 *** (4.58)	-
L3	-	0.089 (1.00)	-	0.058 (0.66)	-
L4	-	-	-	-	-
CPerfA	0.517 *** (3.80)	−7.657 *** (−4.13)	0.428 *** (2.95)	−12.667 *** (−3.37)	−7.400 *** (−3.43)
SusDeve	-	-	0.025 (0.17)	1.046 ** (2.04)	5.893 *** (3.61)
CPerfA*SusDeve	-	-	3.426 *** (4.81)	−53.346 ** (−2.25)	218.821 *** (3.73)
FiRa	0 (0.10)	−0.002 (−1.08)	−0.002 (−1.00)	−0.001 (−0.65)	0.004 ** (2.10)
QuiRa	0.001 (0.02)	−0.061 (−1.25)	0.022 (1.17)	0.012 (0.20)	−0.009 (−0.29)
ToTuRa	0.215 ** (2.07)	−0.113 ** (−2.12)	0.189 ** (2.13)	−0.086 (−1.39)	0.008 (0.26)
Asset	0 *** (4.12)	0 * (1.73)	0 *** (3.96)	0 * (1.65)	0 (1.11)
Liab	0 *** (−2.95)	0 * (−1.78)	0 *** (−4.88)	0 * (−1.69)	0 (−1.14)
ROA	0.002 (0.59)	−0.002 (−0.69)	−0.001 (−0.43)	−0.009 (−1.09)	−0.004 (−1.56)
Sales	0 (−0.43)	0 *** (2.91)	0 (0.35)	0 *** (3.21)	0 *** (2.70)
EC1	−1.187 *** (−6.21)	0.387 ** (1.98)	−1.222 *** (−4.58)	0.331 (1.63)	0.355 *** (2.79)
EC5	−0.616 (−0.81)	0.612 (0.95)	−0.217 (−0.45)	0.673 (1.21)	1.190 *** (3.51)
EC10	−1.369 (−1.52)	−1.112 (−1.36)	1.082 ** (1.98)	−1.126 * (−1.85)	−1.438 *** (−3.19)
EC11	−0.011 (−0.43)	−0.089 * (−1.94)	0.002 (0.08)	−0.104 ** (−2.03)	−0.030 (−1.28)
TrdaSP	0.006 *** (4.50)	−0.002 (−1.40)	0.008 *** (4.67)	-	0.003 ** (2.1)
TrdhSP	-	−0.002 (−1.40)	-	−0.002 (−1.49)	0.015*** (3.24)
Constant	−0.966 ** (−2.03)	9.191 *** (3.79)	−1.134 *** (−2.61)	13.603 *** (3.41)	−3.296 *** (−3.57)
Number of obs	36	54	36	54	44
Numberof instruments	60	76	60	80	71
Wald chi2	701.045 ***	9965.622 ***	937.624 ***	3095.462 ***	4,106,394.440 ***

*** p < 0.01, ** p < 0.05, * p < 0.1.

and

Table 11. Robust results of carbon emission performance in the form of amount and CID.

	Model 1 (HC)	Model 1 (LC)	Model 2 (HC)	Model 2 (LC)
	CarboDisc_-CPerf_EA
L	0.478 *** (3.05)	−0.025 (−0.11)	0.474 *** (3.80)	−0.082 (−0.45)
L2	0.323 *** (2.64)	0.554 *** (6.22)	0.264 ** (2.01)	0.615 *** (9.01)
L3	-	0.090 (0.64)	-	−0.017 (−0.14)
CPerfA	0.569 *** (4.46)	−7.235 *** (−3.50)	0.462 *** (3.37)	−16.645 *** (−5.25)
SusDeve	-	-	0.061 (0.45)	1.851 *** (4.41)
CPerfA*SusDeve	-	-	3.565 *** (4.26)	−92.801 *** (−5.00)
FiRa	0.001 (0.63)	−0.001 (−0.65)	−0.001 (−0.35)	−0.001 (−0.50)
QuiRa	0.004 (0.15)	−0.074 (−1.54)	0.026 (1.30)	0.022 (0.37)
ToTuRa	0.159 * (1.82)	−0.121 * (−1.68)	0.133 ** (2.02)	−0.100 (−1.43)
Asset	0 *** (4.62)	0 (1.49)	0 *** (4.92)	0 (1.30)
Liab	0 *** (−3.88)	0 (−1.53)	0 *** (−6.48)	0 (−1.33)
ROA	0.003 (0.75)	0 (0.02)	−0.002 (−0.62)	−0.012 (−1.49)
Sales	0 (0.26)	0 *** (2.71)	0 (1.37)	0 *** (3.57)
EC1	−1.374 *** (−8.35)	0.222 (1.31)	−1.415 *** (−5.63)	0.179 (0.92)
EC5	−0.766 (−0.96)	0.174 (0.25)	−0.323 (−0.69)	0.181 (0.35)
EC10	1.683 * (1.89)	−0.540 (−0.63)	1.361 *** (2.58)	−0.521 (−0.96)
EC11	−0.024 (−1.08)	−0.108 * (−1.86)	−0.008 (−0.34)	−0.140 ** (−2.43)
TrdaSP	0.007 *** (7.24)	-	0.009 *** (6.47)	-
H	-	−0.002 (−1.32)	-	−0.003 * (−1.87)
Constant	−1.020 *** (−2.91)	8.947 *** (3.20)	−1.211 *** (−3.99)	17.454 *** (4.93)
Number of obs	36	54	36	54
Number of instruments	60	76	60	80
Wald chi2	34,101.256 ***	45,207.295 ***	721.076 ***	16,428.888 ***

*** p < 0.01, ** p < 0.05, * p < 0.1.

) show that the carbon emission amount performance positively affects CID in carbon-intensive industries, and sustainable growth enhances the promotion effect of CPerf_EA on CID. CPerf_EA significantly (negatively) affects CID, and sustainable growth enhances the relationship between CPerf_EA and CID. Additionally, sustainable growth enhances the negative effect between CPerf_EA and CID. Hypothesis 1, hypothesis 5, and hypothesis 7 were thus verified. Other insignificant results can be seen in

Table A2. Regression results of carbon performance in form of amount and CID.

	Model 1 (HC)	Model 1 (LC)	Model 2 (HC)
	CarboDisc_-CPerf_RA
L	0.515 * (1.83)	0.582 *** (2.64)	0.524 * (1.66)
L2	−0.301 ** (−1.97)	-	−0.298 (−1.61)
L3	0.242 (1.34)	-	0.245 (1.28)
L4	0.064 (0.32)	-	0.061 (0.29)
CPerfA	0.048 (1.57)	−0.109 (−0.31)	0.038 (0.38)
SusDeve	-	-	−0.036 (−0.13)
CPerfA*SusDeve	-	-	0.067 (0.05)
FiRa	−0.002 *** (−3.25)	0.001 (0.77)	−0.003 *** (−3.24)
QuiRa	−0.028 (−1.31)	−0.028 (−0.84)	−0.027 (−1.03)
ToTuRa	0.033 (0.49)	0.01 (0.17)	0.027 (0.32)
Asset	0 * (1.65)	0 * (1.74)	0 (1.33)
Liab	0 * (−1.79)	0 * (−1.79)	0 (−1.00)
ROA	0.007 *** (2.97)	−0.003 (−0.92)	0.008 (0.75)
Sales	0 (−0.22)	0 (−0.07)	0 (−0.22)
EC1	−0.064 (−0.77)	0.275 ** (1.98)	−0.068 (−0.84)
EC5	0.034 (0.05)	1.021 ** (2.41)	0.022 (0.03)
EC10	0.119 (0.19)	−1.038 ** (−2.05)	0.133 (0.22)
EC11	0.043 (1.23)	−0.045* (−1.91)	0.044 (1.32)
TrdaSP	0.007 (1.05)	0.004 ** (2.36)	0.007 (0.98)
TrdhSP	0.001 (0.21)	0.020 *** (4.32)	0.002 (0.24)
Constant	−1.038 (−1.33)	0.16 (0.60)	−1.071 (−1.41)
Mean (dep var)	0.168	0.172	0.168
SD (dep var)	0.137	0.178	0.137
Number of obs	45	44	45
Number of instruments	76	71	76
Wald chi2	4195.857 ***	716,948.590 ***	943.905 ***
AB test for ZA	Order (2) = −0.444, p = 0.657	Order (2) = 1.405, p = 0.160	Order (2) = −0.382, p = 0.702
Sargan test of OVERID	chi2 (57) = 28.625, p = 0.9994	chi2 (55) = 24.513, p = 0.9999	chi2 (55) = 23.830, p = 0.9999

*** p < 0.01, ** p < 0.05, * p < 0.1.

and

Table A3. Robust results of carbon performance in the form of amount and CID.

	Model 1 (HC)	Model 2 (HC)
	CarboDisc_-CPerf_RA
L	0.0010	−0.058 (−0.26)
L2	−0.215 *** (−3.39)	−0.203 ** (−2.26)
L3	0.399 (1.61)	0.387 (1.45)
L4	0.291 (1.57)	0.307 (1.56)
CPerfA	0.074 (1.37)	0.117 (0.95)
SusDeve	-	0.152 (0.54)
CPerfA*SusDeve	-	−0.353 (−0.46)
FiRa	−0.002 *** (−3.30)	−0.002 *** (−3.41)
QuiRa	−0.029 *** (−2.81)	−0.033 ** (−2.62)
ToTuRa	0.032 (0.68)	0.055 (0.83)
Asset	0 (−0.87)	0 (−0.81)
Liab	0 (0.26)	0 (0.02)
ROA	0.006 *** (2.90)	0 (0.04)
Sales	0 *** (2.84)	0 *** (3.13)
EC1	−0.026 (−0.41)	−0.007 (−0.10)
EC5	−0.177 (−0.82)	−0.071 (−0.20)
EC10	0.226 (1.06)	0.10 (0.28)
EC11	0.034 ** (2.43)	0.028 * (1.84)
TrdaSP	−0.006 (−0.85)	−0.008 (−1.36)
H	−0.01 (−1.46)	−0.011 ** (−2.08)
Constant	0.301 (0.42)	0.543 (0.88)
Mean (dep var)	0.133	0.133
SD (dep var)	0.104	0.104
Number of obs	45	45
Number of instruments	76	76
Wald chi2	3198.921 ***	7086.540 ***
AB test for ZA	Order (2) = −0.889, p = 0.374	Order (2) = −1.042, p = 0.297
Sargan test of OVERID	chi2 (57) = 26.931, p = 0.9998	chi2 (55) = 22.416, p = 1.0000

*** p < 0.01, ** p < 0.05, * p < 0.1.

.

To sum up, the better the carbon emissions performance (both in terms of measurement of the intensity and amount), the more carbon information is disclosed. Sustainable growth has enhanced the promotion effect of CEP on CID of enterprises in carbon-intensive industries. In low-carbon industries, the better the CEP measured via amount, the less was the CID. Moreover, sustainable growth enhances the inhibitory effect of CEP measured via amount on CID.

4.6. Other Tests

It can be expected from the above results that circulating A-shares may have a certain impact on carbon performance and CID. Further, circulating A-shares were treated as a new moderating variable. The results, including robustness tests, showed that circulating A-shares enhanced the promotion effect of carbon emission intensity performance on CID of enterprises in carbon intensive industries, while circulating A-shares weakened the inhibition effect of carbon emission intensity performance on CID in low-carbon industries. That is, circulating A-shares contribute to CID, which may verify that the financing needs (or emphasis) of enterprises in the stock market play a positive role in the corporate CID (

Table 12. Regression results of carbon emission intensity performance, circulating A-shares, and CID.

	HC (Carbon-Intensive Industries)	LC (Low Carbon Industries)
	CarboDisc_-CPerf_E
L	0.305 * (1.76)	−0.033 (−0.13)
L2	0.200 (1.22)	0.322 *** (3.25)
L3	-	0.061 (0.55)
CPerf_E	23.456 *** (3.15)	−0.657 ** (−2.31)
TrdaSP	−0.023 (−1.60)	-
CPerf_E*TrdaSP	0.897 ** (2.09)	0.057 *** (2.90)
FiRa	0.001 (0.66)	0.006 (1.43)
QuiRa	0.006 (0.15)	0.082 (1.22)
ToTuRa	−0.093 (−0.51)	−0.064 (−0.91)
Asset	0 (0.84)	0 ** (2.32)
Liab	0 (−1.01)	0 ** (−2.38)
ROA	0.003 (0.97)	−0.004 (−1.41)
Sales	0 (1.20)	0 *** (2.78)
EC1	−0.713 ** (−2.17)	0.357 * (1.93)
EC5	−0.050 (−0.08)	0.724 (1.07)
EC10	0.442 (0.66)	−0.784 (−0.92)
EC11	0.005 (0.14)	−0.037 (−0.95)
TrdhSP	-	0.001 (0.35)
Constant	−21.300 *** (−3.14)	0.987 (1.33)
Number of obs	36	54
Number of instruments	60	78
Wald chi2	1401.547 ***	2360.559 ***

*** p < 0.01, ** p < 0.05, * p < 0.1.

).

5. Discussion and Conclusions

Taking A-share companies listed on the main board market as research objects, this paper explores the effects of corporate carbon performance, sustainable growth, industrial carbon intensity, and stock characteristics on corporate carbon information disclosure. Dynamic panel data models were established. Carbon emission performance both measured via intensity performance and amount performance has positive effects on carbon information disclosure, while carbon emission performance only measured via amount performance has the opposite effect. From the perspective of carbon-intensive industries, the essence of increasing carbon information disclosure is the improvement of carbon emission performance, which indicates that the interaction between high-carbon industries and capital markets will be more affected by the mitigation of carbon information asymmetry. The improvement of enterprises’ carbon emission performance in carbon-intensive industries verifies the positive effect of enterprises’ financing demand (or emphasis) in the stock market on enterprises’ carbon information disclosure. Carbon information disclosure in carbon-intensive industries can be well explained by legitimacy theory and shareholder theory. However, carbon emission performance measured only via amount performance’s inhibitory effect on carbon information disclosure in low-carbon industries may be related to corporate risk aversion to uncertainty. Namely, carbon information disclosure is more of a hindrance than a help to stock funding for low-carbon industries, or low-carbon industries have less of a preference for equity financing.

The potential reasons for the insignificant impact of carbon emission reduction performance on carbon information disclosure may be related to the fact that carbon emission reduction performance has not become the mainstream or the main carbon performance measurement index, or it may be the lack of consistency and comparability in carbon emission reduction data. As the task for carbon emission performance accounting and supervision is gradually carried out, corporate carbon emissions management has become overwhelming. Carbon emission reduction performance is somewhat not as important as carbon emission performance.

This study adds and verifies a new important factor in the framework of carbon information disclosure—sustainable growth. Sustainable growth negatively affects carbon information disclosure in carbon-intensive industries, while has a positive impact on carbon information disclosure in low-carbon industries. From the perspective of sustainable growth, carbon-intensive industries have conservative carbon information disclosure, while low carbon industries have aggressive carbon information disclosure.

Further, the results showed a synergistic effect between carbon emission performance (both in terms of intensity performance and amount performance) and sustainable growth on carbon information disclosure in carbon-intensive industries. There is a synergistic effect between carbon emission performance (amount performance rather than intensity performance) and sustainable growth on the carbon information disclosure of enterprises in low-carbon industries. Therefore, enterprise growth may be one factor in the internal driving force for carbon information disclosure. From another perspective, carbon emission performance has weakened the effect of sustainable growth on carbon information disclosure in both carbon-intensive industries and low-carbon industries. This shows the dominance of carbon emission performance on carbon information disclosure.

Based on a consideration of the capital market, carbon information disclosure not only reflects the legitimacy of enterprises, but may also be related to financing constraints. These lead to differences between carbon information disclosure in carbon-intensive industries and carbon disclosure in low-carbon industries. Further tests show that the circulation of A-shares contributes to the carbon information disclosure of enterprises in both carbon-intensive and other industries. This also verifies the positive effect of financing needs (or emphasis) on the carbon information disclosure of enterprises listed in the stock market.

However, this study has some limitations. This study adopts equity concentration to measure corporate governance. However, considering the importance of management in carbon information disclosure decisions, future studies need to add more indicators of corporate governance. In this paper, solvency, capital utilization, and tradable share ratio are included to reflect the financing method and its demand for carbon information. However, research on the role of the asset–liability ratio and debt cost seems obscure. In this way, improving carbon emission performance by consolidating the value of carbon information disclosure in reducing financing costs and financing restrictions will be helpful to explain corporate carbon information disclosure behavior. The potential mechanisms behind the relationship between sustainable growth and carbon information disclosure should include analysts’ reporting and predictions as well as firms’ share prices from the perspective of investors and firms.

With the gradual openness to corporate greenhouse gas reports in China, the future research direction for carbon information disclosure as well as carbon performance will involve the necessity and informative contents of mandatory carbon information disclosure. Under the uniform disclosure format, the contents of enterprise carbon information disclosure are highly similar, which means that dependent variables, such as carbon information disclosure, commonly used in previous research may be studied as control variables. In addition, ownership concentration or stock liquidity is significantly different in the influence mechanisms of the two types of carbon performance and carbon information disclosure. Further exploration of the reasons leading to these differences is conducive to gaining an in-depth understanding of corporate carbon information disclosure behaviors.