Symbolic or Substantive? The Effects of the Digital Transformation Process on Environmental Disclosure

Abstract

This study improves our comprehension of the relationship between the digital transformation process and environmental disclosure in emerging economies. Specifically, it delves into the effects of digital transformation on both symbolic and substantive environmental disclosure types through the application of text mining methods. Additionally, this research examines how these impacts are moderated by the political experience of senior managers. Drawing on a panel dataset of 2033 listed Chinese manufacturing firms over the period from 2009 to 2020, the findings reveal that (1) the digital transformation process is negatively associated with substantive environmental disclosure and (2) the senior managers’ political experience positively moderates the relationship between the digital transformation process and symbolic environmental disclosure. Several supplementary analyses were also conducted to enrich these results. The implications of this work may have substantial value for practitioners, policymakers, and researchers operating within the industrial sector.

1. Introduction

Over the last decade, environmental issues have gained increasingly widespread recognition and prominence in the public’s consciousness on a global scale. In response to growing stakeholder pressure, many firms have acknowledged this shift and begun to disclose information regarding their environmental initiatives. However, the quality of such environmental disclosure has been sharply criticized by scholars due to the tendency to center on vision and environmental strategy claims rather than providing specific environmental performance indicators or credible environmental reports [1,2,3]. This issue has been observed in both developed countries and emerging economies [4,5], and is particularly significant in emerging economies, where robust market supervision mechanisms are not yet fully established [6,7]. Researchers have employed the terms “symbolic” and “substantive” to differentiate between these two types of environmental disclosure [7,8,9]. Enhancing the substantive environmental disclosure efforts while curbing symbolic disclosure efforts has become a pivotal concern in the field of environmental management.

Digital transformation, a concept characterized by the adoption of digital technologies, has gained significant research attention in environmental management in recent years. It has the potential to profoundly impact how businesses operate and innovate, with far-reaching consequences on markets, economies, and societies [10,11]. Previous studies have yielded optimistic results regarding the potential of the digital transformation process to enhance corporate social responsibility (CSR) [12,13,14], especially the performance of environmental disclosure efforts [15,16]. They argue that the digital transformation process may have CSR performance benefits through mechanisms such as enhancing the operation efficiency, reducing the information costs, and allowing swift responses to stakeholders’ demands [17,18,19,20].

However, these contributions often overlook the multi-dimensional nature of CSR, leading to controversial empirical evidence [13,21] and restricting our understanding of the relationship between the digital transformation process and environmental disclosure, a typical CSR practice. Environmental disclosure can be classified into different types based on its symbolic or substantive nature. However, there is a lack of empirical evidence on how a digital transformation might affect one or both of these types of disclosure. Addressing this knowledge gap is crucial to comprehending the complex role of the digital transformation process in today’s CSR field, particularly in the context of emerging economies.

This study aims to bridge this research gap by utilizing text analysis methods to examine the effects of digital transformations on both substantive and symbolic environmental disclosure types. Considering the significant role of the senior managers’ political background in the digital transformation process, as well as environmental disclosure, especially in emerging economies, we also explore the moderating effect of the senior managers’ political experience on these relationships. In a series of supplementary analyses, we further explore the impacts of different digital transformation types on environmental disclosure, as well as other potential boundary conditions.

To test our theoretical framework, we use panel data composed of 2033 publicly listed Chinese manufacturing firms spanning from 2009 to 2020. China serves as our research setting due to its standing as one of the world’s most polluted countries and its notable progress in terms of sustainable development [22]. In the meanwhile, China has also made significant achievements in the area of industrial digitalization in recent years [23,24]. The findings of this paper can be extrapolated to comparable transitional and emerging economies, where the economic growth patterns are transitioning towards sustainable development.

The remainder of this paper is organized as follows. Section 2 reviews the literature and presents the hypotheses. Section 3 introduces the methodology applied in this study, including descriptions of the variables, the sample data, the measurements, and the fixed effects model we used to test the main hypotheses. Section 4 presents the empirical results, consisting of descriptive statistics, a regression analysis, a robustness analysis, and additional analyses. In Section 5, we present our conclusions and a further discussion about the paper.

2. Literature Review and Hypotheses

2.1. Environmental Disclosure as a Strategic Response

Environmental disclosure plays a crucial role in corporate social responsibility (CSR), encompassing a firm’s environmental achievements. It serves as a means of product differentiation, appealing to specific market segments due to the attractive attributes it provides [25], as well as a tool for gaining legitimacy. Legitimacy is essential for firms to acquire valuable resources such as technology, talent, financial resources, and government support [26,27]. Previous studies have indicated that environmental disclosure can complement administrative rules and encourage firms to improve their environmental performance by subjecting them to public scrutiny [28,29,30]. Others suggest, however, that environmental disclosure is a poor substitute for binding regulations and allows firms to benefit from a positive reputation without making real or lasting environmental investments [3,31,32]. This phenomenon is particularly significant in emerging markets, where stable and credible legal frameworks are often lacking compared to developed economies [6,33].

The strategic responsiveness literature [34,35] provides models describing a variety of strategies for responding to institutional pressure. The current literature primarily contains the terms “symbolic” and “substantive” to differentiate the two types of environmental disclosure [8,9,36,37]. Symbolic environmental disclosure represents a socially constructed interpretation aimed at projecting conformity with societal norms, although it often lacks substantial implementation efforts. In contrast, substantive environmental disclosure reflects a firm’s commitment to formulating effective response measures that mitigate the negative environmental impacts and help in pursuing tangible solutions. Scholars have identified several drawbacks to symbolic environmental disclosure, including the tragedy of the commons, free riding, information asymmetries, and greenwashing [1,38,39]. Accordingly, scholars have extensively investigated the driving and inhibiting factors of both symbolic and substantive environmental disclosure from economic and institutional perspectives [1,40]. Some have also attempted to measure these two types of environmental disclosure as a basis for empirical research [9,41,42].

Amidst the digital transformation researchers, there are high expectations for the potential for digital transformation efforts to provide enhanced CSR performance [13,14,43]. Specifically, they argue that digital transformation efforts can lead to improved environmental disclosure performance through various mechanisms, such as by enhancing a business’ resource utilization efficiency, operational capability, and information transparency [15,16]. However, these studies often overlook the heterogeneity of CSR, treating it as a composite index with limited consideration of its subdimensions. This research gap has led to mixed and controversial empirical evidence. Some studies argue that digital transformation efforts can lead to improved CSR performance by creating efficiency advantages [12,44], while others maintain that the high investment costs and uncertainty associated with digital transformation efforts create tensions and impede other firm-level projects [21,24]. Consequently, additional research is necessary to deepen the understanding of the link between digital transformation efforts and environmental disclosure. It remains unclear whether the digital transformation process positively affects symbolic or substantive environmental disclosure efforts, which calls for more in-depth research.

2.2. Digital Transformation Efforts and Environmental Disclosure

This paper argues that the digital transformation process may positively influence symbolic environmental disclosure efforts. The digital transformation process encompasses the integration of digital technologies to enhance or innovate a firm’s current market offerings or to develop novel business models [20]. Hence, it may enrich a firm’s environmental strategy claims and long-term planning vision, which are key components of symbolic environmental disclosure [9,41]. For instance, implementing artificial intelligence (AI) initiatives can help firms better identify stakeholders’ interests, making symbolic environmental disclosure efforts more attractive to the market [45]. Additionally, digital transformation efforts can help reduce information costs by optimizing the information production, gathering, analysis, and reporting processes [20,46]. This efficiency improvement can benefit a firm in broadcasting its environmental initiatives.

H1a. 

The digital transformation process is positively related to symbolic environmental disclosure.

This paper further suggests that the digital transformation process may have a curvilinear relationship with substantive environmental disclosure. The digital transformation process may have efficiency and energy-saving advantages, which would benefit substantive environmental disclosure efforts. For instance, applying digital technologies such as big data and analytics initiatives can lead to improved production flexibility [47,48], reduced set-up costs and errors [49], and shortened machine downtimes [50]. Applying digital technologies such as digital twinning and AI may also benefit environmental innovation initiatives by fastening and broadening the knowledge creation process [46]. However, digital technologies are composed of a series of enabling technologies; their adoption not only involves the modification or replacement of existing production or operation systems [51,52] but also requires skilled and multidisciplinary work teams to minimize the risks inherent to emerging technologies [53]. As a result, excess digital transformation efforts may cause increased risks and dilute the firm’s resources for existing projects, which may lead to reduced environmental protection investments.

H1b. 

The digital transformation process has an inverted-U shaped relationship with substantive environmental disclosure.

2.3. Moderating Role of Senior Managers’ Political Experience

A firm’s strategic behaviors are significantly influenced by the characteristics of their senior managers, including their demographics (e.g., age, gender, education, career background), psychological attributes (e.g., hubris, narcissism, overconfidence), and managerial cognition and abilities [54,55,56]. Recent studies have highlighted the role of senior managers’ attributes in digital transformation and environmental disclosure efforts [57,58]. For instance, Ji et al. [52] found that the social capital of senior managers can used to positively leverage the relationship between digital transformation efforts and financial performance.

Firms can achieve legitimacy and consequently obtain government endorsements by synchronizing their strategic approaches with government bodies [6,59]. In the past decade, government agencies in many countries have encouraged digital transformation efforts; examples include Germany’s “High-Tech Strategy 2020”, the “Advanced Manufacturing Partnership” in the United States, and “Made in China 2025” in China [47]. The adoption and implementation of digital technologies, therefore, is an important way to gain favorable policies and exclusive resources. In emerging economies such as China, governments not only direct their economic activities by crafting industry development strategies and establishing regulatory measures [33] but also manage substantial shares of scarce resources, including land, bank credits, subsidies, and tax incentives [7]. Therefore, we believe political experience within senior management helps firms in building stronger connections with their governments, thereby leveraging government support for improved business outcomes.

“Senior managers’ political experience” refers here to the members of a firm’s top management team having working experience in government agencies. On one hand, senior managers’ political experience can help firms better understand policy intentions and government expectations, thereby informing their optimization efforts for digital transformation models, which may ultimately result in symbolic environmental disclosure benefits. On the other hand, substantive environmental practices such as environmental initiatives naturally face externality problems [60]; that is, substantive environmental practices can yield communal advantages, although the cost cannot be fully shouldered by the focal firm’s competitors. Moreover, as mentioned above, digital transformation efforts can require significant investment. Having senior managers with strong political experience may help firms in building formal and informal connections with government agencies [61], helping them gain more government support and reducing the digital transformation costs, as well as helping in substantive environmental practices.

H2a. 

Senior managers’ political experience positively moderates the relationship between digital transformation efforts and symbolic environmental disclosure.

H2b. 

Senior managers’ political experience positively moderates the relationship between digital transformation efforts and substantive environmental disclosure.

The conceptual framework utilized in this study is depicted in Figure 1.

3. Methodologies

3.1. Sample Data

The sample encompassed manufacturing firms publicly traded on China’s stock exchanges between 2009 and 2020. This period covers the 12th and 13th Five-Year Plans, during which the central government issued a series of national programs to promote industrial upgrades towards sustainable development [11]. Our sample was collected according to the following criteria: (1) we selected manufacturing firms from the China Stock Market and Accounting Research Database (CSMAR) using the industry classification guidelines issued by the China Securities Regulatory Commission; (2) we excluded firms with data available for less than half of the observation period, in line with previous studies [62,63]; (3) we removed firms labeled as “special treatment” (ST) and *ST to prevent extreme values. The designation ST is assigned to high-risk firms that have experienced consecutive losses for a certain number of years. The final sample included 2033 firms with 15,861 observations.

3.2. Measures

3.2.1. Dependent Variables: Symbolic and Substantive Environmental Disclosure

We employed the text mining method to measure the variables relevant to the two types of environmental disclosure and digital transformation. This approach allowed for a comprehensive and structured analysis of the data, avoiding a reliance on subjective survey responses or other potentially biased items. Text mining methods, coupled with natural language processing techniques, have seen widespread use in strategic management research to extract keywords from documents and capture firms’ strategic activities [52,64].

We applied this method to develop the variables for symbolic and substantive environmental disclosure in a step-wise process. First, we constructed separate term dictionaries for symbolic and substantive environmental disclosure. In line with Li [65] and Li et al. [66], who established a term dictionary for environmental disclosure in the context of China, we used the keywords related to environmental strategic information to represent symbolic environmental disclosure and the keywords related to environmental action information to represent substantive environmental disclosure. A list of the English versions of these keywords is given in Appendix A.

Second, we uploaded the term dictionary into the “jieba” toolkit in the Python package. We then applied a machine learning approach for the text analysis based on the annual report for each firm. The words from an annual report reflect a firm’s strategic investments and blueprint [52] and will contain abundant information regarding symbolic and substantive environmental disclosure.

Third, in line with previous studies [52], we used the sum of the keyword frequency rates divided by the length of a firm’s annual report to measure the dependent variables. As not all keywords were equally weighted, we constructed the following equation to adjust the frequency rates of each keyword in a given report [67]:

$$w_{it}=\left\{\begin{array}{c}\frac{1+log{f}_{it}}{1+log{word}_i}log\frac{N}{n_t},f_{it}\ge 1\\ 0f_{it}<1\end{array}\right.$$

(1)

where $w_{it}$ refers to the weighted frequency rates of keyword $t$ in the annual report of firm $i$, $f_{it}$ refers to the frequency rates of keyword $t$ in the annual report of firm $i$, ${word}_i$ is the total word count for the annual report of firm $i$, $N$ refers to the sum of the annual reports, and $n_t$ is the sum of the annual reports that contain keyword $t$.

Given the fact that we employed text mining methods to measure both the dependent and independent variables, in our supplementary analyses, we used other proxies to measure the symbolic and substantive environmental disclosure to avoid any potential bias.

3.2.2. Independent Variable: Digital Transformation

We also employed text mining techniques and followed the same procedures used to evaluate the symbolic and substantive environmental disclosure in order to assess the digital transformation process. In particular, drawing on the studies by Ji et al. [52] and Sun et al. [22], we developed a digital dictionary composed of five dimensions—artificial intelligence, big data, cloud computing, blockchain, and digital technology applications. Subsequently, we utilized Python to calculate the frequency rates of keywords correlated with digital transformation based on the digital dictionary, and we adjusted the frequency rates of each dimension using Equation (1). Finally, we measured the digital transformation process by dividing the weighted frequency rates of digital terms by the length of a firm’s annual report.

3.2.3. Moderating Variable: Senior Managers’ Political Experience

In line with previous studies [52,68], we constructed a dummy variable to measure whether there are former or current government officials; National, Provincial, or Municipal People’s Congress deputies; or National, Provincial, or Municipal People’s Political Consultative Conference members in the firms’ senior management ranks. If so, this result was recorded as 1, otherwise it was recorded as 0. The relevant data came from the CSMAR database and were statistically analyzed based on the annual and firm reports.

3.2.4. Control Variables

We controlled for a number of factors that may have affected the symbolic and substantive environmental disclosure efforts according to prior studies [6,69]. For a concise overview of these variables, refer to

Table 1. Definitions of variables.

Variable	Measurement
Symbolic environmental disclosure	Please refer to “dependent variables” section for details
Substantive environmental disclosure	Please refer to “dependent variables” section for details
Digital transformation	Please refer to “independent variables” section for details
Senior managers’ political experience	Please refer to “moderating variables” section for details
Firm size	Ln (total assets)
Asset/liability ratio	Total liabilities/total assets
Chairman’s shareholdings	Shareholding ratio of chairman
Firm growth	Growth rate of return on net assets
Board size	Ln (numbers of boards members)
Duality	1 if chairman serves as CEO, 0 otherwise
Ownership	1 if firm is state-owned, 0 otherwise

. The controlled variable data were all from the CSMAR database. The year and firm fixed effects were incorporated in the regression models to exclude the impacts of economic cycles and firm-specific factors. We also Winsorized all variables at the first percentile in each tail to prevent the potential effects of outliers.

3.3. Models

We built a fixed-effects model to test the main hypotheses:

$$Y_{it}={\beta }_0+{\beta }_1{X}_{it}+\theta {Control}_{it}+{\alpha }_i+z_t+{\epsilon }_{it}$$

(2)

where the firm is denoted by $i$; time is denoted by $t$; dependent variables, such as symbolic and substantive environmental disclosure, are represented by $Y_{it}$; all explanatory variables, including the digital transformation variable, the senior managers’ political experience variable, and the interaction terms of these variables, are denoted by $X_{it}$; the statistical impacts of explanatory variables on symbolic and substantive environmental disclosure are denoted by ${\beta }_1$, which is considered significant when it is larger than 0; control variables are denoted by ${Control}_{it}$. It should be noted that all variables involved in the interaction terms were standardized to mitigate issues of multicollinearity. An ordinary least squares (OLS) regression method with heteroscedasticity for robust standard errors was employed to prevent any effects of heteroscedasticity [70].

4. Results

4.1. Descriptive Statistics

The descriptive statistics for the variables are presented in

Table 2. Descriptive statistics for the variables.

Variable	Mean	Std. Dev.	Min	Max
Symbolic environmental disclosure	0.604	0.698	0	6.899
Substantive environmental disclosure	0.860	1.065	0	6.197
Digital transformation	2.173	3.279	0	16.018
Senior managers’ political experience	0.444	0.497	0	1
Firm size	21.844	1.145	18.760	25.306
Asset/liability ratio	0.373	0.188	0.025	0.804
Chairman shareholdings	0.115	0.161	0	0.093
Firm growth	0.010	0.033	0	0.249
Board size	2.190	0.225	1.386	3.219
Duality	0.315	0.464	0	1
Ownership	0.281	0.449	0	1
High-tech industries	0.513	0.500	0	1
Heavily polluting industries	0.436	0.496	0	1

. There is considerable variance in the variables for symbolic and substantive environmental disclosure. The values of these two independent variables range from 0 to 6.899 and from 0 to 6.197, respectively, and the standard deviations both exceed the mean values. In accordance with previous research [52], it can be seen that the variable results for digital transformation are highly skewed. Additionally, 28.1% of the sample firms are controlled by government agencies, 51.3% belong to high-tech industries, and 43.6% are classified as heavily polluting firms.

The Pearson’s correlation coefficients of the variables are shown in

Table 3. Correlation coefficient matrix for the variables.

Variables	1	2	3	4	5	6	7	8	9	10	11
1. Symbolic environmental disclosure	1.000
2. Substantive environmental disclosure	0.175 ***	1.000
3. Digital transformation	−0.017 ***	−0.136 ***	1.000
4. Square of digital transformation	−0.024 **	−0.136 ***	0.935 ***	1.000
5. Senior managers’ political experience	−0.002	−0.001	−0.032 ***	−0.031 ***	1.000
6. Firm size	0.062 ***	0.245 ***	0.066 ***	0.037 ***	0.041 ***	1.000
7. Asset/liability ratio	0.015 *	0.147 ***	−0.038 ***	−0.037 ***	−0.009	0.536 ***	1.000
8. Chairman shareholdings	−0.016 *	−0.122 ***	0.048 ***	0.043 ***	−0.017 **	−0.336 ***	−0.267 ***	1.000
9. Firm growth	−0.015 *	0.008	0.034 ***	0.032 ***	−0.029 ***	−0.002	0.076 ***	−0.057 ***	1.000
10. Board size	0.020 ***	0.121 ***	−0.051 ***	−0.048 ***	0.062 ***	0.231 ***	0.149 ***	−0.228 ***	0.028 ***	1.000
11. Duality	−0.023 ***	−0.109 ***	0.070 ***	0.055 ***	−0.033 ***	−0.171 ***	−0.148 ***	0.275 ***	−0.027 ***	−0.153 ***	1.000
12. Ownership	0.022 ***	0.193 ***	−0.100 ***	−0.077 ***	0.016 **	0.351 ***	0.311 ***	−0.422 ***	0.100 ***	0.244 ***	−0.277 ***

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

. The correlations indicate a significant negative relationship between digital transformation efforts and symbolic environmental disclosure, as well as between digital transformation efforts and substantive environmental disclosure. We also examined the variance inflation factors (VIFs) of each model and found that the maximum VIF was less than the threshold of 10; therefore, multicollinearity did not affect our regression results [63].

4.2. Regression Analysis

Table 4. Fixed-effects models predicting symbolic and substantive environmental disclosure.

	Symbolic Environmental Disclosure			Substantive Environmental Disclosure
Variables	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Firm size	0.040 ***	0.041 ***	0.041 ***	0.106 ***	0.113 ***	0.112 ***
	(3.45)	(3.51)	(3.49)	(5.62)	(−6.00)	(−5.99)
Asset/liability ratio	−0.072	−0.073	−0.073	0.068	0.054	0.054
	(−1.19)	(−1.21)	(−1.21)	(0.68)	(−0.54)	(−0.54)
Chairman shareholdings	0.000	0.000	0.001	−0.001	−0.001	−0.001
	(0.11)	(0.11)	(0.09)	(−0.28)	(−1.09)	(−1.10)
Firm growth	−0.001	−0.001	−0.001	0.001	0.001	0.001
	(−0.62)	(−0.58)	(−0.59)	(1.11)	(0.45)	(0.45)
Board size	0.012	0.012	0.011	0.178 ***	0.177 ***	0.177 ***
	(0.30)	(0.30)	(0.27)	(3.11)	(3.12)	(3.11)
Duality	−0.011	−0.010	−0.010	−0.084 ***	−0.082 ***	−0.082 ***
	(−0.51)	(−0.49)	(−0.48)	(−2.77)	(−2.72)	(−2.72)
Ownership	0.005	0.005	0.003	0.232 ***	0.230 ***	0.229 ***
	(0.17)	(0.16)	(0.10)	(4.73)	(4.71)	(4.69)
Senior managers’ political experience	0.013	0.013	0.013	−0.004	−0.004	−0.004
	(0.67)	(0.67)	(0.67)	(−0.15)	(−0.14)	(−0.14)
Digital transformation		−0.003	−0.002		−0.017 **	−0.015 **
		(−0.92)	(−0.88)		(−2.47)	(−2.35)
Square of digital transformation					−0.001	−0.001
					(−1.58)	(−1.60)
Interaction term			0.011 **			0.005
			(2.41)			(0.80)
Year	Yes	Yes	Yes	Yes	Yes	Yes
Firm	Yes	Yes	Yes	Yes	Yes	Yes
Adjusted R2	0.057	0.057	0.057	0.238	0.242	0.242
Observations	14841	14841	14841	14841	14841	14841

Notes: (1) ** p < 0.05, *** p < 0.01; (2) interaction term = senior managers’ political experience, digital transformation; (3) t values in parentheses.

presents the results of the models examining the effects of digital transformation efforts on symbolic and substantive environmental disclosure. Model 1 involves the control and moderating variables that may potentially affect the symbolic environmental disclosure results. The empirical findings indicate that the firm size, measured as the total assets, has a positive and significant effect. Model 2 contains an additional digital transformation term, the coefficient of which is negative and not significant. Therefore, H1a is not supported. A plausible explanation is that some firms may not have fully established a mature digital transformation strategy, which involves aligning existing strategies with selected digital components [71] and adjusting the managerial and operational structures [72]. This lack of readiness could impede firms from fully capitalizing on the advantages of digital technologies. In Model 3, an interaction term composed of senior managers’ political experiences and digital transformation efforts is added; its coefficient is positive and significant. Therefore, H2a is supported, namely that senior managers’ political experience can positively moderate the relationship between digital transformation efforts and symbolic environmental disclosure.

Model 4 involves control and moderating variables that may potentially affect the substantive environmental disclosure results. The firm size, board size, duality, and ownership aspects are significant, as expected. Model 5 additionally contains the linear and quadratic terms of the digital transformation process. The digital transformation process has a negative and significant linear term and a negative and insignificant quadratic term. These results do not support H1b. It is possible that the implementation of digital technologies is still in its early stages, requiring significant investments that may not yet bring positive spillover benefits to other firm projects. An interaction term composed of the senior managers’ political experience and digital transformation efforts is added in Model 6 but its coefficient is positive and insignificant. Thus, H2b is not supported. This finding suggests that in comparison to the relationship between digital transformation efforts and substantive environmental disclosure, the senior managers’ political experience may have a stronger influence on the impacts of digital transformation efforts on symbolic environmental disclosure.

4.3. Robustness Analysis

Considering that both the dependent and independent variables were measured with a text mining analysis, we employed alternative proxies to measure the symbolic and substantive environmental disclosure to test for robustness. Following the approach of Clarkson et al. [41], Li [65], and Li et al. [66], we scored the symbolic environmental disclosure as 1 if a firm disclosed its environmental protection concepts, goals, or training and 0 otherwise. Similarly, we scored the substantive environmental disclosure as 1 if a firm disclosed its environmental protection management system, acts, rewards, or emergency mechanisms for environmental events and 0 otherwise. The relevant data were collected from the CSMAR database. The results, as shown in

Table 5. Results of using alternative proxies when measuring environmental disclosure.

Variable	Symbolic Environmental Disclosure	Substantive Environmental Disclosure
Digital transformation	0.000	−0.001 *
	(−0.21)	(−1.79)
Square of digital transformation		−0.001
		(−0.42)
Controls	Yes	Yes
Year	Yes	Yes
Firm	Yes	Yes
Adjusted R2	0.028	0.219
Observations	14355	14509

Notes: (1) * p < 0.1; (2) t values in parentheses.

, provide further support for our main findings.

Using the method described by Cheng et al. [73], we selected the mean value for digital transformation at the industrial level as the instrumental variable and constructed the two-stage models to examine the fitted value for the effect of digital transformation (fitted_digital transformation) on substantive environmental disclosure. The results are shown in

Table 6. Results of the instrumental variable approach.

	First Stage	Second Stage
Variables	Digital Transformation	Substantive Environmental Disclosure
Fitted_Digital transformation		−0.048 ***
		(−3.15)
Instrumental variable	0.542 *** (33.72)
Observations	14434	14434
Kleibergen–Paap rk LM statistic (Underidentification test)	429.155 (p = 0.000)
Cragg–Donald Wald F statistic (Weak identification test)	2932.571

Notes: (1) *** p < 0.01; (2) t values in parentheses.

, and our finding is valid.

4.4. Additional Analyses

To further enrich our understanding of the relationship between digital transformation efforts and environmental disclosure, we conducted several additional analyses.

First, drawing from the literature, the digital transformation efforts could also be classified into technology-oriented or market-oriented categories. The former category focuses on improving existing production or operation systems while the latter targets creating new business models [20,52,72]. Given this distinction, it is theoretically plausible that these two types of digital transformation efforts have different impacts on symbolic and substantive types of environmental disclosure. To measure these types of digital transformation, we followed the methods outlined by Ji et al. [52] and employed text mining techniques. The regression results are shown in

Table 7. Results for the two types of digital transformation efforts.

Variables	Symbolic Environmental Disclosure	Substantive Environmental Disclosure
Technology-oriented digital transformation	0.010 *	−0.009
	(1.74)	(−0.99)
Market-oriented digital transformation	−0.015 **	−0.019 ***
	(−2.53)	(−2.55)
Controls	Yes	Yes
Year	Yes	Yes
Firm	Yes	Yes
Adjusted R2	0.058	0.242
Observations	14,841	14,841

Notes: (1) * p < 0.1, ** p < 0.05, *** p < 0.01; (2) t values in parentheses.

.

We note two main findings here that complement and support our main results: (1) symbolic environmental disclosure is positively and significantly affected by technology-oriented digital transformation efforts but negatively and significantly affected by market-oriented digital transformation efforts; (2) substantive environmental disclosure is negatively and significantly affected by market-oriented digital transformation efforts. Compared to technology-oriented digital transformation initiatives, a market-oriented digital transformation usually comes with greater strategic challenges and may lead to reduced investment in environmental resources due to resource allocation issues [52].

Second, we investigated other potential boundary factors that may affect the relationship between digital transformation efforts and environmental disclosure. Prior research has suggested that the impact of digital transformation efforts on a firm’s performance is stronger in low- and medium-tech (LMT) industries than in high-tech (HT) industries [74,75]. This is because LMT firms often lack formal R&D processes; therefore, they rely more on external resources to improve their capabilities. We reexamined our framework based on a subsample classifying firms as either HT or LMT based on government accreditation data from the CSMAR database. The results, as shown in

Table 8. Results from HT and LMT industries.

Variables	HT Industries		LMT Industries
	Symbolic Environmental Disclosure	Substantive Environmental Disclosure	Symbolic Environmental Disclosure	Substantive Environmental Disclosure
Digital transformation	−0.004	−0.060 **	−0.009 ***	−0.065 ***
	(−1.18)	(−7.62)	(−2.34)	(−7.22)
Square of digital transformation		0.001		0.002
		(0.97)		(1.27)
Controls	Yes	Yes	Yes	Yes
Year	Yes	Yes	Yes	Yes
Firm	Yes	Yes	Yes	Yes
Adjusted R2	0.010	0.123	0.014	0.148
Observations	7677	7677	6757	6757

Notes: (1) ** p < 0.05, *** p < 0.01; (2) t values in parentheses.

, confirm our expectations; the effect of digital transformation efforts is stronger in LMT industries than HT industries.

Environmental disclosure is a mandatory behavior for heavily polluting industries, so the relationships we explored in this study may be differ between heavily polluting industries and other manufacturing industries. We further reexamined our framework while making this distinction. In line with previous studies [76,77], heavily polluting industries were identified according to the existing literature and policy documents. The results are shown in

Table 9. Results from heavily polluting industries and other manufacturing industries.

Variables	Heavily Polluting Industries		Other Manufacturing Industries
	Symbolic Environmental Disclosure	Substantive Environmental Disclosure	Symbolic Environmental Disclosure	Substantive Environmental Disclosure
Digital transformation	−0.001	−0.060 ***	−0.011 ***	−0.043 ***
	(−0.34)	(−6.35)	(−2.70)	(−4.18)
Square of digital transformation		0.003 **		−0.002
		(2.23)		(−0.94)
Controls	Yes	Yes	Yes	Yes
Year	Yes	Yes	Yes	Yes
Industry	Yes	Yes	Yes	Yes
Adjusted R2	0.008	0.129	0.020	0.113
Observations	6316	6316	8118	8118

Notes: (1) ** p < 0.05, *** p < 0.01; (2) t values in parentheses.

. The effects of digital transformation efforts on symbolic environmental disclosure appear to be weaker in heavily polluting industries than in other manufacturing industries.

5. Discussion and Conclusions

In order to attain a holistic comprehension of the intricate nexus between digital transformation efforts and CSR, with a special focus on the context of emerging economies, we empirically examined correlations between digital transformation efforts and two types of environmental disclosure in this study. We applied text mining methods to analyze the impacts of digital transformation efforts on both symbolic and substantive environmental disclosure and explored the moderating role of the senior managers’ political experience in shaping these relationships. This research was based on a panel dataset consisting of 2033 publicly listed Chinese manufacturing firms covering the period from 2009 to 2020. The main findings can be summarized as follows.

(1) The digital transformation process is negatively and significantly associated with substantive environmental disclosure. This suggests that despite potential efficiency and energy-saving advantages, the adoption of digital technologies may hinder firms from making substantial investments in environmental protection initiatives.

(2) The relationship between digital transformation efforts and symbolic environmental disclosure is positively and significantly moderated by the senior management team’s political experience. This implies that firms with politically experienced senior managers may leverage digital transformation initiatives to enhance their symbolic environmental disclosure results, aligning their strategies with government expectations and gaining support accordingly. We also conducted a series of robustness checks and further analyses to strengthen the validity and reliability of our findings.

This research makes two main contributions to the literature. First, it delves deeper into the relationship between digital transformation efforts and CSR, which has garnered high expectations from researchers. By addressing the oversight of heterogeneity with CSR initiatives, this study offers a more refined insight into the role of digital transformation efforts in the CSR landscape. Our findings reveal that the digital transformation process has distinct impacts on two types of environmental disclosure, with both positive and negative effects, challenging prior studies. Specifically, it is negatively associated with substantive environmental disclosure, which contradicts certain prior studies [14,21,24,43] in suggesting that the implementation of digital technologies may not directly translate into substantive environmental investments but may instead encourage solely symbolic behaviors. Moreover, we have shown that such a negative relationship is largely motivated by market-oriented digital transformation efforts, while symbolic environmental disclosure is positively affected by technology-oriented digital transformation efforts. In the context of emerging economies, the implementation of digital technologies may not generate positive spillover effects to substantive activities requiring actual investments, as the digital transformation process has a more complex effect on CSR.

Second, this research enriches our understanding of the boundary conditions involved in the correlation between digital transformation efforts and CSR. Previous studies have extensively investigated context factors in the links between digital transformation efforts and firm performance indicators [52,69,78]. Our work extends these insights by highlighting the moderating role of senior managers’ political experience. Our additional analyses also indicate that the impacts of digital transformation efforts on symbolic and substantive environmental disclosure may vary across HT and LMT industries, as well as across heavily polluting and other manufacturing industries.

Our findings carry significant managerial implications for both industrialists and policymakers. First, they indicate that digital transformation efforts may not improve actual environmental practices but may only partly benefit symbolic behaviors. Currently, the digital transformation approach (especially when market-oriented) is not an effective tool for firms from emerging economies seeking to upgrade towards sustainable development. Second, our moderating effect examinations show that firms’ close relationships with government agencies may not result in substantive environmental disclosure benefits but do appear to result in symbolic environmental disclosure benefits. Governments from emerging economies should strengthen their regulation and monitoring systems while penalizing greenwashing behavior; they also should reduce the barriers to digital transformation in order to maximize its advantages.

This research has some limitations warranting consideration. First, although we carefully selected keywords to describe the digital transformation process, as well as the symbolic and substantive forms of environmental disclosure, based on existing studies in emerging economies, future research studies could further enrich the term dictionary by including terms relevant to developed countries. Second, while our study provides insights based on the Chinese context, the findings may not be fully generalizable to other countries with different institutional settings and business environments. To enhance the consistency and robustness of our findings, further investigations in developed countries would be valuable. We believe that studying the relationship between digital transformation efforts and environmental disclosure in various countries would help us gain a more thorough understanding of the subject. Third, future research studies could further enhance this study by investigating additional contingency factors, such as regulatory policies and the age or educational background of senior management members, in order to enhance our comprehension of the connection between digital transformation efforts and environmental disclosure. Fourth, future research studies could take other theories into consideration, such as agency theory, to enrich the diversity of perspectives on this relationship.