Can digital finance curb corporate ESG decoupling? Evidence from Shanghai and Shenzhen A-shares listed companies

Baseline regression result

Baseline regression is performed to identify the impact of digital finance on corporate ESG decoupling, and the result is presented in Table 2.

Column (1) in Table 2 presents the regression result for the impact of digital financial development on corporate ESG decoupling practices. The estimation coefficient for \(DF\) is −2.465 and shows a significant negative correlation to ESG decoupling at the 1% level, which means advances in digital finance can effectively prevent enterprises from ESG decoupling. In Column (2), the industry-time interaction fixed effect is introduced to control the specific development trends of different industries at different time points, and the result shows that the regression coefficient of \(DF\) to \(DECOUPLING\) is negative and significant at the 5% level, which means digital finance makes up for the shortcomings of traditional financial services, improves exchange of financial elements between regions, and optimizes allocation of financial resources across time and space, hence meeting the needs of micro-enterprises for financial support during their green transformation. Meanwhile, the big data technology embodied in digital financial services alleviates the information asymmetry between the public and enterprises, so digital finance prevents businesses from committing ESG decoupling by enabling public supervision. Internationally, digital finance facilitates cross-border financial services and improves connectivity between regional financial markets across the globe; it boosts global collaboration in international financial supervision and risk management, contributing to the long-term stability of the global financial system; it also allows enterprises across the world to move toward green and quality development, thereby stemming the practice of ESG decoupling. Therefore, H1 is verified.

As mentioned before, three sub-dimensional indicators are introduced to interpret the impact of digital finance on corporate ESG decoupling more precisely. The three indicators are the coverage (\(DF\_width\), which reflects the flexibility of digital finance), the usage depth (\(DF\_depth\), which reflects the longitudinal development of digital finance in enterprises), and the digitalization degree (\(DF\_digital\), which highlights the digital characteristics) of digital finance, and the regression results are presented in Columns (3)–(5) of Table 2. As the statistics show, \(DF\_depth\) and \(DF\_digital\) demonstrate no significant effects on curbing corporate ESG decoupling, while the coefficient of \(DF\_width\) is negative and significant at the 5% level, which means increasing the coverage of digital finance can effectively contain the practice of ESG decoupling. The causes behind this are not hard to fathom: as the coverage of digital finance expands, consumers are more willing to pay as digital finance provides easier access to more convenient financing channels and a wider selection of financial derivatives, which makes for higher financial liquidity and more trading vigor in the market. In such an active market, enterprises face fewer challenges in seeking financial sponsorship and will be less likely to resort to ESG decoupling in financial strains. Besides, with the wider adoption of digital finance, corporate information grows more transparent, which further cuts the likelihood of ESG decoupling. However, it should be noted that despite the effort China has made in the development of digital finance, challenges remain in the application of financial big data, and this is why the other two indicators (\(DF\_depth\) and \(DF\_digital\)) demonstrate no tangible effects in curbing ESG decoupling. Through the international lens, digital finance also faces a range of problems in its development, including problems in data quality and reliability, weakness in data processing, insufficient cross-border collaborations and supervision, and difficulties in turning innovations into practical solutions, to name just a few. If we focus merely on the “width” of the adoption of digital finance but leave out the “depth” of its development, digital finance is unlikely to provide a long-term impetus for enterprises to go green. In this sense, the regression result allows us to see the development direction of digital finance in the future.

Robustness tests

Robustness tests are performed by controlling the province-time interaction fixed effects, displacing the explanatory variable, and removing special samples. The results are presented in Table 3.

Regression with province-time interaction fixed effects controlled

In baseline regression, only the age-fixed effect and individual-fixed effect are controlled, but this practice cannot eradicate the influence of policies that change with the province and time on the result. Regional governments often issue a range of policies about green development across the years, which means regions differ in their policies that prohibit ESG decoupling and encourage green transformation. To control the influence of regional differences across time, this paper introduces the province-time interaction fixed effect into Model (4) for regression, and the result is shown in Columns (1) and (2) in Table 3. The statistics show that when the province-time interaction fixed effect is controlled, the regression coefficient of \(DF\) to \(GW\) remains significant and negative, indicating that the conclusions in Section “Baseline regression result” are robust.

Regression with the key explanatory variable displaced

There is possibly a time lag when it comes to the suppressing effect of digital financial development on ESG decoupling; the two variables may interact with each other and result in bilateral causality. Therefore, a lagged variable for digital finance (\(L.DF\)) is introduced to the regression model to displace the original explanatory variable (\(DF\)), and the result is presented in Column (3) in Table 3. It is found that the regression coefficient of the lagged variable to \(DECOUPLING\) remains significant. Moreover, according to Lu et al.’s study (Lu et al. 2023), this paper displaces the key explanatory variable (\(DF\)) by the “digital finance index/100” (\(DF2\)) and the digital financial development index at the city level (\(DF\_city\)) to perform robustness tests, and the results are presented in Columns (4) and (5) in Table 3, respectively. As the statistics show, digital finance remains to play a role in suppressing corporate ESG decoupling even if the explanatory variable is displaced, which verifies the robustness of our conclusions.

Regression with special samples removed

As the four municipalities directly under the central government (Beijing, Tianjin, Shanghai, and Chongqing) are economically special in China, regression results for these four municipalities may be substantially different from other provinces. Therefore, the samples from these four municipalities are removed for regression by Model 4, and the results are presented in Column (6) in Table 3. The statistics show that the conclusion remains robust when the special samples are excluded.

Endogeneity in regression

Endogeneity biases resulting from the omission of variables may be present in empirical regression. To deal with the problem of endogeneity, this paper employs the instrumental variable estimation and the system generalized method of moments (GMM) estimation. The regression results are presented in Table 4.

First, according to previous works (Cheng et al. 2023), this paper employs the instrumental variable estimation method to deal with the problem of endogeneity. Specifically, the variable of Internet penetration (\(Internet\): the logarithm of the number of users with access to broadband Internet in 100 people, which is obtained from the China City Statistical Yearbook) is employed as an instrumental variable to deal with endogeneity bias. The Internet is the fundamental infrastructure for digital finance, so Internet penetration is closely correlated to digital financial development; besides, Internet penetration shows no necessary correlation with corporate ESG decoupling, which rules out the possibility of the exogeneity problem, and hence Internet penetration can be considered an instrumental variable for digital finance. The regression result for the instrumental variable is presented in Columns (1) and (2) in Table 4. Column (1) shows the first-stage regression result, where the coefficient of \(Internet\) is positive and significant at the 1% level, and the Cragg-Donald Wald F statistic from the weak instrument variable test is significantly higher than the estimation at the 10% level, which rules out the problem of the weak instrument and the correlation condition is met. Meanwhile, the P value of the Kleibergen-Paap rk LM statistic is below 0.01, which rules out the problem of under-identification of the instrument. Column (2) of the table shows the result of the second-stage regression, where the coefficient of \(DF\) remains significantly negative.

In addition, the dynamic panel model is employed to address the problem of endogeneity. As corporate ESG decoupling is a continued and persistent behavior, the lagged term of \(DECOUPLING\), i.e., \(L.DECOUPLING\), is introduced to the model for dynamic panel analysis, and the result is displayed in Column (3) in Table 4. The system GMM estimation shows that the coefficient of \(L.DECOUPLING\) is significant at the 1% level, which means corporate ESG decoupling is a persistent behavior, and digital finance remains to show negative significance in its relation to corporate ESG decoupling. Moreover, the P value of AR (1) is below 0.1, while that of AR (2) is above 0.1, indicating the presence of first-order autocorrelation but no second-order autocorrelation in the model disturbance term. It indicates the effective configuration of the model and the robustness of the original model.

Mechanism tests

In this study, the mediation effect theory is employed to empirically test the mechanism for the influence of digital financial development on corporate ESG decoupling. Traditional mediation models suffer the problem of endogeneity, and mere reliance on the traditional assumption of exogeneity is likely to fail to verify causality. Under the interplay between the mediator and the explanatory variable, the estimation of the average effect may show errors (Jiang 2023). Therefore, according to the suggestion provided in Prof. Jiang Ting’s work on the mediation effect in casual inference research, this paper selects a mediator that has a clear causal relation to the explained variable (corporate ESG decoupling) and focuses on the impact of the explanatory variable (digital financial development) on the mediator. To control the regional changes over time, the province and time interaction fixed effect is introduced to Model (4) to construct Model (5) for mediation effect analysis:

In Model (5), the definitions and calculations for the explanatory variable and control variables remain the same as those in the baseline regression model; \(\mu _p,t\) represents the province and time-fixed effect, and \(Mediator_\rmp,\rmi,\rmc,\rmt\) is the mediator.

The results of the mechanism tests are shown in Table 5.

Long-tail effect

In terms of investment efficiency, as mentioned before, digital finance can, through the long-tail effect, increase the Pareto improvement of information exchange between enterprises and financial institutes and alleviate the risk of rent-seeking and bribery in investment programs, thus improving the investment efficiency and stems corporate ESG decoupling. The verification is presented in Column (2) in Table 5. The coefficient of \(DF\) to the mediator \(Invest\) is negative and significant, indicating that digital financial development can prevent firms from ESG decoupling by reducing the corporate inefficient investment, or in other words, by increasing the corporate investment efficiency.

Regarding financing constraints. As mentioned before, digital finance can, through the long-tail effect, expand the firm’s financing channels and reduce their cost for financing attempts, thereby alleviating their financing constraints and making them less likely to turn to ESG decoupling for financial support. To preclude the problem of endogeneity caused by measurement and reduce the statistical significance of the estimated correlation between the explanatory variable and the mediator in Model (4), this paper performs a one-stage lag on the explanatory variable to test the mediation effect (Pan, 2024). The results are shown in Column (4) in Table 5, where the coefficient of digital financial development is negative and significant. Therefore, digital finance and traditional financial services are mutually supplementary, which together reduce the financing cost of firms, alleviate their financing constraints, and thereby reduce their intention to resort to ESG decoupling in financial strains. In this logic, the hypothesis H2b is verified.

Information effect

This section analyzes the mediation effect of ESG disclosure quality. As mentioned before, digital finance can, by dint of the digital technologies it embodies, improve financial institutes’ capacity for information collection, enhance the information-sharing efficiency between enterprises and the public, which increases the cost for violations of environmental regulations, prompts enterprises to disclose their ESG information and thereby reduces their chance of ESG decoupling. The test results are displayed in Columns (5) and (6) in Table 5, where the regression coefficient of digital financial development to ESG information disclosure quality is positive and significant. It indicates that digital finance can improve the quality of corporate ESG disclosure by making the information provided by enterprises more transparent and comparable, which directs the enterprises’ focus from mere information disclosure to the goal of green and sustained development, and to some extent discourages ESG decoupling.

Long-term vision effect

This section analyzes the mediation effect of managerial myopia. As mentioned before, By removing market information asymmetries, increasing competition and providing cheap access to information, digital finance promotes a positive green image by corporate management, thereby reducing short-sightedness. At the same time, the long-tail effect of digital finance can also help to ease the financing pressures faced by managers, thereby reducing the incentives for firms to “go green”. The test results are displayed in Columns (7) and (8) in Table 5, where the coefficient of DF remains negative and significant at the 5% level. Thus, it is verified that digital finance can contain corporate ESG decoupling by alleviating managerial myopia. In this logic, the hypothesis H4 is verified.

Heterogeneity analysis

Ownership nature of enterprises

As the ownership nature of an enterprise may affect its environmental performance, its accessibility to external capital, and policy support, this paper divides the samples into the group of state-owned enterprises (SOEs) and the group of non-SOEs to investigate the changes in the impact of digital finance on corporate ESG decoupling as the ownership nature of the enterprise varies. The results are shown in Columns (1) and (2) in Table 6, where the regression coefficient of \(DF\) for the non-SOE group is negative and significant at the 1% level, while that for the SOE group demonstrates no significance. It means digital finance plays a more prominent role in private enterprises that are dominated by the market logic than in SOEs. In China, SOEs can rely on governmental sponsorship and have easier access to external financial support, so digital finance plays a mere marginal role in their business operation, and their need for digital finance is very limited; besides, as SOEs are susceptible to influence from internal stakeholders and agency problems, they focus more on the political and societal objectives than on the goal of green development, which diminishes the role of digital finance in curbing corporate ESG decoupling in these enterprises.

Technological strength of enterprises

The digital technologies embodied in digital finance provide a key technological foundation for corporate development, so the role that digital finance plays in preventing ESG decoupling may differ as the technological strength of the enterprise varies. According to “China’s Catalog of Strategic Emerging Industries” and relevant documents of Organization for Economic Co-operation and Development (OECD) (Zhang et al. 2023), this paper divides the samples into a high-tech group and a non-high-tech group, and the test results are displayed in Columns (3) and (4) in Table 6. In the high-tech group, the coefficient of digital finance is negative and significant, while no significance is observed in the other group. It suggests that the role of digital finance in curbing ESG decoupling is more prominent in high-tech enterprises than in their non-high-tech counterparts. One possible explanation is that the high-tech industry invests more in research and development of digital technologies and has far more advanced methods of production than other industries, so enterprises in the high-tech sector are more compatible with digital finance and enjoy technological edges over others in green transformation, which makes it easier for them to pursue the green development goal. Therefore, digital finance can, by reducing information asymmetry and improving credit allocation, encourage high-tech enterprises to move toward green transformation and thereby reduce the risk of ESG decoupling.

Polluting nature of enterprises

Because of the information effect of digital finance, heavy-polluting enterprises are under higher pressure than others in green transformation, so whether the enterprise is a polluting entity matters when we measure the role of digital finance in suppressing ESG decoupling. Therefore, according to the Notice of Issuing the Industry Classification & Management Directory of Listed Companies Subject to Environmental Verification (Li et al. 2019), this paper divides the samples into the group of heavy-polluting enterprises and the group of non-heavy-polluting enterprises. The heterogeneity test results are displayed in Columns (5) and (6) in Table 6, where the coefficients of \(DF\) remain negative and significant for both groups, but the absolute value for the heavy-polluting group is larger and more significant, and the difference between the intergroup coefficient observations is significant at the 10% level. It indicates that digital finance plays a more effective role in curbing ESG decoupling in heavy-polluting enterprises. Heavy-polluting businesses, subject to dual pressure from the government and the market, are more driven to go green. Therefore, the long-tail effect of digital finance will encourage them to invest more into green equipment and technologies, thereby accelerating green transformation. In comparison, non-heavy-polluting enterprises are less motivated to go green and pay less attention to the goal of green development. Consequently, the role of digital finance in improving green performance and suppressing ESG decoupling is more pronounced in heavy-polluting enterprises than in businesses without heavy pollution to the environment.

Based on the results of the above analyses, hypothesis H5 is verified.

Investor attention as a possible moderator

In the presence of information asymmetry, investors, in a more disadvantaged position than the enterprises, are likely to make wrong decisions and affect the enterprise’s environmental commitment (Kassinis and Vafeas 2006). Established based on the contractual relationship between the enterprise and the society, the legitimacy theory argues that the society ensures the right of survival of enterprises, and expects them to fulfill the expectations of stakeholders. As digital finance makes information more transparent, investors can monitor the actions of enterprises via digital platforms to pressure them to fulfill green commitments and thereby promote their green development. Specifically, investor attention will subject enterprises to stricter supervision: digital platforms built based on digital finance give investors quicker access to information about enterprises, making it easier to expose ESG decoupling risks, and the constant warnings of risks will direct the supervising authorities’ attention to the enterprises’ environmental performance. Meanwhile, investor attention will increase pressure on the reputation management of enterprises. As reported in research on the reputation theory (Rodriguez-Ariza et al., 2016), the constant attention the investor pays to an enterprise via digital platforms will cause a “halo effect”, attaching an unfavorable label on the enterprise that will be hard to take off and even overshadow the good sides, and subjecting the enterprise to a higher cost of financing and operation in the long run. Moreover, informal interferences like investor attention may make corporate managers more environmentally aware, which will work to improve the enterprise’s environmental performance (Yao et al., 2023). In sum, investor attention will enhance the role of digital finance in curbing ESG decoupling.

This research employs the Baidu search index as the dummy variable for investor attention (\(Attention\)) and cites Li et al.’s (2024) work to verify this mechanism. To test the mediating role of investor attention, this paper introduces the interaction term of investor attention and digital financial development (\(DF* Attention\)), and performs regression by Model (6). The results are presented in Column (3) in Table 7, where the coefficient of \(DF* Attention\) is negative and significant at the 1% level, which means investor attention enhances the role of digital finance in containing corporate ESG decoupling.