The Impact of Carbon Disclosure Intensity on Innovation Behavior in Textile and Apparel Enterprises

Abstract

Under the guidance of China’s “dual carbon” goals, the importance of carbon information disclosure in textile and apparel enterprises has become increasingly prominent, and its mechanism for enhancing corporate innovation behavior requires further clarification. Based on regression analysis, this study examines listed textile and apparel companies in China’s Shanghai and Shenzhen A-share markets from 2012 to 2024 using a fixed-effects model to empirically test the impact of carbon information disclosure intensity on corporate innovation behavior and its underlying mechanisms. The results demonstrate that increased carbon information disclosure intensity significantly promotes growth in corporate innovation behavior, a core conclusion that remains valid even after conducting a series of robustness tests addressing endogeneity issues. Mechanistic analysis reveals that the positive driving effect of carbon information disclosure intensity on innovation behavior is weakened by investor attention, with investor focus playing a negative moderating role in this relationship: high-quality carbon information disclosure should enhance innovation by reducing information asymmetry; however, under heightened investor scrutiny, short-term investment orientation and management pressure for immediate performance may distort this transmission pathway, thereby inhibiting innovation promotion. Heterogeneity analysis further shows significant differences in the impact of carbon information disclosure intensity on innovation behavior across textile and apparel firms with varying ownership structures and industry categories. This study provides theoretical foundations and practical guidance for advancing carbon information disclosure practices in the textile and apparel sector, guiding investor focus appropriately, and fostering corporate innovation development.

1. Introduction

The textile and apparel industry, as a traditional pillar industry of China’s national economy and social development, a fundamental sector for addressing livelihood issues and enhancing quality of life, and a competitive industry for international cooperation and integrated development, plays a pivotal role in driving economic and social progress, creating employment opportunities, and establishing global competitiveness. According to the China Statistical Yearbook released in 2025, by the end of 2024, the number of large-scale industrial enterprises in China’s textile and apparel sector reached 34,754, employing approximately 4.679 million people, with operating revenue exceeding 3.6 trillion yuan and international trade exports surpassing 2.1 trillion yuan. However, the industry also faces increasingly severe challenges regarding resource consumption, environmental pollution, and greenhouse gas emissions. According to the 2024 China Environmental Statistical Yearbook, the textile and apparel sector has become one of China’s most water-intensive manufacturing sectors, with chemical oxygen demand emissions reaching approximately 34,000 tons in 2024, accounting for 17.9% of total emissions across all industries. Additionally, its greenhouse gas emissions account for about 8% of global total emissions. With the introduction of the “dual carbon” goals and growing public environmental awareness, the textile and apparel industry urgently needs to adopt more proactive measures to mitigate its negative environmental impacts.

The uniqueness of the textile and apparel industry lies in three aspects: First, the sector is predominantly composed of small and medium-sized private enterprises (only 17.2% of companies in the China sample), characterized by pronounced family-oriented corporate governance and a short-term profit-driven orientation, with management decisions often driven by immediate performance pressures; Second, the industry features a long supply chain with high interconnectivity between upstream and downstream players, allowing the signaling effects of carbon disclosure to propagate along the supply chain, creating a transmission mechanism distinct from other industries; Third, the industry faces dual pressures from the “dual carbon” goals and green barriers in international trade, creating a unique tension between compliance requirements and innovation demands for carbon disclosure. Therefore, focusing the study on the textile and apparel industry not only ensures representativeness but also helps reveal the heterogeneous patterns of how carbon disclosure impacts innovation—patterns that may be masked by average effects in broader industry-wide research.

Carbon disclosure refers to enterprises’ regulatory obligation to disclose greenhouse gas emissions, carbon management practices, climate risks, and emission reduction performance to stakeholders. It helps strengthen corporate awareness of carbon management and emission reduction, while providing a basis for investors to monitor corporate operations [1]. Current policies require the improvement of the information disclosure system to cover key emission entities, financial institutions, and technical service providers, ensuring the public disclosure of comprehensive information across the entire chain—including quotas, compliance status, and trading activities—while strengthening the carbon emission accounting and measurement framework. However, China’s carbon information disclosure efforts started relatively late, characterized by low disclosure rates, inconsistent quality, and a lack of standardized guidelines. The gap between policy requirements and corporate implementation capabilities remains unresolved, significantly hindering the effective implementation of the “dual-carbon” policies. Existing research primarily focuses on the influencing factors or economic consequences of carbon information disclosure, while discussions on systematic institutional frameworks remain insufficient.

In existing research, some scholars have begun examining the relationship between carbon disclosure and corporate innovation behavior, though no unified conclusion has yet been reached. Currently, this study explores the connection between carbon disclosure and corporate innovation behavior from a disclosure perspective.

First, conduct a background study to identify the factors influencing carbon disclosure. This study examines the determinants of environmental disclosure intensity among listed companies, focusing on how corporate strategy and vision (including environmental audits and the establishment of environmental committees), board diversity (gender diversity and independence), and environmental-related factors (environmental performance and pollution levels) influence such intensity. It also aims to develop a scientific measurement index for environmental disclosure intensity, offering a new perspective for research on non-financial reporting quality [2,3,4,5,6,7,8,9,10,11,12].

The second aspect involves research on economic consequences, focusing on the interrelationships among environmental performance, environmental disclosure, and economic performance. For instance, research focuses on environmental performance and environmental disclosure, examining their interrelationships with economic performance. This approach aims to address theoretical inconsistencies and conflicting empirical findings in previous studies—stemming from failure to account for the endogeneity and joint determination among these three dimensions—and provides a foundation for optimizing corporate environmental strategies and regulatory policies [13,14,15].

Third, conduct research on disclosure quality and authenticity to verify whether voluntary carbon disclosures objectively reflect actual carbon performance. By examining the relationship between voluntary carbon disclosure and corporate actual carbon performance, this study focuses on assessing whether carbon disclosure information under the CDP framework objectively reflects companies’ carbon performance (including carbon emission intensity and emission reduction effectiveness), while comparing the relative explanatory power of the “signal transmission” theory and the “legitimacy theory” in explaining corporate carbon disclosure behavior. The findings provide empirical support for stakeholders to evaluate the credibility of carbon disclosures and for policymakers to standardize related practices [16,17,18,19,20,21,22,23,24,25].

Fourth, a systematic review integrates literature across four dimensions—“theoretical foundations, disclosure characteristics, influencing factors, and economic consequences”—to establish a comprehensive analytical framework for carbon information disclosure research. The literature on carbon information disclosure focuses on research trends, theoretical foundations, disclosure characteristics, influencing factors, and economic consequences in this field; it clarifies existing consensus and controversies in current studies and proposes future research directions, providing a systematic reference for academic research, corporate practice, and policy formulation in the realm of carbon information disclosure [1,26,27,28,29,30].

Current research examines the relationship between investor concerns and corporate behavior (including innovation behavior) from an investor perspective, primarily following these key frameworks.

First, the impact of investor attention on corporate behavior. In practice, China’s rapid economic development is accompanied by severe environmental challenges; carbon emissions from high-tech enterprises have been rising annually and hold dual importance in both economic growth and environmental governance. Meanwhile, investors are paying increasing attention to corporate environmental performance, yet China’s capital market is characterized by a high proportion of retail investors and pronounced irrationality. The shaping effect of media coverage and reporting sentiment on corporate behavior cannot be overlooked. Theoretically, existing research remains controversial regarding the relationship between investor attention and corporate environmental performance, with most studies focusing on internal factors. Based on the aforementioned context and grounded in legitimacy theory and principal-agent theory, this study constructs a main-effect model linking “investor attention → corporate environmental performance.” Media attention is measured using the logarithm of news coverage frequency, while media sentiment is calculated via the Janis-Fadner coefficient. Heterogeneity analysis incorporating industry type, pollution characteristics, and corporate ownership structure provides empirical support for understanding how external attention influences the environmental behavior of high-tech firms [31,32,33,34,35,36,37].

Secondly, investor attention and capital market pricing efficiency. On one hand, using high-tech firms as a sample and examining both investor attention and media coverage, this study focuses on testing whether the investor attention hypothesis can explain how corporate names influence capital markets, thereby providing a new perspective for research in this field [38]; on the other hand, building upon the limited attention theory, it constructs a logical framework comprising “institutional attention measurement (AIA) → attention differences (between institutions and retail investors) → information integration and price efficiency (price drift),” revealing how variations in attention allocation affect information pricing efficiency [39].

Third, investor attention and financial asset returns. On one hand, the study examines the direct impact of investor attention on asset returns, focusing specifically on testing its significant effect on exchange rate returns, systematically investigating the relationship between investor attention and stock returns, and assessing the predictive value of a composite attention index [31,40,41,42,43,44,45,46,47].

Fourth, retail investor attention and stock price risk. This study examines the impact of retail investor attention on stock price crash risk in the China stock market, constructs a dynamically evolving comprehensive indicator of retail investor attention, and tests its relationship with stock price crash risk [48].

Introducing investor attention as a moderating variable in the relationship between carbon disclosure and corporate innovation is based on the following rationale: The core function of carbon disclosure is to reduce information asymmetry, while investor attention determines the intensity and direction of information reception and interpretation. When investor attention is low, carbon disclosure primarily serves a “signal transmission” function, lowering financing costs and promoting innovation investment; however, when attention is excessively high, short-sighted investors combined with short-term performance pressures may distort the information transmission pathway, transforming signals that originally foster innovation into short-term compliance pressures, thereby inhibiting innovation. This mechanism is particularly pronounced in China’s capital market, which is dominated by retail investors and characterized by short-term trading patterns, with the private-sector nature of enterprises in the textile and apparel industry further amplifying this effect. Therefore, investor attention is not merely an exogenous environmental variable but also a critical boundary condition influencing how carbon disclosure impacts corporate innovation.

Finally, regarding innovative practices of textile enterprises. The literature review analyzes the drivers and challenges of green transformation, focusing on elucidating how external pressures and internal motivations propel the low-carbon transition. It examines constraints at levels such as power structures, resource conditions, and institutional support, and employs the “just transition” framework to provide empirical evidence and policy recommendations for achieving environmentally sustainable and socially equitable green development in this sector [49,50,51]. There is a significant positive correlation between innovation and the performance of small enterprises in the textile manufacturing sector. The spirit of innovation, manifested in product, market, and process innovations, serves as a key determinant of success [52,53].

A review of the aforementioned literature reveals several shortcomings in existing research: First, studies on the economic consequences of carbon disclosure predominantly focus on industry-wide or heavily polluting sector samples, lacking specialized analysis of specific industries—particularly the textile and apparel sector—which exhibits distinct characteristics such as high water consumption, significant emissions, extensive supply chains, and a high proportion of private enterprises. Consequently, the mechanism by which carbon disclosure influences innovation may fundamentally differ from findings observed across entire industries. Second, while existing research acknowledges the impact of external attention on corporate behavior, few systematically examine investor attention as a moderating variable affecting corporate innovation through carbon disclosure, nor do they elucidate the underlying rationale for its negative moderating effect from the perspectives of principal-agent relationships and short-term decision-making pressures. Third, traditional multi-dimensional scoring methods for measuring carbon disclosure suffer from high subjectivity and poor interfirm comparability, necessitating more objective and replicable alternative indicators.

Compared with existing research, the marginal contributions of this paper are primarily as follows: First, in terms of research perspective, the study focuses on more representative textile and apparel enterprises to examine how carbon disclosure intensity influences innovation behavior within this industry. Second, regarding theoretical mechanisms, building upon an analysis of the impact of carbon disclosure intensity on innovation behavior, the paper proposes and validates that the relationship between carbon disclosure intensity and corporate innovation behavior is moderated by investor attention, which weakens the positive impact of carbon disclosure on innovation. Finally, concerning indicator selection, the total word frequency of “carbon information disclosure” serves as the measurement metric for carbon disclosure intensity. Compared to traditional indicators, this approach provides a more comprehensive and intuitive reflection of companies’ carbon disclosure practices, offering valuable insights for subsequent research.

2. Theoretical Analysis and Hypothesis Formation

2.1. The Intensity of Carbon Information Disclosure and Innovation in Textile and Apparel Enterprises

Carbon disclosure exerts a significant positive impact on the value of textile and apparel enterprises. Its core mechanism extends beyond mere reductions in financing costs and improvements in corporate reputation; rather, it operates through profound principles rooted in legitimacy theory and signaling theory, combined with industry-specific mechanisms such as low-carbon technological upgrades and green supply chain adaptations, thereby achieving long-term value enhancement for companies via multi-dimensional effects.

Specifically, from the perspective of legitimacy theory, the textile and apparel industry, as a traditional manufacturing sector, faces increasingly stringent environmental regulatory policies and demands for green transformation. High-quality carbon disclosure by enterprises is fundamentally a key initiative to align with institutional frameworks and enhance organizational legitimacy, effectively alleviating compliance pressures, mitigating policy penalty risks, and establishing an institutional foundation for sustainable business operations. From the perspective of signal theory, carbon information disclosure serves as a visible signal conveying an enterprise’s sustainable development capabilities, effectively reducing information asymmetry among enterprises, investors, consumers, and supply chain partners. Its signaling efficiency far exceeds that of mere financing cost optimization or reputation building, enabling it to communicate the company’s core competitiveness in low-carbon management and technological innovation to the market and thereby influence the decision-making behaviors of market participants [54]. Furthermore, the value enhancement effect of carbon information disclosure is achieved through a long-term mechanism involving “acquisition of legitimacy—signal transmission—optimization of resource allocation”: by disclosing standardized and transparent carbon information, enterprises first gain institutional legitimacy at the policy level and social legitimacy at the market level; subsequently, they send a clear signal of their commitment to low-carbon development, attracting ESG investors and securing financial support, while enhancing consumer environmental awareness and strengthening green alignment within supply chains; ultimately, this drives enterprises to allocate resources such as capital and technology toward core areas like low-carbon technological upgrades and green supply chain adaptation, fostering R&D investment and technological innovation, improving risk management capabilities, and thereby achieving dual improvements in both long-term market valuation and financial performance. Research on heavily polluting industries and the textile and apparel sector indicates a significant positive correlation between the intensity of environmental information disclosure and investor attention. As a core component of environmental information, enhanced carbon disclosure quality not only attracts diversified financial support for enterprises but also amplifies value enhancement effects through industry-specific mechanisms, incentivizing companies to pursue low-carbon innovation and solidify their green competitive position [8,54,55,56,57,58].

Based on the above analysis, this paper proposes the following hypothesis:

H1: 

Increased intensity in carbon information disclosure helps foster innovative practices among textile and apparel enterprises.

2.2. The Moderating Role of Investor Attention

Information asymmetry is one of the core frictions in capital markets; the information gap between firms and external investors may lead investors to demand higher premiums for “unknown risks,” thereby reducing capital allocation efficiency [59]. The Investor Attention Hypothesis posits that investors tend to allocate their limited attention to signals that reduce information uncertainty [38], with their active pursuit of corporate information serving as a direct manifestation of this attention pattern [32].

According to principal-agent theory, management teams in private textile and apparel enterprises act as agents whose compensation levels and job stability are not only directly linked to the company’s short-term performance but also closely tied to its short-term cash flow conditions. Since owners (principals) are typically company founders or family members, short-term profitability directly determines the enterprise’s survival and development; consequently, performance evaluations of management place greater emphasis on immediate profit returns while neglecting long-term innovation value [60].

Within the short-term management pressure framework, the profit-driven objectives of private textile and apparel companies interact synergistically with heightened investor scrutiny.

Under intense investor scrutiny, short-sighted investors (including some private enterprise owners) tend to overemphasize the short-term environmental compliance costs reflected in carbon disclosures—such as investments in energy-saving equipment for textile dyeing processes and initial expenditures for green supply chain upgrades. These costs directly increase operational burdens and compress profit margins, creating significant conflicts with private enterprises’ immediate profitability objectives. Investors often overlook the long-term value derived from integrating carbon management with innovation, including breakthroughs in low-carbon dyeing technologies, cost savings from recycled textile utilization, and premium pricing for eco-friendly products. This leads to negative expectations regarding corporate innovation investments and exacerbates external financing constraints. Given that private textile and apparel companies already face limited funding channels and weak capital reserves, heightened financing pressures further restrict innovation funding sources. Coupled with management’s focus on short-term performance metrics, this forces executives to prioritize immediate profits and cash flow stability over innovation, thereby undermining the positive driving effect of high-quality carbon disclosure on innovation. As a traditional manufacturing sector closely tied to consumer lives, the textile industry’s carbon management practices attract substantial public and media attention. Any compliance gaps in carbon reporting or conflicts between environmental investments and short-term profitability can easily trigger negative public sentiment. Given private enterprises’ weaker resilience to such risks, adverse publicity may directly impact order acquisition and brand reputation, ultimately threatening their survival. In the layered principal-agent relationship, management’s short-term performance orientation and private enterprises’ pursuit of immediate profitability lead them to proactively control R&D investment scale. This strategy aims to mitigate the dual pressures of carbon management expenditures and innovation investments on short-term performance while avoiding negative public opinion risks, prioritizing limited resources for carbon compliance and short-term profit-generating projects rather than long-term innovation initiatives [61]. This study identifies investor attention as a moderating variable rather than a mediating mechanism. Specifically, investor attention influences the strength and direction of the relationship between carbon disclosure intensity and corporate innovation behavior. When investor attention increases, short-term market pressure intensifies managerial myopia, thereby weakening the positive signaling effect of carbon disclosure on innovation investment.

Based on the above analysis, this paper proposes the following hypothesis:

H2: 

Investor attention exerts a negative moderating effect on the relationship between carbon disclosure intensity and corporate innovation behavior; that is, the higher the investor attention level, the weaker the positive promoting effect of high-quality carbon disclosure on corporate innovation behavior.

3. Model Specification and Variable Description

3.1. Data Sources

According to the definition of the China National Textile and Apparel Council and the national economic industry classification standards, the textile and apparel industry primarily comprises three subsectors: the textile industry (C17), the apparel and accessories industry (C18), and the chemical fiber manufacturing industry (C28). Therefore, this study selects listed companies in these three sectors on China’s Shanghai and Shenzhen A-share markets from 2012 to 2024 as research samples. To ensure data accuracy, the original data were processed as follows: (1) excluding enterprises with a debt-to-asset ratio ≥ 1; (2) excluding enterprises classified as ST, *ST, or PT in the current year; (3) to mitigate potential impacts of outliers, all the continuous variables underwent a 1% bilateral tail-trimming procedure; (4) excluding enterprise data with missing key variables. After these adjustments, a final sample of 105 textile and apparel enterprises was obtained, yielding 956 observed values of unbalanced panel data. All the data in this study are sourced from the CSMAR database.

3.2. Variable Selection

3.2.1. Dependent Variable

This study adopts a closed-loop definition for corporate innovation behavior, specifically using the number of authorized invention patents granted by an enterprise in the current year as the primary measurement indicator. To avoid measurement bias caused by zero data values, this indicator is adjusted by adding 1 and then subjected to natural logarithmic transformation. Additionally, keyword frequency in corporate annual reports serves as a supplementary metric; keywords reflecting corporate innovation activities—namely exploration, search, change, risk-taking, experimentation, flexibility, discovery, and innovation—are extracted from these reports. Word frequency analysis is employed to enhance the reliability of research conclusions. Detailed definitions of these indicators are presented in

Table 1. Definitions of Major Variables.

Variable Type	Variable Name	Variable Symbol	Variable Declaration
explained variable	Enterprise Innovation	Innovation	This paper measures corporate innovation (invention) by taking the natural logarithm of the number of invention patents granted in the current year plus one.
explanatory variable	Carbon Information Disclosure Intensity	Score	The total word frequency of carbon information disclosure is used as the measurement indicator for the intensity of carbon information disclosure.
Moderating Variable	Investor Interest Level	Attention	The Baidu Search Index for listed companies is calculated using the natural logarithm function.
controlled variable	Working Capital	NWC	The difference between the total current assets and the total current liabilities after logarithmic transformation (unit: ten thousand yuan)
	Operating Cost Ratio	OCR	Operating Cost/Operating Revenue × 100%
	Total Operating Cost Ratio	TOCR	Total Operating Costs/Total Operating Revenue
	Equity Ratio	DER	Total Liabilities/Total Owner’s Equity
	Debt-to-equity market value ratio	DMER	Total Liabilities/Market Value
	Comprehensive Revenue Growth Rate	CIGR	(Total comprehensive income for the current period—Total comprehensive income for the same period last year)/(Total comprehensive income for the same period last year)
	Ownership Nature	SOE	State-owned enterprises: 1; Non-state-owned enterprises: 0
	Company Size	Size	The natural logarithm of the enterprise’s total assets at year-end is a continuous variable.

.

3.2.2. Explanation Variables

The core explanatory variable in this study is carbon information disclosure intensity (Score). We measure carbon information disclosure intensity using the total frequency of carbon-related disclosure terms appearing in firms’ annual reports. A higher value of Score indicates a higher intensity of carbon information disclosure.

To construct the indicator, this study develops a structured carbon-related keyword dictionary based on prior literature and commonly used carbon and ESG reporting terminology. The keyword dictionary is constructed at three levels: (1) carbon emission-related terms (e.g., “carbon emission,” “CO₂,” “greenhouse gas”); (2) carbon management terms (e.g., “carbon accounting,” “emission reduction target,” “carbon audit”); and (3) carbon policy and market terms (e.g., “carbon trading,” “carbon quota,” “ETS”). We then count the total occurrences of these keywords in each firm’s annual report to obtain the carbon information disclosure intensity measure.

This text-based measure provides a transparent and replicable proxy for the extent to which firms discuss carbon-related issues in their annual reports and is consistent with the definition of Score reported in Table 1.

3.2.3. Control Variables

Drawing on existing research on the factors influencing corporate innovation behavior, this study selects working capital (NWC), operating cost ratio (OCR), total operating cost ratio (TOCR), equity ownership ratio (DER), debt-to-equity market value ratio (DMER), comprehensive return rate (CIGR), equity nature (SOE), and firm size (Size) as control variables, with their specific definitions presented in Table 1.

3.2.4. Moderating Variables

The moderating variable is investor attention. Given the high proportion of retail investors in China’s capital market, where most investors rely on online platforms for stock information, the frequency of online searches by listed companies objectively reflects actual investor interest levels. Investor attention measures the degree of focus investors have on a company, indicating market participants’ interest, importance attached to the company, and willingness to track and analyze relevant corporate information. This metric plays a crucial role in capital markets: high levels of attention may attract greater capital inflows and enhance market influence, whereas low attention can place a company in a marginalized position. In terms of methodology, this study adopts Zhao Ruwei’s approach, utilizing Baidu search keyword frequency data for measurement. Due to the right-skewed distribution of raw data, logarithmic transformation was applied to approximate normality and mitigate the impact of outliers [62].

Current research has not established a unified standard for measuring investor concerns and related variables; different fields employ various indicators such as online search indices, the number of analysts following an individual, institutional ownership ratios, and media attention. Given the textile and apparel industry’s characteristics—dominated by retail investors and characterized by pronounced short-term trading activity—the online search index effectively captures short-term market focus and investment trends, aligning with the theoretical framework of this study [63]. Its specific definition is presented in Table 1.

3.3. Model Configuration

To investigate the impact of carbon disclosure intensity on innovation behavior in textile and apparel enterprises, this study employs the following fixed-effects model:

$${\mathit{Innovation}}_{it}={\alpha }_0+{\alpha }_{1}Scor{e}_{it}+{\alpha }_2{X}_{it}+{\lambda }_i+{\eta }_t+{\epsilon }_{it}$$

(1)

In the formula: the subscript i represents ${\mathit{Innovation}}_{it}$ a firm, t represents the year; denotes the innovation behavior of firm i in year t; $Scor{e}_{it}$ is the total word frequency of carbon information disclosure by firm i in year t; $X_{it}$ represents a set of control variables that may influence firm innovation behavior. ${\lambda }_{it}$ represents the firm-specific fixed effect, ${\eta }_{it}$ represents the year-specific fixed effect, and ${\epsilon }_{it}$ is the random error term. ${\alpha }_1$ is the core coefficient of this study, measuring the impact of carbon information disclosure intensity on innovation behavior in textile and apparel firms; if ${\alpha }_1$ is greater than 0, it indicates that increased carbon information disclosure intensity enhances innovation behavior in these firms.

Beyond the mechanisms of influence, theoretical analysis further indicates that investor attention may exert a negative moderating effect on the innovation behavior of textile and apparel enterprises as carbon disclosure intensity increases. To test this moderating effect, this paper proposes the following empirical model:

$$\begin{array}{l}{\mathit{Innovation}}_{it}={\gamma }_0+{\gamma }_{1}Scor{e}_{it}+{\gamma }_{2}Attentio{n}_{it}+{\gamma }_{3}Scor{e}_{it}\times Attentio{n}_{it}\\ +{\gamma }_4{X}_{it}+{{\lambda }^{″}}_i+{{\eta }^{″}}_t+{{\epsilon }^{″}}_{it}\end{array}$$

(2)

In the formula $Attentio{n}_{it}$: represents the measure of market participants’ interest and emphasis on a specific investment asset; all the other symbols have the same meanings as in Equation (1).

3.4. Descriptive Statistics

Based on the descriptive statistics in

Table 2. Descriptive statistics of main variables.

Variable	Obs	Mean	Std. Dev.	Min	Max
Innovation	457	1.358	1.360	0	5.283
Score	380	1.174	0.989	0	3.892
Attention	419	12.66	0.680	10.21	14.75
NWC	379	20.518	1.178	15.351	23.157
DER	465	1.017	2.544	0.0310	39.59
DMER	443	0.271	0.191	0.0111	0.839
CIGR	447	−0.033	3.591	−14.45	16.64
OCR	465	0.691	0.214	0.243	1.007
TOCR	465	0.918	0.110	0.629	1.277
Size	463	22.191	1.225	19.545	26.210
SOE	465	0.172	0.378	0	1

, the variable distribution in this study sample exhibits the following characteristics: the mean value of the dependent variable—corporate innovation level (Innovation)—is 1.358 with a standard deviation of 1.360, indicating a well-distributed pattern that reflects certain variations in innovation capabilities among firms; the core explanatory variable—carbon disclosure intensity (Score)—has a mean of 1.174 and a standard deviation of 0.989, falling within the range [0, 3.892], suggesting generally low carbon disclosure levels across firms with significant inter-individual differences; the moderating variable—investor attention (Attention)—has a mean of 12.66 and a standard deviation of 0.680, reflecting marked disparities in market attention toward different firms. Regarding control variables, the mean value of company size (Size) was 22.191 with a standard deviation of 1.225, conforming to the typical distribution of logarithmized total assets for listed companies; the equity ratio (DER) averaged 1.017 with a standard deviation of 2.544, indicating significant variations in capital structures among sample firms; the comprehensive revenue growth rate (CIGR) averaged −0.033 with a standard deviation of 3.591, demonstrating considerable data volatility that requires careful consideration in regression analysis; both operating cost ratio (OCR) and total operating cost ratio (TOCR) were positive values, consistent with financial indicator definitions, though some firms exhibited cost inversion phenomena; working capital (NWC) averaged 20.518 with a standard deviation of only 1.178, reflecting overall stable working capital management across sample enterprises; equity ownership nature (SOE) averaged 0.172, aligning with the industry’s predominant composition of private enterprises. Given the variability among observed variable values, subsequent regression analyses will use non-imputed core variables to construct samples and employ unbalanced panel data estimation methods.

4. Empirical Result Analysis

4.1. Benchmark Regression

Column (1) of

Table 3. Benchmark Regression Results.

Variable	(1)	(2)	(3)
	Innovation	Innovation	Innovation
Score	0.029 ***	0.032 ***	0.021 ***
	(0.010)	(0.011)	(0.008)
Score ∗ Attention			−0.015 **
			(0.008)
Attention			0.0003 ***
			(0.000)
Controlled variable	NO	YES	YES
Constant term	1.217 ***	0.701	1.922 ***
	(0.071)	(0.631)	(0.252)
R2	0.573	0.588	0.594
Individual/Year	NO	YES	YES
Sample	439	439	425

Standard errors in parentheses: *** p < 0.01, ** p < 0.05.

presents the regression results incorporating only core explanatory variables. Without controlling for other factors, the coefficient of Score was 0.029 and statistically significant at the 1% level, indicating a significant positive correlation between carbon disclosure intensity and corporate innovation, providing preliminary evidence for Hypothesis H1. Column (2) further incorporates all the control variables while accounting for individual firm heterogeneity and year-specific effects. The result shows a coefficient of 0.032, also significant at the 1% level, demonstrating that after controlling for firm heterogeneity and temporal trends, carbon disclosure’s positive impact on innovation performance becomes more pronounced and robust. Column (3) examines moderating effects through an interaction term, yielding a coefficient of −0.015 that is significant at the 5% level. This suggests that while high investor attention suppresses this positive effect, carbon disclosure intensity exerts a baseline positive influence on corporate innovation—a finding that highlights the negative moderating role of investor attention in the carbon-disclosure-innovation relationship. Its underlying mechanism can be further elucidated by considering the unique governance characteristics of private enterprises in the textile and apparel industry, thereby validating Hypothesis 2.

Specifically, heightened investor scrutiny constrains corporate innovation through three intermediary mechanisms: First, short-term performance pressure—private enterprises in the textile and apparel sector often operate under intense end-market competition and cash-flow-sensitive governance structures. Excessive investor attention compels management to prioritize quantifiable metrics like immediate revenue and profits, while reducing investments in long-term, uncertain R&D initiatives; Second, stock price volatility pressure—given the industry’s vulnerability to raw material price fluctuations and consumer demand changes, heightened investor attention amplifies stock price sensitivity to short-term performance, leading management to prioritize avoiding R&D-related earnings swings for price stability; Third, managerial short-sightedness—most private textile and apparel firms exhibit family-oriented governance practices, where investor scrutiny drives management toward short-term arbitrage strategies over sustained innovation investments, ultimately undermining the positive impact of carbon disclosure on corporate innovation. The coefficient of Attention appears numerically small due to scaling and logarithmic transformation; however, it remains statistically significant at the 1% level.

4.2. Robustness Test

Table 4. Robustness Test.

Variable	(1)	(2)	(3)
	Replace Fixed Effect	Delayed First Period Inspection	Substitute Variable
Score	0.031 ***	0.029 *	3.069 **
	(0.006)	(0.016)	(1.32)
Controlled variable	YES	YES	YES
Constant term	0.750	1.690	1.833
Individual	YES	YES	YES
Year	YES	YES	YES
Sample Size	452	299	369

Standard errors in parentheses: *** p < 0.01, ** p < 0.05, * p < 0.1.

presents the results of three robustness tests. In column (1), after replacing the fixed-effects specification, the coefficient of Score is 0.031, with a standard error of 0.006, and remains significantly positive at the 1% level. In column (2), using the one-period lagged value of Score, the estimated coefficient is 0.029, with a standard er-ror of 0.016, and is significantly positive at the 10% level. In column (3), we replace the original measure with an alternative emissions-based indicator constructed from firm-level CO₂ emissions. Because higher CO₂ emissions indicate poorer carbon per-formance, which is directionally opposite to the baseline Score variable, CO₂ emissions are first logarithmically transformed and then inversely transformed so that a higher value of the alternative indicator represents a more favorable carbon-related outcome. This transformation also helps reduce the influence of the highly skewed distribution of absolute emissions. The estimated coefficient of the alternative indicator is 3.069, with a standard error of 1.320, and is significantly positive at the 5% level. The sample size in column (3) is 369, mainly because firm-level CO₂ emissions data are unavailable for some firm-year observations.

All three specifications include the full set of control variables as well as firm and year fixed effects, with sample sizes of 452, 299, and 369, respectively. Although both the statistical significance levels and sample sizes vary across the three robustness tests, the estimated coefficients remain positive and statistically significant, and their directions are consistent with the baseline regression results. Overall, the results remain generally robust to alternative fixed-effects specifications, the use of a lagged explanatory variable, and alternative variable measurement, providing further support for the robustness of the main conclusions.

4.3. Endogeneity Test

To address the bidirectional causality and endogeneity bias of omitted variables present in the core explanatory variable Carbon Disclosure Intensity Score, this study employs the first-period lagged carbon disclosure intensity (L1_Score) as an exogenous instrumental variable and conducts a two-stage least squares (2SLS) regression. Column (1) presents the first-stage regression: strictly adhering to the 2SLS specification, it uses current carbon disclosure level Score1 as the dependent variable and L1_Score as the core instrumental variable; the results show that the coefficient of L1_Score is significantly positive. Column (2) shows the second-stage 2SLS regression: with Innovation1 as the dependent variable and after controlling for individual and year fixed effects along with all the control variables, the coefficient of the core variable Score1 remains significantly positive (

Table 5. Endogeneity Test Results.

Variable	(1)	(2)
	Stage I (Tool Variable Validity Test)	Stage II (2SLS Main Effect Estimate)
L1_Score	0.624 ***
	(0.127)
Score		0.044 **
		(0.022)
Constant term	8.056 *
	(4.722)
F_stat	10.285
controlled variable	YES	YES
Individual/Year	YES	YES
Sample Size	305	305

Standard errors in parentheses: *** p < 0.01, ** p < 0.05, * p < 0.1.

).

4.4. Heterogeneity Analysis

This paper categorizes textile and apparel enterprises based on their distinct ownership structures, operational scales, and industry sectors to examine the heterogeneity of how carbon disclosure intensity influences their innovation behaviors.

4.4.1. Ownership Nature

State-owned enterprises, leveraging their natural resource endowments, policy support, and institutional advantages, inherently possess the willingness and conditions for environmental information disclosure. However, heterogeneous group regression analysis reveals no significant differentiation in the positive driving effect of carbon disclosure on corporate innovation within the state-owned sector. This phenomenon can be attributed to stringent carbon assessment requirements: state-owned enterprises face rigorous administrative oversight, low-carbon performance evaluations, and mandatory compliance obligations, leading to carbon disclosures primarily as passive compliance under regulatory frameworks—resulting in homogeneous disclosure practices that fail to effectively address information asymmetry or optimize R&D resource allocation through signaling mechanisms, thereby significantly limiting the marginal improvement potential for innovation. In contrast, non-state-owned enterprises lack rigid carbon assessment constraints and institutional support, exhibiting weaker foundations and lower initial disclosure levels. By enhancing carbon disclosure intensity, they can effectively communicate incremental signals of green transformation and sustainability to capital markets, attract ESG investors, alleviate financing constraints, and thereby strengthen innovation incentives. Group analysis demonstrates that compared to state-owned enterprises subject to strict carbon assessments with standardized compliance-driven disclosures, increased carbon disclosure intensity yields more pronounced benefits for innovation-driven development among non-state-owned textile and apparel firms (

Table 6. Heterogeneity Analysis Results of Ownership Nature.

Variable	State-Owned Enterprises	Non-State-Owned Enterprises
Score	−0.005	0.030 ***
	(0.014)	(0.008)
Constant term	−1.010	0.384
	(2.431)	(1.147)
Controlled variable	YES	YES
Sample Size	70	341
Individual Effect	YES	YES
Time Fixed Effects	YES	YES

Standard errors in parentheses: *** p < 0.01.

).

4.4.2. Enterprise Size

Firm size reflects differences in resource endowments, governance capabilities, and firms’ capacity to adapt to external environmental pressures. To examine whether the relationship between carbon information disclosure intensity and corporate innovation varies with firm size, we first classified firms according to China’s official Statistical Classification Criteria for Large, Medium-sized, Small and Micro Enterprises. As the sample consists of textile and apparel manufacturing firms, the classification criteria for industrial enterprises were applied. Given the relatively limited sample size, further dividing the sample into all individual size categories would result in insufficient observations in some subsamples and could reduce the statistical power and stability of the subgroup regressions. Therefore, large and medium-sized enterprises were combined into one group, while the remaining smaller enterprises were included in the small-enterprise group. This grouping balances the economic differences associated with firm size with the need to maintain sufficient observations for reliable subgroup estimation. Large and medium-sized enterprises generally possess relatively stronger resource endowments, financing capacity, governance capabilities, and innovation capacity, whereas smaller enterprises tend to face greater resource and financing constraints. The regression results are reported in

Table 7. Results of Enterprise Size Heterogeneity Analysis.

Variable	Large and Medium-Sized Enterprises	Small Enterprises
Score	0.008	−0.005
	(0.010)	(0.022)
Constant term	−0.386	1.523
	(1.262)	(1.389)
Controlled variable	YES	YES
Sample Size	293	118
Individual Effect	YES	YES
time effect	YES	YES

.

Specifically, industrial enterprises are classified according to both the number of employees and operating revenue. Large enterprises have at least 1000 employees and operating revenue of at least RMB 400 million; medium-sized enterprises have 300–999 employees and operating revenue of RMB 20–400 million; and small enterprises have 20–299 employees and operating revenue of RMB 3–20 million. Firms must satisfy the lower-bound requirements for both indicators; otherwise, they are classified into the next lower category.

Table 7 shows that, in the large and medium-sized enterprise subsample, the estimated coefficient for carbon information disclosure intensity (Score) was 0.008, indicating a positive trend; however, it failed to pass the significance test. This suggested that a higher level of carbon information disclosure intensity (Score) might have been positively associated with innovation activities, yet the available evidence was insufficient to establish a statistically significant effect. One possible explanation for the positive but statistically insignificant coefficient is that large and medium-sized enterprises may be better positioned to translate the monitoring and reputational benefits associated with information disclosure into innovation incentives. However, this potential effect may be weakened by complex organizational structures, longer decision-making processes, and organizational inertia.

In the small enterprise subsample, the estimated coefficient for carbon information disclosure intensity (Score) was −0.005, displaying a negative trend, but likewise did not reach statistical significance. One possible explanation for the negative but statistically insignificant coefficient is that smaller enterprises may face greater financing constraints, resource shortages, and limited technological accumulation. Moreover, enhancing carbon information disclosure intensity (Score) entailed costs associated with information compilation, auditing, and compliance, which could have crowded out limited innovation resources. Additionally, small enterprises prioritised short-term survival, whereas the financing improvements and governance benefits brought by information disclosure were characterised by long-term horizons and time lags, making them insufficient to offset current-period costs. Regulatory constraints may also have compressed the space available for high-risk innovation. As the variation in carbon information disclosure intensity (Score) among small enterprises was relatively limited, and as standardised disclosure may have alleviated some financing pressures, this negative effect remained weak and did not achieve significance.

Overall, the coefficient of Score is positive in the large and medium-sized enterprise group and negative in the small-enterprise group, but neither coefficient is statistically significant. These results suggest that the current subgroup regressions do not provide sufficient statistical evidence of a pronounced firm-size-specific effect of carbon information disclosure intensity on corporate innovation.

4.4.3. Subsectors

According to the National Economic Industry Classification Standards, the textile and apparel industry encompasses three subsectors: the textile industry(C17), the apparel and accessories sector(C18), and the chemical fiber manufacturing sector(C28). The impact of carbon disclosure on innovation behaviors may vary across these subsectors. To empirically examine this difference, this study reestimated parameters using samples of listed companies categorized by subsector from the “Industry Classification Results for Listed Companies in the Third Quarter of 2021.”

Table 8. Results of heterogeneity analysis by industry segment.

Variable	Clothing and Apparel Industry (C18)	Chemical Fiber Manufacturing Industry (C28)
Score	0.010	0.019 **
	(0.017)	(0.008)
constant term	0.547	1.529 ***
	(0.343)	(0.427)
controlled variable	YES	YES
Sample Size	250	177
Individual Effect	YES	YES
time effect	YES	YES

Note: C17 is excluded due to insufficient sample size (N = 9). Standard errors in parentheses: *** p < 0.01, ** p < 0.05.

reveals that compared to enterprises in the apparel and accessories sectors, chemical fiber manufacturers exhibit a significantly stronger positive correlation between carbon disclosure intensity and innovation behavior. This finding stems from the fact that the textile industry as a whole—particularly in processes such as dyeing and finishing—generates higher energy consumption and pollution emissions. Conversely, this result demonstrates that proactive carbon disclosure enables companies in this sector to more effectively address societal concerns, enhance corporate image, and create greater market value. These findings provide theoretical support and actionable insights for refining carbon disclosure practices to drive innovation improvements across different industry segments. Given the extremely limited number of observations (N = 9, N denotes the number of observations.), the textile industry (C17) is not included in the heterogeneity regression analysis and is not used for inferential interpretation.

5. Conclusions and Implications

5.1. Research Conclusion

Based on data from A-share textile and apparel companies listed on the Shanghai and Shenzhen stock exchanges in China from 2012 to 2024, this paper empirically examines the impact of carbon disclosure intensity on innovation behavior in these enterprises and its underlying mechanisms through theoretical analysis. The main conclusions are as follows: An increase in carbon disclosure intensity significantly enhances corporate innovation behavior among textile and apparel firms; Mechanism analysis shows that investor attention exerts a negative moderating effect, rather than a mediating role, on the relationship between carbon disclosure intensity and innovation behavior. Additionally, heterogeneity analysis reveals that carbon disclosure intensity has a more pronounced value-enhancing effect on non-state-owned enterprises and chemical fiber manufacturing firms. Firm size does not significantly moderate the relationship between carbon disclosure intensity (Score) and corporate innovation.

5.2. Policy Recommendations

Based on the above conclusions, this paper proposes the following policy recommendations: (1) Optimize the carbon disclosure system to strengthen positive incentives for innovation in textile and apparel enterprises. The government should implement comprehensive policy measures that balance incentives and regulatory constraints, effectively reducing corporate carbon disclosure costs while significantly enhancing disclosure intensity. These measures will improve transparency in the industry, boost public trust, and ultimately enhance overall competitiveness. (2) Strengthen support for green technological innovation and improve carbon disclosure requirements, which will positively drive innovation growth in the sector. Enterprises should be incentivized to increase investment in green technology R&D and stimulate innovative vitality. The government may utilize fiscal policies to guide companies in advancing green technology development and application, thereby elevating their overall value. (3) Guide investors to adopt a rational perspective and mitigate the negative impact of excessive focus on corporate innovation. Through investor education initiatives by regulators and information dissemination by industry associations, investors should be encouraged to recognize the “long-term value dimensions” of corporate carbon disclosures (e.g., low-carbon R&D, green supply chain transformation) rather than focusing solely on short-term compliance, thus preventing short-sighted practices. (4) Implement differentiated policy support measures: provide tax incentives and financial subsidies for non-state-owned enterprises and small-to-medium businesses; impose stricter regulatory standards for these entities; and establish unified industry standards for chemical fiber manufacturers to elevate overall disclosure practices across the sector.

5.3. Research Limitations and Prospects

It should be noted that this study still has certain limitations that warrant further exploration in future research: (1) With the rapid development of the digital economy, digitalization may significantly influence corporate carbon information disclosure and innovation behaviors. Due to constraints related to thematic relevance, this paper did not examine the impact of digitalization on corporate innovation behaviors. Future studies could incorporate digital elements such as data factors or digital technologies to broaden the scope of research on innovation behavior in textile and apparel enterprises. (2) This study only evaluated the mechanism by which carbon information disclosure intensity affects innovation behavior in textile and apparel enterprises using investor attention as a metric; subsequent research could explore this relationship from other perspectives, such as financing costs and environmental regulations.