China’s Digital Economy and Enterprise Labor Demand: The Mediating Effects of Green Technology Innovation

Abstract

In the context of the new revolution of science and technology, the digital economy not only exerts a significant and profound impact on the scale of enterprise labor demand, but also increasingly becomes a new engine driving green technology innovation in companies. In this paper, we use the micropanel data of Chinese A-share-listed companies from 2011 to 2020 to assess the impact of the digital economy on the scale of enterprise labor demand and the role played by green technology innovation within it. The study finds that: (1) the digital economy significantly expands the labor demand of enterprises, with a more pronounced expansion effect observed at the lower and higher quantiles of the labor demand scale. (2) Green technology innovation can exert a significant intermediary effect between the digital economy and the scale of enterprise labor demand. (3) There is significant heterogeneity in the impact of the digital economy on the labor demand of enterprises and the mediating effect of green technology innovation. the expansion effect of the digital economy on the scale of labor demand is more pronounced in enterprises located in the western region and characterized by rapid industry technological changes in small-to-medium-sized companies, while the mediating effect of green technology innovation is more pronounced in enterprises located in the eastern region, characterized by slow industry technological changes in large-sized companies. (4) Among the segmented indicators of the digital economy, cloud computing technology, big data technology, and digital technology applications significantly expand the scale of enterprise labor demand. The empirical results of this paper have important theoretical and policy implications for understanding the role of the digital economy in promoting labor force employment and achieving green sustainable development.

1. Introduction

The new round of technological revolution and industrial change is in full effect at present, and the digital economy has developed into a new stage of an intelligent economy with artificial intelligence as the core driving force. As a new driving force for promoting the sustainable development of enterprises, the development of the digital economy has garnered widespread attention. According to the China Digital Economy Development Report (2022) issued by the China Academy of Information and Communication Technology (CAICT) [1], the scale of China’s digital economy reached CNY 45.5 trillion in 2021, with a nominal growth of 16.2% year-on-year and a share of 39.8% of the GDP, providing crucial support to economic development. The digital economy is increasingly becoming a driving force for development in China as it grows in scale and becomes more integrated in all sectors. Strong flexibility and employment flexibility make it a powerful force in the information revolution and will play an increasingly important role in future social development.

Meanwhile, the pressure on the country to stabilize and secure employment remains unrelenting, due to the recurring impact of the pandemic and the international environment. It has been established that information technology has become the main driving force of economic growth in all countries [2,3]. Acemoglu and Restrepo showed that the development of digital technology had an impact on the industrial structure, but also on the total demand of the labor force [4]. As a “stabilizer” and “accelerator” of national economic development, the digital economy has become a key factor in employment-related research, as it has brought about economic development and has had an impact on the scale of demand for labor in enterprises [5]. Meanwhile, green technology innovation, as an essential path for micro-enterprises to transition towards sustainable development, has become a significant and practical research topic in terms of its influencing factors and outcomes. The existing studies demonstrate that the digital economy can significantly promote green technology innovation in enterprises [6,7]. As a vital component of high-quality economic development, the synergistic development of green technology innovation and the digital economy has received considerable attention. Therefore, in the face of the pressure of solving the problem of employment and people’s livelihoods, how does the digital economy impact the demand scale of enterprise labor force? Additionally, what role does green technology innovation play in this context? In this regard, we hope to answer this question based on the micro-enterprise data. By investigating the mechanism of the impact of digital economy development on the scale of labor demand in enterprises and the role played by green technology innovation, we provide a policy basis and targeted policy pathways for fully harnessing the expansion effect of the digital economy and achieving high-quality green economic development.

There is, at present, a wide range of empirical research on digital economy issues, such as carbon emissions [8], market structure [9], and sustainable enterprise development [10], among which the issues of green development and labor force are important areas of digital economy research. To better cope with the opportunities and challenges brought by the digital economy, in this paper, we place the digital economy, green technology innovation, and the scale of labor demand in enterprises within the same research framework, examining the impact of the digital economy on the scale of labor demand and the role played by green technology innovation in between.

There are, at present, different views on the impact of the digital economy on the scale of labor demand. Karabarbounis et al. and Bessen et al. found that technological advances may lead to a reduction in labor market shares [11,12]. Boustanifar et al. suggested that IT did not have a significant impact on employment in the financial sector where IT was heavily applied [13]. This shows that, with the development of the digital economy, there are both risks and opportunities in the employment market and that there is still some uncertainty concerning the employment of labor force. Of these, the effect of the digital economy on the scale of labor demand is mainly achieved through two aspects: direct and indirect effects. In terms of the direct effect of the digital economy, its development affects the employment demand through creation [14,15] and substitution [16,17] effects. On the one hand, under the backdrop of the digital economy, changes, such as job diversification, enhanced talent attraction capacity [18], and improved efficiency of job matching, have a positive impact on employment demand [19]. On the other hand, technological advancements will also inevitably replace repetitive, mechanized tasks, and the changing skill requirements can lead to a mismatch between workers’ capabilities and job vacancies, resulting in structural unemployment and substantial job reductions [20]. Ultimately, the digital economy has both positive and negative effects on the demand scale; however, its impact depends on the net effect of both creation and substitution effects.

In terms of the indirect effect of the digital economy, on one hand, the research has shown that the digital economy has a positive promoting effect on green technology innovation in enterprises [21,22]. Its impact pathway is mainly reflected in the following four aspects: from the perspective of enterprises, the digital economy facilitates talent mobility and capital circulation, promoting optimized resource allocation. It not only provides a better foundation of innovative elements for green technology innovation, but also reduces innovation costs [23]. From the market perspective, the digital economy can alleviate financing constraints and optimize market conditions. The low financing costs brought about by financial technology enable companies to allocate more of their income towards enhancing business and operational capabilities, thus providing a positive incentive for green technology innovation within enterprises [24,25]. From the spillover perspective, the use of the digital economy weakens the role of geographical distance in the flow of factors and goods, strengthening communication and cooperation among enterprises [26,27]. Its characteristics of spillover and sharing make green technology innovation in enterprises more likely to have a driving effect on other companies [28]. From the innovation perspective, to meet the ever-changing consumer demands, companies need to collaborate with external stakeholders, including the government, through open innovation to keep small- and medium-sized enterprises competitive [25]. The digital economy provides favorable conditions for open innovation, thereby promoting the integration and development of businesses with other industries, leading to an enhancement in green technology innovation. Moreover, the existing research indicates that open innovation has evolved from being primarily adopted by high-tech companies to being embraced by firms in the low-tech sector, at present [29]. On the other hand, green technology innovation in enterprises can influence labor demand through different avenues, such as production costs, operational scope, and research and development management. Firstly, during the application of green technology, the optimization of high-cost and high-consumption production processes and equipment reduces resource consumption in products. Lower production costs lead to reduced product costs, increased consumer demand, and an increase in capital available for human capital investment in enterprises, thereby expanding the demand for labor [30,31]. Secondly, green technology innovation in enterprises not only increases the likelihood of companies engaging in new production activities, but also drives the emergence of new green industries, such as energy-efficient and environmentally friendly sectors, and promotes the growth of clean production industries. The expansion of the operational scope significantly and positively impacts labor demand in enterprises [32], as market expansion in emerging sectors compels companies to make decisions regarding operations and labor hiring. Thirdly, the research and development of green technologies and production management in enterprises increase the demand for highly educated and skilled personnel, further increasing the demand for labor in enterprises.

Overall, numerous existing studies explore the impact of the digital economy on the scale of enterprise labor demand, indicating the coexistence of creation and substitution effects on labor demand [14,16], with heterogeneous total effects [33]. Meanwhile, the digital economy presents significant opportunities for green technology innovation, leading to a considerable improvement in enterprise-level green technology innovation [21]. Enterprise green technology innovation can enhance labor demand through organizational innovation, improvements in employee skill structures, and increased end-of-pipe efficiency [34,35]. However, some studies suggest that green technology innovation may enhance production efficiency and lead to labor substitution, resulting in job losses [36]. Based on the existing research and theoretical analysis, this study proposes the following research hypotheses:

Hypothesis 1. 

The digital economy significantly expands the scale of enterprise labor demand.

Hypothesis 2. 

The digital economy influences enterprise labor demand through green technology innovation.

Hypothesis 3. 

There is significant heterogeneity in the impact of the digital economy on enterprise labor demand and the intermediary effect of green technology innovation.

To date, there is a considerable number of research papers on the digital economy, green technology innovation, and the labor demand of enterprises; however, there is still much room for development. It is mainly reflected in the following ways: (1) the measurement methods regarding the degree of digital transformation are not uniform, and the accuracy needs to be improved; (2) the research and calculation of the digital economy and employment are mostly concentrated on the macrolevel, with less research on micro-enterprises; (3) most studies focus on the pairwise analysis of factors, overlooking the transmission mechanism of green technology innovation in the impact of the digital economy on the scale of labor demand in enterprises; and (4) the heterogeneity analysis on the impact of the digital economy on enterprise labor demand, such as region and enterprise type, is not comprehensive.

As such, this study conducts a more in-depth and comprehensive study on the impact of the digital economy on the scale of labor demand. Specifically, by using the micropanel data of A-share-listed companies in China from 2011 to 2020, we explore the impact of the digital economy on labor demand in different quantiles. It also employs mediator and moderator variable analyses to examine the mechanisms through which the digital economy affects the scale of labor demand. The possible marginal contributions of this paper are mainly as follows: firstly, in terms of the theoretical analysis, it comprehensively elucidates the underlying mechanisms of the interaction between digital economy development and the scale of labor demand in enterprises, indicating that the ultimate effect depends on the net effects of both creation and substitution effects. It also systematically analyzes the role played by green technology innovation in this interaction. Secondly, in terms of the research content, it evaluates the influence of digitalization at different quantiles on the scale of labor demand, and systematically examines the mechanisms through which green technology innovation affects the impact of the digital economy on labor demand. The empirical findings of this study indicate that the digital economy exhibits more significant expansion effects on the scale of labor demand at high and low quantiles, with green technology innovation playing an intermediary role, significantly influencing the expansion of labor demand. Furthermore, this study conducts a heterogeneity analysis by classifying the samples based on the factors, such as geographical region, industry, type of enterprises, and specific indicators related to the digital economy, providing stronger evidence to support the hypotheses. Thirdly, in terms of the research data, the analysis is shifted from the macro- to the microlevel, and the empirical results demonstrate that the digital economy significantly expands the scale of labor demand in enterprises. This provides some reference on the topic of how the digital economy affects the labor demand of enterprises, with a view to making rational suggestions for the development of the digital economy and the transformation and upgrading of enterprises.

The reminder of this study is structured as follows: Section 2 introduces the data sources and variable settings; Section 3 presents the empirical research model, examining the impact and effects of the digital economy on the scale of labor demand in companies, as well as the influence pathway of green technology innovation; it then conducts an endogeneity analysis, robustness tests, and heterogeneity analysis; subsequently, the empirical results are discussed; and, finally, we present a conclusion and the policy implications.

2. Data

2.1. Sample Selection and Data Sources

In this paper, we obtained the panel data of all A-share-listed companies in China from 2011 to 2020 as the initial research sample. As the digital economy has become the new driving force of China’s economic growth in recent years, it is more typical to analyze the impact of the development of the digital economy on the scale and structure of corporate labor demand using data from 2011 to 2020.

The main data used in this paper were as follows: (1) data describing the digital transformation degree of listed companies, which were obtained from the annual reports of relevant companies through the official websites of the Shenzhen Stock Exchange (SZSE) and the Shanghai Stock Exchange (SSE), as well as the textual analysis of the content of the business situation in the annual reports and the calculation of key words by the entropy value method after the frequency statistics of the key words; (2) data reflecting the labor demand scale of enterprises, which was obtained by collecting, sorting, and calculating the relevant data in Wind and CSMAR databases; and (3) data for the control variables was obtained from the China National Bureau of Statistics, Wind, and CSMAR databases.

After sorting out the data, the data were treated as follows: (1) we excluded samples with excessive missing values of important indicators; (2) excluded samples from the financial industry; (3) excluded samples influenced by CSRC or the stock exchange due to information disclosure during the sample period; (4) excluded samples for ST and period delisting; and (5) excluded samples for IPOs during the survey years; (6) 1% and 99% tailings were performed for all microlevel continuous variables to reduce the effect of outliers. In this paper, a total of 3950 “enterprise-year” observation samples were obtained for 51 industries and 359 enterprises in 30 provinces (autonomous regions and municipalities directly under the Central Government).

2.2. Meaning of Variables

• Explained variables (Labor_i,t). The explained variables in this paper were the scale of demand. The demand scale was measured by the logarithm of the total number of employees employed by the enterprise.
• Explanatory variable (DIG_i,t). The explanatory variables in this paper were the degree of digital transformation of enterprises, which is still measured with varying degrees of weakness. Considering the applicability of the research subject and the availability of the data, we mainly drew on the study of Wu et al. and Zhao to obtain the degree of digital transformation through a combination of text analysis and entropy value methods [37,38]. First, we used a format converter to convert all the annual reports of the listed companies into a TXT format; then, we used Python to extract the text of the analysis of the operations of the listed companies in the annual reports and processed the text with the Jieba Chinese word separation function. Then, based on the Digital Transformation Trends Report (2020) and a summary of specific keywords related to digital transformation by Wu et al. [37,39], we identified the high-frequency terms related to digitalization; based on the high-frequency words, we subsequently processed the text and cleaned the data, excluding the negative expressions, such as non, don’t, no, nothing, not, ignore, not yet, not at all, as well as keywords for “digital transformation” that were not related to the company (including profiles of shareholders, executives, etc.). Finally, we constructed a digital word segmentation dictionary based on the relevant documents and screening results, including the keyword frequency statistics of artificial intelligence technology, block-chain technology, cloud computing technology, big data technology, and digital technology applications. The selection of keywords is shown in

  Table 1. Construction of digital development indexes for enterprises and selection of keywords.

  Indicator Classification	Indicator Name	Indicator Attribute
  Artificial intelligence technology	Artificial intelligence, business intelligence, image understanding, investment decision aids, intelligent data analysis, intelligent robotics, machine learning, deep learning, semantic search, biometrics, face recognition, voice recognition, identity verification, autonomous driving, natural language processing	+
  Block-chain technology	Digital currencies, smart contracts, distributed computing, decentralization, bitcoin, federated chains, differential privacy technologies, consensus mechanisms	+
  Cloud computing technology	In-memory computing, cloud computing, streaming computing, graph computing, Internet of Things, multi-party secure computing, brain-like computing, green computing, cognitive computing, converged architectures, billion-level concurrency, EB-level storage, information physical systems	+
  Big data technology	Big data, data mining, text mining, data visualisation, heterogeneous data, credit, augmented reality, mixed reality, virtual reality	+
  Digital technology application	Mobile Internet, industrial Internet, Internet healthcare, e-commerce, mobile payment, third-party payment, NFC payment, B2B, B2C, C2B, C2C, O2O, Internet connection, smart wear, smart agriculture Smart transportation, smart medicine, smart customer service, smart home, smart investment advisor, smart cultural tourism, smart environmental protection, smart grid, smart energy, smart marketing, digital marketing, unattended retail, Internet finance, digital finance, Fintech, quantitative finance, open banking	+

  . The second step was the entropy value method, i.e., weighting the index according to data dispersion. The five categories of digital indicators in this paper were all positive, and the result of the entropy processing of the data was the degree of digital transformation (DIGI).

3.

Control variables (X_i,t). To improve that reliability of the research, we added a series of control variables in the paper. At the macrolevel, we selected the level of regional economic development (to some extent, reflecting the regional digital economy and employment development degree), the foreign direct investment level (reflecting the external environmental factors in the enterprise development), and the industrial structure (reflecting the present employment structure of the enterprise). At the microlevel, we took into account the selection of enterprise-level control variables by Sun et al. and focused on the age of the enterprise, the nature of ownership (whether it was a state-owned enterprise), the shareholding ratio of the top-ten shareholders, and the number of board of directors [33].

4.

Mediating variable (Z_i,t). In this article, the mediating variable was the level of green technology innovation in companies. According to the existing research, green technology innovation is often measured through the number of green patent applications and grants. This article adopted the research approach of Zhang et al. and Shu et al. [40,41], measuring the level of green technological innovation in companies by taking the natural logarithm of the number of green patent grants plus one. The number of green patent grants was derived from the compilation of the International Patent Classification Green Inventory published by the World Intellectual Property Organization (WIPO) and the patent application data published by the China National Intellectual Property Office [42]. The specific variables are defined in

Table 2. Variable names and definition methods.

Variable		Symbol	Calculation Method
Explained variable	Demand scale	Labor	Log (total number of employees employed by the enterprise)
Explanatory variable	The degree of digital transformation of enterprises	DIGI	Calculated by text analysis and entropy method
Control variable	The level of regional economic development	GDP	Provincial GDP/provincial population
	Foreign direct investment level	Trade	Foreign direct investment/provincial GDP
	Industrial structure	Stru	Value added of the tertiary industry/provincial GDP
	Age of the enterprise	FirmAge	Log (current year–time of establishment + 1)
	Shareholding ratio of the top-ten shareholders	Top10	Number of shares held by top-ten shareholders/total number of shares
	Nature of shareholding	SOE	Whether it is a state-owned enterprise: 1 for state-owned enterprise and 0 for non-state-owned enterprise
	Number of board of directors	Board	Log (number of board of directors)
Mediating variable	The level of green technology innovation	GTI	Log (number of green patent grants + 1)

.

2.3. Descriptive Statistics

The descriptive statistics of the calculated variables are shown in

Table 3. Descriptive statistics.

	(1)	(2)	(3)	(4)	(5)
Variable	Obs	Mean	Std. Dev.	Max	Min
Labor	3590	8.0432	1.2048	10.9628	4.4067
DIGI	3590	0.0025	0.0065	0.0782	0.0000
GDP	3590	71.3351	32.3845	164.2200	25.6590
Trade	3590	0.0188	0.0137	0.0948	0.0007
Stru	3590	1.3680	0.8973	5.1543	0.5985
FirmAge	3590	2.9087	0.3047	3.6889	1.0986
Top10	3590	0.5562	0.1524	0.9097	0.1988
SOE	3590	0.4655	0.4989	1	0
Board	3590	2.1552	0.1977	2.7081	1.6094
GTI	3590	0.9446	1.1958	6.8997	0

. The minimum value of the demand scale (Labor) is 4.4067, with a standard deviation of 1.2048, indicating significant differences in the demand scale among different enterprises. The average digital development level (DIGI) is 0.0025, the standard deviation is 0.0065, and the maximum value is 0.0782, indicating significant differences in the development level of the digital economy among different enterprises. The average green technology innovation level (GTI) in companies is 0.9446, the maximum value is 6.8997, and the minimum value is 0, indicating significant differences in the green technology innovation level among different enterprises.

3. Analysis of Empirical Results

3.1. Model Setting

On the one hand, to test the impact of the digital economy on the labor demand of enterprises, we constructed the following measurement model in this paper:

$$Labo{r}_{i,t}={\alpha }_0+{\alpha }_{1}DIG{I}_{i,t}+{\alpha }_2{X}_{i,t}+{\mu }_i+{\delta }_{\mathrm{t}}+{\epsilon }_{i,t}$$

(1)

where i and t represent the enterprise and year, respectively, Labor_i,t denotes the explained variables, DIGI_i,t denotes the digital transformation degree of the enterprise i in the t-th year, X_i,t denotes a series of control variables, μ_i and δ_t refer to the individual and year fixed effects, respectively, and ε_i,t denotes the random disturbance term.

On the other hand, to further investigate the effect of the digital economy on enterprises with different scales of labor demand, this study employed the quantile regression for panel data (QRPD) with non-additive fixed effects model to estimate Equation (1), thus obtaining the heterogeneity of the marginal effects of the digital economy. Drawing on the methods of Powell and Ma et al. [43,44], this study applied the adaptive Monte Carlo method (adaptive MCMC) to estimate the QRPD model, and selected five quantiles (10%, 25%, 50%, 75%, 90%) to construct the panel quantile function of the QRPD:

$$Q_{Labo{r}_{i,t}}=\theta (\tau )DIG{I}_{\mathrm{i},t}+\beta (\tau )X_{i,t}$$

(2)

Among them, τ represents the corresponding quantile; Q_Labori,t represents the labor demand scale of enterprises at the respective quantile; DIGI_i,t represents the degree of digital transformation of enterprises at the respective quantile; and X_i,t represents a series of control variables at the macro- and microlevels for the respective quantile, which are the same as Equation (1).

3.2. Baseline Regression

Pan et al. empirically showed that the improvement of digitalization levels in small- and medium-sized enterprises (SMEs) could increase the overall employment demand of companies; however, they did not analyze the regression results at different quantiles [45]. This study investigated the results of baseline and quantile regression estimations at different quantiles using the two-way panel fixed effect, as shown in

Table 4. The effect of digital economy development on labor demand scale.

	Model (1)		Model (2)
Variable			Q = 0.1	Q = 0.25	Q = 0.5	Q = 0.75	Q = 0.9
DIGI	7.7824 ***	5.9537 ***	12.8196 **	7.5206 ***	2.6087 *	18.6464 ***	9.5221 ***
	(1.6160) 1	(1.4911)	(5.2641)	(2.6741)	(1.5417)	(5.5648)	(2.2222)
trade		−1.4946 *	−0.7497	−0.7285	4.7738	8.7482	6.0739 ***
		(0.8103)	(1.2465)	(2.0996)	(3.1416)	(6.3191)	(1.3522)
stru		−0.0165	−0.1213 ***	−0.0331	0.1969 ***	−0.0714 ***	0.2513 ***
		(0.0467)	(0.0439)	(0.0319)	(0.0470)	(0.0275)	(0.0238)
GDP		0.0037 ***	−0.0008	−0.0008 *	0.0042 **	−0.0019 **	−0.0048 ***
		(0.0008)	(0.0010)	(0.0005)	(0.0018)	(0.0007)	(0.0005)
FirmAge		1.1400***	0.7543 ***	0.6831 ***	−0.2002	0.3549 ***	0.8766 ***
		(0.1527)	(0.0663)	(0.0337)	(0.2106)	(0.0572)	(0.0417)
Top10		0.4685 ***	0.2634	0.7975 **	−0.2118	0.3256 **	1.0515 ***
		(0.1509)	(0.1948)	(0.3730)	(0.1776)	(0.1308)	(0.1349)
SOE		0.0905	0.7498 ***	0.2538 ***	0.3853 ***	0.2936 ***	0.3611 ***
		(0.0735)	(0.1455)	(0.0301)	(0.0481)	(0.0664)	(0.0420)
Board		0.1536 **	0.0127	0.3522 ***	0.4118 ***	−0.0239	0.7194 ***
		(0.0696)	(0.1143)	(0.0747)	(0.1249)	(0.1336)	(0.2767)
Individual-year fixed effects	YES	YES	YES	YES	YES	YES	YES
_cons	8.0235 ***	3.8624 ***
	(0.0074)	(0.5124)
N	3590	3590	3590	3590	3590	3590	3590
R2	0.9194	0.9222

***, **, * indicate significance at the levels of 1%, 5%, 10%, respectively. The same applies to the following table. ¹ The robust standard errors are in parentheses, and the same applies to the following tables.

. The estimation results in columns (2) and (3) in Table 4 indicate the positive and significant effects of the digital economy on the scale of labor demand for companies at the statistical level of 1%. With a 1-unit increase in the degree of digital transformation, on average, employment demand increases by 5.9537%. The estimation results for model (2) show that the direction of the estimated coefficient of the digital economy on the scale of labor demand for companies is consistent with model (1), and the regression coefficients at the 0.25, 0.75, and 0.9 quantiles are statistically significant at the statistical level of 1%. This indicates that the regression results for model (1) are highly reliable.

The quantile regression results are shown in Figure 1, illustrating the effects of the digital economy on the scale of labor demand for companies at the 0.1, 0.75, and 0.9 quantiles. The effects were significant at these quantiles, and the impact of the digital economy on labor demand exhibited non-linear changes as the scale of labor demand expanded. There are two possible reasons for the greater promotion of the digital economy on labor demand in smaller-sized companies. Firstly, the digital economy provides opportunities for small-scale enterprises to explore and promote their products on the market. Through the flexibility and integration of information technology, it promotes open innovation [46], which is more conducive to increasing labor demand. Secondly, the digital economy offers convenient analytical tools for companies, leading to an increased demand for highly skilled talents to utilize these tools for decision-making analysis and market optimization. It is worth noting that, although smaller-scale companies benefit from the dividends of the digital economy, their limited capabilities may hinder significant increases in labor demand, despite these opportunities. Therefore, the amplification effect of the digital economy on labor demand may be reduced for companies situated at lower quantiles compared to those at higher quantiles. Additionally, there were several reasons why the digital economy had a strong promotion effect on the labor demand in companies with larger-scale labor demands. Firstly, advancements in the digital economy led to the development of the Internet and e-commerce. Companies with higher labor demands could expand their markets and sales channels through Internet and e-commerce platforms, and an increase in business activities generates a higher demand for labor. Secondly, large companies had abundant social capital, effective risk diversification, and a faster pace of product innovation [47]. They often had the responsibility of technological research and development and innovation. As research and development activities progresses, the demand for highly skilled and highly educated talents will expand further. Lastly, companies with higher labor demands tended to have a higher degree of digital transformation. To increase market competitiveness, they were more likely to explore new markets and optimize business processes, and the expansion of the company’s scale increases the demand for labor. However, the scale of labor demand in companies was influenced by various factors; therefore, the impact of the digital economy on labor demand did not present a linear relationship as the labor demand scale expanded.

3.3. The Results of the Mediation Effect Analysis

Based on the theoretical analysis, it can be inferred that the digital economy has both a direct impact on the scale of labor demand for companies and an indirect impact through the green technology innovation it generates. This transmission mechanism aligns with the mechanism of mediation effects. Mainly, the digital economy has shown a significant ability to promote green technology innovation within enterprises [21,22]. Concurrently, green technology innovation can influence the scale of enterprise labor demand by reducing production costs, expanding business operations, increasing the likelihood of engaging in new production activities, and generating demands for research and development management. The structural model of the assumed relationships between variables is shown in Figure 2. To further analyze the mechanism of the digital economy’s impact on the scale of labor demand in companies, this study selected the level of green technology innovation as the mediator variable and constructed the following mediation effect model:

$$Labo{r}_{i,t}={\rho }_0+{\rho }_{{}_1}DIG{I}_{i,t}+{\rho }_2{X}_{i,t}+{\mu }_{\mathrm{i}}+{\delta }_{\mathrm{t}}+{\epsilon }_{i,t}$$

(3)

$$GT{I}_{i,t}={\lambda }_0+{\lambda }_{{}_1}DIG{I}_{i,t}+{\lambda }_2{X}_{i,t}+{\mu }_{\mathrm{i}}+{\delta }_{\mathrm{t}}+{\epsilon }_{i,t}$$

(4)

$$Labo{r}_{i,t}={\psi }_0+{\psi }_{{}_1}DIG{I}_{i,t}+{\psi }_{2}GT{I}_{i,t}+{\psi }_3{X}_{i,t}+{\mu }_{\mathrm{i}}+{\delta }_{\mathrm{t}}+{\epsilon }_{i,t}$$

(5)

The regression results for the mediation effect are shown in

Table 5. The examination results of the mediation effect.

	(1)	(2)	(3)
Variable	Labor	GTI	Labor
DIGI	5.9537 ***	11.2751 ***	4.7560 ***
	(1.4911)	(2.6952)	(1.5115)
GTI			0.1062 ***
			(0.0118)
trade	−1.4946 *	2.9674 **	−1.8098 **
	(0.8103)	(1.2684)	(0.7856)
stru	−0.0165	0.0790	−0.0249
	(0.0467)	(0.0669)	(0.0461)
GDP	0.0037 ***	−0.0014	0.0039 ***
	(0.0008)	(0.0013)	(0.0008)
FirmAge	1.1400 ***	0.2835	1.1099 ***
	(0.1527)	(0.2120)	(0.1466)
Top10	0.4685 ***	0.1629	0.4512 ***
	(0.1509)	(0.1685)	(0.1444)
SOE	0.0905	0.0884	0.0811
	(0.0735)	(0.0965)	(0.0720)
Board	0.1536 **	−0.0689	0.1609 **
	(0.0696)	(0.0995)	(0.0684)
Individual-year fixed effects	YES	YES	YES
Sobel Z			3.939 ***
_cons	3.8624 ***	0.0436	3.8578 ***
	(0.5124)	(0.6941)	(0.4940)
N	3590	3590	3590
R2	0.9222	0.7809	0.9247

***, **, * indicate significance at the levels of 1%, 5%, 10%, respectively.

. When green technology innovation is included as the mediator variable, the results indicate a significant mediation effect at the company level. The results in column (2) in Table 5 demonstrate that, when green technology innovation is the dependent variable, holding other factors constant, a one-unit increase in the degree of digital transformation leads to an 11.2751% increase in the level of green technology innovation in companies, with statistical significance at the 1% level. When both the digital economy and green technology innovation are included as explanatory variables, the regression coefficient for the digital economy decreases from 5.9537 to 4.7560, suggesting that green technology innovation has a certain impact on the scale of labor demand in companies. Holding other factors constant, a one-unit increase in the number of green patents presented results of a 0.1062% expansion in labor demand for companies. Additionally, the results of the Sobel test show a p-value significantly less than 0.01, and the proportion of the mediated effect is approximately 20.12%. This indicates a significant influence of the digital economy on green technology innovation, with green technology innovation playing an important role in the impact of the digital economy on the scale of labor demand, demonstrating a significant partial mediation effect. This is because the digital economy itself entails technological development and innovation, and an increase in the degree of digital transformation benefits the improvement of the company’s level of green technology innovation, thereby promoting the expansion of labor demand in companies. The empirical research results indicate that green technology innovation has a significant stimulating effect on the scale of enterprise labor demand, with the mediating effect accounting for approximately 20.12%. However, studies have shown that the impact of innovation on employment varies significantly across different industries, subject to the conditions of technological innovation [5]. As technological innovation in enterprises involves certain innovation costs, the mediating effect of green technology innovation may be relatively less evident in some companies. Nonetheless, the negative correlation between innovation and employment levels is primarily observed in small businesses, while there is a significant positive lag effect of innovation on employment growth, surpassing the immediate impact. Thus, in the long run, innovation contributes to promoting employment growth [48].

3.4. Endogeneity Analysis

The endogeneity issue is a critical concern that the empirical research must address. On one hand, given the research theme of this paper, endogeneity problems may exist due to the reverse causality between the digital economy and labor demand scale. The impact of a company’s digital economy on the labor demand scale can simultaneously be influenced by the company’s demand for labor through various means, such as digital economy talent, a highly educated and skilled labor force, and gender, affecting the extent of the digital transformation. On the other hand, there are numerous factors influencing both the development of the digital economy in companies and the scale of labor demand, potentially leading to omitted variable bias. To address this endogeneity issue, the paper employed the instrumental variable analysis to ensure the reliability and accuracy of the results. The core explanatory variable, lagged by one period, was used as an instrumental variable to tackle the endogeneity problem. The results of the IV regression can be observed in

Table 6. The estimation results of the instrumental variable.

	(1) First Stage	(2) Second Stage	(3)
Variable	DIGI	Labor	Labor
L.DIGI	0.6614 ***		−0.5718
	(0.0466)		(2.1302)
DIGI		6.0710 ***	6.9356 ***
		(2.2560)	(2.1391)
Control variables	YES	YES	YES
Individual-year fixed effects	YES	YES	YES
Constant	0.0112	3.0805 ***	3.1898 ***
	(0.0074)	(0.4552)	(0.6085)
Kleibergen–Paap rk LM statistic	54.9300 ***
Kleibergen–Paap rk Wald F statistic	201.8490
N	3231	3231	3231
R2	0.8205	0.9331	0.9331

*** indicates significance at the levels of 1%.

.

The first-stage regression results presented in Table 6 indicate that the lagged explanatory variable significantly promotes the digital transformation of enterprises, with an LM statistic of 54.9300 for the over-identification test and a Wald F statistic of 201.8490 for the weak instruments test. Passing both the over-identification and weak instruments tests ensures the effectiveness of the instrumental variable without issues of under-identification, weak identification, or over-identification. The second-stage regression shows that the digital economy has a positive and significant impact on the scale of enterprise labor demand, which aligns with the benchmark regression results, further confirming that digital transformation significantly expands labor demand.

Although the over-identification test, to some extent, indicated the exogeneity of instrumental variables, there remains a possibility of persistent effects from the previous phase of the digital transformation, suggesting partial endogeneity. Furthermore, this study incorporated both instrumental variables and the degree of digital transformation in the regression equation. If the estimated coefficient of instrumental variables was not significant, it enhanced the confidence in their exogeneity results. The estimation results are shown in Table 6, column (3), where the estimated coefficients of instrumental variables are all insignificant and close to zero, indicating the exogeneity of the instrumental variables.

3.5. Robustness Test

Since the results of the previous regression may be biased to some extent, this paper further tested the robustness by controlling the multi-dimensional fixed effect, the transformation explanatory variables’ measure, and the sample sub-interval estimation.

3.5.1. Controlling Multi-Dimensional Fixed Effects

Regions with a high level of economic development have a relatively rapid technological development of the Internet, and thus have a relatively great advantage in the development of the digital economy. Considering that the labor demand scale of enterprises is also affected by regional and industry levels, we added the province and industry fixed effects to the regression in this paper to further test the robustness of the regression results. Column (1) in

Table 7. The regression results of controlling multi-dimensional fixed effects and transforming the measures of explanatory variables.

	(1)	(2)	(3)
Variable	Labor	Labor	Labor
DIGI	5.9537 ***
	(1.4911)
DIGIW		4.1870 ***
		(1.1109)
L.DIGI			4.0153 **
			(1.5670)
trade	−1.4946 *	−1.5125 *	−1.6462 **
	(0.8103)	(0.8103)	(0.7989)
stru	−0.0165	−0.0173	−0.0090
	(0.0467)	(0.0467)	(0.0466)
GDP	0.0037 ***	0.0038 ***	0.0034 ***
	(0.0008)	(0.0008)	(0.0008)
FirmAge	1.1400 ***	1.1338 ***	1.3682 ***
	(0.1527)	(0.1529)	(0.1848)
Top10	0.4685 ***	0.4659 ***	0.6555 ***
	(0.1509)	(0.1509)	(0.1651)
SOE	0.0905	0.0899	0.0819
	(0.0735)	(0.0735)	(0.0780)
Board	0.1536 **	0.1525 **	0.1576 **
	(0.0696)	(0.0696)	(0.0695)
Individual-year provincial-industry fixed effects	YES
Individual-year Fixed effects		YES	YES
_cons	3.8624 ***	3.8844 ***	3.0896 ***
	(0.5124)	(0.5129)	(0.6093)
N	3590	3590	3231
R2	0.9222	0.9222	0.9328

***, **, * indicate significance at the levels of 1%, 5%, 10%, respectively.

presents the regression results for the impact of digital economic development at the demand scale after increasing provincial and industry fixed effects. According to the results presented in column (1) in the table, at the statistical level of 1%, the digital economy has a significant positive effect on the labor demand scale of firms. An increase of 1 unit in the level of digital transformation leads to an average expansion of 5.9537% in the labor demand scale. Therefore, even after considering macrolevel factors, the research conclusion still holds true.

3.5.2. Transformation of Explanatory Variable Measurements

To test the robustness of the regression results, the measures of the degree of digital transformation of explanatory variables were transformed. Given the different emphases of the five categories of indicators in each enterprise’s digital transformation process, each category was given a different weight when measured by the entropy value method. Hence, the weights of the five categories of indicators were changed and each of the standardized categories was assigned an average weight, and the degree of digital transformation measured by the average weighting method was referred to as DIGIW. Column (2) in Table 7 reports the regression results of the digital economic development level on the scale of enterprise labor demand under the average weighting method, and the results show that the impact coefficient of digital economic development on it is significant at the significant 5% level. This result further confirms the robustness of the conclusion.

Furthermore, considering the cyclical and lagging nature of digital transformation, the current period’s digital transformation may have a more noticeable impact on the scale of labor demand in the following period. Therefore, this study included a lagged variable of enterprise digital transformation in the regression analysis. Column (3) in Table 7 presents the regression estimation results with a lag of one period for the degree of digital transformation. The results demonstrate that the coefficient of the lagged degree of digital transformation is statistically significant at the 5% level for the scale of labor demand in enterprises, indicating the long-term impact of digital economic development on the scale of labor demand.

3.5.3. Sample Subinterval Model Estimation

As listed companies in the manufacturing sector have been listed for a relatively long time and have a large sample size and comprehensive annual report data, the development of listed companies in the information transmission, software, and information technology services sector is closely linked to digital transformation and is more significant for the research on the digital economy than other enterprises. Hence, this paper only kept the sample data of the manufacturing industry, information transmission, and software and information technology service industry, and further examined the model estimation effect of the sample sub-interval. The results show that the estimation results of the model of the sample sub-interval (see

Table 8. The regression results for the sample subinterval model.

	(1) Sub-Interval Sample	(2) Shortened Sample Range
Variable	Labor	Labor
DIGI	5.9536 ***	5.8419 ***
	(1.4634)	(1.9309)
trade	−2.5492 ***	−0.3438
	(0.8476)	(1.0896)
stru	0.0382	0.0361
	(0.0513)	(0.0515)
GDP	0.0019 **	0.0010
	(0.0009)	(0.0009)
FirmAge	1.2093 ***	0.8627 **
	(0.1552)	(0.3359)
Top10	0.1056	1.3832 ***
	(0.1199)	(0.2817)
SOE	0.0372	0.0988
	(0.0747)	(0.0642)
Board	0.2013 ***	0.2483 ***
	(0.0769)	(0.0840)
Individual-year fixed effects	YES	YES
_cons	3.8876 ***	4.0218 ***
	(0.5100)	(1.0864)
N	2894	2154
R2	0.9300	0.9607

*** and ** indicate significance at the levels of 1% and 5%, respectively.

) do not change the conclusions of this paper, and the results remain robust to some extent.

In addition, this study borrowed Pan et al.’s method to shorten the sample window to test the regression robustness [45]. The digital economy has expanded in China since 2015, and its penetration and impact on social life has increased, with the digital transformation of enterprises entering an increase in development. Hence, the sample range was adjusted to 2015–2020 for the regression analysis. The regression results are shown in Table 8. Column (2) in Table 8 reports the result of the regression estimation in the case of shortening the sample range. The results suggest that the coefficient of the degree of digital transformation after shortening the sample range has a significant impact on the labor demand of the enterprise at a statistical level of 1%. This indicates that the development of the digital economy has a significant impact on the scale and structure of labor demand.

3.6. Heterogeneity Analysis

The impact of the digital economy on labor demand, as well as the role of green technology innovation, may vary across different regions, industries, and types of enterprises. In order to thoroughly explore this phenomenon and provide policy suggestions for the future development of enterprises, we analyzed the heterogeneity of enterprises based on different division criteria in this paper.

3.6.1. Heterogeneity Analysis Based on Firm Characteristics

Region-Based Heterogeneity Test

First, we examined the impact of the digital economy on the labor demand scale of enterprises in different regions. We divided Chinese enterprises into three groups according to their locations, eastern, central, and western, and ran separate regressions for enterprises in different regions. The estimation results are shown in

Table 9. The results of the region-based heterogeneity test.

	(1) Eastern Region	(2) Central Region	(3) Western Region
Variable	Labor	Labor	Labor
DIGI	4.5495 **	4.8382 *	13.4076 ***
	(1.9204)	(2.7256)	(4.4044)
Control variables	YES	YES	YES
Individual-year fixed effects	YES	YES	YES
_cons	4.4130 ***	3.5081 **	3.6844 **
	(0.5841)	(1.4195)	(1.7074)
N	2180	800	590
R2	0.9171	0.9257	0.9450
Inter-group coefficient differences	Eastern–central region 0.2887 (0.4450) 1	Central–western region 8.5694 ** (0.0500)	Western–eastern region −8.8581 *** (0.0000)

¹ The p-values of inter-group coefficient differences are in parentheses. The same applies to the following table. ***, **, * indicate significance at the levels of 1%, 5%, 10%, respectively.

. From the perspective of the scale of labor demand in enterprises, the digital economy has a positive promoting effect on the scale of labor demand in the eastern, central, and western regions. The estimated coefficients were 4.5495, 4.8382, and 13.4076, respectively, and they were statistically significant at the 5%, 10%, and 1% levels, respectively. Moreover, the intergroup analysis results indicate that the coefficient differences between the central–western and the eastern–western regions are statistically significant at the 5% level or higher, indicating significant variations in the role of the digital economy in different regions. A further comparison of the coefficients revealed that, compared to the eastern and central regions, the western region experienced a greater positive effect of the digital economy on labor demand (+13.4076%). The main reason for this phenomenon was that the eastern and central regions had relatively higher levels of economic development and a more advanced digital economy, while the growth potential of the digital economy in the western region has not been fully realized, leading to a stronger promoting effect on the scale of labor demand in enterprises. Dahlman et al. and Bukht et al. provided theoretical policy recommendations on how developing countries develop their digital economies [49,50]. The abovementioned region-based heterogeneity analysis can provide data to support policy formulations in China and developing countries in regions with development situations similar to China, allowing countries to formulate more effective development policies based on the actual development situation of the region.

Heterogeneity Test Based on the Industry Perspective

Due to the nature of the industries, the demand for data elements varies and so does the impact on the labor demand of enterprises. Unlike the existing articles that focus on industry differences by studying specific industries or categorize them into three major sectors [51,52], this study adopts a different approach. This study adopted the approach of Agrawal et al. by categorizing the computer, communications, and other electronic equipment manufacturing industry (industry code: 39), electrical machinery and equipment manufacturing industry (industry code: 38), and instrumentation manufacturing industry (industry code: 40) as industries with a rapid pace of technological renewal [53]. The enterprises were then divided into two categories based on their industry types: fast- and slow-changing technology industries. The estimation results from the sample regression are presented in columns (1) and (2) in

Table 10. Heterogeneity test results based on industry perspective and enterprise type.

	(1) Firms with Fast-Changing Technology Industries	(2) Firms with Slow-Changing Technology Industries	(3) Large-Scale Enterprises	(4) Small- and Medium-Sized Enterprises
Variable	Labor	Labor	Labor	Labor
DIGI	7.6333 ***	4.0861 **	4.7753 ***	19.1940 ***
	(2.2619)	(2.0807)	(1.5360)	(4.7880)
Control variables	YES	YES	YES	YES
Individual-year fixed effects	YES	YES	YES	YES
_cons	3.3848 ***	4.8765 ***	4.0177 ***	2.0417 *
	(0.8867)	(0.6307)	(0.5638)	(1.2023)
N	680	2910	3110	480
R2	0.9385	0.9191	0.9079	0.7520
Inter-group coefficient differences	Firms with fast-changing technology industries–firms with slow-changing technology industries −3.5472 *** (0.0000)		Large-scale enterprises–small- and medium-sized enterprises 14.4187 * (0.0900)

***, **, * indicate significance at the levels of 1%, 5%, 10%, respectively.

. In Table 10, column (1) presents the regression results for firms with fast-changing technology industries, while column (2) shows the regression results for firms with slow-changing technology industries. The results indicate that the digital economy has a significant impact on both types of firms at the 1% and 5% levels of statistical significance, respectively. The test results for inter-group coefficient differences reveal significant variations in the effects of the digital economy among firms in different industries at a 1% level of significance. Holding other conditions constant, a one-unit increase in the level of digital transformation is associated with a 3.5472% higher impact of the digital economy on labor demand for firms with fast-changing technology industries compared to other firms. This may be attributed to the fact that industries, such as the computer industry, experience rapid technological changes, and with the rapid development of the digital economy, as key providers of digital technology, they continue to push the technological frontier. Consequently, they have a higher demand for high-skilled labor. On the other hand, firms with slow-changing technology industries primarily benefit from the spillover effects of digital technology. While enjoying the benefits of the digital economy, their expansion of labor demand is relatively smaller in scale.

Heterogeneity Test Based on Enterprise Type

The sensitivity of different-sized firms to the impact of the digital economy on labor demand may vary. This study classified the firms into three types, large-, medium-, and small-sized, based on the Classification of Enterprises by Size (2017) issued by the National Bureau of Statistics of China [54]. The following analysis investigated the differences in the impact of the digital economy on labor demand between large- firms and medium–small-sized firms. The results of the grouped regression, shown in columns (3) and (4) in Table 10, indicate significant positive effects of the digital economy on labor demand for both large- and medium–small-sized firms at a 1% level of significance. The analysis of inter-group coefficient differences revealed significant differences in the effects of the digital economy between large- and medium–small-sized firms at a 10% level of statistical significance. The increase in the level of digital transformation had a greater impact on labor demand for medium–small-sized firms (+19.1940%) compared to large-sized firms (+4.7753%). This result suggests that digital transformation does not have a scale-expansion effect on labor demand for firms. On the contrary, smaller-sized firms had a greater impact on labor demand.

Heterogeneity Analysis Based on Segmented Indicators

To further explore the impact of different dimensions of digital transformation on the scale of enterprise labor demand, this study conducted regressions based on the digital transformation index for various dimensions, and the regression estimation results are presented in

Table 11. Heterogeneity testing results based on segmented indicators.

	(1)	(2)	(3)	(4)	(5)
Variable	Labor	Labor	Labor	Labor	Labor
AIT	0.0410
	(0.5403)
BCT		−0.2012
		(0.3901)
CCT			0.9410 ***
			(0.2956)
BDT				3.2543 ***
				(0.8695)
DTA					2.7621 ***
					(0.5626)
Control variables	YES	YES	YES	YES	YES
Individual-year fixed effects	YES	YES	YES	YES	YES
_cons	3.6981 ***	3.6950 ***	3.8201 ***	3.7329 ***	3.8453 ***
	(0.5113)	(0.5114)	(0.5173)	(0.5109)	(0.5052)
N	3590	3590	3590	3590	3590
R2	0.9219	0.9219	0.9220	0.9223	0.9223

*** indicates significance at the levels of 1%.

. Columns (1) to (5) represent the effects of artificial intelligence technology, block-chain technology, cloud computing technology, big data technology, and digital technology applications on enterprise labor demand scale, respectively. Feng X.L. et al. (2023) found that artificial intelligence technology innovation significantly negatively affected the scale of enterprise labor demand [55]. In contrast, the empirical results in column (1) show a positive but insignificant coefficient for the impact of artificial intelligence technology on labor demand. This may be attributed to the fact that the effects of business transformation caused by deep learning and machine learning at present have not yet fully materialized, leading to an insignificant impact on the labor demand scale. Column (2) indicates that block-chain technology has a negative and insignificant impact on enterprise labor demand. On the other hand, columns (3) to (5) demonstrate that cloud computing technology, big data technology, and digital technology applications can significantly promote enterprise labor demand scale, similar to the effects of the digital economy on enterprise labor demand. Cloud computing technology, represented by the Internet of Things and cyber-physical systems, increases the demand for high-skilled labor. Big data technology enables companies and job seekers to have a low-cost, comprehensive two-way understanding through job platforms, enhancing the labor-matching efficiency [56] and increasing the labor demand from enterprises. Digital technology applications, such as Internet finance, can reduce a company’s financing costs and increase investments in human capital [25]. Chen Q. found that the application of modern information technology had a significant positive relationship with the total labor demand in China [57], consistent with the conclusions obtained from the empirical results in this study.

3.6.2. Heterogeneity Analysis of the Mediating Effect

Regional-Based Heterogeneity Testing of Mediating Effect

There were issues of imbalance in the digital economic development among different regions within a country. Constrained by factors such as regional development levels, the level of green technology innovation also varied across different regions. This study divided the sample companies into three areas, eastern, central, and western, and conducted a regional heterogeneity analysis of the “digital economy–green technology innovation–labor demand scale of enterprises” path. The results are presented in

Table 12. Heterogeneity testing results of mediating effects based on regions.

	Eastern Region		Central Region		Western Region
Variable	(1) GTI	(2) Labor	(3) GTI	(4) Labor	(5) GTI	(6) Labor
DIGI	12.6810 ***	3.3986 *	3.9241	4.3111	23.7447 **	10.6951 **
	(3.3768)	(1.9490)	(3.9694)	(2.7175)	(12.0314)	(4.2740)
GTI		0.0908 ***		0.1343 ***		0.1142 ***
		(0.0137)		(0.0292)		(0.0256)
Control variables	YES	YES	YES	YES	YES	YES
Individual-year fixed effects	YES	YES	YES	YES	YES	YES
Sobel Z		3.2620 ***		0.7859		2.3620 **
_cons	0.2966	4.3861 ***	0.9953	3.3744 **	−4.1328	4.1565 **
	(0.8421)	(0.5723)	(1.4928)	(1.3403)	(2.7240)	(1.6334)
N	2180	2180	800	800	590	590
R2	0.8105	0.9189	0.7382	0.9297	0.6685	0.9476
Inter-group coefficient Differences in the impact of DIGI on GTI	Eastern– central region −8.7569 (0.1200)		Central–western region 19.8205 *** (0.0050)		Western–eastern region −11.0636 *** (0.0000)

***, **, * indicate significance at the levels of 1%, 5%, 10%, respectively.

.

The estimation results for step (4) of the mediating effect model are shown in columns (1), (3), and (5) in Table 12. For companies in the eastern and western regions, the digital economy had a significant positive effect on green technology innovation at the 1% and 5% levels of statistical significance, respectively. Moreover, the intergroup coefficient difference results of DIGI’s impact on GTI indicate significant differences between western and the eastern and central regions. The digital economy had the greatest promoting effect on green technology innovation in enterprises located in the western region. One possible reason was that the market size in the western region was relatively smaller, and the spread of digital technology and the development of the Internet provided opportunities to overcome traditional regional limitations [28]. Companies can target a broader market, expand their sales scale, and subsequently increase their demand for labor.

In addition, the results for the Sobel test indicate that the p-values for the eastern and western regions are significantly lower than 0.01 and 0.05, respectively, while the central region did not pass the significance test. Furthermore, the proportions of the mediating effects in the eastern and western regions were 0.2530 and 0.2023, respectively, indicating that the mediating effect of green technology innovation was greater in the eastern region than in the western region. There may be two reasons for this: firstly, it could be due to the technological barriers and specialized knowledge. Companies in the western region may face challenges of technological and talent shortages in green technology innovation, making it difficult for them to engage in such an innovation. Secondly, financial and resource constraints play a role. Green technology innovation often requires significant investments and resource support. However, the level of economic development in western China is relatively lower, and companies in the region may face limitations in terms of funds and resources. Consequently, they may find it challenging to undertake large-scale green technology innovation projects. As a result, the mediating role of green technology innovation in the relationship between the digital economy and labor demand scale is relatively smaller in western China.

Heterogeneity Testing of Mediating Effects Based on the Industry Perspective

Due to different industry types in which companies operate, there were variations in their levels of green technology innovation and the mediating effects they exerted. Following the same classification method as the heterogeneity test based on the industry perspective mentioned earlier, this study categorized the companies into two types: those in industries with rapid technological changes and those in industries with slow technological changes. The heterogeneity testing results of the mediating effects are shown in

Table 13. Heterogeneity testing results of mediating effects based on the industry perspective.

	Firms with Fast-Changing Technology Industries		Firms with Slow-Changing Technology Industries
Variable	(1) GTI	(2) Labor	(3) GTI	(4) Labor
DIGI	8.0770 *	6.8609 ***	11.3582 ***	2.8731
	(4.3548)	(2.1527)	(3.4847)	(2.2158)
GTI		0.0956 ***		0.1068 ***
		(0.0187)		(0.0141)
Control variables	YES	YES	YES	YES
Individual-year fixed effects	YES	YES	YES	YES
Sobel Z		1.7410 *		2.9880 ***
_cons	0.7446	3.3136 ***	1.0544	4.7639 ***
	(1.4275)	(0.8379)	(0.7801)	(0.6271)
N	680	680	2910	2910
R2	0.7972	0.9407	0.7658	0.9215
Inter-group coefficient differences in the impact of DIGI on GTI	Firms with fast-changing technology industries– firms with xlow-changing technology industries 3.2813 *** (0.0050)

*** and * indicate significance at the levels of 1% and 10%, respectively.

.

Based on the results in columns (1) and (3) in Table 13, it is evident that both types of companies, those in industries with rapid technological changes and those in industries with slow technological changes, have a positive promoting effect on green technology innovation. The estimation results for the intergroup coefficient differences indicate a significant difference, at a 1% level of significance, in the impact of the digital economy on green technology innovation across industries with different rates of technological changes. The effect of the digital economy on green technology innovation in industries with slow technological changes is more significant. The p-values from the Sobel test were significantly lower than 0.1 and 0.01, and the proportion of the mediating effect in companies with rapid technological changes was 0.1012, which was smaller than the proportion of the mediating effect in companies with slow technological changes (0.2969). The main reason for this phenomenon was that companies in industries with slow technological changes made slower advancements in the technological frontier, requiring greater integration with other fields to continuously broaden their technological scope. Therefore, the mediating effect of developing new green technologies in expanding the labor demand scale was more significant for companies in industries with slow technological changes.

Heterogeneity Testing of Mediating Effects Based on Enterprise Type

Companies were classified into large-, medium-, and small-sized enterprises based on their types, and heterogeneity testing of the mediating effects was conducted according to different enterprise types. The results of the grouped regression are presented in

Table 14. Heterogeneity testing results of mediating effects based on the enterprise perspective.

	Large-Scale Enterprises		Small- and Medium-Sized Enterprises
Variable	(1) GTI	(2) Labor	(3) GTI	(4) Labor
DIGI	13.5081 ***	3.3974 **	−4.9983	19.6004 ***
	(2.7814)	(1.5558)	(8.8415)	(4.5973)
GTI		0.1020 ***		0.0813 **
		(0.0124)		(0.0356)
Control variables	YES	YES	YES	YES
Individual-year fixed effects	YES	YES	YES	YES
Sobel Z		4.2580 ***		−0.6540
_cons	0.4183	3.9751 ***	−2.3730	2.2347 *
	(0.7549)	(0.5439)	(1.7771)	(1.2060)
N	3110	3110	480	480
R2	0.7878	0.9108	0.5844	0.7554
Inter-group coefficient differences in the impact of DIGI on GTI	Large-scale enterprises–small- and medium-sized enterprises −18.5064 * (0.0850)

***, **, * indicate significance at the levels of 1%, 5%, 10%, respectively.

. The results in columns (1) and (3) in Table 14 indicate a significant positive effect of the digital economy on green technology innovation in large-sized enterprises. The inter-group coefficient differences in the impacts of DIGI on GTI suggest that the effect of the digital economy on green technology innovation significantly varies across different enterprise types at a 10% level of significance. The Sobel test revealed a clear mediating effect of green technology innovation on large-sized enterprises, while the test was not passed for small- and medium-sized enterprises. In large-sized enterprises, when the digital economy affected the labor demand scale, green technology innovation played a mediating role of 28.86%. The reason for this could be that large-sized enterprises have a higher degree of digital transformation and make greater investments in green technology innovation, resulting in an amplifying effect on the labor demand scale. On the other hand, small- and medium-sized enterprises mostly benefit from the digital economy, but face limitations, such as limited funding and difficulties in risk diversification, which restrict their investments in green technology innovation [25,58,59]. Consequently, their promoting effect on the labor demand scale through this pathway is relatively weaker.

4. Discussion

This study not only empirically demonstrated that the improvement in the digitalization level of small- and medium-sized enterprises could increase the overall employment demand, consistent with the findings of Pan Y.R. et al. [45], but also conducted empirical analyses on the regression results at different quantiles. At present, some research remains at the theoretical analysis level without quantifying the impact of technological innovation on total employment. However, the empirical results in this paper show that green technology innovation plays a significant intermediary role between the digital economy and the scale of enterprise labor demand, with the main pathway being that the digital economy promotes green technology innovation within companies [22], and enterprise green technology innovation further increases the labor demand scale [34]. Song’s research found that progress in environmentally biased technologies can facilitate the supply and demand of regional labor [60], while Yang Xue et al. demonstrated that technological innovation did not result in significant substitution pressure to Japan’s overall labor employment [61], both of which aligned with the conclusions of this study. Furthermore, the research indicates that environmental protection has evolved into an industry capable of creating jobs [62], suggesting that policies promoting green technology innovation can, to some extent, increase the demand for enterprise labor. Moreover, the impact of the digital economy and technological innovation on total labor demand may lead to structural shocks [61], and thus, the digital economy may increase the demand for high-skilled labor through enhanced green technology innovation.

5. Conclusions and Policy Insights

Based on the background of “digital+”, we used the micropanel data of all A-share-listed companies in China from 2011 to 2020, and systematically discussed the impact of the improvement of the digital level of enterprises on the labor demand scale of enterprises and the role of firms’ green technology innovations in this process. It was found that the digital economy could expand the demand for a labor force; the amplification effect of the digital economy on labor demand was more significant at the lower and higher quantiles; green technology innovation played a significant mediating role; the digital economy had a more pronounced effect on labor demand expansion for companies located in the western region, those in industries with faster technological changes, and small- and medium-sized enterprises (SMEs), while green technology innovation had a more prominent mediating effect on companies in the eastern region, those in industries with slower technological changes, and large enterprises; among the segmented indicators of the digital economy, cloud computing technology, big data technology, and digital technology applications significantly expanded the scale of enterprise labor demand. The findings of this study to some extent provide insights into and support for addressing the impact of the digital economy on the scale of labor demand in enterprises and the role played by green technology innovation within it. Furthermore, they offer guidance and support for harnessing the new potential of the digital economy. In order to solve the people’s livelihood issues of employment and give full play to the boosting role of the digital economy, combined with the research conclusion, this paper put forward the following policy suggestions:

First, the foundation of digital construction should be improved and the pace of digital transformation should be accelerated. This paper concluded from the empirical analysis that the digital economy has a significant impact on employment demand scale, with small- and medium-sized enterprises being more significantly affected. Hence, it was very important to make full use of the new energy of the digital economy. On the one hand, the digital development level in some regions was low and the development among regions was unbalanced in China. Hence, it was necessary to improve the digital development level, strengthen digital construction, and increase investment in relevant industries. On the other hand, the digital transformation of enterprises served as an important way to improve the production efficiency and accelerate industrial structural upgrading. On the one hand, enterprises may promote the renewal and transformation of technological resource elements through internal investments, and the popularization and application of digital technology may explore the new business models of enterprises. The expansion of an enterprise’s business scope and the promotion of production and operation efficiency can promote the improvement of an enterprise’s digital level. Additionally, the government should issue relevant industrial support policies to promote the strong integration of the digital economy with technological innovation, industrial innovation, and real economy; promote the digital transformation of different types of enterprises; and ensure the full utilization of the dividend of digital economic development. Finally, the promotion of the enterprise digital development level and the exploration of innovation ability had a radiation driving effect. Enterprises can further expand the scale of the technology market and give full play to the effects of knowledge spillovers and industrial agglomeration through external innovation environment optimizations, knowledge and technology diffusions, and cooperative learning.

Secondly, the driving force of green technology innovation should be enhanced and the technological innovation of enterprises should be encouraged. It is essential to recognize that technology serves as a vital productivity driver, and the requirements of green development and technological innovation are inevitable in the new era. Therefore, expediting the advancement of green technology innovation and fully harnessing its mediating effects are of utmost importance. On the one hand, the development of green technology innovation necessitates enterprises to strengthen the relevant research and fully leverage the spillover effects of green technological innovation, thereby propelling the green development of related industries and technologies. On the other hand, the government must enhance policy support, not only by implementing corresponding subsidies and tax incentives, but also by striving to improve the market conditions at present and promote fair and transparent competitions in the market. Additionally, various sectors of society should fully utilize the development dividends provided by digital technology applications, utilizing shared platforms and transmission channels to promote collaborative and shared initiatives in green technological innovation. In doing so, the organic integration of green technology and the digital economy can be continuously advanced, fully leveraging the mediating and regulatory roles of green technology.

Third, the security of the employment system for workers needs to be guaranteed and diversified employment needs to be developed. The development of the digital economy has accelerated societal progress. It has not only enhanced the ability of enterprises to absorb a labor force, but has also given rise to a series of challenges, such as the need for improvement in employment systems and the diversification of employment methods. Therefore, it is essential for the government to improve employment system safeguards for workers, ensuring the full utilization of the digital economy. At the government level, it is necessary to constantly explore and improve the relevant enterprise recruitment regulations and employment social security systems, and to strengthen the family friendly policies in line with new employment patterns, such as telecommuting, flexible working, and job-sharing. At the level of the enterprise, it is necessary to establish a perfect standard of employment and recruitment as the standard of enterprise recruitment, to cooperate with the government to reward the introduction and settlement of relevant talents to a certain extent, to explore the balanced development of work and life, and to expand the scale of digital talent. At the individual level, it is important to continuously enhance and improve one’s own skills and abilities in order to remain competitive in the labor market. The development of the digital economy is undoubtedly the main direction of future social development at present. Through continuous improvements, the digital economy in all aspects must move in the directions of high efficiency, high quality, and high kinetic energy, and will continue to serve the interests of employment and people’s livelihoods by improving the employment system for workers.

Finally, it is important to review the main limitations of the study in order to provide recommendations for the future research. Firstly, this article only examined the mechanism of green technology innovation concerning the impact of the digital economy on demand scale, and the exploration of this mechanism was not sufficiently comprehensive. If the data are available, the research can expand to other multiple influencing pathways in the future. Secondly, this paper obtained the degree of digital transformation through comprehensive calculations based on the relevant literature research. However, there was a significant problem in that the word frequency statistics related to digital transformation obtained through the text analysis did not accurately reflect the extent of the digital transformation in an organization. The future research can further improve the measurement system of the degree of enterprise digital transformations. Thirdly, enterprise green technology innovation involves higher costs, which can lead many companies to opt-out of innovative activities. Consequently, the mediating effect of green technology innovation might be non-existent or limited in these firms. In the future research, we will conduct more in-depth analyses to explore this aspect. Fourthly, the study’s content may be susceptible to certain issues, such as temporal and spatial autocorrelations, as well as multi-level variables. In the future research, advanced econometric models and methods can be employed to address these related concerns more effectively.