Driving Impact of Digital Transformation on Total Factor Productivity of Corporations: The Mediating Effect of Green Technology Innovation

ABSTRACT

Based on the data of Chinese A-share listed companies from 2010 to 2020, this article examines the driving effect of Digital Transformation of Corporations (DTC) on Total Factor Productivity (TFP) and selects Green Technology Innovation of Corporations (GTIC) as the mediating variable to test the mechanism of action of Digital Transformation of Corporations driving corporate TFP improvement. It is found that Digital Transformation of Corporations can significantly promote corporate TFP improvement; mechanism analysis shows that Digital Transformation of Corporations drives TFP improvement by promoting Green Technology Innovation of Corporations.

KEYWORDS:

• Digital Transformation of Corporations
• corporate TFP
• Green Technology Innovation of Corporations
• mechanism of action

JEL:

• D24
• O33

1. Introduction

As an acting point and an important engine of China’s economic growth under the new normal, Total Factor Productivity (TFP) has become a key indicator in China’s economy with regard to its move from input-based growth to efficiency-based growth. At present, China is facing the dilemma of “low-end locking” and “two-way extrusion,” which seriously restrict the quality and efficiency of real economy development and hinder the improvement of the production efficiency of corporations. As a new development strategy for deep integration of digital technology and the real economy (Wu et al. 2021), digitalization, with its high flexibility, self-reliance, and integration (Wang and Wu 2023; Yoo et al. 2012), not only improves the development resilience and innovation ability of the real economy but also reshapes the original organizational environment and business processes, constantly rewrites the social production function, spawns a new organizational operation logic and value creation method ‎ (Nambisan et al. 2017; Wu, Lou, and Hitt 2019), and improves the quality and efficiency of digital technology serving the real economy, thereby improving corporate TFP (Pan et al. 2021).

However, China’s economic development model has been dependent on input-driven characteristics, with typical characteristics of “high energy consumption, high emissions, and high pollution.” The continuous improvement of the economic development scale is accompanied by the deterioration of the natural environment coupled with the shortage of energy resources. What is more, the contradiction between economic growth and environmental protection is becoming increasingly severe (Wu et al. 2021). In view of the externality of environmental pollution and the nature of environmental public goods, it is difficult to completely solve the problem by relying solely on market mechanisms. Some studies have introduced environmental regulation as a breakthrough point to make up for market failures (Jin et al. 2019). However, from the long-term perspective, to thoroughly solve the contradiction between economic growth and environmental protection and improve corporate TFP, there is still a need to rely on technological progress (Griliches 1979), especially green development-oriented GTIC (Acemoglu et al. 2012; Shao et al. 2016). GTIC follows the laws of ecological economy, saves resources and energy, can effectively avoid, eliminate, or reduce ecological pollution and damage, and can help achieve coordinated progress in the economy and in the environment. Its ultimate goal is to serve the sustainable and high-quality development of the economy, especially enabling the economy to improve its TFP from investment-oriented growth to efficiency-oriented growth‎ (Feng et al. 2022; Wu et al. 2021).

The enabling mechanism of green technology innovation on enterprise TFP is as follows: First, green technology innovation can effectively reduce energy consumption and emissions per unit output and improve enterprise TFP levels. Especially with the increasingly stringent environmental red line and emission standards, green technology innovation improves resource and energy use efficiency, reduces energy consumption and waste emissions, releases a more efficient incentive effect (Shao et al. 2016), and further improves enterprise TFP by virtue of alternative clean energy and carbon capture and storage technology. Second, green technology innovation enables industrial upgrading to improve enterprise TFP. Specifically, green technology innovation can help promote the emergence of a scientific and technological achievement such as new energy, gradually eliminate enterprises with “high energy consumption, high emissions, and high pollution,” increase the proportion of green industries with “low energy consumption, low pollution, and high added value,” enable traditional industries to extend and penetrate into intelligent and low-carbon industries, and broaden the industrial value chain to a certain extent. The green manufacturing system of “high efficiency, clean, low carbon, and cycle” is constructed (Shao et al. 2016) so as to improve the TFP of enterprises. Third, enterprises can enjoy preferential policies and support with their green technology innovation, which help to improve their reputation, reduce environmental risks, and make it easier to obtain government subsidies. Government subsidies can weaken the negative impact of financing constraints on enterprise productivity to a certain extent (Xue et al. 2022), so as to promote enterprise sustainable development and TFP improvement (Wu et al. 2021). Fourth, green technology innovation helps green and environmental protection enterprises to expand their market share, increase their revenue level, and thus increase their TFP. The innovation benefits and competitive advantages brought by green technology innovation have improved the market share of green environmental protection enterprises and increased their revenue levels. On one hand, the improvement of revenue levels can help increase the cash flow of enterprises and ease the internal liquidity constraints of enterprises. Enterprises with abundant-free cash flow have higher investment intensity and innovation willingness in green innovation, thus improving their TFP level. On the other hand, abundant-free cash flow can effectively improve the debt-paying ability of enterprises, defuse loan default risk, reduce enterprise financing costs, and alleviate financing constraints to a certain extent (Berrone et al. 2013; Fan and Fu 2021; Xue et al. 2022). The easing of financing constraints is conducive to enterprises’ green technology innovation and further improves enterprises’ TFP (Xue et al. 2022). Therefore, improving enterprise TFP through green technology innovation is necessary to boost the high-quality development of China’s economy as well as a strategic choice for enterprises to enhance market competitiveness (Feng et al. 2022).

Existing literature shows that there is little research on the mechanism of Digital Transformation of Corporations (DTC) driving corporate TFP improvement, and that the linkage of GTIC between DTC and corporate TFP is ignored. There are few studies that take GTIC as a mediating mechanism for testing, and, generally, there is a need to include DTC, GTIC, and corporate TFP in the same research framework for systematic research. In view of this, this article attempts to integrate the three into the same research framework for systematic research, and to further explore the internal mechanism of digital transformation driving enterprise TFP improvement. The following are our research questions: Can DTC improve corporate TFP? What role does GTIC play in this path? Answering the above questions has important theoretical and practical significance for choosing the path of DTC to enable corporate TFP.

This article’s contribution to the existing body of literature is as follows: On one hand, it brings DTC, GTIC, and corporate TFP into the same research framework to deeply tap the internal logic between the three, so as to make up for the limitedness of focusing on the pair relationship in previous research and to enrich the research on corporate TFP. On another hand, based on corporate micro data, it identifies and verifies the micro-mechanism of corporate DTC to improve corporate TFP, clearly revealing that DTC promotes corporate TFP improvement through GTIC. Additionally, it expands the research on the mechanism of action of DTC to drive corporate TFP improvement.

2. Literature Review

2.1. Study on the Impact of DTC on Corporate TFP

Existing research on corporate DTC mostly focuses on stock liquidity, input and output, integration mode, corporate performance, etc. to explore its economic consequences (Matt, Hess, and Benlian 2015; Wu et al. 2021; Yuan et al. 2021). In contrast, a small amount of literature focuses on the impact of DTC on corporate TFP and carries out mechanism analysis and heterogeneity tests on the impact. Chou, Chuang, and Shao (2014) investigated the impact of information technology (IT) on TFP, and found that IT drives TFP improvement by virtue of IT externalities and IT amplification innovation effects, suggesting that IT plays a more pivotal role than input consumption and accumulation that neoclassical growth theory assumes for it. Zhao, Wang, and Li (2021) found that DTC drives corporate TFP improvement by improving innovation capability, optimizing human capital structure, promoting the integration of “two industries,” reducing costs, etc. It was found through heterogeneity tests that this effect is more significant in state-owned corporations, large-scale corporations, as well as corporations in areas with a high degree of intellectual property protection and service openness (Tian and Liu 2021) found that DTC drives corporate TFP improvement, such that the degree of corporate capitalization, asset–liability ratio, profitability, staff quality, management efficiency, and per capita R&D investment have an important impact on this effect. Again, heterogeneity tests show that this driving effect is more obvious in the eastern region. Based on DID model research, Zhang et al. (2022) found that DTC promotes corporate TFP by reducing costs, improving efficiency, innovating performance, etc. Zhao et al. (2022) found that DTC promotes corporate TFP and tested how innovation performance plays a mediating role in this effect. It was found through heterogeneity tests that this effect is more significant in private corporations.

2.2. Study on the Impact of DTC on GTIC

DTC enables corporations to share information and accumulate knowledge, enhances their competitive advantages in GTIC (Feng et al. 2022), expands the scope of innovation resource allocation, encourages corporations to carry out joint innovation, optimizes corporate innovation technology resources (Rigby and Zook 2002), eases financing constraints caused by asymmetric information and incomplete contracts of GTIC corporations or projects, and helps to increase GTIC investment, thereby improving the GTIC level (Goldfarb and Tucker 2019). By taking the data of industrial robots (IRA) in 34 countries from 1993 to 2019 as the research sample, Lee, Qin, and Li (2022) that the application of IRA can effectively drive the GTIC of corporations, which further verified the regulatory role of environmental regulation in this effect. Xue et al. (2022) found that DTC drives corporate GTIC by easing financing constraints, increasing government subsidies, etc., and it was found through heterogeneity tests that this effect is more obvious in state-owned corporations and large-scale corporations. Based on technology integration capability, Wang et al. (2022) verified the impact of DTC on GTIC, and its mechanism by taking resource-based corporations as the object and pointing out that technology integration capability positively regulates this effect. Feng et al. (2022) found that DTC drives corporate GTIC by increasing R&D investment, offering government subsidies and reducing tax burden, and heterogeneity tests show that this effect is more obvious in high-tech corporations and state-owned corporations. Wang et al. (2022) verified the driving effect of DTC on GTIC by taking environmental regulation as a mediating variable, and further verified its marginal effect and spatial spillover effect.

2.3. Study on the Impact of GTIC on Corporate TFP

GTIC follows the laws of ecological economy, saves resources and energy, can effectively avoid, eliminate or reduce ecological environment pollution and damage (Wu et al. 2021), which helps realize coordinated development of economy and environment, and its ultimate goal is to serve the sustainable and high-quality development of economy (Feng et al. 2022; Wu et al. 2021). Berrone et al. (2013) preliminarily verified that green technology of corporate could help reduce production costs. Du and Li (2019) verified that GTIC is conducive to improving TFP. The empirical analysis of Fan and Fu (2021) shows that corporate environmental information disclosure drives corporate TFP improvement by easing financing constraints and promoting GTIC. The empirical analysis of Wu et al. (2021) proves that GTIC can improve corporate TFP by reducing their production costs, improving their reputation and popularity, improving its financing capabilities, weakening environmental risks, etc. Song et al. (2022) found that GTIC drives corporate TFP improvement by improving unit labor productivity, and heterogeneity tests show that this effect is more obvious in corporations with better financial resources and human resource foundations.

It can be seen from the existing literature that in recent years, the academic world has made a good exploration of the relationship between DTC and corporate TFP, providing an important theoretical reference and logical starting point for the research in this article. However, the following aspects still need to be improved. First, existing literature focuses on research of the relationship between DTC, GTIC, and corporate TFP but generally neglects to bring the three into the same research framework for systematic research. Second, research on the transmission mechanism of DTC driving corporate TFP improvement needs to be deepened; not much of the existing literature analyzes the impact mechanism of DTC on corporate TFP from GTIC, and fails to explore the mechanism of action of GTIC in this process. Especially for the special attributes of corporate TFP, how can the driving factors be reasonably and accurately transmitted to corporate TFP? The existing results have not yet given a precise answer, which needs to be further tested.

3. Theoretical Analysis and Hypothesis Development

3.1. DTC and Corporate TFP

First, DTC alleviates the financing constraints of corporations and improves corporate TFP. First of all, corporate DTC is highly consistent with such national strategies as building a “digital China” and “smart society,” and the positive signals released are easy to be “favored” (Wu et al. 2021). This positive “exposure effect” not only obtains national strategic support but also helps corporations obtain preferential policies of financial institutions, thereby easing financing constraints. Secondly, corporate DTC reduces the costs of information analysis, process optimization, etc., to achieve a higher level of corporate performance (Matt, Hess, and Benlian 2015). This positive signal is an important factor in attracting external investors, and helps to broaden the sources and channels of funds and expand the scale of fundraising. Easing financing constraints not only effectively reduces the liquidity risk of corporations but also helps improve the incentive systems of corporations, stimulate the innovation spirit of entrepreneurs, and effectively weaken the “passivation” effect of corporations’ entry and withdrawal, force low-efficiency corporations to withdraw from the market, thus promoting corporate TFP improvement.

Second, DTC reduces corporate costs and improves corporate TFP. First, DTC has nurtured many platform corporations (Nambisan, Zahra, and Luo 2019), strengthened the degree of supply chain integration of platform corporations, broken the geographical space restrictions and market segmentation barriers of traditional transactions, and reduced transaction costs and search costs (Agarwal et al. 2010). Secondly, DTC drives the reconstruction of boundaries between the internal departments of corporations, improves the efficiency of internal information transmission and processing, and reduces corporate management costs, transportation costs, and information transmission costs (Wu et al. 2021). Finally, DTC helps to accelerate the process of automation and intelligence in corporations (Kromann et al. 2020), optimize the configuration of traditional production factors and reduce the labor cost, production cost, and service costs of corporations (Banalieva and Dhanaraj 2019). The reduction of corporate costs will, to a certain extent, diversify the liquidity risk of investment projects, facilitate the optimization of investment portfolios of corporations and help corporations to develop from low to high returns on investment, thereby improving their TFP.

Third, DTC reduces information asymmetry and improves corporate TFP. First of all, DTC improves the ability of corporations to obtain “soft information” (Sia, Weill, and Zhang 2021). The multi-dimensional description of the digital footprint of customer consumption behavior helps reduce the degree of information asymmetry between corporations and consumers, so as to achieve accurate matching between supply and demand (He et al. 2018). Secondly, DTC accelerates the modernization of the industrial chain and supply chain of corporate (Zhao, Wang, and Li 2021), enables the upstream and downstream collaboration and communication of the industrial chain and supply chain, unblocks the resource integration channel, and improves the efficiency of resource allocation. Finally, DTC helps corporations to integrate data elements and traditional elements (Holmstrom 2018), accelerates the process of corporate informationization, helps corporations clear the information transmission barriers in many links of the real economic ecosystem (Zhao, Wang, and Li 2021), accelerates the flow of resources among corporations, and reduces the element redundancy, thereby improving corporate TFP. Based on the above analysis, this article proposes:

H1:

DTC helps to improve corporate TFP.

3.2. Study on the Mechanism of DTC Driving Corporate TFP Improvement

GTIC follows the laws of ecological economy, saves resources and energy, can effectively avoid, eliminate or reduce ecological environment pollution and damage, and is conducive to achieving coordinated progress of economy and environment. Its ultimate goal is to serve the sustainable and high-quality development of the economy, especially enabling the economy TFP improvement from investment-oriented growth to efficiency-oriented growth (Wu et al. 2021). Therefore, it is the objective need for China to transform the development mode of national economy and the strategic choice for corporations to enhance their market competitiveness by improving corporate TFP through GTIC (Feng et al. 2022). Therefore, this article further discusses how DTC improves corporate TFP by influencing GTIC.

First, corporate DTC indirectly drives corporate TFP improvement through the promotion effect of GTIC investment. First of all, the liquidity constraints faced by corporations will inhibit investment in GTIC and negatively act on corporate TFP. DTC can ease the external financing constraints of corporations, which can broaden the sources and channels of funds and reduce financing costs through “incremental supplement” and “stock optimization,” and thus increase the investment in corporate GTIC (Xue et al. 2022), thereby indirectly driving corporate TFP improvement. Secondly, the distortion of capital elements not only leads to the misallocation of resources but also inhibits technological progress, weakens the intensity of corporate green innovation investment, causes insufficient motivation for corporate green technology upgrading, and inhibits corporate TFP improvement. In the context of better processing and output of effective information, corporate DTC can “push” information to the financial institutions (Wu et al. 2021), and the financial institutions can simultaneously generate user corporations characteristics “portrait” with the help of information sharing systems, reduce the degree of information asymmetry between financial institutions and corporations, improve the adaptability of credit resources and risk characteristics of green innovation projects, increase investment in GTIC, and indirectly drive corporate TFP improvement. Finally, since GTIC is closely related to corporate production and low-carbon management process, it needs to integrate information on resource consumption, environmental impact, manufacturing systems, etc. DTC enables knowledge elements and data elements to accelerate the sharing and flow among various systems within the organization, improve the level of corporate information sharing, accelerate the integration and sharing of information related to internal and external resources and environment of corporations, and increase the investment in GTIC (Feng et al. 2022; Goldfarb and Tucker 2019; Rigby and Zook 2002), indirectly driving corporate TFP improvement.

Second, corporate DTC indirectly drives corporate TFP improvement through the promotion effect of GTIC willingness. First of all, GTIC is characterized by strong commonweal and a weak economy, which results in poor willingness of corporations to innovate green technology (Marchi 2012). Corporate DTC accelerates the marketization process, is conducive to the flow of GTIC resources among industries, corporate and projects, improves the transmission speed of product and factor market data information, enhances the market trading ability of GTIC resources, and can alleviate the pressures of a poor economy to a certain extent, thereby helping to enhance the GTIC willingness, and indirectly driving corporate TFP improvement. Secondly, GTIC involves the integration, creation, and diffusion of multi-disciplinary and interdisciplinary knowledge in an organization, including such elements as corporate production and operation, energy control, and resource emission reduction (D’Aspremont and Jacquemin 1988). It is difficult to achieve results by relying only on one corporation’s knowledge capital in a single technical field (Mubarak et al. 2021), which inhibits corporate willingness for GTIC. The application of digital technology accelerates the collection, exchange, and integration of internal and external information of corporations, promotes the integration, dissemination, and sharing of information or knowledge among the innovation subjects, so as to realize the integration and reconstruction of knowledge in different technical fields and disciplines. Wang et al. (2022)stimulate corporate willingness to innovate in green technology, and thereby indirectly driving the improvement of corporate TFP. Based on the above analysis, this article proposes:

H2:

DTC indirectly drives corporate TFP improvement by influencing GTIC.

4. Design Development

4.1. Sample Selection and Data Source

This article chooses Chinese A-share listed companies from 2010 to 2020 as the research sample. The basic characteristics and financial data of listed companies are from Wind and CSMAR databases; the green patent data of corporations comes from the database of CNRDS. Based on existing literature, the original data are processed as follows: the samples of financial companies are excluded; PT and ST listed companies are eliminated; the samples of insolvent companies are excluded; the missing samples of main variables are eliminated. In addition, to avoid the influence of abnormal values on the empirical results, 1% tail reduction is applied to all continuous variables, and 19,782 observation samples are finally obtained.

4.2. Definition of Variables

4.2.1. Explained Variable: TFP

In this article, OP method and LP method are used to estimate corporate TFP, and OP method is used to build the following model (Olley and Pakes 1996):

(1) $\mathit{Ln}{Y}_{\mathit{i},\mathit{t}}={\theta }_0+{\theta }_1\mathit{Ln}{K}_{\mathit{i},\mathit{t}}+{\theta }_2\mathit{Ln}{L}_{\left(\mathit{i},\mathit{t}\right)}+{\theta }_3\mathit{Ln}{I}_{\mathit{i},\mathit{t}}+{\theta }_4\mathit{Ag}{e}_{\mathit{i},\mathit{t}}+{\theta }_5\mathit{SO}{E}_{\mathit{i},\mathit{t}}+{\theta }_6\mathit{Exi}{t}_{\mathit{i},\mathit{t}}+{\delta }_j+{\phi }_t+{\epsilon }_{\mathit{i},\mathit{t}}$(1)

Where, $Y_{\mathit{i},\mathit{t}}$ is corporate business income; $K_{\mathit{i},\mathit{t}}$ is expressed by fixed asset investment; $L_{\mathit{i},\mathit{t}}$ is expressed by the number of corporate staff; $I_{\mathit{i},\mathit{t}}$ is intermediate input; $\mathit{Ag}{e}_{\mathit{i},\mathit{t}}$ is corporate age (logarithm); $\mathit{SO}{E}_{\mathit{i},\mathit{t}}$ is a dummy variable of the nature of corporate property rights; $\mathit{Exi}{t}_{\mathit{i},\mathit{t}}$ is a dummy variable indicating whether the corporation withdraws from market. In this article, a corporation whose abbreviation and industry have changed simultaneously is regarded as withdrawing from the market; ${\delta }_j$ and ${\phi }_t$ are industry and time fixed effects, respectively; ${\epsilon }_{\mathit{i},\mathit{t}}$ is a residual term. Meanwhile, this article also uses the LP method to estimate corporate TFP (Levinsohn and Petrin 2003), and the form of its production function is as follows:

(2) $\mathit{Ln}{Y}_{\mathit{i},\mathit{t}}={\sigma }_0+{\sigma }_1\mathit{Ln}{K}_{\mathit{i},\mathit{t}}+{\sigma }_2\mathit{Ln}{L}_{\left(\mathit{i},\mathit{t}\right)}+{\sigma }_3{M}_{\mathit{i},\mathit{t}}+W_{\mathit{i},\mathit{t}}+{\epsilon }_{\mathit{i},\mathit{t}}$(2)

Where, $Y_{\mathit{i},\mathit{t}}$ is the total output; $K_{\mathit{i},\mathit{t}}$ refers to capital input; $L_{\mathit{i},\mathit{t}}$ refers to labor input; $W_{\mathit{i},\mathit{t}}$ refers to the intermediate input products; $M_{\mathit{i},\mathit{t}}$ is the monotone increasing function of $W_{\mathit{i},\mathit{t}}$, and $M_{\mathit{i},\mathit{t}}$ = $M_t\left(K_{\mathit{i},\mathit{t}},W_{\mathit{i},\mathit{t}}\right)$.

4.2.2. Explanatory Variable: DTC

In this article, DTC is divided into “bottom technology application” and “technology practice application” according to the functional realization. At the same time, “bottom technology application” is divided into four categories: artificial intelligence, block chain, cloud computing, and big data. Python crawler technology is used to count the frequency of five keywords in the annual reports of Chinese A-share listed companies, including artificial intelligence technology, big data technology, cloud computing technology, block-chain technology and digital technology, and to sum up the frequency of the above five categories of words related to DTC as the original measurement standard of corporate DTC (Wu et al. 2021). In view of the differences in the length of the annual reports of listed companies, the ratio of the total number of words related to corporate digitalization to the total number of words in the annual reports of corporate is further used to measure the DTC of micro corporations (Yuan et al. 2021). For the convenience of coefficient interpretation, the index is selected to be expanded by 100 times. If the DTC is high, the level of corporate DTC will be high.

4.2.3. Mediating Variable: GTIC

Existing research mostly measures GTIC by the number of green patent applications and green patent authorizations of corporations. However, testing and payment of annual fees are required for corporate patent granting. There are many uncertainties and instabilities due to multiple factors, and patent technology is likely to have an impact on a corporation’s performance during the application process, which makes the number of patent applications more stable, more reliable and more timely than the number of grants. Therefore, this article selects the number of green patent applications of listed companies as the original measurement standard of corporate GTIC and further adds 1 to the number of green patent applications of corporations to take the natural logarithm to measure the GTIC. If the GTIC is high, the GTIC ability will be strong.

4.2.4. Control Variables

Based on the existing literature (Wu et al. 2021; Zhao et al. 2022), this article adds the following control variables: corporate age (AGE), asset–liability ratio (LEV), rate of return on total assets (ROA), corporate growth (GROWTH), total assets turnover (TOA), ownership concentration (TOPEE), board size (BOARD), ratio of independent directors (INDEP), duality (DUAL) and other corporate-level characteristic variables, as well as time (YEAR), industry (IND) and corporate (ID) fixed effects. The above variables are defined in Table 1:

Table 1. Definition of variables.

Variable		Meaning	Method of Measurement
Explained variable	TFP_LP	LP method for calculating total-factor productivity of enterprises	Calculated by LP method
	TFP_OP	OP method for calculating total-factor productivity of enterprises	Calculated by OP method
Explanatory variable	DTC	Digital transformation	Total number of digital-related words/total number of words in enterprise annual report
Mediating variable	GTIC	Green technology innovation	Natural logarithm after the number of green patent applications plus 1
Control variable	AGE	Enterprise age	Natural logarithm after enterprise establishment period plus 1
	LEV	Asset-liability ratio	Total liabilities/total assets at the end of the period
	ROA	Return on assets	Net profit/total average assets
	GROWTH	Enterprise growth	Growth rate of operating income
	TOA	Total assets growth rate	Main business income/average balance of total assets
	TOPI	Ownership concentration	Shareholding ratio of the largest shareholder
	BOARD	Board size	Natural logarithm of the number of directors
	INDEP	Ratio of independent directors	Number of independent directors/number of directors
	DUAL	Duality	1 if the chairman and the general manager are the same person, otherwise 0
	YEAR	Time fixed effect	Dummy variable
	IND	Industry fixed effect	Dummy variable
	ID	Enterprise fixed effect	Dummy variable

4.3. Model Building

To test whether DTC can improve corporate TFP, i.e. H1, the following model is built in this article:

(3) $\mathit{TF}{P}_{\mathit{i},\mathit{t}}={\alpha }_0+{\beta }_1\mathit{DC}{G}_{\mathit{i},\mathit{t}-1}+{\gamma }_1\mathit{Control}{s}_{\mathit{i},\mathit{t}}+{\mu }_i+{\delta }_j+{\phi }_t+{\epsilon }_{\mathit{i},\mathit{t}}$(3)

Where, $\mathit{i}$ represents corporate, $\mathit{t}$represents time; $\mathit{TF}{P}_{\mathit{i},\mathit{t}}$ is the explained variable, that is, corporate TFP, which is measured by LP method and OP method; $\mathit{DC}{G}_{\mathit{i},\mathit{t}}$ is the core explanatory variable, namely, DTC, in view of the fact that DTC depends on infrastructure improvement, information technology application and promotion, and its impact on corporate GTIC and TFP has a certain lag, so this article chooses to lag the explanatory variable by one period; $\mathit{Control}{s}_{\mathit{i},\mathit{t}}$ represents a series of control variables; ${\mu }_i$ refers to the individual corporate fixed effect, ${\delta }_j$ refers to the industry fixed effect; ${\phi }_t$ is time fixed effect; ${\epsilon }_{\mathit{i},\mathit{t}}$ is the random disturbance term. This article focuses on the coefficient ${\beta }_1$, if ${\beta }_{1\mathit{\hspace{0.17em}}}$ is significant and positive, it indicates that DTC can significantly improve corporate TFP, which further indicates that H1 is satisfied.

To test whether GTIC can play a mediating role in the DTC and corporate TFP improvement, namely H2, the following mediation model is built in this article:

(4) $\mathit{GTI}{C}_{\mathit{i},\mathit{t}}={\alpha }_1+{\beta }_2\mathit{DC}{G}_{\mathit{i},\mathit{t}-1}+{\gamma }_2\mathit{Control}{s}_{\mathit{i},\mathit{t}}+{\mu }_i+{\delta }_j+{\phi }_t+{\epsilon }_{\mathit{i},\mathit{t}}$(4)

(5) $\mathit{TF}{P}_{\mathit{i},\mathit{t}}={\alpha }_2+{\beta }_3\mathit{DC}{G}_{\mathit{i},\mathit{t}-1}+{\beta }_4\mathit{GTI}{C}_{\mathit{i},\mathit{t}}+{\gamma }_3\mathit{Control}{s}_{\mathit{i},\mathit{t}}+{\mu }_i+{\delta }_j+{\phi }_t+{\epsilon }_{\mathit{i},\mathit{t}}$(5)

Where model (3) is used to examine whether there is a significant relationship between DTC and corporate GTIC; model (4) is used to re-test after GTIC is added based on model (2) to judge whether there is a mediating effect of GTIC. This article mainly focuses on ${\beta }_2$ and ${\beta }_4$; if ${\beta }_2$ and ${\beta }_4$ are significantly positive, it indicates that GTIC plays a mediating effect, and it satisfies H2. If at least either one of ${\beta }_2$ or ${\beta }_4$ is not significant, the Sobel test is required. If it is significant, it means that the mediating effect is significant, it indicates that H2 is satisfied. Conversely, if the mediating effect is not significant, it indicates that H2 is not satisfied.

5. Empirical Results and Analysis

5.1. Descriptive Statistics

The descriptive statistical results of the main variables in this article are shown in Table 2. The mean of explained variable TFP_LP calculated by LP method is 9.154, with a median of 9.054; the mean of explained variable TFP_OP calculated by OP method is 8.189, with a median of 8.071, which is slightly lower than the results calculated by LP method. Moreover, the mean and median of corporate TFP calculated by LP method and OP method are close to each other, indicating that the data obtained is unbiased. The mean value of the core explanatory variable DTC is 0.132, and the standard deviation is 0.234, which indicates that the overall DTC level of Chinese listed companies is low, and there are also some differences among sample individuals. The mean value of GTIC is 0.920, the minimum value is 0, the maximum value is 7.319, and the standard deviation is 1.217, which indicates that the overall GTIC level of Chinese A-share listed companies is low. There are significant differences between sample individuals. The control variables are basically consistent with the results in the existing literature.

Table 2. Descriptive statistics.

Variable	Number	Mean	Median	StdDev	Min	Max
TFP_LP	19,782	9.154	9.045	1.142	5.717	13.533
TFP_OP	19,782	8.189	8.071	1.018	4.415	12.553
DTC	19,782	0.132	0.054	0.234	0.000	4.362
GTIC	19,782	0.920	0.000	1.217	0.000	7.319
AGE	19,782	2.887	2.935	0.334	1.792	3.502
LEV	19,782	0.437	0.432	0.205	0.057	0.895
ROA	19,782	0.041	0.039	0.064	−0.247	0.219
GROWTH	19,782	0.183	0.114	0.427	−0.561	2.780
TOA	19,782	0.670	0.568	0.458	0.075	2.737
TOPEE	19,782	0.343	0.321	0.149	0.085	0.741
BOARD	19,782	2.131	2.197	0.197	1.609	2.708
INDEP	19,782	0.375	0.357	0.054	0.333	0.571
DUAL	19,782	0.277	0.000	0.448	0.000	1.000

5.2. Benchmark Regression

Table 3 shows the test results of the impact of DTC on corporate TFP. Columns (1) and (2) are regression results that exclude the control variables, but only include individual—fixed effects, industry-fixed effects, and time-fixed effects. The results show that the estimated results of core explanatory variable DTC on corporate TFP are significantly positive at the level of 1%. Columns (3) and (4) are the control variables that may affect the regression estimation results added on the basis of columns (1) and (2). The results show that the regression coefficient of the core explanatory variable DTC is 0.394 for TFP_LP, the estimated coefficient is 0.313 for TFP_OP, and both of them are significantly positive at the level of 1%, which indicates that DTC can significantly improve corporate TFP. If the DTC level is high, corporate TFP will be high, and H1 is verified. The reason may be that DTC can encourage corporations to improve their own GTIC level by virtue of digital information technology, provide effective solutions for green products design, production, and service processes of corporations, support green manufacturing processes of corporations and enhance their own product development capabilities (Bai and Sarkis 2017), thereby improving corporate TFP. At the same time, DTC helps corporations to improve their information transmission and processing efficiency, and reduces corporate transaction costs, search costs, management costs, transportation costs, and information transmission costs (Wu et al. 2021), thereby improving corporate TFP.

Table 3. Results of benchmark regression.

Variable	(1)	(2)	(3)	(4)
	TFP_LP	TFP_OP	TFP_LP	TFP_OP
DTC	0.408*** (6.97)	0.324*** (6.01)	0.394***	0.313***
			(7.77)	(6.76)
AGE			0.273**	0.248**
			(2.54)	(2.50)
LEV			0.831***	0.696***
			(11.63)	(11.09)
ROA			1.149***	1.013***
			(11.93)	(11.51)
GROWTH			0.174***	0.156***
			(16.93)	(15.18)
TOA			0.936***	0.843***
			(24.21)	(25.57)
TOPEE			−0.009	−0.018
			(−0.07)	(−0.16)
BOARD			0.322***	0.247***
			(5.27)	(4.38)
INDEP			0.305*	0.268*
			(1.85)	(1.79)
DUAL			−0.007	−0.009
			(−0.47)	(−0.64)
Constant	8.669***	7.588***	5.987***	5.286***
	(53.30)	(46.06)	(17.29)	(16.12)
YEAR	Yes	Yes	Yes	Yes
IND	Yes	Yes	Yes	Yes
ID	Yes	Yes	Yes	Yes
N	16,956	1,6956	16,956	16,956
R2	0.326	0.334	0.602	0.587

Note: *、**、***represent the significance levels of 10%, 5%, and 1% respectively. The items in brackets are the t-value test of coefficients, which are the same in the following tables.

The regression results of control variables show that the estimated coefficient of AGE is significantly positive at the level of 5% because mature corporations have high size effect, strong financial strength, a high-quality labor force and a perfect management system, which is conducive to promoting corporate TFP improvement. The estimated coefficients of ROA and GROWTH are both significantly positive at 1%, because corporations with strong profitability and good growth will promote corporate innovation and efficient production by virtue of their abundant capital and cash flow (Khanna and Iansiti 1997), thereby helping to improve corporate TFP. The estimated coefficient of BOARD is significantly positive at the level of 1%, because a large board of directors is conducive to improving the level of corporate governance, thereby promoting corporate TFP improvement. The regression coefficient of INDEP to corporate TFP is significantly positive at the level of 10%, because independent directors can ease the principal-agent problem by virtue of their independence and improve the governance effect of the board of directors, thereby helping to improve corporate TFP. In addition, LEV and TOA have a significant positive correlation with corporate TFP, which is consistent with existing research conclusions.

5.3. Mechanism Analysis

The benchmark regression part has verified the promotion effect of DTC on corporate TFP. Next, we will further explore its mechanism to provide more in-depth empirical evidence for accelerating the process of DTC, improving the GTIC capability of corporations, and further improving TFP. Based on model (2), this article uses models (3) and (4) to investigate whether DTC can improve corporate TFP by promoting GTIC. The regression results are shown in Table 4. According to column (1), the estimated coefficient of the core explanatory variable DTC to the corporate GTIC is 0.392 and is significantly positive at the level of 1%, which indicates that DTC can significantly drive corporate GTIC. Columns (2) and (3) show that the estimated coefficients of core explanatory variable DTC on corporate TFP are 0.372 and 0.296, respectively, and both are significantly positive at the level of 1%. The estimated coefficients of mediating variable GTIC on corporate TFP are 0.057 and 0.042, respectively, and both are significantly positive at the level of 1%, which indicates that DTC can indirectly drive corporate TFP improvement by promoting corporate GTIC. The reason is that DTC can increase investment in GTIC, stimulate GTIC willingness, help realize the coordinated progress of the economy and environment, drive the sustainable and high-quality development of China’s economy, and thus improve corporate TFP (Wu et al. 2021). Furthermore, the Sobel test shows that the z statistical values of GTIC are 9.395 and 9.325, respectively, which are significant at the 1% confidence level, and the proportion of mediating effect in the total effect are 27.35% and 31.30%, respectively.

Table 4. Results of mechanism inspection.

Variable	(1) GTIC	(2) TFP_LP	(3) TFP_OP
DTC	0.392*** (4.76)	0.372*** (7.60)	0.296*** (6.55)
GTIC		0.057*** (9.30)	0.042*** (7.86)
AGE	0.060 (0.32)	0.269** (2.55)	0.245** (2.50)
LEV	0.281*** (3.23)	0.815*** (11.58)	0.684*** (11.02)
ROA	0.352** (2.46)	1.129*** (11.89)	0.998*** (11.44)
GROWTH	0.051*** (3.06)	0.171*** (16.77)	0.154*** (14.99)
TOA	−0.150*** (−3.35)	0.944*** (24.86)	0.849*** (26.14)
TOPEE	−0.233 (−1.53)	0.004 (0.03)	−0.009 (−0.07)
BOARD	0.139 (1.48)	0.314*** (5.20)	0.241*** (4.31)
INDEP	−0.036 (−0.12)	0.307* (1.89)	0.270* (1.81)
DUAL	0.012 (0.49)	−0.008 (−0.52)	−0.009 (−0.68)
Constant	0.131 (0.19)	5.979*** (17.47)	5.280*** (16.22)
YEAR	Yes	Yes	Yes
IND	Yes	Yes	Yes
ID	Yes	Yes	Yes
N	16956	16956	16956
R^2	0.186	0.608	0.591
Sobel		Z = 9.395***	Z = 9.325***

5.4. Endogenous Problem

5.4.1. Instrumental Variable Method

Although this article uses the fixed effect model in the benchmark regression part and adds control variables that may affect the regression results, it may still be affected by some missing variables or other endogenous problems, which could result in biased DTC coefficients estimated in this article. Therefore, the instrumental variable method is selected to alleviate possible endogenous problems. This article selects the average value of corporate DTC in the same year, province, and industry as a tool variable for two-stage regression (Benkmasr, Boubaker, and Rouatbi 2015). First, in terms of relevance, the digital transformation of enterprises in the same province and in the same industry may put transformation pressure on the enterprises. In order to maintain their competitiveness and market share, enterprises will choose to follow the transformation pace of the same industry, which accelerates the process of digital transformation for enterprises. At the same time, the digital transformation of enterprises in the same province and in the same industry can provide enterprises with successful transformation experiences and promote the completion of digital transformation. Second, from the aspect of exogeneity, descriptive statistics show that the digital transformation of listed enterprises in China is still in the initial stages and that the overall level is low. It is difficult for the digital transformation of other enterprises in the same industry to produce spillover effects on the TFP of target enterprises. Therefore, the selected instrumental variables conform to the principles of correlation and exogeneity. Meanwhile, the weak instrumental variable test is also carried out in this article. The F value in the first stage is greater than the critical value at the level of 10%, and the regression coefficient of instrumental variables on the digital transformation is significant at the level of 1%, which better proves that there is no weak instrumental variable problem. Table 5 shows the two-stage regression results of the tool variables, in which column (1) is the regression result of tool variable IV to the core explanatory variable DTC, and column (2) is the estimated result of the core explanatory variable DTC to corporate TFP after the endogenous problem is considered. The results show that the estimated coefficient of the core explanatory variable DTC to the tool variable is 0.473 and is significantly positive at the level of 1%, which indicates that there is a significant positive correlation between the DTC of other corporations in the same industry and the DTC of target corporations. As shown in columns (2) and (3), the estimated coefficient of DTC on corporate TFP is significantly positive, at least at the level of 5%, which indicates that the above research conclusions are still satisfied after the endogenous problem is alleviated.

Table 5. Instrumental variable method.

Variable	(1) DTC	(2) TFP_LP	(3) TFP_OP
IV	0.473*** (4.14)
DTC		0.503*** (3.22)	0.347** (2.55)
AGE	0.051 (1.47)	0.266** (2.47)	0.245** (2.48)
LEV	0.002 (0.11)	0.831*** (11.65)	0.695*** (11.09)
ROA	−0.047* (−1.65)	1.153*** (12.00)	1.015*** (11.52)
GROWTH	−0.007 (−1.62)	0.174*** (16.93)	0.156*** (15.19)
TOA	0.008 (0.71)	0.935*** (24.27)	0.843*** (25.58)
TOPEE	−0.093*** (−3.65)	0.003 (0.02)	−0.015 (−0.13)
BOARD	0.004 (0.29)	0.321*** (5.26)	0.246*** (4.37)
INDEP	−0.061 (−1.50)	0.314* (1.91)	0.271* (1.80)
DUAL	−0.002 (−0.45)	−0.007 (−0.47)	−0.009 (−0.64)
Constant	−0.011 (−0.10)	5.990*** (17.31)	5.287*** (16.13)
YEAR	Yes	Yes	Yes
IND	Yes	Yes	Yes
ID	Yes	Yes	Yes
N	16956	16956	16956
R^2	0.265	0.601	0.587
First-stage F	58.46
p-value	0.000***

5.4.2. Heckman Two-Stage Model

In this article, Heckman’s two-stage model is further used for endogenous treatment. First, a dummy variable that represents the high or low level of corporate DTC is constructed based on the grouping of annual median. That is, if the DTC of the corporate is greater than the median of the DTC of all sample corporations in the current year, it is defined as having a high DTC level, and the value is assigned as “1.” On the contrary, if the DTC of the corporate is less than the median of the DTC of all sample corporate in the current year, it is defined as having a low DTC level and the value is assigned as “0.” Secondly, take the generated dummy variable as the explained variable in the first stage of Heckman’s two stages, and introduce all control variables in the model (2) for Probit regression to calculate the Inverse Mills Ratio (IMR). Finally, the IMR estimated in the Probit regression in the first stage is added to the regression in the second stage as the control variable for further regression analysis. The regression results of the second stage are shown in Table 6. The results show that the estimated coefficient of core explanatory variable DTC on corporate TFP is significantly positive at the level of 1%, which indicates that DTC can significantly improve corporate TFP. That is, after endogenous effects are controlled by the Heckman two-stage model, the above research conclusions are still satisfied.

Table 6. Results of Heckman two-stage model regression.

Variable	(1) TFP_LP	(2) TFP_OP
DTC	0.292*** (6.66)	0.234*** (5.83)
IMR	−0.617 (−1.61)	−0.561* (−1.67)
AGE	0.483*** (3.45)	0.309** (2.43)
LEV	0.895*** (8.17)	0.777*** (8.09)
ROA	0.801*** (4.50)	0.651*** (4.09)
GROWTH	0.156*** (10.17)	0.141*** (9.39)
TOA	0.779*** (8.52)	0.700*** (9.83)
TOPEE	−0.138 (−0.77)	−0.138 (−0.94)
BOARD	0.229*** (3.01)	0.176** (2.40)
INDEP	0.131 (0.63)	0.076 (0.41)
DUAL	−0.030 (−1.23)	−0.025 (−1.14)
Constant	7.058*** (10.58)	6.247*** (9.79)
YEAR	Yes	Yes
IND	Yes	Yes
ID	Yes	Yes
N	8442	8442
R^2	0.649	0.635

5.5. Robustness Test

5.5.1. Replacement of Explained Variables

There are many methods to measure corporate TFP, such as the OP method, LP method, GMM method, etc. In the benchmark regression part, LP method and OP method are used to calculate corporate TFP. To verify the reliability of the conclusion, GMM method (Wooldridge 2009) is used to calculate corporate TFP (TFP_GMM) instead of the originally explained variable to carry out the robustness test. The results are as shown in column (1) of Table 7. When the control variables are added and the fixed effects of individuals, industry and time are controlled, the estimated coefficient of the core explanatory variable DTC to corporate TFP is 0.273, and it is significantly positive at the level of 1%. It shows that DTC can significantly promote corporate TFP improvement, providing robust evidence for the previous research conclusions.

Table 7. Robustness test.

Variable	(1) TFP_GMM	(2) TFP_LP	(3) TFP_OP	(4) TFP_LP	(5) TFP_OP	(6) TFP_LP	(7) TFP_OP
DTC	0.273*** (4.47)			0.395*** (6.81)	0.317*** (5.86)	0.395*** (7.74)	0.315*** (6.77)
DIG		0.064*** (9.06)	0.051*** (7.76)
AGE	0.006 (0.06)	0.321*** (2.99)	0.286*** (2.89)	0.181 (1.58)	0.187* (1.73)	0.333*** (3.11)	0.288*** (2.94)
LEV	0.245*** (3.72)	0.832*** (11.67)	0.697*** (11.12)	0.834*** (10.18)	0.718*** (9.93)	0.837*** (11.62)	0.704*** (11.22)
ROA	0.991*** (10.03)	1.140*** (11.82)	1.006*** (11.41)	1.203*** (10.28)	1.090*** (10.01)	1.118*** (11.43)	0.972*** (11.04)
GROWTH	0.112*** (10.77)	0.173*** (16.78)	0.156*** (15.06)	0.191*** (17.02)	0.174*** (15.16)	0.172*** (15.93)	0.156*** (14.11)
TOA	0.966*** (30.96)	0.936*** (24.17)	0.843*** (25.64)	0.902*** (21.56)	0.817*** (22.78)	0.932*** (23.31)	0.842*** (24.91)
TOPEE	−0.195* (−1.68)	−0.024 (−0.18)	−0.030 (−0.26)	0.157 (1.11)	0.106 (0.84)	−0.108 (−0.78)	−0.082 (−0.67)
BOARD	0.049 (0.91)	0.304*** (5.02)	0.233*** (4.15)	0.335*** (4.93)	0.271*** (4.32)	0.296*** (5.17)	0.226*** (4.25)
INDEP	0.056 (0.36)	0.271* (1.66)	0.241 (1.62)	0.390** (2.14)	0.321* (1.93)	0.266* (1.65)	0.218 (1.50)
DUAL	−0.012 (−0.83)	−0.006 (−0.37)	−0.008 (−0.55)	0.018 (0.99)	0.008 (0.53)	−0.006 (−0.39)	−0.009 (−0.64)
Constant	2.034*** (5.11)	5.845*** (16.78)	5.173*** (15.64)	6.188*** (14.11)	5.312*** (10.99)	5.990*** (16.97)	5.249*** (15.88)
YEAR	Yes	Yes	Yes	Yes	Yes	Yes	Yes
IND	Yes	Yes	Yes	Yes	Yes	Yes	Yes
ID	Yes	Yes	Yes	Yes	Yes	Yes	Yes
N	16956	16956	16956	14059	14059	15691	15691
R^2	0.450	0.602	0.587	0.591	0.574	0.613	0.601

5.5.2. Replacement of Explanatory Variables

In this article, five characteristic keywords involving artificial intelligence technology, big data technology, cloud computing technology, block-chain technology, and the application of digital technology in the annual reports of listed companies are used as the keywords for corporate DTC. The final word frequency is formed based on the frequency of each keyword as the measurement index for corporate DTC (Wu et al. 2021). The proportion of DTC-related word frequency in the total number of words in the annual reports of listed companies in the alternative benchmark regression is replaced by robustness test. At the same time, considering the right bias of this kind of data, in this article, 1 is added to the final total word frequency, and then the natural logarithm is taken, which is expressed as DIG. If the DIG value is high, the corporate DTC level will be high. The regression results of replacing core explanatory variables are shown in columns (2) and (3) of Table 7. The estimated coefficients of DIG for corporate DTC on corporate TFP are 0.064 and 0.051, respectively, and both of them are significantly positive at the level of 1%. It shows that DTC can significantly promote corporate TFP improvement, providing robust evidence for the previous research conclusions.

5.5.3. Exclusion of Corporate Strategic Behaviors

In this article, the frequency ratio of digital-related words in the annual reports of listed companies is used to measure corporate DTC. Although it can comprehensively capture the true operation of DTC at the corporate level, it may also be affected by corporate strategic information disclosure behavior (Yuan et al. 2021), which results in more digital information disclosed in the annual reports than implemented. To eliminate this possible impact, the following tests are carried out in this article: first, because most of the listed corporations on the GEM are high-tech corporations, and their business models are closely connected with such digital technologies as the Internet, we choose to exclude the sample of companies listed on the GEM. Second, the disclosure of DTC information in the corporation’s annual report may be affected by the strategic disclosure behavior at the management level. Therefore, we choose to exclude the sample of DTC data is zero for the robustness test. The regression results are shown in columns (4–77) of Table 7. The estimated coefficients of corporate DTC on corporate TFP are significantly positive at 1%, which indicates that DTC can significantly promote corporate TFP improvement, providing robust evidence for the previous research conclusions.

6. Conclusion and Recommendations

This article has investigated the impact of DTC on corporate TFP by taking Chinese A-share listed companies from 2010 to 2020 as the research sample. The research shows that DTC can significantly improve corporate TFP. The research conclusion is still valid after a series of robustness tests, such as using the instrumental variable method and Heckman’s two-stage model to alleviate endogenous problems, replacing explained and core explanatory variables, and excluding strategic behavior of corporations, etc. Mechanism analysis shows that DTC can indirectly drive corporations to improve TFP by improving the investment intensity of GTIC and stimulating the willingness of corporate GTIC. Based on the above conclusions, the following policy recommendations are suggested:

First, given the promotion effect of DTC on corporate TFP, government should issue related policies to encourage and support corporations to accelerate the process of DTC and also accelerate the organic integration of digital technology and traditional production modes to achieve TFP improvement. Corporations need to seize the opportunities offered by DTC, make full use of the dividends of the digital economy and digital finance development, improve financing efficiency, enhance their core competitiveness, and cultivate new driving forces of high-quality development.

Second, given the mediating effect of GTIC in DTC to improve corporate TFP, government should encourage and guide corporations to implement the concept of green development and actively carry out GTIC activities through government subsidies, tax incentives, and other ways to promote high-quality development. At the same time, as the practitioners of GTIC, corporations should fully implement the concept of green development, increase investment in R&D and innovation, and enable their own virtuous circle with higher quality input in production factors.

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.

Additional information

Funding

This research was supported by: The National Social Science Fund of China [22BTJ002], [20BGL130]; Soft Science Special Project of Gansu Basic Research Plan under Grant [No. 22JR11RA101]; Higher Education Innovation Fund Project of Gansu Province [2023A-074]; Key Teaching Research Project of Lanzhou University of Finance and Economics [LJZ202106].