2.1. Development of DE Level

The DE spans three major sectors, including individuals, enterprises, and other inter-connected levels, forming a broad and highly complex integrated economic system. The report of the 20th CPC National Congress emphasized accelerating the implementation of the “Digital China” strategy, expediting the development of the DE, and promoting the deep integration of the DE with the real economy. Following the principles of scientific rigor, systematic design, and operational feasibility, this study constructs a regional digital economic development level system based on four categories of indicators: digital infrastructure development, digital industrialization, industrial digitalization, and the innovation environment supporting digital economic development, as shown in Table 1. Compared with frameworks developed by institutions such as the China Academy of Information and Communications Technology and the Tencent Research Institute, the digital economic development measurement system proposed in this study offers distinct advantages. These advantages are grounded in the core principles of the DE, which ensure stability throughout the sample period. The entropy weight method is employed to assess the digital economic development level of each province.

This study employs the entropy weight method to determine indicator weights, enabling objective weight allocation for indicators and significantly reducing the influence of subjective judgment on indicator evaluation. Compared with alternative weight determination methods, the entropy weight method is not constrained by the volume or composition of the indicator data, making it applicable to all scenarios that require weight determination. Furthermore, this method fully extracts information embedded in raw data and can be combined with other weighting approaches to enhance evaluation accuracy and reliability. By analyzing time-series data from 31 provinces, municipalities, and autonomous regions in China over the period 2013–2020, we estimated DE development indicators of each province. The empirical data sources include the China Statistical Yearbook and China Labor Statistical Yearbook.

2.2. Model Construction and Indicator Selection

This study adopts a two-way fixed-effects model to comprehensively examine the macro-level impact of the DE on total employment, employment structure, and employment quality. This approach enhances the understanding of the influence of the DE on employment and broadens the research perspective. It provides a comprehensive analytical framework for understanding the employment-promoting effects of the DE and its role in shaping changes in the structure and quality of labor markets. Based on the theoretical analysis and research hypotheses, this study first evaluates the impact of the DE on total employment. Drawing on Cong and Yu ¹⁵, this study assumes that technological progress is a function of digital progress in accordance with the Cobb–Douglas production function, and uses the development of DE to represent digital technological progress. The baseline empirical model is specified as follows:

Based on this specification, the overall employment model is further defined as

In Eq. (1), \(i\) and \(t\) denote provinces and periods, respectively. \(u_t\) and \(\lambda_i\) represent the year and provincial fixed effects, respectively, and \(\varepsilon_{\textit{it}}\) denotes the residual term. \(\textit{EMP}_{\textit{it}}\) denotes total regional employment, \(\textit{DIG}_{\textit{it}}\) represents the level of regional digital economic development, and \(X_{\textit{it}}\) represents the set of control variables influencing employment. These controls comprise regional economic development levels ¹⁶, regional human capital, foreign direct investment, trade openness, regional infrastructure, and urbanization level. The data sources include the China Statistical Yearbook and the China Labor Statistical Yearbook.

This study further examines the impact of the DE on employment structure by measuring the proportions of high-, medium-, and low-skilled workers. Skill levels were classified according to educational attainment, as shown in Table 2. Although skills can be distinguished by occupational characteristics, Berman et al. ¹⁷ found no significant differences between the conclusions derived from the two classification methods. Given data availability, the present study uses educational attainment as a proxy for skill level to examine the impact of the DE on the distribution of workers across tiers with different skills. The following model was employed to investigate the effects of digital economic development on the employment structure:

In Eqs. (3)–(6), \(i\) and \(t\) denote provinces and time periods, respectively. \(u_t\) and \(\lambda_i\) represent the fixed effects for the year and province, respectively, and \(\varepsilon_{\textit{it}}\) denotes the residual term. \(\textit{LOW}_{\textit{it}}\), \(\textit{MID}_{\textit{it}}\), \(\textit{HIGH}1_{\textit{it}}\), and \(\textit{HIGH}2_{\textit{it}}\) denote the proportions of high-, medium-, and low-skilled workers, respectively. \(\textit{DIG}_{\textit{it}}\) represents regional digital economic development level, and \(X_{\textit{it}}\) denotes a set of control variables.

This study incorporates multiple control variables influencing employment in the model construction:

1. (1)

   Regional economic level (ECO). The regional marketization index was selected to measure regional economic development using data sourced from the China Marketization Index Database.

2. (2)

   Trade openness (OPEN). The ratio of total imports and exports to GDP was used to analyze the impact of international technology and labor flows on employment changes. Data were sourced from the China Trade and Foreign Economic Statistics Yearbook and the China Statistical Yearbook.

3. (3)

   Urbanization level (UAR). The proportion of the urban population was used to measure the urbanization levels. Data were sourced from the China Statistical Yearbook.

4. (4)

   Human capital level (HC). The average years of education per capita at each province were used to measure regional human capital using data sourced from the China Education Statistical Yearbook and the China Statistical Yearbook.

5. (5)

   Infrastructure development (INF). Infrastructure development was measured using the logarithm of per capita postal and telecommunications service volume in each province, calculated as the logarithm of the ratio of total service volume to year-end resident population. Data were sourced from the National Bureau of Statistics.

6. (6)

   Foreign direct investment (INVEST). Foreign direct investment is measured as the proportion of FDI to regional GDP using data sourced from the National Bureau of Statistics.

Descriptive statistical analysis was conducted on all indicators used in this empirical study, and the results are presented in Table 3.

2.3. Model Estimation and Robustness Testing

Model (2) evaluates the impact of the DE on total employment. Table 4 presents the model estimation results.

The results indicate that the development level of the DE exerts a significant positive promotional effect on the overall employment scale of China. The job creation effect generated by digital economic development outweighs its direct substitution effect, resulting in an overall positive promotional effect. Consequently, DE increases employment opportunities and elevates the overall employment levels within the economy and society.

The results presented in Table 5 indicate that the development of the DE exerts a positive influence on the proportion of both high- and low-skilled labor, while producing a negative effect on the share of medium-skilled workers. Specifically, a 0.1 increase in the DE level increases the proportion of workers with postgraduate degrees, undergraduate and associate degrees, and primary education or below by 0.329%, 0.173%, and 0.174%, respectively. Conversely, the shares of workers with junior high and senior high school graduates decrease by 0.371%. Notably, the development of the consumer internet has spawned numerous new employment forms, creating additional job opportunities for low-skilled workers. Overall, the development of the DE significantly increases demand for highly skilled labor while reducing reliance on medium-to-low skilled workers, driving the employment structure of China toward a shift from medium-to-low to high-skill roles. This trend indicates that although the DE promotes employment structure optimization, it also generates an “employment polarization” effect on the skill level structure of the labor market, exacerbating the polarization of skill demand.

Robustness tests were conducted to evaluate the reliability of research findings (Table 6). These tests primarily examine the impact of changing conditions, such as indicator selection and model evaluation methods, on the research conclusions. This study aims to verify the explanatory power of the selected indicators and rationality of the evaluation methods to ensure the stability and consistency of the conclusions under different conditions. The consistency between modified and baseline results indicate the scientific validity and reliability of the research findings. To verify the robustness of the benchmark regression, we substituted explanatory variables for testing.

This study employs the entropy weight method to calculate the DE development index. This evaluation approach may introduce variable measurement errors, potentially imparting a degree of randomness to the research findings. To demonstrate the robustness of the conclusions, we draw upon Zhao et al.’s ¹⁸ use of principal component analysis (PCA) to recalculate the DE development level as a substitute for the core explanatory variables. Based on the eigenvalues and variance contribution rates of each principal component, the composite function is derived as follows:

In Eq. (7), \(F\) represents the composite score, whereas \(F_1\) and \(F_2\) denote the scores of the first and second principal components, respectively. Based on the composite function, panel data for the DE development index (DIG_PCA) are constructed.

Substitution of the variables into the aggregate employment model yielded magnitude and sign of the regression coefficients consistent with the benchmark regression results, indicating the robustness of the benchmark regression. The positive “creation effect” on employment offset the “direct destruction” effect, producing a net positive impact on employment. After variable substitution in the employment structure model, the coefficients and signs of the core explanatory variables remained consistent with the preliminary regression results, validating the robustness of the benchmark regression model. The results indicate that the development of DE significantly increases the demand for highly skilled labor while reducing reliance on medium-to-low-skilled labor. This drives the employment structure of China toward a shift from medium-to-low- to high-skilled labor, simultaneously confirming the existence of an “employment polarization” effect.

3.1. Industrial Structure as a Mechanism Variable

The upgrading and rationalization of the industrial structure represents the key pathway through which digital economic development impacts overall employment, reflecting an industrial structure effect. Upgrading industrial structures captures improvements in the quality and efficiency of the entire industrial system. This process involves not only the adjustment of the proportion of primary, secondary, and tertiary industries in the national economy but also enhancements in inter-industrial collaboration and coordination. To comprehensively capture the essence of industrial structure upgrading, this study adopts two indicators following the approach of Cai and Xu ¹⁹. The first indicator is upgrading the industrial structure. The rapid development of DE has stimulated the emergence of new digital industries, including high-tech sectors, live streaming, and e-commerce. These developments have not promoted industrial transformation and simultaneously influenced the evolution of the labor employment structure. These emerging industries have introduced new job opportunities into the labor market, promoting diversification and upskilling in employment structures. However, traditional industries undergoing digital transformation are gradually advancing toward higher levels of development and quality. This shift drives labor demand toward skill restructuring, reducing the need for low-to-medium skill positions and increasing the reliance on highly skilled labor. This dual dynamic highlights the pivotal role of the DE in advancing industrial upgrading and improving employment quality, while underscoring the importance of enhancing labor skills and adaptability in the digital era. This study constructs an industrial structure upgrading index (OIS) following the methodology of Fu ²⁰ to reflect the degree of industrial structure optimization. The second indicator is industrial structure rationalization. The rationalization of industrial structure refers to the process of improving the overall economic efficiency and competitiveness via complementary and coordinated development across industries, achieved by adjusting and optimizing of their composition and proportion during a specific period. This process not only involves the rational allocation of proportions among primary, secondary, and tertiary industries but also includes structural adjustments within each industry across sub-sectors, in conjunction with the optimization of technological structures, product portfolios, organizational frameworks, and other sectors. Following the methodology of Gan and Zheng ²¹, this study constructs an industrial structure rationalization index (TL) to measure the quality of industrial clustering. This indicator is measured using the Tier index.

Based on the above analytical framework and drawing on the research methodology of Jiang ²², the following model is established to further examine the mechanism through which digital economic development influences employment in China via industrial structure transformation:

Here, \(\textit{OIS}_{it}\) and \(\textit{TL}_{it}\) represent the OIS and TL respectively; \(\textit{DIG}_{it}\) denotes the regional level of digital economic development; and \(X_{it}\) denotes a set of control variables.

The instrumental variables for the industrial structure effect comprise the OIS and TL. The OIS precisely maps the impact of digital economic development on employment structure upgrading by emphasizing the secondary and tertiary industries, demonstrating its use as a theoretically grounded and empirically valid measure of industrial structure upgrading. The OIS is calculated using the cosine method ²⁰, using the following steps:

1. Step 1:

   Construct a set of three-dimensional \(X_0=(x_{1,0},x_{2,0},x_{3,0})\) vectors, where \(x_{1,0}\), \(x_{2,0}\), and \(x_{3,0}\) represent the respective shares of primary, secondary, and tertiary industry value added in GDP.

2. Step 2:

   Calculate the angles \(\theta_1\), \(\theta_2\), \(\theta_3\), and between each vector \(X_0\) and the unit vectors \(X_1=(1,0,0)\), \(X_2=(0,1,0)\), and \(X_3=(0,0,1)\), respectively, using the following equations:

   \begin{equation} \label{eq:10} \theta_j=\arccos\left(\dfrac{x_0 x_j}{\left\Vert x_0\right\Vert\times \left\Vert x_j\right\Vert}\right), \quad j=1,2,3. \tag{10} \end{equation}
3. Step 3:

   Calculate the industrial upgrading index using the following expression:

   \begin{equation} \label{eq:11} \textit{OLS}=\sum_{k=1}^3 \sum_{j=1}^k \theta_j. \tag{11} \end{equation}

A higher industrial upgrading index value indicates a higher level of industrial upgrading and higher-quality industrial structure transformation.

Additionally, we selects the TL as the mechanism variable. From an inter-industry perspective, it reflects the overall optimization and quality of the industrial structure, facilitating a deeper understanding of the rationalization degree of the industrial structure and its underlying mechanism of influencing economic development. This indicator was measured using the Theil index, calculated as follows:

The index \(Y_i\) represents the industrial output value of industry \(i\), \(L_i\) represents the employment in sector \(i\), and \(n\) represents the number of industrial sectors. The values of the inverse indicator TZ closer to 0 indicate a more rational industrial structure, whereas higher TZ values indicate less rational industrial structure.

3.2. Human Capital as a Mechanistic Variable

Human capital constitutes a key channel through which the DE influences employment quality. The DE promotes the accumulation of regional human capital across multiple sectors, thereby expanding the employment scale and improving job quality. The development of digital communication technologies, big data applications, and digital platforms has significantly accelerated the dissemination of knowledge and information, broadened access to advanced technologies, and positively impacted the job market. The core components of human capital include education, vocational training, and healthcare. Driven by the DE, these domains have achieved substantial development at the regional level. Specifically, by enhancing educational quality, increasing investment in skills training, and optimizing health conditions, the DE has effectively enhanced labor supply, thereby providing robust support for regional employment growth. This process not only highlights the positive impact of the DE on human capital accumulation but also underscores its crucial role in driving regional employment expansion. Furthermore, the widespread application of new technologies provides new solutions to the challenges introduced by the decline of the demographic dividend, thereby serving as a new driver for economic transformation and upgrading. These findings reveal the profound impact of digital technologies on education and employment, providing crucial insights into talent development and employment policies in the DE era. Therefore, this study employs human capital dividends (HCD) to measure the human capital structure in China and examines the mechanism through which digital economic development influences employment via human capital. The following model is constructed for empirical analysis:

Drawing on the methodology for constructing human capital dividend variables by Yu and Cong ²³, and integrating the classification of employment skill structures discussed earlier, this study adopts the ratio of highly skilled to low-skilled labor as a representative variable for human capital dividends. This ratio reflects the relative quantitative relationship between the labor forces at different skill levels and is calculated as follows:

3.3. Mechanism Regression Results Analyze

Empirical tests were conducted to examine whether the development of the DE influences employment fluctuations through a rationalized industrial structure, high-level industrial layout, and human capital dividends. Estimation results are presented in Table 7.

The results presented in Table 7 show that DE development of the DE exerted a significant positive effect on industrial structure upgraded. As DE advanced, the level of industrial structure upgradation was substantially enhanced. Specifically, for each one-unit increase in the DE development level, the industrial structure upgrading level increased by 0.209 units, demonstrating that DE development effectively promoted the optimization and transformation of the industrial structure from the primary sector to the secondary and tertiary sectors. Moreover, the expansion of DE broadens the scope and scale of the tertiary sector in addition to driving greater service orientation within the industrial structure, thereby providing robust support for advancing industrial upgrading to higher levels. Therefore, when using the OIS as a measure of industrial quality and efficiency improvement, the development of the DE significantly propelled the industrial restructuring in China. The level of digital economic development exerts a significant negative effect on industrial structure rationalization. The industrial structure rationalization index is an inverse indicator, where a lower index value (closer to 0) indicates enhanced inter-industrial coordination and a more rational industrial structure, whereas a higher index value indicates a less rational one, which suggests that the development of the DE has a significantly positive impact on improving the rationality of the industrial structure. Specifically, a one-unit increase in the DE development index decreases the TL by 0.492 units, indicating that digital economic advancement helps optimize and enhance the quality of coordination and integration among industries.

Digital economic development exerts a significantly positive effect on human capital dividends. As a positive indicator, an increase in the human capital dividend index reflects a trend toward the rationalization of the human capital structure, revealing that DE maturity plays a crucial role in optimizing human resource allocation and enhancing human capital quality. Therefore, DE promotes the accumulation of human capital via multiple channels. Drawing on the methodology of Jiang ²², it is evident that digital economic development influences both the quantity and structure of employment through the upgrading and rationalization of the industrial structure as well as the human capital dividend.