Impact of Digital Transformation on Carbon Performance of Industrial Firms Considering Performance–Expectation Gap as a Moderator

Abstract

The swift advancement of the industrial economy has depleted resources and degraded the environment, hindering global economic growth. Digitalization provides a novel approach to decrease carbon emissions and enhance the environment. This study utilized panel data from 2012 to 2021 of listed A-share industrial enterprises as the research sample. It employed suitable measures to assess digitalization and corporate carbon performance. Furthermore, a double fixed-effects regression model was constructed to examine the correlation between digitalization and corporate carbon performance. The findings indicate that digital transformation and corporate carbon performance varied widely across different firms, but there was notable overall progress. Adopting digital transformation in the industrial sector had a substantial and favorable effect on enterprises’ carbon performance. This effect remained substantial despite multiple robustness tests. An examination of the mechanisms involved indicated that digital transformation enhances the carbon performance of industrial sector enterprises by improving the clarity and accessibility of company information. Corporations may intentionally seek difficulties and take strategic risks due to performance–expectation discrepancies. Due to the digital transformation, this behavior may improve the carbon performance of listed industrial businesses. The carbon performance of industrial businesses after digital transformation depends on elements like property rights, market rivalry, industry pollution, and capital investment.

1. Introduction

Since the mid-20th century, the issue of rapid industrial economic growth causing resource depletion and environmental damage has become even more pronounced. Nevertheless, the advancement of the industrial economy is both the foundation for economic stability and the crucial factor in establishing an “industrial powerhouse”. Hence, it is imperative to consider environmental advantages and advocate for converting conventional industries into environmentally friendly and low-emission production methods while striving for economic growth. In 2020, China’s Gross Domestic Product (GDP) surpassed CNY 100 billion for the first time, while its carbon emissions nearly hit 10 billion tons, accounting for 27.92% of the global total. Considering the environmental consequences of socio-economic progress and the obligation to decrease carbon emissions in light of climate change, Xi Jinping suggested at the 75th session of the UN General Assembly that “carbon dioxide emissions should aim to reach their highest point by 2030 and make endeavors to achieve carbon neutrality by 2060”. In addition to its “dual-carbon” objective, China has implemented a series of robust steps to address the escalating environmental issues and enhance the quality of the ecological environment. In 2021, the national carbon market initiated online trading in the power generation sector, a significant milestone in achieving China’s dual-carbon objective and managing overall carbon emissions. The 20th National Congress of the Communist Party of China (CPC) report reiterated the significance of environmental control. It promoted the establishment of green and low-carbon methods of production and lifestyles. China acknowledges that progress encompasses not only economic expansion but also the enhancement of environmental conditions. Nevertheless, the intricate nature of environmental problems presents numerous obstacles to effective governance. The transition towards environmentally friendly and low-carbon industries, which play a crucial role in economic growth and account for a significant portion of energy usage, is of utmost importance. Addressing the urgent need to reduce carbon emissions from the industrial sector and promote industrial energy conservation and emission reduction has become crucial.

In recent years, with the continuous progress of information technology and the arrival of the digital technology revolution, digitization has rapidly swept through all aspects of society and become an effective development path [1]. Digitalization facilitates energy consumption optimization to achieve supply–demand equilibrium and substantially mitigates carbon emissions. It fosters the transition of the economic development paradigm from “black growth” to “green growth”, leading to continuous enhancement of the quality of the ecological environment and ultimately green and low-carbon emissions [2]. The goal is to achieve green, low-carbon, and sustainable development by improving the quality of the ecological environment. Emerging digital technologies such as big data, the Internet of Things, and artificial intelligence are gradually being integrated into numerous economic and social sectors. This integration is helping to significantly improve and adjust the structures of different industries. Enterprises have incorporated digital transformation into their production processes, resulting in the digitization and timely recording of the production processes. This information is visibly shown to internal managers. Furthermore, digitizing financial management not only enhances the transparency of financial information but also facilitates its accessibility to shareholders. Based on the stakeholder and information asymmetry theories, digital transformation will significantly decrease stakeholders’ access challenges, enhancing the effectiveness of information transfer and leading to stronger external oversight of managers and reductions in potential moral hazards and adverse selection problems [3,4]. The performance–expectation gap necessarily influences the beneficial strategy change known as digital transformation. According to the theory of corporate behavior, decision-makers are only partially rational while making strategic decisions. Instead of seeking the optimal outcome, they prioritize satisfaction in the decision-making process. The consequences of performance feedback can influence the risk preferences of decision-makers. When a company’s performance falls short of its goals, managers may engage in risky behaviors to return its performance to the targeted level, impacting carbon performance.

This study focused on 1172 industrial enterprises that are publicly listed. The carbon performance of these firms was determined by considering parameters such as industry operating costs, industry carbon emissions, enterprise operating costs, and enterprise operating revenues. Additionally, a Python 3.9 crawler was used to measure the level of digital transformation within these enterprises. This study aimed to analyze many facets of this topic. The relationship between enterprise digital transformation and enterprise carbon performance was thoroughly examined through benchmark regression analysis. Simultaneously, the variable of corporate information transparency was incorporated as a mediating factor to establish a model that investigated whether corporate digital transformation can influence corporate carbon performance through its effect on corporate information transparency. Based on the principle of finite rational choice, the performance–expectation gap can have a favorable influence on digital transformation and its connection with carbon performance. This study investigated the potential impact of performance–expectation fallout as a moderating factor. This was performed by developing a model of the moderating mechanism and considering the industry and the year as control variables. Furthermore, companies were categorized into various types depending on whether they were highly polluting, the degree of market competition, the degree of capital intensity, and the nature of the ownership of the firms. This classification helped us examine how differences among companies impact the connection between digitalization and their carbon performance. This study aimed to evaluate the validity of three hypotheses: H1 proposes that the process of corporate digital transformation has a beneficial impact on corporate carbon performance. H2 proposes that digital transformation can decrease corporate carbon emissions and thus enhance carbon performance by improving the transparency of corporate information. H3 proposes that the performance–expectation gap has a beneficial moderating effect on the connection between digital transformation and the carbon performance of companies in the industrial sector.

This paper’s findings demonstrate digital transformation’s substantial and favorable impact on industrial organizations’ carbon performance. Compared with previous studies, this study was innovative in the following ways: (1) This study calculated corporate carbon performance based on industry operating costs, industry carbon emissions, corporate operating costs, and corporate operating revenues and used a Python crawler to measure the level of corporate digital transformation, facilitating in-depth analysis of the relationship between enterprise digitization and corporate carbon performance. (2) This research incorporated information transparency into the research framework, which enhanced the understanding of how digital transformation impacts enterprise carbon performance. It also revealed the mechanism of the “dark box” between digital transformation and enterprise carbon performance. (3) Previous studies have not investigated whether the relationship between digital transformation and corporate carbon performance is affected during periods of declining corporate performance. By introducing the performance–expectation gap as a moderating variable, this study further defined the impact of digital transformation on carbon performance. This enriched the research on the relationship between digital transformation and performance feedback.

2. Literature Review

2.1. The Literature on Digital Transformation

The evolution of the digitalization concept can be categorized into three main phases: informatization, digitalization, and digital transformation [5]. There has yet to be a consensus within the academic community on the definition of enterprise digitization [6]. Digitization is commonly defined as the utilization of digital technology, but in a broader sense, it refers to the complete restructuring of businesses by implementing digital technology. This study provides a precise definition of the concept of digital transformation. Digital transformation refers to the process in which a firm implements a series of comprehensive digital upgrades to its internal management, production, operation, and other aspects using digital technology. This results in the construction of a new business model. The growing utilization of digital technology and the rapid development of the digital economy have made digitalization an increasingly popular subject. The digitization process is being progressively implemented across several domains and has substantially influenced diverse sectors. At a small scale, industries and organizations are implementing digital technology and promoting digitalization to react to external changes and enhance their overall competitiveness when accomplishing the “dual-carbon” goal. This shift has revolutionized the conventional business model and significantly enhanced energy utilization efficiency, directly or indirectly reducing carbon emissions from industries and organizations and improving carbon performance. Digitalization has become the best tool to reduce carbon emissions and achieve China’s “dual-carbon” goal [2]. Scholars have different ways of measuring enterprise digitization, and their methods can be broadly divided into two types. One is the direct measurement method represented by multidimensional comprehensive evaluation. The primary purpose of this method is to divide digitalization into different levels and select different secondary indicators for each level, which are used as the basis for comprehensive evaluation. The comprehensive assessment is then analyzed according to the chosen model, resulting in a more reliable indicator [7,8]. The other is to select proxy variables that are directly related to digitization. Compared with the comprehensive indicator evaluation method, this method does not need to divide digitization into multiple indicator dimensions, making it much less difficult, and the data are also easy to obtain [9,10]. Doaa Shohaieb et al. [11] utilized the Eikon database to gather financial data on non-financial companies that are part of the FTSE Composite Index in the United Kingdom. Doaa Shohaieb et al. utilized the Corporate Financial Information Environment (CFIE) software developed by Lancaster University to evaluate the narrative disclosure of diversity management (DMD) in annual reports. This automated content analysis method was employed to extract and score the relevant narrative component. In this research, we opted to employ text analysis as a substitute for gauging the extent of digital transformation at firms. We accomplished this by quantifying the frequencies of specific keywords in the annual reports of publicly traded corporations. The annual reports of A-share businesses listed from 2012 to 2021 were gathered utilizing Python technology. The outcome was a glossary of words consisting of five distinct categories: “Artificial Intelligence”, “Big Data”, “Cloud Computing”, “Blockchain”, and “Digital Technology Application”. Ultimately, the frequency of phrases linked to corporate digitalization was used as a proxy indicator to measure the extent of enterprise digital transformation.

2.2. The Literature on Carbon Performance

The quantification of carbon performance has evolved from a single indication to a comprehensive system of many indicators, showing gradual improvement over time. Due to the growing concern about global climate change and environmental protection, carbon performance has emerged as a crucial criterion for assessing the progress of organizations in adopting environmentally friendly practices and transitioning to a low-carbon economy. Carbon performance refers to the correlation between the carbon emissions generated by businesses or individuals during production and consumption and their economic gains. There is a limited number of pertinent studies conducted by international scholars on carbon performance. Kaya [12] first proposed the concept of carbon productivity, which she defined as the ratio of total carbon emissions to gross product. This measurement method is straightforward to comprehend and simple to use but ignores a region’s multidimensional characteristics, etc. [13]. Hoffmann and Busch [14] categorized carbon performance into physical and monetary domains. They also proposed four comprehensive and systematic metrics to measure corporate carbon performance: carbon intensity, carbon reliance, carbon exposure, and carbon risk estimation. As research progressed, scholars started directly measuring carbon performance using DEA evaluation, SFA models, three-stage DEA models, and SE-SBM [15,16,17,18]. Zhou Luliu and Wen Subin [19] developed a carbon performance evaluation approach called the “five-ring” model based on a balanced scorecard framework. An alternative method involves using proxy indicators, such as the ratio between firm revenue and carbon emissions, to assess corporate carbon performance [20,21]. This study used the corporate income to carbon emissions ratio as a metric to assess the carbon performance of corporations, taking into account the challenges associated with data acquisition. While the assessment criteria for carbon performance have evolved from single-indicator evaluations to multi-indicator comprehensive evaluations, they are still in the experimental phase. A fully developed and standardized evaluation system has not been established.

2.3. The Literature on the Influence of Digital Transformation on Carbon Performance

Academics have yet to reach a unanimous view on the impact of digital transformation on carbon emissions. The first view is that digital transformation can effectively improve energy efficiency, reduce energy intensity, and thus reduce carbon emissions [22,23]. With the development of the digital economy, digital technology-based applications will bring about multidimensional optimization, such as production technology enhancement and industrial structure adjustment, thus improving energy efficiency and reducing carbon emissions [2,24,25,26]. The second viewpoint is that digital transformation increases carbon emissions. Scholars holding this view believe that the implementation of digitalization in enterprises relies more on electronic devices and their accessories, which requires the introduction of a large number of digital devices. Digital devices require large amounts of energy for their proper functioning during their life cycles, which leads to a sharp increase in the demand for energy, resulting in higher energy intensity and an increase in corporate carbon emissions [27,28]. Asongu et al. (2018) showed that in an interactive regression, an increase in the penetration of digital technologies, especially the rapid development of information and communication technologies, increased per capita carbon emissions [29]. The third view is that there is a “U”-shaped relationship between digital transformation and carbon emissions. Although digital transformation reduces energy intensity, it can cause an energy rebound effect if a certain threshold is exceeded [30,31,32]. Miao Lun et al. (2022) found that the digital economy affects a city’s carbon emission level by influencing the city’s innovation efficiency and that the development of the digital economy and the carbon emission level have a nonlinear correlation, showing an inverted U-shaped relationship [33]. Xiao Renqiao et al. (2023) found a U-shaped relationship between the level of digital construction, application, and development and enterprise carbon performance [34]. The application of digital technology enables enterprises to better meet the requirements of green development. It can not only help enterprises change from the traditional production mode of high input, high output, high energy consumption, and high pollution to a low-carbon, energy-saving, and high-efficiency production mode but can also allow enterprises to improve their own production efficiency while reducing their adverse impacts on the environment, thus realizing sustainable green development.

2.4. The Literature on Corporate Information Transparency

Information disclosure can alleviate enterprises’ financial limitations and effectively monitor and deter management’s unethical profit-seeking practices, prompting organizations to prioritize sustainable development [35]. Based on the information asymmetry and signaling theories, enhancing information transparency can decrease the information imbalance between companies and external investors, allowing external stakeholders to access more comprehensive information resources [36]. According to the stakeholder theory, enhancing information transparency in a company reduces the monitoring costs for external stakeholders, which, in turn, substantially improves supervision intensity and efficiency [37]. Through external oversight, it is possible to mitigate management’s “hedonistic” mindset, minimize the undisclosed expenses incurred by executives while working for a company, and achieve cost savings [38], which typically helps generate more cash to address the financial limitations of a firm, mitigate the speculative behaviors of enterprises, and provide enterprises with greater opportunities for digital transformation [39]. This study examined the concept of information transparency by analyzing how information is disclosed in different phases, focusing on the viewpoint of information users. Based on the theory of information asymmetry and signaling, digital transformation can enhance the transparency of enterprise information disclosure, enabling stakeholders to access valuable information about an enterprise easily and ultimately improving carbon performance. This paper utilized research conducted by Xin Qingquan et al. [40] to develop comprehensive indicators for measuring information transparency. The aim was to reduce measurement error and obtain more reliable research results, and the specific indicators are presented in Section 4.2.

2.5. The Literature on Performance–Expectation Gaps

Limited rationality leads decision-makers to simplify the process by comparing actual and predicted enterprise performance. According to this comparison, they categorize a company’s activity as either a “loss” or a “gain” [41]. The expectation level quantifies the extent to which an organization experiences a decrease or increase in value. It represents the minimum degree of managerial contentment and is crucial in challenging and modifying corporate practices [42]. Managers establish objectives or expectations and implement organizational restructuring and strategic changes based on the disparity between their business’s performance and their goals. Based on corporate behavior theory, firms with unfulfilled expectations aggressively strive to enhance their businesses or reverse their declines [41]. Firms engage in the “problem search” paradigm by evaluating different business procedures to enhance performance [43,44]. Underperforming is considered a “failure” for a company. Managers are motivated by loss aversion to seek solutions for company difficulties and enhance performance [45]. The underperformance of individuals or organizations drives decision-makers to actively pursue remedies and embrace calculated risks in their strategic choices [46]. Digital transformation is a proactive process that involves reevaluating strategy and structure in response to digital technology [47]. This strategy may enhance organizational performance [48]. Digitalization enables firms to grow and generate profits. Enhancing performance, rectifying poor practices, and enhancing operations are crucial organizational strategies [49]. Resource limitations cause companies to exercise greater caution when making strategic decisions. In cases where operational performance falls below an anticipated target, a company will sell its assets to avoid engaging in mergers or acquisitions that require significant resources, particularly if the gap between expected and actual performance widens [50]. Unfavorable performance evaluations negatively impact organizations’ reputations. When a company’s performance deteriorates over time, stakeholders raise concerns about its effectiveness and dependability. A negative performance evaluation has a detrimental impact on a company’s reputation in society. In order to regain trust and confidence, a business needs to adopt and execute strategic enhancements [51]. Managers derive satisfaction from surpassing expectations since they are cautious and reluctant to embrace change. They may not have the urge to engage in novel experiences or demonstrate creativity [52]. Managers may use past experiences and inertia to make decisions and may oppose strategic changes when they face high expectations [53]. The most successful organizations are content with their present circumstances, capitalizing on obstacles that hinder new competitors from entering the market, and they exhibit a reduced inclination to alter their strategies. Excessive performance expectations weaken the principal–agent dilemma.

2.6. Research Gap

Based on a review of the literature, pertinent studies have been conducted that demonstrate the relationship between digital transformation and carbon performance; nonetheless, several limitations remain, primarily in the following areas: (1) Previous research primarily examined the relationship between these two factors through the lens of internal control, green innovation, and information disclosure quality. However, further investigation is required to widen the analysis of their internal impact mechanism. (2) The evaluation techniques have been constrained. The existing approaches employed to evaluate the extent of digital transformation within a company are quite uniform, which might result in biased evaluation outcomes, particularly when corporate disclosure is incorrect. Hence, creating more thorough and multidimensional assessment techniques is imperative to precisely gauge the true impact of digital transformation. (3) Carbon emissions solely consider environmental advantages, while carbon performance aims to decrease carbon emissions while maintaining stable economic benefits [54], highlighting the dual growth of economic and environmental advantages. A body of research has examined how to address the carbon problem through digitization. However, most of this research has primarily investigated the effect on corporate carbon emissions rather than corporate carbon performance [55]. Furthermore, the majority of this research has followed a linear approach [56], which has not yet fully uncovered the intricate and comprehensive relationship between the level of digitization and corporate carbon performance. (4) Despite the significant potential for digital transformation to improve carbon performance, there is still uncertainty surrounding the requisite circumstances for effectively harnessing the empowering impacts of this transition in actual applications. What are the essential prerequisites for properly harnessing the empowering impacts of digital transformation? To effectively harness the promise of digital transformation, firms must consider these factors and establish precise plans to facilitate a seamless transition and enhance carbon performance. This study added corporate information transparency and the performance–expectation gap to the research framework to investigate if digital transformation can improve corporate carbon performance levels. This study aimed to enhance and broaden the current body of knowledge on the mechanisms involved in digital transformation, elucidating the precise influence of digital transformation on carbon performance in the context of a performance–expectation gap. It helps companies minimize pollution and boost competitiveness for green and sustainable development. The relevant authorities will receive specific policy assistance and guidance on digital development and environmental protection.

3. Theoretical Analysis and Research Hypothesis

3.1. The Enhancement Effect of Digital Transformation on Corporate Carbon Performance

The fundamental principles of the resource-based theory encompass heterogeneity, irreplaceability, permanence, and scarcity of resources. According to the resource-based approach, firms consistently enhance their utilization of digital technology to efficiently integrate and leverage green resources and knowledge [57], thereby fostering green innovation within organizations. Consistently improving the degree of green innovation ultimately aids firms in achieving sustainable development [58]. The resource-based theory suggests that enterprises can gain a sustainable competitive advantage by utilizing resources that are valuable, rare, irreplaceable, and difficult to imitate, such as advanced technologies [59]. Technological resources play a crucial role in enabling organizations to enhance their existing technology and bolster their research and development (R&D) and innovation skills [60]. Karim A. E et al. discovered that companies allocating more funds to capital expenditures tend to communicate more extensively with their stakeholders. This strategic approach is used to gain a competitive edge in the market [61]. Companies are expected to increase their disclosure of carbon-related information to convey a favorable image to stakeholders, enhance credibility, establish a strong reputation, and gain a competitive edge. This paper examines the resource-based view and asserts that enterprise digital transformation can facilitate sustainable development by effectively utilizing and integrating valuable and irreplaceable digital technological resources. It also emphasizes the importance of enhancing original production and management technologies and improving an enterprise’s green innovation capability to establish a long-term competitive advantage. To summarize, this paper puts forward the following hypothesis:

H1. 

enterprise digital transformation has a positive impact on enterprise carbon performance.

3.2. The Mediating Effect of Corporate Information Transparency

Implementing enterprise digital transformation can facilitate the exchange of information within and outside an organization, improve the visibility of corporate information, significantly enhance information quality, and expedite the availability of information. Specifically, it enhances the clarity of non-financial information, enabling investors and analysts to have a more comprehensive understanding of a company’s status [62]. According to the information asymmetry and signaling theories, Moussa et al. [63] proposed that companies utilize ESG and carbon disclosure to convey their dedication to sustainability and responsible practices. This communication influences the market’s expectations and perceptions regarding a company’s prospects and performance. Effective transparency of external information accurately mirrors a firm’s internal production situation. Simultaneously, the implementation of new technologies and patents in internal production, as well as investments in environmental protection, are transparently communicated to the public, thereby impacting the rating choice [64,65]. Furthermore, firms that disclose environmental information well lower the chance of a stock market meltdown and are more likely to be preferred by investors [66]. Hence, digital transformation enhances the transparency of corporate carbon performance, thereby incentivizing corporations to more effectively fulfill their environmental obligations and attain mutually beneficial outcomes with both ecological and economic advantages. Companies encounter pressure from multiple stakeholders (the government, the community, and the media) concerning external monitoring. Enhancing corporate information transparency will subject corporations to increased external scrutiny. Companies will prioritize environmental issues and actively fulfill their social duties, thus enhancing their carbon performance. This approach considers the importance of social reputation, brand influence, and the personal images of executives. Consequently, following a firm’s adoption of digital transformation, its information becomes more visible. As a result, a company becomes more proactive or reactive when implementing environmental governance measures, enhancing its carbon performance compared to the past. To summarize, this paper puts forward the following hypotheses:

H2a. 

digitization can improve the transparency of company information.

H2b. 

digitization can enhance carbon performance by enhancing corporate information transparency.

3.3. The Moderating Effect of the Performance–Expectation Gap

An enterprise’s strategic decision-making is susceptible to the impact of the performance–expectation gap. When actual performance fails to meet expectations, it enhances an enterprise’s drive for strategic change and further accelerates its digital transformation. During challenging business circumstances, enterprises engage in problem-driven organizational searches and tend to take risks. To overcome these obstacles, managers will stray from their original plans and take proactive steps to improve their situations [67]. When the actual business performance exceeds the expected performance, positive feedback encourages an organization to continue following its established strategic path, which can result in a mindset of prioritizing stability over innovation, reducing efforts to identify and address problems and favoring less risky actions [68,69]. However, this condition is not beneficial for implementing a digital transformation strategy. To summarize, this paper puts forward the following hypothesis:

H3. 

the performance–expectation gap has a positive moderating effect on the relationship between digital transformation and carbon performance in industrial sector companies.

4. Data Sources and Descriptive Statistical Analysis

4.1. Sample Selection and Data Sources

This study took listed industrial enterprises from 2012 to 2021 as the research sample. In order to avoid the influence of abnormal data, the raw data were subjected to the following processing steps: (1) enterprises that transitioned from industrial to non-industrial sectors during the observation period were removed; (2) enterprises that received warnings about the risk of being delisted were excluded; (3) samples that were missing data were eliminated; and (4) the variables were adjusted by decreasing and increasing them by 1% to mitigate the influence of extreme values on the regression outcomes. After processing, 1172 companies and 11,720 data points were obtained. The reference data were sourced from the China Statistical Yearbook, the China Energy Statistical Yearbook, the Cathay Pacific Database (CSMAR), and annual financial reports of firms obtained from the Juchao Information Network. In an analysis of heterogeneity, the identification of heavily polluting industries was primarily based on the Guidelines for Industry Classification of Listed Companies revised by the China Securities Regulatory Commission in 2012, the Management Directory for Classification of Listed Companies in Environmental Verification Industries formulated by the Ministry of Environmental Protection in 2008 (Huanban Letter [2008] No. 373), and the Guidelines for Environmental Information Disclosure of Listed Companies (Huanban Letter [2010] No. 78). These guidelines encompass 16 heavily polluting industries, such as coal, mining, textiles, tanning, papermaking, petrochemicals, pharmaceuticals, chemicals, metallurgy, and thermal power [70].

4.2. Selection of Variables

4.2.1. Dependent Variable

The explanatory variable was corporate carbon performance (CP), which can be defined as the ratio of corporate operational revenue to carbon emissions. Due to the absence of a comprehensive monitoring system for corporate carbon emissions in China, it is currently impossible to directly access corporate carbon emissions data. Therefore, this study relied on research conducted by Yan Huahong et al. [20] and Zhao Yuzhen et al. [21] to estimate corporate carbon emissions. Estimation was based on factors such as an industry’s carbon emissions and operating cost and an enterprise’s operating cost and income. An industry’s carbon emissions were determined by its energy consumption data published in the National Statistical Yearbook multiplied by the carbon emissions trading website’s listed carbon emission coefficients. The formula for calculating carbon performance was as follows:

$$\mathrm{C}\mathrm{a}\mathrm{r}\mathrm{b}\mathrm{o}\mathrm{n} \mathrm{P}\mathrm{e}\mathrm{r}\mathrm{f}\mathrm{o}\mathrm{r}\mathrm{m}\mathrm{a}\mathrm{n}\mathrm{c}\mathrm{e}={\displaystyle \frac{\mathrm{B}\mathrm{u}\mathrm{s}\mathrm{i}\mathrm{n}\mathrm{e}\mathrm{s}\mathrm{s} \mathrm{o}\mathrm{p}\mathrm{e}\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{n}\mathrm{g} \mathrm{i}\mathrm{n}\mathrm{c}\mathrm{o}\mathrm{m}\mathrm{e}}{\frac{\mathrm{I}\mathrm{n}\mathrm{d}\mathrm{u}\mathrm{s}\mathrm{t}\mathrm{r}\mathrm{y} \mathrm{C}\mathrm{a}\mathrm{r}\mathrm{b}\mathrm{o}\mathrm{n} \mathrm{E}\mathrm{m}\mathrm{i}\mathrm{s}\mathrm{s}\mathrm{i}\mathrm{o}\mathrm{n}\mathrm{s}}{\mathrm{I}\mathrm{n}\mathrm{d}\mathrm{u}\mathrm{s}\mathrm{t}\mathrm{r}\mathrm{y} \mathrm{o}\mathrm{p}\mathrm{e}\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{n}\mathrm{g} \mathrm{i}\mathrm{n}\mathrm{c}\mathrm{o}\mathrm{m}\mathrm{e}}\times \mathrm{B}\mathrm{u}\mathrm{s}\mathrm{i}\mathrm{n}\mathrm{e}\mathrm{s}\mathrm{s} \mathrm{o}\mathrm{p}\mathrm{e}\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{n}\mathrm{g} \mathrm{c}\mathrm{o}\mathrm{s}\mathrm{t}\mathrm{s}}}$$

(1)

4.2.2. Independent Variable

The primary independent variable was enterprise digital transformation (DCG). A study by Xiao Hongjun et al. [9,71,72,73] is cited in this paper. We used text analysis to estimate the frequency of matching keywords in the annual reports released by listed firms as a proxy indicator of the level of enterprise digital transformation. The precise sequence of actions was as follows: (1) Python technology was utilized to gather the annual reports of publicly traded A-share businesses spanning from 2012 to 2021. All textual information was extracted from these reports using the Java PDFbox module. This extracted content served as a data pool for further feature word screening. (2) A comprehensive lexicon of terminology related to digital transformation was created. First, we consulted the established scholarly literature on digital transformation to condense the essential terms associated with this concept [9,74,75,76]. Conversely, keywords were derived from significant policy documents about the digital economy that were published at the national level between 2012 and 2021. These keywords were used to enhance the existing feature thesaurus and ultimately create a terminology dictionary consisting of five categories: “Artificial Intelligence”, “Big Data”, “Cloud Computing”, “Blockchain”, and “Application of Digital Technology”. (3) The total occurrence of phrases linked to enterprise digitization was calculated. Since this type of data has a typical “right skewed” characteristic, the natural logarithm of the data plus one was used to construct an index of the degree of enterprise digital transformation.

4.2.3. Intermediate Variable

The intermediate variable was corporate information transparency (Trans). According to the methods proposed by Lang et al. [77] and Xin Qingquan et al. [40], the sample percentile ranks of whether the auditor was from a Big-Four accounting firm (BIG4), the number of analysts tracking the enterprise (ANALYST), the accuracy of the analysts’ surplus forecasts (ACCURACY), the disclosure assessment scores of Shenzhen-listed firms (DSCORE), and the surplus quality index (DD) were used to find the mean value of corporate transparency.

The first transparency indicator was the surplus quality indicator (DD) calculated based on the adjusted Dechow and Dichev [78] model. To calculate this variable, we first used the following model for an industry and a year:

$${TCA}_{i,t}={\alpha }_0+{\alpha }_1{CFO}_{i,t-1}+{\alpha }_2{CFO}_{i,t}+{\alpha }_3{CFO}_{i,t+1}+{\alpha }_4{∆REV}_{i,t}+{\alpha }_5{PPE}_{i,t}+e_{i,t}$$

(2)

TCA is the total current accrued profit, which equals operating profit minus operating cash flow plus depreciation and amortization expenses. CFO is the operating cash flow (from the statement of cash flows). $∆\mathrm{R}\mathrm{E}\mathrm{V}$ refers to the amount of change in the operating income. PPE refers to the value of fixed assets at the end of the year. Respectively, i and t stand for the firm and the year, and e refers to the error term. All the above variables were divided by the average total assets for deflation. After regressing model (2) on “industry-year” groups, we could obtain the regression residuals ($e_{i,t}$, the manipulated accrued profit for the year) for each firm for each year. Then, we calculated the standard deviation of the five values based on the regression residuals of year t and the previous four years and thus obtained the surplus quality indicator (DD) of the enterprise in year t. To facilitate comparisons with other transparency indicators, we multiplied this indicator by −1. As a result, the larger the DD, the higher the quality of the surplus.

The second indicator was the Shenzhen Stock Exchange’s disclosure assessment score (DSCORE) for companies listed in Shenzhen each year. The assessment results of listed companies’ disclosures were divided into four grades of A, B, C, and D (i.e., excellent, good, passing, and failing) based on the quality of the listed companies’ disclosures from high to low, which were publicly disclosed on the website of the SZSE. Through manual collection, we obtained the disclosure index (DSCORE) of SZSE companies for each year, the value of which ranged from 1–4 points, where the larger the score, the better the quality of the information disclosure. The third and fourth indicators were the number of analysts tracking an enterprise (ANALYST) and the accuracy of the analysts’ surplus forecasts (ACCURACY), respectively. ANALYST was expressed as the number of analysts who made forecasts of a company’s annual surplus in that year. The actual EPS was subtracted from the median EPS forecast by different analysts for the same year and divided by the previous year’s stock price per share. The resulting value was taken as an absolute value and multiplied by −1. This value was used to express ACCURACY. The fifth transparency indicator was whether a company hired a Big-Four firm as the auditor of its annual report in that year (BIG4).

4.2.4. Moderating Variable

The moderating variable was the performance–expectation gap. The expectation level (A) consisted of historical expectations (HA) and social expectations (SA) [42]. For the measurement of historical expected performance (${HA}_{i,t}$), this study referred to Chen et al.’s [46,79,80] research method and used the classical recursive metric formula to explore the gap between actual and expected performance. The independent variable was taken to be lagged by one period relative to the dependent variable. The specific formula was as follows:

$${HA}_{i,t}={\alpha }_{1}H{A}_{i,t-1}+(1-{\alpha }_1)P_{i,t-1}$$

(3)

In particular, ${HA}_{i,t}$ and ${HA}_{i,t-1}$ denote the historical expected performance level of firm i in year t and year t−1. $P_{i,t-1}$ denotes the actual performance level of firm i in year t − 1. ${\alpha }_1$ denotes the weight of the historical expected performance ($H{A}_{i,t-1}$) in year t − 1, which has a range of [0, 1]. Calculations showed that an ${\alpha }_1$ of 0.6 could produce a log-likelihood value that represented the best model fit. In this study, the average performance of the other firms in a firm’s industry was used as the socially desirable performance (${SA}_{i,t}$). After determining ${HA}_{i,t}$ and ${SA}_{i,t}$, the expected level of performance ($A_{i,t}$) of firm i in year t − 1 was calculated using Equation (4) and the weighted average method.

$$A_{i,t}={\alpha }_2{HA}_{i,t}+(1-{\alpha }_2){SA}_{i,t}$$

(4)

In this equation, ${\mathsf{\alpha }}_2$ was determined in the same way as ${\mathsf{\alpha }}_1$, and the highest log-likelihood value of the model was produced when α₂ was 0.3, indicating that the best fit was achieved. Then, the gap between the firms’ actual performance (${\mathrm{P}}_{\mathrm{i},\mathrm{t}}$) and the desired level (${\mathrm{A}}_{\mathrm{i},\mathrm{t}}$) was calculated. It should be noted that ${\mathrm{A}}_{\mathrm{i},0}$ is the historical expected performance of company i in period 0, which was replaced by the actual performance in period 0. Meanwhile, the performance–expectation gap was truncated by drawing on the method of Wang Jing et al. [81]: ${\mathrm{I}}_1$ = 1 when ${\mathrm{P}}_{\mathrm{i},\mathrm{t}-1}$−${\mathrm{A}}_{\mathrm{i},\mathrm{t}-1}$ < 0, and ${\mathrm{I}}_1$ = 0 when ${\mathrm{P}}_{\mathrm{i},\mathrm{t}-1}$−${\mathrm{A}}_{\mathrm{i},\mathrm{t}-1}$ ≥ 0. $\mathrm{I}$ is a dummy variable indicating whether a firm faces a reduction in business expectations. To facilitate the subsequent empirical study, absolute values were calculated for the performance–expectation gap. A larger absolute value indicated a larger performance–expectation gap for a firm.

4.2.5. Control Variables

In order to reduce the impacts of omitted variables on the regression results, drawing on studies by He Fan and Liu Hongxia (2019) [48] and Yuan Chun et al. (2021) [10], the number of years a firm was listed (listage), gearing ratio (Lev), profitability (Mark), cash flow ratio (Cflow), return on assets (Roa), company size (Size), independent director ratio (Indir), two positions in one (Dual), and Nature of Ownership (Soe) were used as control variables in the regression model. Detailed variable definitions and calculations are shown in

Table 1. Definitions of variables.

Variable Type	Variable Name	Variable Symbol	Measurement Method
Dependent variable	Corporate carbon performance	CP	The ratio of operating income to carbon emissions
Independent variable	Corporate digital transformation	DCG	Logarithmization based on text analysis and word frequency statistics
Mediating variable	Corporate information transparency	TRANS	See variable measurement above.
Moderating variable	Performance–expectation gap	PEG	The absolute value of the difference between the actual performance and the desired level of performance. Otherwise taken as 0.
Control variables	Number of years the firm has been listed	Listage	Natural logarithm of the company’s IPO life
	Gearing ratio	Lev	Total liabilities/total assets
	Profitability	Mark	Net profit to total revenue
	Cash flow ratio	Cflow	Cash flow to total assets
	Firm size	Size	Natural logarithm of total assets of the enterprise at the end of the year
	Independent director ratio	Indir	The number of independent directors of the enterprise as a proportion of the total number on the board of directors
	Two positions in one	Dual	Taken as 1 if the chairman and general manager are the same person. Otherwise taken as 0.
	Nature of Ownership	Soe	Taken as 1 if the enterprise is state-owned. Otherwise taken as 0.
Dummy variable	Industry	Industry	According to the 2012 version of the Securities and Futures Commission (SFC) industry classification, the “B”, “C”, and “D” header codes of the industrial sector take two digits.
	Year	Year	Year dummy variable: 1 for the current year and 0 for other years

.

4.3. Model Setting

In order to study the relationship between enterprise digitization and enterprise carbon performance and test hypothesis 1, a regression model was set up as follows, drawing on existing research:

$${\mathrm{C}\mathrm{P}}_{\mathrm{i},\mathrm{t}}=\mathsf{\alpha }+\beta {DCG}_{i,t}+\sum \mathsf{\delta }\mathrm{C}\mathrm{O}\mathrm{N}\mathrm{T}\mathrm{R}\mathrm{O}\mathrm{L}\mathrm{S}+\mathrm{Y}\mathrm{e}\mathrm{a}\mathrm{r}+\mathrm{I}\mathrm{n}\mathrm{d}\mathrm{u}\mathrm{s}\mathrm{t}\mathrm{r}\mathrm{y}+{\mathsf{\epsilon }}_{\mathrm{i},\mathrm{t}}$$

(5)

In this equation, i represents individual listed companies, t denotes time, ${\mathrm{C}\mathrm{P}}_{\mathrm{i},\mathrm{t}}$ is the dependent variable, the core explanatory variable is ${\mathrm{D}\mathrm{C}\mathrm{G}}_{\mathrm{i},\mathrm{t}}$, and β is the regression coefficient of the core explanatory variable (digital transformation). $\mathrm{C}\mathrm{O}\mathrm{N}\mathrm{T}\mathrm{R}\mathrm{O}\mathrm{L}\mathrm{S}$ represents a series of control variables, which are factors other than the explanatory variables that affect the carbon performance of enterprises. The result of the Hausman test (p < 0.01) indicated that the fixed regression model was the optimal choice. In order to attenuate the endogeneity effect, a fixed-effect model was chosen for the regression and the industry ($\mathrm{I}\mathrm{n}\mathrm{d}\mathrm{u}\mathrm{s}\mathrm{t}\mathrm{r}\mathrm{y}$) and time ($\mathrm{Y}\mathrm{e}\mathrm{a}\mathrm{r}$) were fixed.

4.4. Descriptive Statistics

Table 2. Summary statistics.

Variables	Observations	Mean	SD	Min	Median	Max
CP	11,720	45.722	41.400	0.533	33.958	180.520
DCG	11,720	1.156	1.237	0.000	0.693	5.737
Lev	11,720	0.404	0.184	0.032	0.403	0.838
Listage	11,720	2.701	0.371	2.197	2.639	3.367
Mark	11,720	0.071	0.128	−1.121	0.063	0.485
Size	11,720	22.407	1.267	19.876	22.225	26.497
Cflow	11,720	0.055	0.061	−0.134	0.052	0.250
Indir	11,720	37.418	5.427	30.000	33.330	60.000
Dual	11,720	0.258	0.438	0.000	0.000	1.000
Soe	11,720	0.388	0.487	0.000	0.000	1.000

reports the results of the descriptive statistical analysis. The results show that the mean value of carbon performance (CP) was 45.72, the minimum value was 0.53, and the maximum value was 180.52, indicating that there was a large gap in carbon performance across the sampled companies. The mean value of digital transformation (DCG) was 1.156, the minimum value was 0, and the maximum value was 5.737. This suggests significant disparities in the level of digitization among the companies in the sample, with certain companies taking prominent roles in driving digital transformation in the industrial sector. The results of the control variables were basically consistent with the results of existing studies.

An analysis of the correlations between the variables is presented in

Table 3. Correlation analysis and multicollinearity test.

	CP	DCG	Lev	Listage	Mark	Size	Cflow	Indir	Dual	Soe	Mean VIF
CP	1
DCG	0.445 ***	1
Lev	−0.203 ***	0.011	1
Listage	−0.237 ***	−0.088 ***	0.330 ***	1
Mark	0.103 ***	−0.035 ***	−0.345 ***	−0.019 **	1
Size	−0.193 ***	0.084 ***	0.528 ***	0.375 ***	0.023 **	1
Cflow	−0.071 ***	−0.042 ***	−0.093 ***	0.069 ***	0.284 ***	0.143 ***	1
Indir	0.096 ***	0.084 ***	−0.024 ***	−0.053 ***	−0.022 **	−0.003	−0.007	1
Dual	0.146 ***	0.093 ***	−0.101 ***	−0.205 ***	0.013	−0.156 ***	−0.033 ***	0.131 ***	1
Soe	−0.232 ***	−0.119 ***	0.287 ***	0.560 ***	−0.037 ***	0.373 ***	0.007	−0.030 ***	−0.267 ***	1
VIF		1.60	1.88	1.70	1.39	2.05	1.21	1.06	1.13	1.72	2.44
Cells in the lower triangle report Pearson’s correlation coefficients, while cells in the upper triangle report Spearman’s rank correlations.

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

. There was a significant positive correlation between the level of digital transformation and carbon performance. This relationship was significant at the 1% level. The correlation coefficient between the control variable (profitability) and corporate carbon performance was positive and significant at the 1% significance level. There were negative correlations between the gearing ratio, the number of years a company had been listed, and the company’s size and the company’s carbon performance, and the correlation coefficients were significant at the 1% level. Furthermore, this study assessed the presence of multicollinearity in the regression model by employing variance inflation factors. The vif values of all the important variables did not exceed 10, which proved that there was no serious multicollinearity problem among the variables selected for this study.

5. Empirical Analysis and Test

5.1. Baseline Regression Results

First, without considering the effects of other variables on corporate carbon performance, only the impact of corporate digital transformation on corporate carbon performance was analyzed. Only digital transformation was regressed as an explanatory variable of firms’ carbon performance, controlling for the year and the industry. Column (1) in

Table 4. Impact of digital transformation on corporate carbon performance.

	(1)	(2)
	CP	CP
DCG	0.819 ***	0.889 ***
	(0.178)	(0.164)
Lev		−11.255 ***
		(1.126)
Listage		−3.735 ***
		(0.474)
Mark		27.107 ***
		(2.333)
Size		−0.516 ***
		(0.165)
Cflow		23.271 ***
		(2.803)
Indir		0.117 ***
		(0.029)
Dual		0.722 **
		(0.356)
Soe		−1.552 ***
		(0.349)
_Cons	−4.932 ***	16.504 ***
	(0.541)	(3.919)
N	11,720	11,720
Adj. R2	0.846	0.865
Industry	Yes	Yes
Year	Yes	Yes

Standard errors are provided in parentheses. ** p < 0.05, *** p < 0.01.

reports the regression results, and the regression coefficient of firms’ digital transformation (DCG) was significantly positive at the 1% level, indicating that firms’ digitalization could promote the level of firms’ carbon performance. Considering the robustness of the regression results, control variables were added to the model where the effects of the year and the industry were taken into account, and the results are shown in column 2. The results indicate that the estimated coefficient of DCG was 0.889, demonstrating a positive and statistically significant relationship at the 1% level. These findings suggest that the digital transformation of enterprises has a substantial beneficial influence on carbon performance, confirming the validity of premise 1. In the context of digital transformation, enterprises have easy access to information, and the efficiency of regulating and using resources is greatly improved. At the same time, digital transformation also brings certain external pressures to enterprises, forcing them to reduce their carbon emissions and enhance their good green images to cope with pressure from various stakeholders, which is consistent with hypothesis 1. An enterprise’s profitability was positively correlated with its carbon performance, and this was significant at the 1% level, indicating that the higher the profitability (Mark), the more the carbon performance of an enterprise could be improved. The coefficient of the gearing ratio (Lev) was significantly negative, which indicated that the higher the gearing ratio of an enterprise, the more unfavorable it was to improve its level of carbon performance. The coefficient of enterprise age was negatively significant at the 1% level, which indicated that compared with enterprises with more extended histories, enterprises with shorter histories had more green concepts and emphasized improving the efficiency of resource utilization so as to build up a green image and promote the improvement of their carbon performance level.

5.2. Robustness

This study employed several methods to validate the robustness of the results, including (1) replacing the core explanatory variable, (2) replacing the explanatory variable, (3) using carbon performance in the next period, (4) regressing over a reduced sample period, (6) regressing using a new sample, (7) using higher-order fixed effects, (8) adding control variables, and (9) using instrumental variables and PSM to mitigate potential endogeneity problems.

5.2.1. Replacing the Core Explanatory Variable

Based on previous research, an enterprise digitization measurement index was established by constructing a digitization vocabulary utilizing machine learning text analysis techniques and Python crawler technology. First, digitization keywords were extracted based on the semantic framework of national policies to create a comprehensive lexicon of terminology related to company digitization. Second, annual reports of listed industrial enterprises were obtained from the Juchao Information Network. Specifically, the section titled “Management Discussion and Analysis (MD&A)” was extracted and subjected to text analysis to determine the frequencies of digitized keywords. The Dig index, as defined by Xiao Tusheng et al. (2022) [82], was calculated by dividing the total frequency of the keywords by the length of the “MD&A” section and multiplying it by 100. This index served as a measure of the digital transformation of an enterprise, with a higher Dig value indicating a higher level of digital transformation. The regression results in column (1) of

Table 5. Robustness test.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
	CP	lnCP	F.CP	CP	CP	CP	CP	CP
DCG		0.015 ***	0.864 ***	0.813 ***	0.889 ***	0.866 ***	0.866 ***	0.885 ***
		(0.002)	(0.179)	(0.182)	(0.164)	(0.143)	(0.164)	(0.164)
Dig	0.505 **
	(0.241)
Lev	−11.345 ***	−0.195 ***	−13.173 ***	−13.860 ***	−11.255 ***	−11.206 ***	−11.059 ***	−11.325 ***
	(1.128)	(0.018)	(1.240)	(1.320)	(1.126)	(1.040)	(1.122)	(1.125)
Listage	−3.699 ***	−0.051 ***	−3.408 ***	−3.894 ***	−3.735 ***	−3.874 ***	−3.936 ***	−3.841 ***
	(0.474)	(0.008)	(0.538)	(0.539)	(0.474)	(0.496)	(0.485)	(0.485)
Mark	27.057 ***	0.509 ***	24.138 ***	20.216 ***	27.107 ***	26.743 ***	27.238 ***	33.142 ***
	(2.338)	(0.039)	(2.640)	(2.256)	(2.333)	(1.274)	(2.336)	(3.952)
Size	−0.402 **	−0.005 *	−0.517 ***	−0.262	−0.516 ***	−0.667 ***	−0.475 ***	−0.436 ***
	(0.166)	(0.003)	(0.185)	(0.193)	(0.165)	(0.159)	(0.167)	(0.166)
Cflow	23.082 ***	0.518 ***	27.085 ***	27.849 ***	23.271 ***	25.350 ***	24.008 ***	27.247 ***
	(2.801)	(0.042)	(3.081)	(3.390)	(2.803)	(2.521)	(2.836)	(2.928)
Indir	0.119 ***	0.001 **	0.115 ***	0.122 ***	0.117 ***	0.098 ***	0.116 ***	0.113 ***
	(0.029)	(0.000)	(0.032)	(0.035)	(0.029)	(0.027)	(0.029)	(0.029)
Dual	0.799 **	0.014 ***	0.784 **	0.869 **	0.722 **	0.892 ***	0.755 **	0.727 **
	(0.357)	(0.005)	(0.387)	(0.424)	(0.356)	(0.339)	(0.356)	(0.356)
Soe	−1.677 ***	−0.045 ***	−1.484 ***	−1.867 ***	−1.552 ***	−2.257 ***	−1.634 ***	−1.648 ***
	(0.348)	(0.006)	(0.398)	(0.403)	(0.349)	(0.385)	(0.349)	(0.348)
Capin							−11.331 ***	−10.918 ***
							(3.938)	(3.945)
Roa								−17.389 **
								(7.470)
_Cons	13.560 ***	1.208 ***	20.908 ***	20.512 ***	16.504 ***		16.891 ***	15.906 ***
	(3.922)	(0.062)	(4.332)	(4.695)	(3.919)		(3.916)	(3.907)
N	11,720	11,720	10,548	8204	11,720	11,720	11,720	11,720
Adj. R2	0.865	0.969	0.856	0.871	0.866	0.869	0.866	0.866
Industry	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Province	No	No	No	No	No	Yes	No	No

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

indicate that the coefficient of Dig was 0.506. This coefficient was significantly positively correlated at the 5% level and was consistent with the benchmark regression results.

5.2.2. Replacing the Explained Variable

To address the issue of heteroskedasticity in the carbon performance values and regression estimates, we substituted carbon performance (CP) with carbon performance plus one and applied the natural logarithm (lnCP) for the primary regression analysis. Table 5 displays the outcomes of estimating this regression in column 2. The regression results were consistent with the baseline regression results, verifying the robustness of the above findings.

5.2.3. Carbon Performance Using the Next Stage

To consider the likelihood of potential reverse causality issues between digital transformation and carbon performance and to ensure the sustainability of the positive impact of digital transformation on firms’ carbon performance, this study used the subsequent phase of the carbon performance in the regression analysis, the results of which are presented in column 3 of Table 5. The results show that digital transformation continued to exhibit a significant positive impact on firms’ carbon performance, which was consistent with the results of previous studies.

5.2.4. Reducing the Sample Period for the Regression

Considering that China proposed its digitalization strategy in 2015, this study narrowed the research sample and utilized the data from 2015 to 2021 for regression analysis. The results are shown in column 4 of Table 5, and they indicate that there were no significant anomalous alterations in the link between enterprise digitalization and enterprise carbon performance after shortening the sample period. The regression coefficient of the effect of enterprise digitalization on enterprise carbon performance passed a 1% statistical test, suggesting that enterprise digital transformation was an effective driver of the improvement of enterprise carbon performance.

5.2.5. New Sample Regression

Compared with other industries, enterprises in the computer industry have an “inherent” advantage in digitization, which may have impacted the results of this study. Therefore, a sample of computer, communication, and other electronic equipment manufacturing enterprises was eliminated, and the results were re-estimated. The regression results are shown in column 5 of Table 5. The findings indicate that even after removing the computer industry sample, the regression results remained considerably positive at the 1% level. This confirmed the conclusion, showing that digital transformation’s enhancement of organizations’ carbon performance was extremely resilient.

5.2.6. Higher-Order Fixed Effects

To enhance the robustness of the baseline regression model, this study integrated individual fixed effects and industry fixed effects, taking into account potential estimation mistakes caused by unobservable elements. Additionally, in order to account for the impacts of provinces on the extent of digital transformation in businesses, a dummy variable for provinces was included in the higher-order fixed effects. The results are shown in column 6 of Table 5. There was a strong positive association at the 1% significance level between digital transformation and carbon performance.

5.2.7. Adding Control Variables

Variables that could be omitted such as the net profit on total assets (ROA) and capital intensity (Capin) were added. The results are presented in columns 7 and 8 of Table 5, and the conclusions of this study remained unchanged with the addition of these control variables.

5.2.8. Instrumental Variable Method

The empirical investigation in this work could have faced an endogeneity problem caused by reverse causality. Furthermore, enterprises that had high levels of carbon performance were more likely to engage in digital transformation. Carbon performance may have been the cause, rather than the result, of firms’ digital transformation. We tackled this problem by employing two methodologies. Initially, taking into account the possibility of a delay in the effects of companies’ digital transformations [83], we conducted a regression analysis using the main explanatory variables with lags of one and two periods. The results can be observed in

Table 6. Instrumental variable approach and PSM.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)
Variables	DCG	CP	DCG	CP	DCG	CP	CP
L.DCG	0.794 ***
	(0.006)
L2.DCG			0.694 ***
			(0.008)
MobileNum					0.002 ***
					(0.000)
DCG		0.984 ***		1.103 ***		13.595 ***	0.7141 ***
		(0.193)		(0.233)		(3.131)	(2.9689)
Lev	−0.025	−12.677 ***	−0.051	−13.363 ***	−0.124 *	−9.266 ***	−15.8242 ***
	(0.046)	(1.105)	(0.059)	(1.160)	(0.067)	(1.432)	(−8.4286)
Listage	0.009	−3.757 ***	0.019	−3.813 ***	0.054 *	−4.240 ***	−4.4725 ***
	(0.022)	(0.517)	(0.028)	(0.538)	(0.032)	(0.648)	(−5.1931)
Mark	0.075	25.027 ***	0.096	22.790 ***	−0.035	27.664 ***	26.2249 ***
	(0.055)	(1.326)	(0.069)	(1.346)	(0.083)	(1.667)	(12.6284)
Size	0.043 ***	−0.427 **	0.072 ***	−0.379 **	0.142 ***	−2.412 ***	−0.3685
	(0.007)	(0.168)	(0.009)	(0.177)	(0.010)	(0.509)	(−1.2493)
Cflow	−0.128	24.826 ***	−0.274 *	26.451 ***	−0.405 **	28.365 ***	31.5650 ***
	(0.113)	(2.699)	(0.145)	(2.835)	(0.163)	(3.502)	(7.0104)
Indir	0.000	0.124 ***	−0.001	0.131 ***	0.003 *	0.079 **	0.2020 ***
	(0.001)	(0.028)	(0.001)	(0.029)	(0.002)	(0.036)	(4.2862)
Dual	0.050 ***	0.692 *	0.084 ***	0.705 *	0.105 ***	−0.742	0.9278
	(0.015)	(0.362)	(0.019)	(0.381)	(0.022)	(0.569)	(1.5363)
Soe	−0.066 ***	−1.717 ***	−0.113 ***	−1.785 ***	−0.159 ***	0.527	−1.5946 **
	(0.017)	(0.398)	(0.021)	(0.417)	(0.024)	(0.707)	(−2.3632)
_Cons	−0.854 ***	21.420 ***	−1.379 ***	20.054 ***	−3.828 ***	63.930 ***	6.0534
	(0.172)	(4.127)	(0.219)	(4.319)	(0.243)	(12.657)	(0.8555)
Kleibergen–Paap rk LM statistic		2298.112 ***		1685.800 ***		42.561 ***
Kleibergen–Paap rk Wald F statistic		14,441.76 (16.38)		6504.507 (16.38)		42.716 (16.38)
Observations	10,548	10,548	9376	9376	11,720	11,720	4489
R-squared	0.743	0.867	0.630	0.870	0.378	0.776	0.843
Industry	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year	Yes	Yes	Yes	Yes	Yes	Yes	Yes

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

, specifically in columns (1) and (2) as well as columns (3) and (4). The coefficient estimate for digital transformation remained notably positive at the 1% confidence level. The regression results demonstrate that the conclusions of this research remained strong even after considering the problems of reverse causality and omitted factors.

Furthermore, based on earlier research [77,84,85], we selected the MobileNum per 100 individuals in the city where a firm is situated as an instrumental variable for measuring the digital transformation of an enterprise. The rationale for selecting this instrumental variable was based on the fact that the number of cell phones in the city where an enterprise’s office is situated was indicative of the development level of the network infrastructure in that region. Fluctuations in the amount of this infrastructure impacted the degree to which emergent technologies were utilized in the process of transformation. Consequently, when the number of cell phones in a region increased, the level of digital transformation of the firms also increased, thereby meeting the criterion for relevance. However, the quantity of cell phones in a region did not have a direct impact on the carbon performance of businesses, meeting the condition of exogeneity. The data in column (3) of Table 6 indicate a strong and positive correlation between the number of cell phones in a city, referred to as MobileNum, and the digital transformation of enterprises. Furthermore, the outcome of the weak instrumental variable test in the initial-stage regression exceeded 10, refuting the initial hypothesis regarding the presence of weak instrumental factors. Column (4) indicates that the estimated coefficient of enterprise digital transformation (DCG) remained significantly negative and did not undergo any major changes compared to the baseline regression. These findings strongly indicate that the benchmark regression results of this study remained valid even after employing instrumental factors to address potential endogeneity issues. The Kleibergen–Paap rk LM statistic was significant at the 1% level and passed the unidentified test. The Kleibergen–Paap rk Wald F statistic was 42.561, which was greater than the critical value of the F-test at the 10% level of weak instrumental variable identification (16.38) and passed the test for weak instrumental variables, suggesting that the selection of the instrumental variables in this study was reasonable to some extent. In summary, digital transformation could still promote carbon performance after considering endogeneity issues.

5.2.9. PSM

First, firms were categorized according to the median degree of digital transformation. After selecting all control variables, year dummies, and industry dummies as covariates, the samples were matched using a 1:1 nearest-neighbor matching method. Subsequently, kernel density functions were produced to compare the distributions before and after sample matching. This can be seen in Figure 1 and Figure 2. Upon comparison, it was evident that the kernel density curves of the treatment and control groups were significantly closer to each other following the completion of matching, indicating the effectiveness of the sample matching process. Column 5 of Table 6 displays the results of the regression estimation using the matched samples. The results demonstrate that the conclusions of this research remained valid even after addressing the issue of randomness.

5.3. Further Analysis

Mediating Mechanism Test of Corporate Information Transparency (TRANS)

In order to test the relationship between corporate information transparency in digital transformation and corporate carbon performance, a regression model was set up as follows:

$${\mathrm{T}\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{S}}_{\mathrm{i},\mathrm{t}}={\mathrm{b}}_0+{\mathrm{b}}_1{\mathrm{D}\mathrm{C}\mathrm{G}}_{1,\mathrm{t}}+\sum \mathrm{b}\mathrm{C}\mathrm{O}\mathrm{N}\mathrm{T}\mathrm{R}\mathrm{O}\mathrm{L}\mathrm{S}+\mathrm{Y}\mathrm{e}\mathrm{a}\mathrm{r}+\mathrm{I}\mathrm{n}\mathrm{d}\mathrm{u}\mathrm{s}\mathrm{t}\mathrm{r}\mathrm{y}+{\mathsf{\epsilon }}_{\mathrm{i},\mathrm{t}}$$

(6)

$${\mathrm{C}\mathrm{P}}_{\mathrm{i},\mathrm{t}}={\mathrm{c}}_0+{\mathrm{c}}_1{\mathrm{D}\mathrm{C}\mathrm{G}}_{1,\mathrm{t}}{+{\mathrm{c}}_2\mathrm{T}\mathrm{R}\mathrm{A}\mathrm{N}\mathrm{S}}_{\mathrm{i},\mathrm{t}}+\sum \mathrm{c}\mathrm{C}\mathrm{O}\mathrm{N}\mathrm{T}\mathrm{R}\mathrm{O}\mathrm{L}\mathrm{S}+\mathrm{Y}\mathrm{e}\mathrm{a}\mathrm{r}+\mathrm{I}\mathrm{n}\mathrm{d}\mathrm{u}\mathrm{s}\mathrm{t}\mathrm{r}\mathrm{y}+{\mathsf{\epsilon }}_{\mathrm{i},\mathrm{t}}$$

(7)

Table 7. Mediating mechanism test for corporate information transparency.

	(1)	(2)	(3)
	CP	TRANS	CP
DCG	0.889 ***	0.007 ***	0.869 ***
	(0.164)	(0.001)	(0.165)
Lev			2.920 ***
			(1.043)
Listage	−11.255 ***	−0.101 ***	−10.959 ***
	(1.126)	(0.010)	(1.136)
Mark	−3.735 ***	−0.013 ***	−3.696 ***
	(0.474)	(0.005)	(0.474)
Size	27.107 ***	0.251 ***	26.374 ***
	(2.333)	(0.012)	(2.383)
Cflow	−0.516 ***	0.094 ***	−0.792 ***
	(0.165)	(0.002)	(0.206)
Indir	23.271 ***	0.330 ***	22.308 ***
	(2.803)	(0.025)	(2.853)
Dual	0.117 ***	−0.000	0.118 ***
	(0.029)	(0.000)	(0.029)
Soe	0.722 **	0.004	0.711 **
	(0.356)	(0.003)	(0.356)
_Cons	−1.552 ***	0.010 ***	−1.581 ***
	(0.349)	(0.004)	(0.351)
N	11,720	11,720	11,720
Adj. R2	0.865	0.401	0.865
Industry	Yes	Yes	Yes
Year	Yes	Yes	Yes

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

shows the three-step findings of corporate information transparency testing. Column (1) of the model’s findings demonstrates that industrial firms’ digital transformation level (DCG) enhanced corporate carbon performance (CP) at 99% significance. Column (2) shows that digital transformation (DCG) enhanced corporate information transparency (TRANS), and its regression coefficient was significant at the 1% level, confirming hypothesis 2a. All of the model coefficients of corporate information openness related to enterprises’ carbon performance, which is included as a mediator variable in column (3), were positive and significant at the 1% level. Corporate information transparency partially mediated the regression coefficient of digital transformation’s effect on carbon performance. This showed that company digitalization could boost carbon performance by increasing information openness, which verified hypothesis 2b.

Further testing using the Sobel test revealed a substantial mediating influence of corporate information transparency (Z-value = 4.191, p-value < 0.01). Company information transparency partially mediated 24.5% of the total effect. Bootstrapping was used for the next test. After sampling the same 95% confidence interval 1000 times, the result showed that corporate information transparency mediated the confidence interval (TRANS = [2.475497,6.69153]), which did not include 0. The mediating effect of corporate information transparency was confirmed.

5.4. Moderating Mechanism Test

After finding that corporate digital transformation had an enhancing effect on the level of carbon performance, Model 5 was estimated to test the moderating effect of the performance–expectation gap.

$${\mathrm{C}\mathrm{P}}_{\mathrm{i},\mathrm{t}}={\mathrm{d}}_0+{\mathrm{d}}_1{\mathrm{D}\mathrm{C}\mathrm{G}}_{1,\mathrm{t}}{+{\mathrm{d}}_2\mathrm{P}\mathrm{E}\mathrm{G}}_{\mathrm{i},\mathrm{t}-1}+\sum \mathrm{d}\mathrm{C}\mathrm{O}\mathrm{N}\mathrm{T}\mathrm{R}\mathrm{O}\mathrm{L}\mathrm{S}+\mathrm{Y}\mathrm{e}\mathrm{a}\mathrm{r}+\mathrm{I}\mathrm{n}\mathrm{d}\mathrm{u}\mathrm{s}\mathrm{t}\mathrm{r}\mathrm{y}+{\mathsf{\epsilon }}_{\mathrm{i},\mathrm{t}}$$

(8)

$${\mathrm{C}\mathrm{P}}_{\mathrm{i},\mathrm{t}}={\mathrm{f}}_0+{\mathrm{f}}_1{\mathrm{D}\mathrm{C}\mathrm{G}}_{1,\mathrm{t}}{+{\mathrm{f}}_2\mathrm{P}\mathrm{E}\mathrm{G}}_{\mathrm{i},\mathrm{t}-1}{+{\mathrm{f}}_3{\mathrm{D}\mathrm{C}\mathrm{G}}_{1,\mathrm{t}}\ast \mathrm{P}\mathrm{E}\mathrm{G}}_{\mathrm{i},\mathrm{t}-1}+\sum \mathrm{d}\mathrm{C}\mathrm{O}\mathrm{N}\mathrm{T}\mathrm{R}\mathrm{O}\mathrm{L}\mathrm{S}+\mathrm{Y}\mathrm{e}\mathrm{a}\mathrm{r}+\mathrm{I}\mathrm{n}\mathrm{d}\mathrm{u}\mathrm{s}\mathrm{t}\mathrm{r}\mathrm{y}+{\mathsf{\epsilon }}_{\mathrm{i},\mathrm{t}}$$

(9)

Managers create objectives or expectations and undertake organizational restructuring and strategic changes depending on the gap between their business’s performance and their goals. According to corporate behavior theory, organizations with unmet expectations actively seek ways to improve their businesses or reverse their declines. Digital transformation is strategic and helps companies grow and profit. The performance–expectation gap may moderate the association between digital transformation and corporate carbon performance, according to this study. This study introduced an interaction term (DCG_PEG) between the performance–expectation gap (PEG) and digital transformation (DCG) for testing, specifically in Model 9. Carbon performance is positively correlated with the performance–expectation gap, according to research. Carbon performance improves with the performance–expectation gap. Column (2) of

Table 8. Moderating mechanism test for performance–expectation gap.

	(1)	(2)	(3)
	CP	CP	CP
DCG	0.889 ***	0.892 ***	0.762 ***
	(0.164)	(0.163)	(0.174)
PEG		64.254 ***	56.255 ***
		(12.861)	(13.874)
DCG_PEG			7.165 *
			(3.682)
Lev	−11.255 ***	−9.981 ***	−9.765 ***
	(1.126)	(1.134)	(1.132)
Listage	−3.735 ***	−3.637 ***	−3.611 ***
	(0.474)	(0.470)	(0.471)
Mark	27.107 ***	42.442 ***	43.281 ***
	(2.333)	(4.020)	(3.920)
Size	−0.516 ***	−0.542 ***	−0.554 ***
	(0.165)	(0.164)	(0.164)
Cflow	23.271 ***	23.347 ***	23.009 ***
	(2.803)	(2.772)	(2.778)
Indir	0.117 ***	0.111 ***	0.111 ***
	(0.029)	(0.029)	(0.029)
Dual	0.722 **	0.596 *	0.579
	(0.356)	(0.353)	(0.353)
Soe	−1.552 ***	−1.360 ***	−1.346 ***
	(0.349)	(0.347)	(0.347)
_Cons	16.504 ***	14.458 ***	14.587 ***
	(3.919)	(3.932)	(3.934)
N	11,720	11,720	11,720
Adj. R2	0.866	0.868	0.868
Industry	Yes	Yes	Yes
Year	Yes	Yes	Yes

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

shows a DCG_PEG regression coefficient of 7.165, which was statistically significant at the 10% level. The performance–expectation gap appeared to positively regulate the association between digital transformation and enterprise carbon performance.

5.5. Heterogeneity Analysis

5.5.1. Nature of Enterprises

Referring to a prior investigation [86], different types of ownership were distinguished to assess the impact of digital transformation on corporate carbon performance. Columns (1) and (2) of

Table 9. Heterogeneity analysis.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
	CP	CP	CP	CP	CP	CP	CP	CP
DCG	0.192	1.212 ***	0.210 ***	0.925 ***	0.882 ***	0.959 ***	0.718 ***	1.146 ***
	(0.251)	(0.210)	(0.062)	(0.200)	(0.242)	(0.185)	(0.228)	(0.165)
Lev	−2.604 **	−15.883 ***	0.742 *	−19.710 ***	−19.968 ***	−1.207	−21.747 ***	−0.648
	(1.272)	(1.598)	(0.443)	(1.623)	(1.806)	(1.089)	(1.852)	(0.778)
Listage	−4.573 ***	−2.879 ***	0.527	−4.286 ***	−4.625 ***	−2.322 ***	−4.048 ***	−1.379 ***
	(0.654)	(0.689)	(0.329)	(0.649)	(0.801)	(0.452)	(0.777)	(0.413)
Mark	37.447 ***	24.388 ***	8.661 ***	31.089 ***	27.000 ***	25.584 ***	30.770 ***	17.606 ***
	(3.898)	(2.699)	(1.634)	(3.024)	(3.293)	(2.832)	(3.396)	(2.236)
Size	−0.127	−1.149 ***	−0.297 ***	−0.463 *	−0.883 ***	−0.206	−1.144 ***	0.456 ***
	(0.179)	(0.265)	(0.095)	(0.240)	(0.260)	(0.188)	(0.247)	(0.172)
Cflow	20.175 ***	24.849 ***	3.154 ***	37.116 ***	30.561 ***	15.594 ***	37.523 ***	10.914 ***
	(3.483)	(3.853)	(1.069)	(4.132)	(4.738)	(2.728)	(4.934)	(1.979)
Indir	0.063	0.129 ***	−0.004	0.173 ***	0.065	0.174 ***	0.117 ***	0.124 ***
	(0.040)	(0.040)	(0.009)	(0.040)	(0.045)	(0.034)	(0.044)	(0.033)
Dual	0.075	0.686	0.100	0.800*	1.515 ***	0.020	1.047 *	0.024
	(0.595)	(0.428)	(0.089)	(0.470)	(0.554)	(0.360)	(0.537)	(0.285)
Soe			−0.567 ***	−2.788 ***	−2.718 ***	−0.532	−3.113 ***	0.203
			(0.199)	(0.486)	(0.572)	(0.348)	(0.574)	(0.285)
_Cons	7.609	24.782 ***	6.398 ***	17.924 ***	46.091 ***	−1.075	62.035 ***	−13.686 ***
	(4.787)	(5.931)	(1.544)	(5.601)	(5.774)	(4.488)	(5.561)	(4.207)
N	4544	7176	3886	7834	5895	5825	6271	5449
Adj. R2	0.897	0.849	0.882	0.769	0.829	0.891	0.724	0.858
Industry	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes

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

report the performance of SOEs and non-SOEs in improving corporate carbon performance levels due to digital transformation. In terms of the heterogeneity of property rights, the coefficient of digital transformation for state-owned enterprises was insignificant, while the coefficient of digital transformation for non-state-owned enterprises was significantly positive. This may have been because state-owned enterprises have always been the main force of green and low-carbon development and have actively taken on various social responsibilities, so the level of carbon performance was not affected much after digital transformation. On the other hand, the ability and willingness of non-state-owned enterprises to take on corporate social responsibility increased significantly after digital transformation, which could contribute to the continuous improvement of carbon performance levels.

5.5.2. Whether the Enterprise Is in a High-Pollution Industry

Enterprises in heavily polluting industries are the primary contributors to pollutant emissions and exert a more significant adverse influence on the environment. Hence, based on prior research [87], this study categorized the sample into two groups for regression analysis based on whether the firms belonged to heavily polluting industries. If the industry in which an enterprise operated was characterized by significant levels of pollution, the enterprise itself could be classified as a heavily polluting enterprise. Conversely, if the sector had low levels of pollution, the enterprise could be classified as a non-heavily polluting enterprise. The empirical findings are displayed in column (3) and column (4) of Table 9. The digital transformation of firms in industries with low pollution levels greatly enhanced the environmental performance of these enterprises. The digitalization of businesses enhanced the visibility of enterprise information. In comparison to companies in less-polluted sectors, companies operating in heavily polluted industries experienced greater public scrutiny and environmental accountability. They also faced higher costs for environmental management and achieved fewer improvements in carbon performance. This may have been because heavily polluting industries were more concerned with reducing pollution quickly when faced with environmental regulations and thus mostly resorted to measures such as shutting down or switching production. On the other hand, non-polluting firms had more time to invest in digitalization and green innovation when facing regulatory pressure. Therefore, the interaction between digital transformation and carbon performance was more significant for non-polluting firms.

5.5.3. Degree of Market Competition

Differences in market competitiveness will influence the behavioral choices made by organizations, particularly the effects of their digital transformation on energy conservation and emission reduction. When faced with intense competition in the market, companies focus on attaining long-lasting competitive advantages by adopting advanced digital transformation strategies to achieve environmentally friendly growth. This paper argues that the degree of market competition plays a role in the impact of digital transformation on carbon performance. This study used the Herfindahl index to measure the degree of market competitiveness, making reference to a previous analysis [86]. Afterward, the sample was divided into two groups based on the median value of this indicator. The regression coefficients for digital transformation in columns (5) and (6) exhibited positive values and were statistically significant at the 1% level. Nevertheless, the coefficient was higher in the sample group, which was characterized by a substantial level of market competition, and the disparity between the coefficients of the groups was also statistically significant. As expected, the effect of businesses’ digital transformation on carbon performance was particularly noticeable in fiercely competitive marketplaces.

5.5.4. Capital Intensity

In order to further test the heterogeneity of the impact mechanism by which enterprises’ digital empowerment reduces carbon emissions, the sample enterprises were divided into the three classifications of capital-intensive, technology-intensive, and labor-intensive according to the differences in the factor intensities of the listed companies [88,89]. Based on the given classification, the enterprises were divided into two groups: capital-intensive enterprises and non-capital-intensive enterprises. A group regression was then performed. The higher the capital intensity, the higher the modernization level of the industrial sector and the easier it was to operate in a decarbonization mode. These results are shown in columns (7) and (8) of Table 9. Digital transformation had a larger impact on carbon performance at firms with low capital intensity. A possible reason is that the carbon performance of non-capital-intensive firms was affected more by the same level of digital transformation enhancement.

6. Conclusions and Recommendations

Under the “dual-carbon” goal, the development of enterprises in the industrial sector is facing great challenges. Academic research is increasingly focusing on how to utilize the competitive advantages brought by digital transformation to effectively improve corporate carbon performance and explore a sustainable development path with both economic and social benefits. Therefore, this study aimed to investigate the impact of digital transformation on the carbon performance of listed companies in the industrial sector. It further analyzed the internal mechanism of this impact by considering corporate information transparency. Additionally, it uncovered the variations in the relationship between digital transformation and carbon performance due to the performance–expectation gap. This study offered novel theoretical support and policy suggestions for the comprehensive amalgamation of digital transformation and sustainable development. Additionally, it yielded the following significant discoveries: (1) The implementation of digital transformation had a substantial impact on the carbon performance of industrial sector companies listed on the stock market, demonstrating its beneficial influence on corporate strategies for sustainable development. Viewing digital transformation from a micro-enterprise perspective allowed for more thorough and unbiased comprehension of its effects. (2) The effect mechanism analysis revealed that corporate information transparency acts as a mediator between digital transformation and carbon performance, indicating that enhancing corporate information transparency is crucial for facilitating the integration of these two factors. This further enriches the research on the relationship between digital transformation and corporate carbon performance by providing new empirical evidence of the mechanisms of digital transformation’s impact. (3) Managers are incentivized to engage in a problem search when there is a disparity between the anticipated performance and the real performance. Subsequently, a resolution to the predicament is ascertained to reinstate the organization’s intended level of performance. As the disparity between corporate expectations widens, managers are becoming more active in implementing strategic changes. Digital transformation is an advantageous and deliberate change. In this situation, firms are likely to embrace a more positive attitude towards the implementation of digital transformation and attempt to reduce carbon emissions. As a result, this would accelerate the overall progress of the industry in terms of sustainability. This study elucidated the impact of digital transformation in the presence of a performance–expectation gap and enriched the research in this area. (4) This study analyzed a variety of industries and enterprises by considering their specific traits, degrees of market competitiveness, and levels of capital intensity. The results suggest that the adoption of digital transformation has a notable beneficial effect on the carbon performance of privately owned companies and companies with minimal emission levels. Firms operating in highly competitive marketplaces with lower capital requirements experience more significant impacts on their carbon performance due to digital transformation.

Based on the aforementioned facts, this paper proposes recommendations for enhancing the carbon performance of industrial enterprises in China, focusing on the following four aspects: (1) In light of the rapid advancement of digital transformation, businesses should proactively grasp the trend and consistently advance the process of digital transformation. Enterprises should fully embrace the Internet in their operations, focus on customer needs, and use Internet thinking to revolutionize traditional production processes and bring enterprises and consumers closer. This will enable low-carbon development while advancing digital transformation. (2) In strengthening digital transformation, enterprises should prioritize the development of an information disclosure system to improve their information transparency. Simultaneously, during digitalization, enterprises must fully utilize the resources and information benefits of digital transformation. They should prioritize the development of information technology infrastructure, establish an effective information-sharing platform, facilitate efficient transmission and communication of information, and enhance their capacity to manage risks. These efforts will ultimately improve their level of carbon performance. (3) In the context of a performance–expectation gap, enterprises anticipate overcoming operational challenges and attaining a competitive edge through digital transformation. However, they must recognize that while digital transformation can offer growth opportunities, it also carries significant risks. Enterprises should conduct incremental technological innovation by thoroughly assessing their resources and digital capabilities and progressively enhancing their digital transformation approach. When there are excessive performance expectations, firms tend to stick to their prior strategies and ignore future development prospects due to route dependence and strategic inertia. Enterprises should utilize the excess resources resulting from exceeding performance expectations to implement their digital transformation strategies, ensuring stable operation while seizing future development prospects. (4) In the digital economy era, the government will enhance the legislative and regulatory framework, establish suitable systems, and invest more resources in developing digital technology to facilitate the digital transformation of businesses. Simultaneously, it will build and promote standards and methods for disclosing data assets, assist businesses in constructing and enhancing platforms for data asset disclosure, improve the transparency of company information, and reinforce the impacts of carbon performance improvements for enterprises. (5) To provide managers and policymakers with policies that support the digital transformation of enterprises, based on the nature of different property rights and differences in the digital transformation of enterprises in various industries, actions will be targeted to solve the pain points in the digital transformation of enterprises.