Impact of Digital Technology Adoption on the Similarity of Sustainability Reports

Abstract

Digital technology has transformed sustainability reporting practices, creating both opportunities and homogenization risks. This study analyzes 9903 sustainability reports from Chinese listed companies (2009–2021) through cosine similarity analysis. It reveals high intercorporate similarity (mean = 0.776). Fixed-effects modeling demonstrates that digital adoption increases report similarity, while analyst scrutiny and innovation capacity significantly mitigate this convergence effect. The findings suggest that digital tools promote isomorphic disclosure patterns through template-driven reporting. However, market monitoring (analyst attention) and R&D investment counterbalance this trend by incentivizing unique environmental, social, and governance (ESG) disclosures. This study offers novel insights into information asymmetry theory and social learning theory. The increased similarity in reporting will lead to standardization among Chinese companies, thereby enhancing their comparability in the international market. This will not only help Chinese companies improve their performance assessments for global investors but also facilitate cross-border investments.

1. Introduction

Corporate disclosure mechanisms primarily operate through three channels: annual reports, earnings calls, and sustainability documentation [1,2]. Among these, sustainability reports have evolved into a vital platform for non-financial stakeholder communication [3], with growing academic attention to their effectiveness in conveying ESG information [4,5,6]. A recent analysis of 280 non-financial Pakistani firms by Waris Ali et al. (2024) reveals that sustainability disclosures demonstrate greater diversity and depth compared to traditional annual reports [6]. This observation corroborates Higgins and Coffey’s (2016) conceptualization of sustainability reports as strategic instruments for operationalizing sustainable practices and institutionalizing ethical governance [7].

Contemporary scholarship systematically examines three critical textual dimensions in corporate disclosures: linguistic complexity [8,9,10], tonal attributes [11,12,13,14], and similarity [9,15]. Empirical evidence indicates that systematic variations in these attributes significantly impact capital market information efficiency. Enhanced readability reduces information asymmetry through improved cognitive processing [16], while affirmative linguistic framing amplifies market-responsive information content [11,17]. However, Xu (2024) identifies a paradoxical effect: excessive textual convergence in corporate reports creates informational opacity that obscures firm-specific signals and impairs price discovery mechanisms [18].

Current research identifies four primary determinants of report similarity: (1) corporate financial characteristics [19,20], (2) regulatory frameworks [21], (3) audit conventions [22,23], and (4) environmental shifts [24]. Brown and Knechel (2016) demonstrate that firms sharing financial characteristics and regulatory environments tend to adopt similar audit styles, resulting in standardized reporting templates [19]. This standardization proves problematic during economic disruptions, as highly similar reports fail to convey critical firm-specific risks, potentially misleading investor decisions [24].

The digital transformation of corporate communications introduces novel complexities. While digital tools enhance information collection efficiency and stakeholder engagement [25,26,27], their adoption generates contradictory effects on disclosure practices. Kuzovkova et al. (2021) demonstrate technology-enabled diversification of reporting content through granular stakeholder feedback [27,28], whereas Larson and DeChurch (2020) identify digital dependency as a catalyst for creative stagnation and textual uniformity [29,30,31]. This tension aligns with Xu’s (2024) conceptualization of “informational fuzziness” in technologically mediated disclosures [18]. However, the mechanism by which digital technology impacts the information environment of fuzzy firms remains ambiguous.

Our findings propose that digital technology enhances reporting similarity but introduces ambiguity to corporate information. Nevertheless, it can bridge the gap between transparency and uniqueness. Digital technologies, empowered by firm innovation, facilitate customization and dynamic data integration in sustainability reports, balancing comparability and firm-specificity. Enhancing accountability through analyst reviews could transform sustainability reporting into a market differentiation mechanism. In summary, digital technology can mitigate information asymmetries by standardizing templates while also creating new asymmetries by reducing firm-specific disclosures, thereby substantiating the theory of information asymmetry. From a social learning perspective, firms increase reporting similarity through imitation and observation, while analysts differentiate their reviews and innovation through substantive ESG advancements.

This study makes four substantive contributions to disclosure literature: Firstly, it introduces sustainability reports as novel vectors for textual similarity analysis, revealing paradoxically high similarity (mean > 0.6) despite the absence of standardized disclosure guidelines [32,33]. Secondly, it establishes longitudinal effects of digital adoption on report similarity dynamics. Thirdly, it identifies analyst scrutiny and innovation capacity as effective counterweights to digital-induced homogeneity. Finally, this study extends Akerlof’s framework [34] by demonstrating how digital technologies have transformed the information market from a “lemon” equilibrium to a sub-quality market, where analyst scrutiny and firms’ ability to innovate have become key differentiators. Furthermore, we identify a dual social learning dynamic in disclosure practices: imitative convergence through benchmarking industry leaders, and strategic differentiation driven by investment in innovation to combat ‘greenwashing’ while amplifying firm-specific sustainability advantages.

This study presents three pivotal policy recommendations, with significant implications for enhancing organizational legitimacy and global capital markets. Firstly, the development of adaptive AI architectures that integrate entity-specific sustainability indicators into standardized reporting frameworks is essential. This preserves the unique sustainability nuances of individual entities while ensuring comparability across the board [35]. Secondly, institutionalizing third-party analyst feedback mechanisms in materiality determination processes can considerably enrich the evaluation of sustainability metrics and ensure a more robust and credible assessment [36]. Thirdly, strategically converting R&D outputs into proprietary sustainability metrics can drive innovation whilst maintaining a competitive edge in sustainability reporting [37]. These measures not only enhance report comparability but also suggest improved transparency may mitigate emerging market discounts for Chinese firms in international investments. This research thus contributes substantively to the digital governance literature by advocating for a balanced approach that harmonizes standardization with innovation in sustainability communication.

The subsequent analysis proceeds as follows: Section 2 synthesizes key literature, Section 3 articulates theoretical foundations, Section 4 details methodological architecture, Section 5 presents empirical findings, and Section 6 discusses implications for digital disclosure practices.

2. Literature Review

The term “digital technology adoption” refers to the utilization of digital technologies, including artificial intelligence, blockchain, cloud computing, and big data. This analysis is conducted from three perspectives: the theoretical foundation, the practical application of digital technology, and the similarities in sustainability reports associated with digital technology. By integrating advanced digital tools, organizations can create more transparent, accountable, and data-driven sustainability initiatives, ultimately contributing to a more sustainable future. This intersection of technology and sustainability represents a crucial area for further research and development.

2.1. Theoretical Foundation

2.1.1. Information Asymmetry and Digital Technologies in Sustainability Contexts

The theory of information asymmetry posits that unequal access to information between firms and stakeholders distorts market valuations and governance efficacy, particularly in sustainability contexts [38,39,40]. In sustainability reporting, this asymmetry manifests as stakeholders’ inability to discern substantive environmental actions from superficial claims [41] (e.g., unverified “net-zero” pledges or exaggerated carbon reduction metrics), thereby enabling greenwashing practices [42]. For instance, digital companies often underreport Scope 3 emissions (indirect emissions across value chains), which account for six times their operational emissions on average, creating systemic opacity [43].

Digital technologies as institutional remedies. Digital technologies reduce information gaps by enhancing data granularity, traceability, and stakeholder collaboration [34,44,45]. For example, blockchain-enabled carbon tracking systems and AI-driven analytics improve transparency in supply chain emissions, enabling real-time verification of sustainability claims [46]. Firms further amplify accountability by integrating IoT sensors and cloud-based platforms to monitor energy consumption and emissions, thereby curbing opportunistic distortions [47]. Firms adopting advanced digital tools exhibit higher ESG disclosure quality [28], as seen in Ant Group’s use of AI to optimize data center efficiency and carbon accounting. These digital governance mechanisms create verifiability thresholds that render greenwashing increasingly cost-prohibitive.

Reduced asymmetry diminishes greenwashing incentives by increasing the reputational and financial risks of misleading claims [48]. Standardized frameworks like the ISSB’s IFRS S1/S2, which mandate granular climate-related disclosures, further pressure firms to align reporting with actual performance [49,50]. This phenomenon similarly induces a progressive convergence in the structural and thematic composition of the reports. However, SMEs often lack resources for robust data collection, perpetuating asymmetry in less regulated sectors [51].

2.1.2. Social Learning Theory and Sustainability Reports

Social learning theory posits that firms emulate peers’ behaviors to reduce uncertainty and gain legitimacy [52], particularly in evolving regulatory landscapes [53]. This imitation is driven by similarity [54]—the alignment of observable attributes (e.g., industry, size) between the observed and observing entities.

Two distinct analytical frameworks derived delineate the operational mechanisms governing mimetic behavior in reporting. Regulatory and market pressures: Firms in sensitive industries (e.g., energy, finance) face heightened scrutiny, prompting them to adopt reporting frameworks validated by high-performing peers to avoid penalties or attract ESG-focused investors [55]. For example, Malaysian firms in non-sensitive industries exhibit stronger sustainability reporting effects on financial performance, suggesting selective emulation to meet investor expectations [53]. Path dependency in disclosure practices: Economically disadvantaged SMEs exhibit strategic isomorphism through replication of market leaders’ reporting architectures, a risk-aversion mechanism particularly amplified by institutionalized sustainability frameworks (GRI, SASB). This convergence dynamic systematically erodes disclosure heterogeneity through dual channels of normative compliance and cognitive benchmarking. Additionally, firms replicate prior reporting structures (e.g., GRI or SASB standards) to maintain consistency and reduce adaptation costs, leading to elevated content similarity across consecutive reports [56]. Empirical analyses of corporate sustainability reporting reveal this dynamic: firms exhibit heightened susceptibility to imitating prior reporting frameworks, particularly within temporally adjacent reporting cycles. This behavioral mimicry manifests as elevated content similarity across consecutive reports, reflecting institutional path dependency in sustainability disclosure practices [57].

While mimicry enhances comparability, it risks superficial compliance. For instance, companies may adopt “checklist” ESG metrics without substantive action, as seen in cases where firms tout carbon offsets without verifying their efficacy. Paradoxically, homogenization may also stifle innovation: firms prioritizing conformity over tailored sustainability strategies often overlook localized environmental challenges (e.g., water scarcity in agriculture).

2.2. The Practical Adoption of Digital Technology

The perpetual advancement of digital infrastructures catalyzes the emergence of dynamic information ecosystems that optimize information processing efficiency through two synergistic mechanisms. On the one hand, visualization tools demystify complex datasets, enabling stakeholders to decode multidimensional corporate disclosures with higher accuracy in risk assessment tasks [58]. Chen et al. (2024) empirically demonstrate that machine learning algorithms reduce demand forecasting errors through real-time sentiment analysis of unstructured market data [59]. On the other hand, with regard to information dissemination, the employment of data-tracking technology and digital platforms has been demonstrated to enhance the timeliness of information [60].

Paradoxically, digital technology adoption can negatively impact reporting [61]. A total of 73% of S&P 500 firms now replicate industry-standard AI reporting templates, correlating with a 22% decline in sustainability innovation patents since 2020 [62]. This innovation erosion stems from two sides. Firstly, overreliance on GRI-aligned KPIs suppresses contextual problem-solving (e.g., water stewardship metrics overlook regional hydrological stressors). Secondly, digital benchmarking tools inadvertently prioritize conformity over creativity, as shown by some firms verbatim copying peers’ disclosure frameworks [53].

2.3. The Similarity in Sustainability Reports Associated with Digital Technology

Digital technologies are assuming an increasingly significant role in the contemporary business environment, particularly in the domain of sustainability report preparation and presentation. As companies continue to adopt technologies such as artificial intelligence, the Internet of Things, and big data in the process of digital transformation, these technologies have not only impacted their operating models [63,64], but have also exerted a profound effect on the content [60].

The emergence of digital technologies has been shown to enhance data collection and analysis efficiency while simultaneously strengthening the ability to trace and provide feedback on information [27,65]. Digital tools can monitor key sustainable development indicators, such as resource utilization and carbon emissions [66], in real time. This reduces the consumption of human and material resources in report preparation. Additionally, the immediacy and accuracy of information delivery are significantly improved [67]. The integration of digital technologies, including blockchain, facilitates feedback provision to consumers and investors [68], thereby enabling monitoring of corporate sustainability practices.

The promotion of standardization in sustainable development reporting has been significantly influenced by the advent of digital technology. By leveraging artificial intelligence, enterprises can now automatically generate compliant sustainability reports using predefined templates and standards [69,70,71], thereby minimizing human error and enhancing operational efficiency [72]. Additionally, the integration of big data and cloud computing enables organizations to analyze large datasets, identify key sustainability disclosure indicators across various industries, and further advance the standardization of these indicators [73].

In the future, digital technology adoption in corporate sustainability reporting is expected to rise significantly. This trend will be fueled by technological advancements, leading to more precise and standardized reporting practices within the industry. As stakeholders increasingly demand transparency and accountability, companies are likely to leverage real-time data analytics and blockchain technology to deliver verifiable sustainability metrics. This approach will enhance consumer trust and engagement.

3. Hypothesis Construction

3.1. The Direct Effect of Digital Technology Adoption on the Similarity of Sustainability Reports

Recent advancements in digital technology enable businesses to swiftly identify reliable information [74]. On one hand, these developments assist corporate sustainability report writers in obtaining data on key performance indicators [75]. By leveraging cloud computing and a network of edge nodes, Ma et al. (2024) proposes a comprehensive sustainability benchmark that provides essential indicators across environmental, social, and corporate governance domains to meet investor demands [76]. On the other hand, reports are generated based on predefined templates to enhance similarity [62]. Wu et al. (2022) [28] demonstrate that digital technologies, including blockchain, IoT, and AI, can streamline the processing of firms ESG reports, enhance their transparency, security, and credibility. Deloitte’s Industry Chain tool integrates big data analysis and NLP technology to automatically produce reports that adhere to GRI standards, thereby improving consistency [77,78,79]. However, SMEs often face resource constraints and may resort to templated tools or outsourced services, leading to elevated text similarity [53].

To summarize, digital technology adoption can help firms acquire vital information to harmonize key sustainability report indicators and content. This paper derives the following hypotheses:

H1. 

Digital technology adoption can significantly improve the similarity of firms’ sustainability reports.

3.2. Influencing Mechanism of Digital Technology Adoption on the Similarity of Sustainability Reports

Analysts are in possession of considerable resources and expertise, and their monitoring and feedback can exert a significant influence on corporate disclosure behaviors. The diagnostic scrutiny exerted by analysts serves as a critical counterbalance to symbolic sustainability disclosures, addressing the pervasive issue of institutional decoupling—where reported practices diverge from operational realities [80].

Through iterative feedback loops, analysts catalyze substantive reporting improvements by two sides. Firstly, sector-specific benchmarking identifies materiality gaps, prompting firms to replace vague assertions with quantifiable metrics. For instance, the Italian fisheries industry has continually refined and elevated its sustainability report in response to analysts’ recommendations, resulting in an increasingly substantial and scientifically rigorous presentation [81]. Secondly, differentiated disclosure incentives: monitoring intensity systematically shapes industry-specific reporting priorities. Financial institutions exhibit 2.8-fold greater disclosure density in climate risk quantification frameworks relative to non-financial counterparts [82].

This oversight function generates countervailing pressures against similarity in reporting environments [83]. Idiosyncratic signaling: In accordance with the principles of social learning theory, corporate entities tend to emulate the practices of their industry peers, with a particular emphasis on those demonstrating strong market performance. The focus of analysts can exert a significant influence on firms, prompting them to draw parallels between their own practices and those of these exemplary “role models”. This can subsequently lead to an enhancement in the quality and innovation of their own reporting. Machine learning analysis of 10-K filings reveals analyst-tracked firms report more unique sustainability KPIs than peers, circumventing template-driven homogenization [84]. Innovation-compliance synergy: Natural language processing shows analyst feedback reduces boilerplate content while increasing science-based target disclosures in blockchain-augmented reports [85].

H2. 

Analyst attention negatively moderates the relationship between digital technology adoption and sustainability report similarity.

Firm innovation, while essential for competitive differentiation [86], engenders strategic disclosure tensions within sustainability reporting ecosystems. Innovative reporting mechanisms (e.g., AI-driven materiality mapping, blockchain verification) disrupt traditional information asymmetries. Bronzetti et al. (2023) have enhanced the comprehensibility of sustainability reports by innovating their content and improving research and development capabilities [87]. Haffar and Searcy (2020) discover that companies address social concerns by creatively disclosing information from sustainability reports in order to cultivate a favorable corporate image [88]. However, open innovation ecosystems drive firms to co-create sustainability-oriented solutions with external stakeholders (suppliers, customers, regulatory bodies, NGOs), leveraging cross-domain knowledge integration to resolve complex challenges while enhancing disclosure depth and innovation efficacy [89]. These collaborations systematically reduce reporting isomorphism through context-specific information architectures that prioritize material stakeholder concerns over standardized templates [90]. The strategic use of intellectual property rights (IPRs), particularly patents, to protect research and development (R&D) outcomes is a critical measure leveraged by modern enterprises. By converting R&D outcomes into proprietary sustainability indicators and employing methodologies protected by IPRs, these businesses can highlight their unique innovation achievements in their reports, thereby maintaining a competitive edge in the marketplace [89].

Overall, business innovation in sustainability reporting not only enhances the content of the reports but also introduces new formats. While these changes contribute to a more diverse and personalized approach to sustainability reporting, they may also diminish the uniformity of the reports. Therefore, this paper derives the following hypothesis:

H3. 

Innovation capacity negatively moderates the relationship between digital technology adoption and sustainability report similarity.

In conclusion, the adoption of digital technologies significantly influences the similarity of sustainability reports, a phenomenon moderated by firms’ innovativeness and analysts’ attention (Figure 1). This interplay indicates that firms should invest not only in digital tools but also in fostering a culture of innovation to differentiate their sustainability narratives. As analysts’ scrutiny intensifies, companies may utilize unique reporting practices to distinguish themselves, thereby enhancing stakeholder engagement and trust. Such strategic positioning can yield a competitive advantage, ultimately improving sustainability outcomes and organizational reputation in a rapidly evolving marketplace.

4. Methodology and Data

4.1. Sample and Data

This study utilizes a sample of Chinese A-share listed companies spanning from 2009 to 2021, ultimately yielding 9903 company-year observations after excluding instances with missing data on pertinent variables. We selected 2009 as a baseline year because voluntary sustainability efforts met stricter compliance rules at this time, driving consistent reporting practices across sectors. This timeframe helps isolate how regulations and standards reshaped corporate strategies before AI-driven reporting tools emerged. The 2008 financial crisis triggered major changes in corporate accountability, pushing governments to strengthen transparency rules. These reforms shifted corporate reporting from focusing solely on financial data to broader narratives about value creation, setting key benchmarks for analyzing reporting trends.

This study requires that the calculation of the similarity of sustainability reports be based on comparable disclosure texts existing in two consecutive years (i.e., it requires enterprises to proactively release reports in adjacent years), the results can be found in the Supplementary Materials. Due to the non-mandatory nature of sustainability report disclosure, some firms have intermittent disclosures during the research period, which may lead to data loss. If an enterprise only releases reports in specific years (such as the year of an IPO or after a public opinion crisis), it can cause the time series to be discontinuous. For example, if a company releases a report in 2020 but discontinues disclosure in 2021, the paired data for 2020–2021 are invalid.

The sample data sources comprise essential information on listed companies, while governance-related data are primarily sourced from the China Stock Market & Accounting Research Database (CSMAR), which is acknowledged as the most extensive and reliable comprehensive economic and financial research database in China. To mitigate the impact of outliers, this study applies winsorization to all continuous variables at both the 1% and 99% thresholds.

4.2. Definition of Variables

Digital technology adoption. In accordance with the findings of prior research conducted by Yang et al. (2024) [91], it is possible to divide digital technology applications into two categories: digital technology breadth and digital technology depth. The term “breadth of digital technology” refers to the extent of utilization of digital technologies, specifically measured by the frequency of use of various digital tools. This study aims to identify and match keywords such as artificial intelligence, big data, Internet of Things, blockchain, and cloud computing within annual reports to ascertain their presence and frequency. Consequently, Breadth is quantified through the frequency of relevant terminology associated with these digital technologies. In contrast, the “depth of digital technology” encompasses four domains: digital production, digital management, digital marketing, and digital products. This analysis utilizes Yang et al.’s (2024) [91] framework to assess the level of integration between digital technologies and the physical industry. The findings suggest that a greater breadth of digital technology often correlates with advancements in Depth, indicating that frequent use of digital tools can facilitate deeper integration across various domains. Furthermore, the interaction between these dimensions may reveal pathways for industries to innovate and adapt, ultimately enhancing their competitiveness. Future research could further explore how this relationship varies across sectors, potentially uncovering unique strategies employed by different industries to effectively leverage digital technologies.

Similarity. Similarity is defined as the extent to which the sustainable development report from a given year corresponds with that of the prior year. In studies concerning text similarity, empirical research frequently entails examining the variations in content among different texts. Among the various metrics for assessing text similarity, cosine similarity is particularly prominent. As demonstrated by Brown and Tucker (2010) [24] and Peterson et al. (2015) [92], the effective utilization of cosine similarity has been employed to evaluate the textual parallels among firms. This methodology involves the initial extraction of pertinent industry and market-specific information, followed by the computation of the cosine similarity of the text vectors generated from the reported data. A cosine similarity score approaching 1, coupled with an angle nearing 0, indicates a high degree of similarity between the two vectors. This study employs the cosine similarity formula for its analyses. It is crucial to systematically categorize documents into distinct folders according to their respective years, with each labeled by the company name and year, for example, ‘Ping An Bank: 2006’. When assessing the similarity between the years 2006 and 2005, the result will be designated as ‘Similarity_2006’. However, if the preceding year’s report is unavailable, the calculation will not be performed. In this paper, the cosine similarity formula is utilized to assess the similarity of sustainability reports:

$$Similarity={\displaystyle \frac{x_1{x}_2+y_1{y}_2}{\sqrt{{x_1}^2+{y_1}^2}\times \sqrt{{x_2}^2+{y_2}^2}}}$$

(1)

We find the number of sustainability reports of listed companies in China increases on an annual basis. In addition, the average value of similarity of annual sustainability reports is high, with a value greater than 0.6 being recorded (Figure 2). This trend suggests a growing recognition among companies of the importance of sustainability practices and transparency. However, the high similarity indicates a potential lack of originality and critical engagement with unique issues that face individual businesses. To foster genuine sustainability, companies must not only report but also innovate and tailor their strategies to address specific environmental and social challenges pertinent to their operations.

Moderator variables. This study measures a firm’s analyst attention (Attention) using data from CSMAR, focusing on both analyst and research report attention. Research report attention (Attention1) reflects the total number of reports tracking and analyzing the company in the same timeframe. Analyst attention indicates the number of analysts (Attention2) monitoring and analyzing the company within a year, with each team counted as one, regardless of its size. The ability of companies to innovate is assessed through their investments in research and development (Innovate1), as well as the number of personnel dedicated to R&D (Innovate2), both of which can be sourced from CSMAR.

Control Variables. Drawing on the previous literature of Bose et al. (2022) and Dhaliwal et al. (2011) [8,93], this paper adds the following control variables to the model: firm size (Size), litigation risk (Litg), return on assets (Roa), industry competition (Hhi), level of internationalization (Global), liquidity (Liquidity), Soe, Listage, Tobin Q, debt ratio (Lev), duality (Dual), board size (Board), and number of independent directors (Indep).

4.3. Model Setting

Drawing on Hoberg and Phillips (2016, 2018) [94,95], this paper constructs the following model to explore the impact of digital technology adoption on firm sustainability reports’ similarity:

$${Similarity}_{i,t}={\alpha }_{i,t}+{\beta }_1{DigiAdop}_{i,t}+{Controls}_{i,t}+{\delta }_t+{\theta }_k+{\gamma }_i+{\epsilon }_{i,t}$$

(2)

where the variable subscripts i, k, and t denote company, industry, and time, respectively. The explanatory variable ${Similarity}_{i,t}$ is the similarity of firm i in year t, the core explanatory variable ${DigiAdop}_{i,t}$ denotes the level of digital technology adoption of firm i in year t, and ${Controls}_{i,t}$ is the control variable mentioned above. Furthermore, this paper incorporates a series of fixed effects into the model. In light of the varying degrees of acceptance of digital technologies across different industries, this paper incorporates a set of fixed effects into the model. These are represented by ${\delta }_t$, ${\theta }_k$, and ${\gamma }_i$, which denote time fixed effects, industry fixed effects, and firm fixed effects, respectively. This paper focuses on the significance of the coefficient ${\beta }_1$ of ${DigiAdop}_{i,t}$ in model (1).

The significantly positive coefficient (${\beta }_1$) for digital technology adoption supports technology-driven disclosure convergence. This aligns with observed industry practices where digital tools standardize data collection (e.g., ESG metrics) and automate template-based report generation. Conversely, a significantly negative coefficient (${\beta }_1$) may indicate that technology empowers firms to transcend conventional disclosure paradigms. Such divergence could arise from innovation-enabling features like dynamic data visualization and interactive design elements, which amplify report distinctiveness through tailored stakeholder engagement interfaces.

On the basis of model (2) and referring to the model settings in Hoberg and Phillips (2016, 2018) [94,95], the moderator effect model (3) is obtained in this paper:

$$\begin{array}{cc}\hfill {Similarity}_{i,t}& ={\alpha }_{i,t}+{\beta }_2{DigiAdop}_{i,t}+{\beta }_3{Moderator}_{i,t}\hfill \\ & +{\beta }_4{DigiAdop}_{i,t}\times {Moderator}_{i,t}+{Controls}_{i,t}+{\delta }_t+{\theta }_k+{\gamma }_i\hfill \\ & +{\epsilon }_{i,t}\hfill \end{array}$$

(3)

This paper investigates the moderator of digital technology adoption on firms’ ESG performance as two moderating effects, including two moderator variables: analyst attention (Attention) and innovation capacity (Innovate). This paper focuses on the significance of the coefficient of the two moderator variables. In the case where ${\beta }_1$ is significant, if ${\beta }_4$ is significant, then the moderating effect is significant. If the sign of ${\beta }_1$ is the same as that of ${\beta }_4$, it is a positive effect, and if the sign is different, it is a negative effect.

5. Results

5.1. Descriptive Statistics

Table 1. Descriptive statistics of main variables.

Variables	Explanation of Variables	n	Mean	SD.	Min	Max
DigiAdop	The utilization of digital technology adoption.	9903	15.064	31.664	0	204
Breadth	The extensive scope of digital technology coverage.	9903	9.025	22.116	0	150
Depth	The degree to which digital technologies are integrated into business operations.	9903	5.734	11.615	0	72
Similarity	Similarity is defined as the extent to which a report from a given year corresponds with that of the prior year.	9903	0.776	0.301	0	0.991
Size	Natural logarithm of total assets.	9903	23.401	1.742	20.404	29.265
Lev	It stands for the balance sheet ratio of liabilities divided by assets.	9903	0.504	0.211	0.069	0.94
Roa	It is a measure of how much net profit is generated per unit of asset.	9903	0.046	0.056	−0.153	0.238
Board	Number of board members.	9903	2.203	0.226	1.609	2.833
Indep	Number of independent directors.	9903	0.375	0.055	0.308	0.571
Dual	It is a binary variable with a value of 1 if the Chairman and CEO are the same individual, and 0 otherwise.	9903	0.194	0.395	0	1
Tobin Q	As an important indicator to measure the company’s performance or the company’s growth.	9903	1.825	1.231	0.82	8.108
Hhi	HHI is a comprehensive index that measures industrial concentration.	9903	0.091	0.097	0.011	0.584
Liquidity	Liquidity is measured by taking the reciprocal of the illiquidity indicator.	9903	0.033	0.043	0.001	0.278
Litg	It is a binary indicator: 1 if litigation occurs, and 0 otherwise.	9903	0.147	0.354	0	1
Global	The level of internationalization is a binary indicator with a value of 1 if the company reports foreign revenue, and 0 otherwise.	9903	0.038	0.118	0	0.693
Soe	State-owned firms have a value of 1, and 0 otherwise.	9903	0.559	0.497	0	1
Listage	Listage = ln(current year − listed year + 1)	9903	13.396	7.158	1	28

presents the descriptive statistics for the primary variables. The mean value of digital technology adoption is 15.064, with a standard deviation of 31.664, indicating that over 75% of businesses have embraced digital technology. As noted by Cai et al. (2023), the adoption of digital technology is becoming increasingly prevalent [96]. This trend not only signifies a shift in operational strategies but also suggests a potential transformation in competitive dynamics. The mean of Similarity is 0.776, with a standard deviation of 0.301. This high similarity score indicates that many businesses are aligning their digital strategies, which could lead to homogenization in the market. As organizations increasingly adopt similar technologies, the unique competitive advantages they once held may diminish.

Table 2. Correlation coefficient of main variables.

	DigiAdop	Breadth	Depth	Similarity
DigiAdop	1
Breadth	0.935 ***	1
Depth	0.811 ***	0.584 ***	1
Similarity	−0.019 *	−0.020 **	−0.0150	1

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

presents the correlation coefficient of main variables. A negative correlation exists between DigiAdop and Similarity (β= −0.019, with p < 0.1), with a more pronounced negative correlation observed between Breadth and Similarity (β = −0.020, with p < 0.05). In contrast, Depth does not exhibit a significant relationship with Similarity. This disparity suggests that while the integration of advanced digital tools fosters unique reporting practices, merely increasing the frequency of technology use may not enhance differentiation. This raises intriguing questions about the potential for innovation in sustainability reporting methodologies.

5.2. Benchmark Regressions

Table 3. Digital technology adoption and firms’ ESG performance.

Variables	(1)	(2)	(3)
	Similarity	Similarity	Similarity
DigiAdop	0.0476 *
	(2.2305)
Breadth		0.0659 **
		(2.9929)
Depth			0.0073
			(0.4401)
Size	0.1173 **	0.1138 **	0.1247 ***
	(3.1599)	(3.0728)	(3.3529)
Lev	−0.0924	−0.0914	−0.0893
	(−0.6827)	(−0.6758)	(−0.6594)
Roa	−2.2850 ***	−2.2766 ***	−2.2925 ***
	(−8.6080)	(−8.5911)	(−8.6189)
Board	−0.0903	−0.0908	−0.0916
	(−0.9212)	(−0.9256)	(−0.9337)
Indep	−0.2169	−0.2080	−0.2398
	(−0.7748)	(−0.7437)	(−0.8559)
Dual	−0.0361	−0.0352	−0.0377
	(−1.0082)	(−0.9858)	(−1.0543)
Tobin Q	0.0350 *	0.0349 *	0.0346 *
	(2.2273)	(2.2259)	(2.1976)
Hhi	0.1113	0.1102	0.1061
	(0.3756)	(0.3722)	(0.3584)
Liquidity	−3.3480 ***	−3.3524 ***	−3.3646 ***
	(−8.7459)	(−8.7696)	(−8.7992)
Litg	0.0647 *	0.0643 *	0.0642 *
	(2.3469)	(2.3309)	(2.3284)
Global	0.1828	0.1705	0.1967
	(1.3104)	(1.2228)	(1.4087)
Soe	0.0319	0.0303	0.0346
	(0.3740)	(0.3547)	(0.4075)
Listage	0.1522 **	0.1546 **	0.1516 **
	(2.8761)	(2.9141)	(2.8674)
Constant	−4.2776 ***	−4.2296 ***	−4.4332 ***
	(−3.7096)	(−3.6662)	(−3.8480)
Firm	YES	YES	YES
Year	YES	YES	YES
Industry	YES	YES	YES
n	9579	9579	9579
R2-adj.	0.2312	0.2318	0.2307

Note: Firm, Year, and Industry represent fixed firm, year, and industry effects. “YES” indicates that the variable is controlled. * p < 0.1, ** p < 0.05, *** p < 0.01.

presents the core empirical evidence from our benchmark regressions, revealing significant dual-channel mechanisms through which digital technology adoption shapes sustainability reporting isomorphism. The coefficient on digital technology adoption (β = 0.0476, t = 2.2305) confirms our hypothesis that technological integration systematically reduces intra-organizational information asymmetries (Akerlof, 1970) [34]—particularly between operational units and strategic decision-makers—enabling more consistent ESG data aggregation and disclosure alignment. This finding extends the conventional information asymmetry paradigm by demonstrating how digital infrastructure transforms not merely information availability but its structural codification processes. Additionally, this evidences Bandura’s (1977) [52] social cognitive theory in action: organizations progressively emulate industry-leading disclosure templates visible through shared digital ecosystems, creating emergent reporting conventions. The magnitude suggests that each 10% expansion in a firm’s digital connectivity breadth increases reporting similarity by 0.66 standard deviations through observational learning channels. As stated by Shahzad et al. (2022) [78], these findings suggest a deeper relationship between digital tools and the standardization of sustainability reporting practices. Future research could explore how specific digital technologies, such as blockchain or AI-driven analytics, further influence reporting transparency and accuracy. This could lead to a more unified approach in sustainability disclosures, facilitating better stakeholder engagement and trust.

5.3. Endogeneity Test

IV method. To address the issue of endogeneity arising from the reciprocal relationship between sustainability reports’ similarity and digital technology adoption, this study utilizes the IV method to validate the robustness of the findings. The instrumental variable employed is DigiAdop_(t+1) and Breadth_(t+1) (

Table 4. Endogeneity test for benchmark regression.

Variables	(1)	(2)	(3)	(4)
	DigiAdop	Breadth	Similarity	Similarity
DigiAdop(t+1)	0.9888 ***
	(65.8526)
Breadth(t+1)		0.9928 ***
		(68.7237)
DigiAdop			0.0087 **
			(2.6529)
Breadth				0.0066 *
				(2.0654)
Constant	−0.6583 ***	−0.2402 *	0.9809 ***	0.9713 ***
	(−5.1863)	(−2.3192)	(13.5679)	(13.4473)
Kleibergen-Paap rk Wald F statistic	√	√	√	√
Cragg-Donald Wald F statistic	√	√	√	√
Stock-Yogo weak ID test critical values: 10% maximal IV size	16.38	16.38	16.38	16.38
Control	YES	YES	YES	YES
Year	YES	YES	YES	YES
Industry	YES	YES	YES	YES
n	8185	8185	8185	8185
R2-adj.	0.8731	0.8615	0.0628	0.0627

Note: Firm, Year, and Industry represent fixed firm, year, and industry effects. “YES” indicates that the variable is controlled. Standard errors in parentheses. A “√” indicates that the examination has been successfully completed. * p < 0.1, ** p < 0.05, *** p < 0.01.

). The results indicate that the effect of firm digital technology adoption on sustainability reports’ similarity remains significant, with a positive regression coefficient.

Exogenous impacts. This paper draws on a series of practices of double-difference model (DID) [97], and adopts the exogenous shock of the implementation of the strategy of “National Big Data Comprehensive Pilot Zone” as a proxy for measuring the level of digital technology adoption. The specific settings are as follows: if the city where the firm is located is the pilot implementation city of “National Big Data Comprehensive Pilot Zone”, ${Post}_t$ is assigned to 1, otherwise it is assigned to 0; in the year of the implementation of the strategy of “National Big Data Comprehensive Pilot Zone” and after, ${Treat}_i$ is assigned to 0, and ${Treat}_i$ is assigned to 1. ${Treat}_i$ is assigned as 1, otherwise it takes the value of 0; the interaction term ${Treat}_i\times {Post}_t$ denotes the firms that are impacted by the implementation of “Smart City Pilot Policy” strategy.

$${Similarity}_{i,t}={\alpha }_{i,t}+{\beta }_1{Treat}_i\times {Post}_t+{Controls}_{i,t}+{\delta }_t+{\theta }_k+{\gamma }_i+{\epsilon }_{i,t}$$

(4)

Table 5. Exogenous impacts.

Variables	(1)	(2)
	Similarity	Similarity
${Treat}_i\times {Post}_t$	0.1277 **	0.1293 ***
	(3.2713)	(3.3237)
Constant	0.0718 ***	−4.3832 **
	(5.0557)	(−2.6339)
Control	NO	YES
Year	YES	YES
Industry	YES	YES
Firm	YES	YES
n	8071	8071
R2-adj.	0.1986	0.2222

Note: Year, Industry, and Firm represent fixed effects. “YES” indicates that the variable is controlled. ** p < 0.05, *** p < 0.01.

shows the impact of digital technology adoption on the similarity of sustainability reports under policy shocks. The interaction term (${Treat}_i\times {Post}_t$) coefficients in columns (1) and (2) are 0.1277 (p < 0.05) and 0.1293 (p < 0.01), respectively, demonstrating a statistically significant positive effect of the National Big Data Comprehensive Pilot Zone policy on firm similarity. This implies a 12.8–12.9 percentage point increase in similarity among treated firms, likely driven by technology standardization, data sharing, or competitive mimicry. The inclusion of control variables strengthens model validity, as evidenced by the adjusted R² improvement from 0.1986 to 0.2222. These findings suggest that digital infrastructure and policy experimentation can foster technological collaboration and market integration. However, policymakers should balance these benefits against risks of excessive homogenization in innovation ecosystems.

5.4. Robustness Test

Four other methods are used in this study to ensure the robustness of the findings. Firstly, VIF is used to test for multicollinearity and all VIFs are less than 10, indicating that the model does not suffer from multicollinearity. Second, we group firms according to the nature of their shareholding.

Table 6. Robustness tests—group testing. (A) Grouping tests according to the nature of the shareholding. (B) Grouping tests according to the technological level of firms.

(A)
Variables	(1)	(2)	(3)	(4)	(5)	(6)
	Similarity	Similarity	Similarity	Similarity	Similarity	Similarity
DigiAdop	0.0297	0.0662 *
	(0.9302)	(2.1960)
Breadth			0.0434	0.0894 **
			(1.2243)	(3.0170)
Depth					0.0082	0.0061
					(0.4091)	(0.2263)
Constant	−1.1392	−7.8879 ***	−1.0949	−7.8865 ***	−1.2181	−8.0681 ***
	(−0.7663)	(−4.6963)	(−0.7375)	(−4.6994)	(−0.8167)	(−4.7882)
Control	YES	YES	YES	YES	YES	YES
Soe	1	0	1	0	1	0
Firm	YES	YES	YES	YES	YES	YES
Year	YES	YES	YES	YES	YES	YES
Industry	YES	YES	YES	YES	YES	YES
n	5422	4119	5422	4119	5422	4119
R2-adj.	0.2491	0.2239	0.2493	0.2250	0.2489	0.2227
(B)
Variables	(1)	(2)	(3)	(4)	(5)	(6)
	Similarity	Similarity	Similarity	Similarity	Similarity	Similarity
DigiAdop	0.0513 *	0.0166
	(1.9759)	(0.4004)
Breadth			0.0769 **	0.0137
			(2.9846)	(0.2918)
Depth					−0.0117	0.0217
					(−0.5032)	(0.9153)
Control	−4.5646	−3.8407 **	−4.4435	−3.8484 **	−5.0192	−3.8270 **
	(−1.4051)	(−2.7938)	(−1.3640)	(−2.8002)	(−1.5478)	(−2.7809)
Control	YES	YES	YES	YES	YES	YES
High-tech industry	1	0	1	0	1	0
Firm	YES	YES	YES	YES	YES	YES
Year	YES	YES	YES	YES	YES	YES
Industry	YES	YES	YES	YES	YES	YES
n	4367	5194	4367	5194	4367	5194
R2-adj.	0.2160	0.2418	0.2174	0.2418	0.2151	0.2419

Note: Year, Industry, and Firm represent fixed effects. “YES” indicates that the variable is controlled. * p < 0.1, ** p < 0.05, *** p < 0.01.

validates the heterogeneous effects of digital transformation through ownership-based subgroup analyses. For non-state-owned enterprises (NSOEs), digital technology adoption (β = 0.0662, p < 0.1) and technology breadth (β = 0.0894, p < 0.05) significantly increase report similarity by 6.6–8.9 percentage points per unit improvement, underscoring market-driven technological diversification as a convergence mechanism. In contrast, SOEs exhibit no statistically meaningful responses, likely constrained by institutional rigidities and administrative interventions that suppress digitalization’s homogenizing forces. These findings necessitate policy differentiation: prioritizing NSOE-oriented digital collaboration platforms while reforming SOE governance structures to mitigate path dependency.

Thirdly, the control variables are adjusted.

Table 7. Robustness tests—changing control variables.

Variables	(1)	(2)	(3)
	Similarity	Similarity	Similarity
DigiAdop	0.0546 *
	(2.5127)
Breadth		0.0703 **
		(3.1381)
Depth			0.0138
			(0.8110)
Size	0.1566 ***	0.1535 ***	0.1650 ***
	(4.3773)	(4.3045)	(4.6079)
Lev	−0.1003	−0.0996	−0.0970
	(−0.7474)	(−0.7419)	(−0.7219)
Roa	−1.9619 ***	−1.9536 ***	−1.9721 ***
	(−7.5542)	(−7.5346)	(−7.5741)
Board	−0.0726	−0.0731	−0.0735
	(−0.7331)	(−0.7385)	(−0.7421)
Indep	−0.1573	−0.1500	−0.1817
	(−0.5606)	(−0.5349)	(−0.6471)
Dual	−0.0490	−0.0483	−0.0508
	(−1.3675)	(−1.3487)
(−1.4193)	0.0238	0.0223	0.0268
	(0.2798)	(0.2620)	(0.3164)
Constant	−3.2183 ***	−3.1483 ***	−3.4080 ***
	(−3.8275)	(−3.7523)	(−4.0502)
Control	YES	YES	YES
Year	YES	YES	YES
Firm	YES	YES	YES
Industry	YES	YES	YES
n	9579	9579	9579
R2-adj.	0.2158	0.2164	0.2152

Note: Year, Industry, and Firm represent fixed effects. “YES” indicates that the variable is controlled. * p < 0.1, ** p < 0.05, *** p < 0.01.

confirms model robustness under constrained controls (n = 7), with DigiAdop (β = 0.0546, p < 0.1) and Breadth (β = 0.0703, p < 0.05) sustaining directional persistence. DigiAdop reveals information asymmetry reduction mechanisms—standardized disclosures lower verification costs through machine-readable formatting. Three theoretically grounded patterns emerge: (1) Scaling as Social Learning. Size effects (β = 0.1535–0.1650, p < 0.01) demonstrate how observational learning dynamics transform large firms into institutionalized templates, creating network effects that marginalize idiosyncratic disclosures. (2) Profit Signaling Equilibrium. The ROA paradox (β = −1.9536–1.9721, p < 0.01) reflects strategic information asymmetry maintenance, where high performers deliberately complexify disclosures using proprietary digital architectures to signal competitive advantage. (3) Governance mechanisms (e.g., board independence or ownership structures) exhibit negligible explanatory power in driving disclosure convergence, challenging agency theory assumptions of monitoring-driven behavioral alignment (p > 0.1 across all specifications).

Finally, model adjustments.

Table 8. Robustness tests—model adjusted.

Variables	(1)	(2)	(3)
	Similarity	Similarity	Similarity
DigiAdop	0.0401
	(1.7098)
Breadth		0.0627 **
		(2.6074)
Depth			−0.0032
			(−0.1808)
Constant	−3.2937 **	−3.2595 **	−3.3929 **
	(−2.8027)	(−2.7716)	(−2.8887)
Firm	YES	YES	YES
Year	YES	YES	YES
Industry	YES	YES	YES
Industry × Year	YES	YES	YES
n	9593	9593	9593
R2-adj.	0.2164	0.2169	0.2160

Note: Year, Industry, and Firm represent fixed effects. “YES” indicates that the variable is controlled. ** p < 0.05.

confirms that Breadth (β = 0.0627, p < 0.05) persistently drives cross-firm similarity even after controlling for industry-year fixed effects, while DigiAdop and Depth show null effects. These results advance information asymmetry theory by demonstrating how broad digital integration resolves interorganizational knowledge disparities: standardized data ecosystems and interoperable platforms enable firms to systematically observe rivals’ capabilities, reducing uncertainty and accelerating social learning that sustains competitive isomorphism. Critically, the insignificance of adoption intensity and specialization depth challenges conventional assumptions about technological lock-in. We posit strategic convergence arises when digital architectures facilitate mass vicarious learning (Bandura, 1977) [52], not incremental technological accumulation.

The effect’s stability across heterogeneous contexts highlights two mechanisms. First, pervasive digital resources establish “information commons” [57], lowering imitation costs while countering Akerlof’s (1970) “lemons market” asymmetries [34]. Second, breadth-driven transparency converts proprietary operational knowledge into public benchmarks, allowing firms to bypass trial-and-error through cross-organizational comparisons. This dual mechanism—simultaneously enabling imitation and disincentivizing differentiation—explains why scope dominates depth in homogenization dynamics.

5.5. Analysis of Moderating Effects

5.5.1. Digital Technology Adoption, Analyst Attention, and the Similarity of Sustainability Reports

The results presented in

Table 9. Digital technology adoption, analyst attention, and the similarity of sustainability reports.

Variables	(1)	(2)	(3)	(4)
	Similarity	Similarity	Similarity	Similarity
DigiAdop	0.0586 *		0.0551 *
	(2.4133)		(2.2424)
Attention1	0.0183	0.0172
	(0.9615)	(0.9106)
DigiAdop × Attention1	−0.0353 *
	(−2.5627)
Attention2			0.0021	0.0003
			(0.1059)	(0.0133)
DigiAdop × Attention2			−0.0235
			(−1.7922)
Breadth		0.0786 **		0.0735 **
		(3.1543)		(2.9047)
Breadth × Attention1		−0.0299 *
		(−2.0986)
Breadth × Attention2				−0.0160
				(−1.2258)
Constant	−6.3359 ***	−6.2752 ***	−6.5769 ***	−6.5226 ***
	(−4.6312)	(−4.5988)	(−4.8026)	(−4.7721)
Control	YES	YES	YES	YES
Year	YES	YES	YES	YES
Industry	YES	YES	YES	YES
Firm	YES	YES	YES	YES
n	7929	7929	7929	7929
R2-adj.	0.2295	0.2299	0.2290	0.2294

Note: Year and Industry represent fixed year and industry effects. “YES” indicates that the variable is controlled. * p < 0.1, ** p < 0.05, *** p < 0.01.

demonstrate that analysts are expressing concerns regarding the repercussions of the diminution in the adoption of digital technologies on the similarity of sustainability reports. The coefficients of the cross-multiplier terms analysts’ concerns, namely −0.0353 and −0.0299, have been determined to be statistically significant at the 10 per cent level. These findings lend support to the predictions made in Hypothesis 2.

Information asymmetry mitigation mechanism. Analyst coverage reduces institutional investors’ information disadvantages (Diamond, 1985) [98], creating market incentives for firms to two sides. On the one hand, deemphasize isomorphic disclosure patterns. By fostering diverse reporting strategies, firms can enhance their competitive edge, allowing for differentiated insights that better reflect their unique value propositions. This shift not only aids in reducing information asymmetry but also encourages a more nuanced understanding among investors. Consequently, firms that embrace tailored disclosure practices may attract a broader investor base, ultimately enhancing market liquidity and stability.

On the other hand, amplify firm-specific ESG innovations through enhanced granular disclosure that highlights their unique sustainability initiatives. By showcasing distinct ESG practices, firms can not only distinguish themselves from competitors but also align their narratives with evolving investor preferences. This strategic communication enables stakeholders to make informed decisions based on comprehensive insights, fostering a culture of transparency. As a result, organizations that prioritize such disclosures are likely to build stronger relationships with their investors, creating a virtuous cycle of trust and engagement that benefits both parties. Furthermore, this divergence in reporting practices may lead to a fragmented understanding of sustainability performance across industries. As analysts emphasize the importance of distinct firm characteristics, it becomes critical to explore how these differences influence stakeholder perceptions and decision-making processes.

5.5.2. Digital Technology Adoption, Innovation, and the Similarity of Sustainability Reports

As demonstrated in

Table 10. Digital technology adoption, innovation, and the similarity of sustainability reports.

Variables	(1)	(2)	(3)	(4)
	Similarity	Similarity	Similarity	Similarity
DigiAdop	0.0667 **		0.0899 *
	(2.5920)		(2.4029)
Innovate1	−0.0077	−0.0063
	(−0.5128)	(−0.3883)
DigiAdop × Innovate1	−0.0233 *
	(−2.1963)
Innovate2			−0.0297	−0.0333
			(−0.9986)	(−1.1023)
DigiAdop × Innovate2			−0.0313 **
			(−3.0008)
Breadth		0.0865 ***		0.1159 **
		(3.2923)		(3.1029)
Breadth × Innovate1		−0.0216 *
		(−2.2479)
Breadth × Innovate2				−0.0317 **
				(−3.1507)
Constant	−5.9156 ***	−5.7763 ***	−8.2437 ***	−8.1628 ***
	(−3.6978)	(−3.6074)	(−3.9979)	(−3.9552)
Control	YES	YES	YES	YES
Year	YES	YES	YES	YES
Industry	YES	YES	YES	YES
Firm	YES	YES	YES	YES
n	6997	6997	4855	4855
R2-adj.	0.2399	0.2407	0.2803	0.2814

Note: Year and Industry represent fixed year and industry effects. “YES” indicates that the variable is controlled. * p < 0.1, ** p < 0.05, *** p < 0.01.

, the innovativeness of firms has a mitigating effect on the impact of digital technologies on the similarity of sustainability reports. In accordance with the social learning theory, corporate innovation disrupts the paradigm of imitative learning, thereby attenuating the similarity of content in sustainability reports. Grounded in information asymmetry perspectives, our analysis suggests that innovative capacity serves as a critical organizational filter that disrupts mimetic isomorphism in sustainability disclosure practices. The negative coefficients of the interaction terms between digital adoption and R&D investment (−0.0233 and −0.0216, p < 0.10) and R&D personnel (−0.0313 and −0.0317, p < 0.05) empirically validate this moderating effect.

This finding extends information asymmetry theory by demonstrating how innovation-driven informational advantages (1) increase firms’ capacity for differentiated sustainability narratives, and (2) raise the costs of imitation for less innovative competitors. The results suggest that technologically advanced firms develop unique network positions that facilitate access to heterogeneous sustainability data, thereby reducing reliance on industry-standard disclosure templates. Consequently, this unique positioning enables these firms not only to craft compelling narratives but also to influence market perceptions and consumer behavior significantly. Moreover, this unique positioning fosters an environment where firms can engage in proactive stakeholder communication. By leveraging their technological prowess, they can tailor their sustainability narratives to resonate deeply with specific consumer segments, thus enhancing brand loyalty. This dynamic not only solidifies their market leadership but also sets a benchmark for sustainability practices within the industry, encouraging a shift towards more innovative and transparent approaches across the board.

This suggests that increased investment in R&D and a larger pool of R&D personnel not only foster innovation but also contribute to a more diverse array of sustainability reporting practices. Such diversity can enhance corporate transparency and stakeholder engagement, as firms differentiate their sustainability narratives. This shift may also encourage industry-wide best practices, where innovative firms lead the way, prompting others to follow suit in a more tailored approach to sustainability, ultimately enriching the discourse around corporate responsibility.

6. Conclusions and Policy Implications

6.1. Conclusions

The digital revolution has fundamentally reconfigured the disclosure mechanism of corporate sustainability, bringing both transformational opportunities and the risk of structural homogenization. Using computational linguistics, this study analyzes 9903 sustainability reports (2009–2021) of Chinese listed companies and finds significant convergence in report content, with an average cosine similarity of 0.776. Our double fixed-effects model demonstrates three key relationships: (1) digital technology adoption is positively correlated with disclosure similarity (β = 0.0476, p < 0.1). (2) Analyst attention plays a negative moderating role, reducing isomorphism by 3.5% (p < 0.1). (3) Firm innovation (R&D investment) (p < 0.1) may have led to metrics transformation and facilitated disclosure differentiation (Figure 3).

These findings advance the information asymmetry theory through how template-driven digital technologies can improve data accessibility while limiting formal uniqueness. Digital technology applications reduce information asymmetries within organizations [34], particularly between operating units and strategic decision-makers, leading to more consistent aggregation and disclosure of ESG data. This finding extends the traditional information asymmetry paradigm by demonstrating that digital infrastructures change not only the availability of information but also the process of structural codification of information.

The similarity of corporate sustainability reports is higher for companies that have been established for a long time and are large-scale. The scale effect (β = 0.1535–0.1650, p < 0.01) demonstrates how observational learning dynamics can transform large-scale firms into institutionalized templates, which creates a network effect that marginalizes personalized disclosure. It also demonstrates Bandura’s (1977) [52] social cognitive theory at work: firms progressively mimic industry-leading disclosure templates through a shared digital ecosystem. If analysts were to increase their focus on companies, they would reformulate their sustainability reports to be differentiated according to the actual situation of the company. This empirical evidence suggests that strategic digital transformation, combined with robust market governance and innovation ecosystems, can build China into a global architect of both standardized and differentiated sustainability reporting standards. This dual advancement will facilitate cross-border investment decisions while preserving the veracity of corporate disclosures in emerging markets.

Our research results make contributions between digital technology adoption and information disclosure through two sides. We demonstrate that digital technology adoption creates a paradoxical duality—while reducing asymmetries within organizations, it also enhances the ability to differentiate between firms. This extends Akerlof’s framework [34] by revealing how digital technology adoption transforms information markets from “lemon” equilibria to differentiated quality signaling markets, where analyst attention and firms’ ability to innovate become the main breakthroughs. Furthermore, we propose new dynamics of social learning. We find two evolutionary patterns of disclosure practices. One is imitative convergence: due to technological constraints or policy interventions, it is possible to imitate and learn from the reporting content of industry-leading firms (Bandura’s modeling approach [52]). The second is strategic differentiation: analysts’ focus on review and firm innovation can drive firm differentiation, emphasizing firms’ own idiosyncratic information while combating greenwashing.

6.2. Policy Implications

This study examines the sustainability reports of listed firms in China, focusing on the impact of digital technology development in emerging economies on the preparation of these reports. The use of digital tools often results in a higher degree of similarity in report content, as firms tend to rely on standardized templates and structures, which can obscure their unique identity. This trend not only highlights the widespread adoption of digital technologies in report preparation but also uncovers potential opportunities for innovation that firms may overlook while pursuing compliance and standardization.

As digital tools become more prevalent, firms frequently adopt industry best practices and commonly used templates for sustainability reporting. While this approach enhances the consistency and comparability of reports, it may also lead to the homogenization of information, thereby inhibiting the expression of a company’s distinct values and strategic objectives. Additionally, the application of digital technology, though it improves reporting efficiency and transparency to some extent, may encourage firms to adopt a conservative stance regarding information disclosure. To meet external expectations, companies may focus on easily quantifiable environmental and social indicators, often neglecting softer, more subjective aspects such as corporate culture, innovation capacity, and social responsibility. This oversight can result in stakeholders misinterpreting a company’s sustainability efforts, preventing the full realization of its intrinsic value.

This study presents three pivotal policy recommendations, with significant implications for enhancing organizational legitimacy and global capital markets. Firstly, the development of adaptive AI architectures that integrate entity-specific sustainability indicators into standardized reporting frameworks is essential. This preserves the unique sustainability nuances of individual entities while ensuring comparability across the board. Secondly, institutionalizing third-party analyst feedback mechanisms in materiality determination processes can considerably enrich the evaluation of sustainability metrics and ensure a more robust and credible assessment. Thirdly, strategically converting R&D outputs into proprietary sustainability metrics can drive innovation whilst maintaining a competitive edge in sustainability reporting. These measures not only enhance report comparability but also suggest improved transparency may mitigate emerging market discounts for Chinese firms in international investments. This research thus contributes substantively to digital governance by advocating for a balanced approach that harmonizes standardization with innovation in sustainability communication.

6.3. Limitations and Future Reaches

This paper identifies several limitations. Firstly, the focus on Chinese A-share listed companies significantly restricts the generalizability of the findings. It overlooks regional policy differences and regulatory requirements. The advent of novel environmental regulations might call for an enlargement in the scope of corporate environmental measures and compliance disclosures, thereby giving rise to a more comprehensive and transparent reporting process. Gunawan et al. (2022) explore the role of policy in shaping report content [99]. Secondly, the paper exclusively examines the impact of industry on robustness tests, while neglecting to provide a comprehensive discussion of the underlying causes of variability. Popkova et al. (2022) find that the focus of corporate reporting may be influenced by economic and environmental policies that differ across regions [100]. Finally, the methodology employed was that of a study by Peterson et al. (2015) [92] which utilized the cosine similarity algorithm to analyze the text similarity in annual 10-K reports. The objective of this analysis was to assess the needs for sustainability reporting, and it was determined that the match requires enhancement.

Scholars have highlighted the importance of advanced digital technologies, such as big data and blockchain, in enhancing data analytics [101]. The utilization of big data analytics and NLP technologies enables companies to generate reports that are consistent. To illustrate this point, consider the processing and analysis of voluminous textual data employing NLP techniques. Such practices can enhance the quality of sustainability reports, ensuring consistency and standardization [79].

Furthermore, the interplay between NLP and data visualization tools can facilitate a more nuanced understanding of sustainability metrics. By transforming complex datasets into accessible visual formats, stakeholders can quickly grasp the implications of a company’s sustainability efforts. This democratization of information not only empowers consumers but also encourages companies to improve their practices in response to informed feedback.

In conclusion, the convergence of big data, blockchain, and NLP is reshaping the landscape of sustainability reporting. As these technologies continue to evolve, they will play a pivotal role in fostering an environment where transparency, accountability, and innovation are paramount in driving sustainable business practices.