Study on the Influence of Low-Carbon Economy on Employment Skill Structure—Evidence from 30 Provincial Regions in China

Abstract

In confronting escalating economic uncertainty, achieving a win–win situation for low-carbon transition and improved employment structure will contribute to economic recovery and sustainable growth but also contribute to building a community with a shared future for mankind. A critical issue for China’s economy and societal welfare, as well as a core component of sustainable development, concerns whether low-carbon economic transition influences employment skill structure. This study utilizes data from 30 provinces (municipalities and autonomous regions) in China from 2006 to 2021. Employing the entropy method, a low-carbon economic development level indicator system was constructed from four aspects: low-carbon output, low-carbon consumption, low-carbon resources, and low-carbon environment to measure the low-carbon economy and explore its direct and indirect effects on employment skill structure and spatial effects. The research findings indicate that low-carbon economies not only directly and significantly promote employment skill structure optimization but also indirectly generate promotional effects through pathways such as industrial structure adjustment, green innovation’s innovative effects, and factor substitution effects of increased pollution control investment. Among these, the indirect impact of industrial structure adjustment contributes most substantially. Low-carbon economies’ influence on employment skill structures exhibits spatial spillover effects, with neighboring regions’ low-carbon economies exerting positive spillover effects on local skill structures. Additionally, significant negative interdependence exists among regional employment skill structures. Based on the aforementioned research conclusions, the following recommendations are proposed: accelerate low-carbon economy development and employment skill structure enhancement in central and western regions to diminish regional disparities; encourage green innovation and promote traditional industry upgrading and transformation; formulate regional coordinated development plans, thereby strengthening the low-carbon economy’s optimizing role upon employment skills structure; and increase educational investment and strengthen labor skill training.

1. Introduction

One of the world’s most pressing environmental issues is global warming, precipitated primarily by high greenhouse gas (GHG) emissions—a critical threat to global sustainable development, as delineated in the United Nations Sustainable Development Goals (SDGs), which prioritize climate action (SDG 13) and responsible consumption and production (SDG 12). As the world’s largest energy consumer, China accounts for a substantial proportion of global carbon emissions. In response to accelerating climate change, the Chinese government has established ambitious dual-carbon targets of carbon peaking and carbon neutrality—a pivotal step in aligning national development strategies with the global sustainable development agenda. The 20th Party Congress report underscored that “promoting low-carbon economic and social development is essential for achieving high-quality growth,” underscoring the strategic importance of this transition as a cornerstone of sustainable development that integrates environmental protection and economic vitality. This transformation has catalyzed the emergence of renewable energy sectors and eco-technological services, fundamentally restructuring employment patterns [1]. Given that employment constitutes the primary income source for most citizens, addressing structural labor market imbalances remains vital for economic stability. In China, the central challenge lies in skills mismatch, which threatens to impede both the low-carbon transition and broader sustainable development advancement by generating inefficiencies in labor allocation.

The conception of a low-carbon economy was initially articulated in the UK’s 2003 energy white paper and has subsequently endeavored to decouple economic growth from emissions [2]. Empirical evidence on the employment consequences of low-carbon transitions has remained contested. A subset of academics perceive the low-carbon transition as amplifying employment opportunities. They have contended that such transitions yield net employment gains, whilst other studies have revealed that innovation incentives in low-carbon economies have fostered substantial job creation [3,4]. In China, analyses of pilot low-carbon city policies’ employment effects show these policies elevated regional employment levels relative to non-pilot areas, with heterogeneous impacts—state-owned enterprises, the eastern region, and the secondary industry experienced more pronounced effects [5,6,7,8]. Parallel findings indicate that industrial agglomeration and innovation, proxied by carbon productivity, have propelled employment expansion [9,10]. Research on the CET policy notes it augments firm-level employment quantity by expanding output and substituting production factors, though it diminishes employment quality through declining employee benefits and a widening wage gap [11]. For the Netherlands, renewable energy transitions are projected to benefit the economy and employment, primarily by boosting jobs in construction and service sectors [12]. Projections suggest electrolysis-based transformation of the hydrogen sector could generate approximately 40,000 jobs by 2050 [13], while direct energy jobs across electricity, heating, transportation, and desalination may surge from 57 million in 2020 to nearly 134 million by 2050 [14]; in the Middle East, renewable energy generation could create 180,000 direct jobs, with 36% in installation, 15% in (local) manufacturing, and 49% in O&M activities [15].

Conversely, another body of work emphasizes detrimental employment impacts from low-carbon transitions. It posits that carbon taxation imposes societal costs and triggers job losses, particularly in fossil energy sectors and areas distant from major employment centers [16,17,18]. Similarly, energy efficiency and emission reduction targets may hinder employment expansion by lowering company pay and raising business tax burdens [19].

Regarding low-carbon policies’ influence on employment skill structures, provincial panel data from China reveals emissions regulation affects employment through two channels: an “innovation offset effect” that promotes both high- and low-skilled employment, and a “compliance cost effect” that reduces employment across skill levels, with high-skilled labor showing greater sensitivity to such policies. City-level panel data in China further indicate carbon emission trading policies boost employment via structural transformation, with uneven impacts: policies in eastern and large cities exert stronger effects than those in central/western and small-to-medium cities [20]. For the EU energy sector, low-carbon transitions are projected to triple demand for highly and moderately trained workers by 2030 compared to 2015 levels [21]. General equilibrium model simulations of climate policies show expanded output in construction (investment) and agriculture (biomass) sectors, increasing low-skilled employment [22,23,24]. Conversely, workers with lower educational attainment are found to be more vulnerable to climate policies, facing elevated attrition rates [23].

Existing research has examined the relationship between low-carbon transition and employment from multiple dimensions, which can be categorized into two aspects: research content and research methods. Regarding research content, current findings primarily focus on the low-carbon transition’s impact on employment numbers, yielding two opposing viewpoints: ‘increased employment’ and ‘decreased employment.’ It merits noting that existing research shares common characteristics in measuring low-carbon transition, frequently utilizing carbon dioxide emissions as a quantitative indicator or employing national-level low-carbon emission restriction policies as proxy variables. Concerning research methods, the difference-in-differences method and general equilibrium models constitute the most widely employed analytical tools, with the former commonly utilized to assess policy interventions’ net effects and the latter suitable for simulating macroeconomic responses under multi-market interactions.

Despite the substantial accumulation of research, three notable limitations persist: first, regarding low-carbon economy measurement dimensions, existing studies tend towards oversimplification. Most literature either treats environmental regulations as the core explanatory variable to comprehensively examine their employment impact or uses single indicators such as low-carbon pilot policy implementation or carbon dioxide emissions to represent regional low-carbon economic development levels, failing to fully reflect a low-carbon economy’s multidimensional nature. Second, concerning impact mechanisms’ analytical depth, existing studies primarily focus on indirect pathways, such as analyzing the transmission mechanisms of low-carbon transition on employment structure through intermediary variables like output effects, cost effects, and innovation effects. However, systematic exploration of the direct causal relationship between low-carbon economy and employment skill structure remains lacking, leading to incomplete understanding of the underlying logic behind differing impacts on employment across various skill levels. Third, regarding spatial perspectives’ incorporation extent, existing research exhibits significant gaps. Low-carbon economy development possesses typical spatial correlations, yet its spatial spillover effects on employment skill structure have not been sufficiently addressed, with relevant empirical tests being particularly scarce.

Based on the aforementioned research gaps, this paper’s marginal contributions are primarily reflected in the following three aspects: (1) Theoretical framework integration and expansion: Building upon existing theories’ systematic review, a more hierarchical analytical framework has been constructed, specifically including deriving the low-carbon economy’s direct impact effects on employment skills structure from a variable factor substitution elasticity production function; identifying indirect transmission pathways through intermediary variables such as industrial structure adjustment, green innovation, and pollution control investment and, for the first time, incorporating a spatial dimension to explicitly propose the hypothesis of the low-carbon economy’s spatial spillover effects on employment skill structure. (2) Core variable measurement optimization: Addressing the single-dimensional low-carbon transition indicators’ issue in existing research, this paper re-evaluates their impact on employment skill structure based on an improved low-carbon economy indicator system, enhancing variable measurement’s comprehensiveness and accuracy. (3) Targeted measurement method improvements: Given that both the low-carbon economy and employment skills structure exhibit significant spatial externalities, this study introduces spatial econometric models to quantitatively analyze spatial spillover effects, thereby more comprehensively revealing the complex relationship between the two and addressing existing research’s shortcomings in spatial dimension analysis.

Based on the aforementioned analysis, this study endeavors to address the following three key questions: (1) Does the low-carbon economy exert a significant impact upon employment skills structure? (2) What is the extent of the low-carbon economy’s impact upon employment skills structure? What are the characteristics of its impact pathways (direct and indirect effects)? (3) Does the low-carbon economy’s impact upon employment skills structure exhibit spatial spillover effects? If so, what are the directions and intensities of these spillover effects?

The remaining sections of this paper are organized as follows. Section 2 discusses the theoretical mechanism of low carbon economy affecting employment skill structure and advances research hypotheses. Section 3 details the model, variables, and data sources. Section 4 provides the direct, indirect and spatial impacts of low carbon economy on employment skill structure. Section 5 summarizes the key findings and presents relevant policy recommendations. Finally, Section 6 presents the limitations and future research directions of this paper.

2. Theoretical Analysis and Hypotheses

2.1. The Direct Impact of the Low-Carbon Economy on the Employment Skill Structure

Production function model of variable factor elasticity of substitution was employed. This model could effectively capture the time-varying characteristics of substitution elasticity and harmonize well with real-world economic conditions. It was employed to analyze the theoretical mechanism underlying the direct effect of the low-carbon economy on the employment skill structure. Given that the production process necessitated diverse labor types, and in accordance with prior studies and this paper’s research focus, the workforce was categorized into two groups based on educational attainment. The first group, L₁, encompassed highly skilled workers with a bachelor’s degree or higher, while the second group, L₂, comprised medium- and low-skilled workers with less than a bachelor’s degree. Based on these categorizations, the following production function was formulated:

$$\mathrm{Y}=\mathrm{A}{\left[{\mathsf{\delta }}_1{\mathrm{K}}^{-\mathsf{\rho }}+{\mathsf{\delta }}_2{\mathrm{L}}_1^{-\mathsf{\rho }}+{\mathsf{\delta }}_3{\mathrm{L}}_2^{-\mathsf{\rho }}\right]}^{-{\displaystyle \frac{\mathrm{m}}{\mathsf{\rho }}}}$$

(1)

In this paper, Y denotes output volume, with the unit being CNY. Technological advancement is represented by A, substitution elasticity by ρ, and production scale compensation by m. K represents capital, with the unit being CNY. L₁ and L₂ constitute the high-skilled labor force and the low-skilled labor force, respectively, with the unit being the number of people. ${\mathsf{\delta }}_{\mathrm{i}}$ represent the intensity of the three factors, ${\mathsf{\delta }}_1+{\mathsf{\delta }}_2+{\mathsf{\delta }}_3=1$. Given China’s development status, the number of high-skilled workers was substantially lower than that of medium–low-skilled workers, so ${\mathsf{\delta }}_2<{\mathsf{\delta }}_3$. σ was introduced as the elasticity of substitution between variable factors and constituted a function of the low-carbon economy (lcel). This was expressed as σ (lcel) = 1/(1 − ρ), with this relationship being available at market equilibrium.

$${\displaystyle \frac{{\mathrm{M}\mathrm{P}}_{\mathrm{L}1}}{{\mathrm{M}\mathrm{P}}_{\mathrm{L}2}}}={\displaystyle \frac{{\mathrm{W}}_1}{{\mathrm{W}}_2}}$$

(2)

$${\mathrm{M}\mathrm{P}}_{\mathrm{L}1}={\displaystyle \frac{\mathrm{d}\mathrm{Y}}{{\mathrm{d}}_{\mathrm{L}1}}}=\mathrm{A}\mathrm{m}{\left[{\mathsf{\delta }}_1{\mathrm{K}}^{-\mathsf{\rho }}+{\mathsf{\delta }}_2{\mathrm{L}}_1^{-\mathsf{\rho }}+{\mathsf{\delta }}_3{\mathrm{L}}_2^{-\mathsf{\rho }}\right]}^{-{\displaystyle \frac{\mathrm{m}}{\mathsf{\rho }}}-1}{\mathsf{\delta }}_2{\mathrm{L}}_1^{-\mathsf{\rho }-1}$$

(3)

$${\mathrm{M}\mathrm{P}}_{\mathrm{L}2}={\displaystyle \frac{\mathrm{d}\mathrm{Y}}{{\mathrm{d}}_{\mathrm{L}2}}}=\mathrm{A}\mathrm{m}{\left[{\mathsf{\delta }}_1{\mathrm{K}}^{-\mathsf{\rho }}+{\mathsf{\delta }}_2{\mathrm{L}}_1^{-\mathsf{\rho }}+{\mathsf{\delta }}_3{\mathrm{L}}_2^{-\mathsf{\rho }}\right]}^{-{\displaystyle \frac{\mathrm{m}}{\mathsf{\rho }}}-1}{\mathsf{\delta }}_3{\mathrm{L}}_2^{-\mathsf{\rho }-1}$$

(4)

The above formula was derived based on relationships. In the formula, W₁ and W₂ represent the compensation of high-skilled and low–middle-skilled labor, respectively. The units of W₁ and W₂ are CNY.

$${\displaystyle \frac{{\mathrm{M}\mathrm{P}}_{\mathrm{L}1}}{{\mathrm{M}\mathrm{P}}_{\mathrm{L}2}}}={\displaystyle \frac{{\mathrm{W}}_1}{{\mathrm{W}}_2}}={\displaystyle \frac{{\mathsf{\delta }}_2{\mathrm{L}}_1^{-\mathsf{\rho }-1}}{{\mathsf{\delta }}_3{\mathrm{L}}_2^{-\mathsf{\rho }-1}}}$$

(5)

Additional rollout:

$${\left({\displaystyle \frac{{\mathrm{L}}_1}{{\mathrm{L}}_2}}\right)}^{-\mathsf{\rho }-1}={\displaystyle \frac{{\mathrm{W}}_1{\mathsf{\delta }}_3}{{\mathrm{W}}_2{\mathsf{\delta }}_2}}$$

(6)

$${\displaystyle \frac{{\mathrm{L}}_1}{{\mathrm{L}}_2}}={\left({\displaystyle \frac{{\mathrm{W}}_1{\mathsf{\delta }}_3}{{\mathrm{W}}_2{\mathsf{\delta }}_2}}\right)}^{\mathsf{\rho }+1}={\left({\displaystyle \frac{{\mathrm{W}}_1{\mathsf{\delta }}_3}{{\mathrm{W}}_2{\mathsf{\delta }}_2}}\right)}^{2-{\displaystyle \frac{1}{\mathsf{\sigma }}}}$$

(7)

In the above equation, ${\displaystyle \frac{{\mathbf{L}}_1}{{\mathbf{L}}_2}}$ denotes the employment skill structure, whilst σ(lcel) constitutes a function of the low-carbon economy. This function was derived from both sides of Equation (7).

$${\displaystyle \frac{\mathrm{d}\frac{{\mathrm{L}}_1}{{\mathrm{L}}_2}}{\mathrm{d}\mathrm{l}\mathrm{c}\mathrm{e}\mathrm{l}}}={\displaystyle \frac{1}{{\mathsf{\sigma }}^2}}{\left({\displaystyle \frac{{\mathrm{W}}_1{\mathsf{\delta }}_3}{{\mathrm{W}}_2{\mathsf{\delta }}_2}}\right)}^{2-{\displaystyle \frac{1}{\mathsf{\sigma }}}}\ln \left(2-{\displaystyle \frac{1}{\mathsf{\sigma }}}\right){\mathsf{\sigma }}^{\prime }(\mathrm{l}\mathrm{c}\mathrm{e}\mathrm{l})$$

(8)

The sign direction of Equation (8) was determined by ln(2 − 1/σ) and σ′(lcel). Given that σ denotes the elasticity of substitution between high-skilled and low-skilled labor and that high-skilled labor typically yields higher output, σ exceeded 1 and proved positive. On the other hand, Porter’s theory posited that low-carbon economic growth stimulates technological innovation. This had increased demand for highly trained workers while also facilitating the replacement of medium- and low-skilled workers, leading to σ′(lcel) > 0. Low-carbon economic development enhances the elasticity of substitution between high-skilled labor and medium-and low-skilled labor. This consequently encourages employment skill structure upgrading. According to Equation (8)’s final result, which could be obtained when the aforementioned analysis was considered, a positive value was derived.

Therefore, Hypothesis 1 is advanced:

H1. 

The low-carbon economy has a direct optimization effect on employment skill structure.

2.2. Indirect Effects of the Low-Carbon Economy on the Employment Skill Structure

High-polluting and high-energy-consuming industries have been progressively supplanted by low-polluting and low-energy-consuming industries, thereby optimizing industrial structure. This transition proves indispensable for propelling the low-carbon economy, which necessitates industries to evolve toward greater environmental friendliness and energy efficiency. The Matey–Clark theorem posits that employment and industrial structures exhibit a dynamic, co-directional correlation. As industrial structures undergo optimization, labor market demands shift: low-pollution, low-energy-consumption industries typically require elevated levels of technical and managerial expertise, thereby driving heightened demand for high-skilled labor.

As the low-carbon economy progresses, national carbon emission constraints have been steadily strengthened. Porter’s hypothesis posits that firms demonstrate propensity to pursue expanded green innovation activities when environmental regulations are reasonably designed. This process influences the employment skill structure through two mechanisms. First, green innovation necessitates a considerable workforce of high-skilled personnel engaged in core technology R&D, amplifying demand for high-skilled labor. Second, the R&D and implementation of new technologies and processes may impose a substitution effect on medium- and low-skilled labor, thereby diminishing demand for low-skilled workers.

Concurrently, low-carbon economic expansion has propelled increased investment in environmental pollution abatement [25], imposing pressures on firms’ regular production and operational costs. According to the factor substitution effect, as expenditures on environmental management escalate, high-carbon and high-polluting enterprises are compelled to accelerate their adoption of low-carbon technologies to reduce dependence on high-emission production methods. Such technological advancements and factor substitutions tend to elevate employment skill structure: they may decrease demand for low-skilled labor while augmenting demand for highly skilled labor.

The following hypotheses are proposed:

H2. 

Low-carbon economy optimizes employment skill structure through industrial restructuring.

H3. 

Low-carbon economy optimizes employment skill structure through the innovation effect of green innovation.

H4. 

Low-carbon economy optimizes employment skill structure through factor substitution effect by increasing investment in pollution control.

2.3. Spatial Spillover Effect of Low Carbon Economy on Employment Skill Structure

According to the theory of “geographic dual economy” and Muldaur and Hirschman’s perspective of polarization and diffusion effects, economic development does not manifest a balanced progression trend but constitutes a non-equilibrium process replete with dynamic transformations. In this process, the polarization effect and diffusion effect operate through economic agents’ profit maximization mechanisms. Polarization effect manifests during economic development, whereby resources, factors and economic activities concentrate in certain advantaged regions. These regions, by virtue of superior infrastructure, abundant human resources, convenient market access and other conditions, attract capital, labor and other factors’ continuous inflow, consequently achieving rapid development and establishing disparities with other regions. The diffusion effect occurs as the core region develops to a certain stage, whereby due to factors such as rising factor costs, certain economic activities, technology, capital, etc., begin diffusing to the surrounding areas, precipitating neighboring regions’ development of. Through this mechanism of convergence, the polarization effect and diffusion effect jointly shape regional economic development patterns.

In the context of low-carbon economic development, this imbalanced development manifests in its impact upon employment skills structure and generates spatial spillover effects. From the polarization effect perspective, regions with elevated low-carbon economic development levels often prove more capable of attracting resource elements from surrounding areas by virtue of their existing industrial foundations, technological innovation capabilities and policy advantages. These regions attract greater capital investment, high-end talent and related industry concentration due to their first-mover advantages in low-carbon technology research and development, clean energy utilization and low-carbon industry cultivation. For instance, through new energy automobile industry development, certain economically developed regions have attracted substantial numbers of battery technology research and development specialists, engineers and related supporting enterprises, further consolidating their leading position in the low-carbon economy, whilst neighboring regions may experience development constraints due to resource factor outflows, rendering employment skill structure improvement challenging. The diffusion effect materializes as low-carbon economic development precipitates related industry transfer and diffusion amongst regions. With low-carbon technology maturation and cost reduction, certain high-energy-consuming and high-polluting industries gradually transfer to less developed regions, simultaneously driving employment opportunity and skill requirement transfers. For example, to satisfy environmental protection requirements, some traditional manufacturing enterprises adopt low-carbon production technology for upgrading and transformation, relocating portions of production processes to areas with lower labor costs and relatively greater environmental capacity. This has precipitated an increase in skill positions related to low-carbon manufacturing within less developed regions’ employment structures, promoting local employment skill structure adjustment and optimization.

The following hypothesis is proposed.

H5. 

There exists a spatial spillover effect of the low-carbon economy’s impact upon employment skill structure.

3. Materials and Methods

3.1. Model

3.1.1. Direct Effect Model

To test Hypothesis 1, it proves essential to account for inter-provincial heterogeneity inherent in each Chinese province’s specific context, alongside potential temporal effects. Accordingly, this study employs a two-way fixed-effects panel data model for regression analysis, aiming to mitigate these concerns. This effectively reduces omitted-variable bias and enhances estimation accuracy, enabling verification of the low-carbon economy’s direct impact upon employment skill structure. The model was formulated as follows:

$$\mathrm{s}\mathrm{s}\mathrm{o}\mathrm{e}={\mathsf{\alpha }}_0+\mathsf{\gamma }{\mathrm{l}\mathrm{c}\mathrm{e}\mathrm{l}}_{\mathrm{i},\mathrm{t}}+\sum {\mathsf{\alpha }}_{\mathrm{j}}{\mathrm{X}}_{\mathrm{i},\mathrm{t}}+{\mathrm{u}}_{\mathrm{i}}+{\mathsf{\lambda }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i},\mathrm{t}}$$

(9)

In the model, lcel denotes the low-carbon economy, and ssoe represents the employment skill structure. X stands for the control variables. ${\mathsf{\alpha }}_0$ is a constant term, and ${\mathsf{\alpha }}_{\mathrm{j}}$ represents the coefficients of the control variables. γ is used to denote the coefficients of the core explanatory variables. ${\mathrm{u}}_{\mathrm{i}}$ represents a province-fixed effect, while ${\mathsf{\lambda }}_{\mathrm{t}}$ indicates a time-fixed effect. ${\mathsf{\epsilon }}_{\mathrm{i},\mathrm{t}}$ is a random error term, with subscripts i and t denoting the province and year, respectively.

3.1.2. Indirect Effect Model

To test transmission channels’ existence, an intermediary effect model was constructed for empirical analysis. A commonly used method for intermediary effect analysis is the stepwise regression method [26]. This method is simple to operate but has encountered significant criticism in recent years. Some scholar have used structural equation models to analyze the mediating effect [27].

Structural equation models are generally categorized into two major classifications. The first category operates upon variables’ covariance matrix to analyze inter-variable relationships, termed Covariance-based Structural Equation Modeling (CB-SEM). This approach suits theoretical model validation, with the objective of achieving the closest possible correspondence between the sample matrix and the model’s expected covariance. The alternative category constitutes Partial Least Squares–SEM (PLS-SEM), which predominantly focuses upon variance explanation and causal relationship significance assessment. It suits theoretical model exploration, with the goal of maximizing endogenous variables’ explanatory power.

Given that this study endeavors to validate a prior theoretical model of how a low-carbon economy influences employment skills structure through green innovation, industrial structure, and pollution control investments, and that the provincial panel data utilized are sufficient in quantity whilst the core variables are approximately normal, CB-SEM is employed for hypothesis testing. The maximum likelihood estimation of CB-SEM provides unbiased parameters within a reflective measurement framework and accurately assesses model specification appropriateness through global fit indices such as RMSEA and CFI; its nested model χ² difference test can also precisely compare the advantages and disadvantages of chain mediation and partial mediation. In contrast, PLS-SEM proves more suitable for exploratory research or small-sample scenarios dominated by formative indicators. Since this study is theoretically mature and possesses sufficient samples, CB-SEM constitutes the preferred choice.

Structural equation modeling not only estimates the model’s overall fit but also effectively identifies mediating paths. Therefore, a multiple mediation effect model based on structural equation modeling was constructed to analyze the transmission mechanism of the low-carbon economy upon employment skills structure. The construction steps are as follows:

Step 1: Model construction: Based on theoretical analysis, determine the dependent variable, mediating variable, and explanatory variable, while establishing the basic model for the mediating effect.

Step 2: Model fit testing and model correction: Assess model fit and correct the basic model to achieve optimal correspondence between the model and data.

Step 3: Estimation of mediating path coefficients: Calculate the segmented path coefficient estimates and obtain complete path coefficient estimates by multiplying the segmented path coefficients.

3.1.3. Spatial Effects Model

Based on the preceding theoretical analysis, regions are not mutually independent and affect each other, necessitating consideration of the low-carbon economy’s spatial spillover effect upon employment skills structure. The most prevalent spatial econometric models currently are the spatial autoregressive model (SAR), spatial error model (SEM) and spatial Durbin model (SDM). The first two models contain only spatial autoregressive term and error term, respectively, whilst the SDM model encompasses both spatial autoregressive term and error term. To determine the appropriate spatial econometric model, a series of tests are required, including the Lagrange multiplier test (LM) and the Hausman test. The Lagrange multiplier test constitutes a test for spatial structure existence, whilst the Hausman test involves selection between random and fixed effects. The results are shown in

Table 1. Test results of the spatial measurement model.

Test	Statistical Quantity	p-Value
LM-lag	65.390	0.000
R-LM-lag	10.301	0.001
LM-err	210.845	0.000
R-LM-err	155.756	0.000
Hausman	26.370	0.003

.

The test results presented in Table 1 demonstrate that the LM test statistic calculation shows the null hypothesis is rejected at the 1% significance level, clearly indicating significant spatial lag and spatial error effects exist in the spatial relationships concerning the research object. Simultaneously, Hausman’s test results also significantly rejected the null hypothesis, signifying that in the comparison between fixed effects and random effects, fixed effects possess more pronounced advantages than random effects in explaining model data and reflecting inter-variable relationships. Combining these two pivotal test results, following rigorous analysis and argumentation, the spatial Durbin model with fixed effects was ultimately selected as this study’s analytical model. The spatial Durbin model constitutes a combined extension of the spatial error model and spatial lag model, which accounts for both the employment skill structure’s spatial autoregressive effect and the low-carbon economy’s spatial lag effect. In summary, the model is constructed as follows:

$$\begin{gathered} \mathrm{s}\mathrm{s}\mathrm{o}\mathrm{e}=\mathsf{\rho }\sum _{\mathrm{i}\ne \mathrm{q}}{\mathrm{W}}_{\mathrm{i}\mathrm{q}}\times \mathrm{s}\mathrm{s}\mathrm{o}\mathrm{e}+{\mathsf{\beta }}_1\mathrm{l}\mathrm{c}\mathrm{e}\mathrm{l}+\mathsf{\eta }\mathrm{X}+{\mathsf{\theta }}_1\sum _{\mathrm{i}\ne \mathrm{q}}{\mathrm{W}}_{\mathrm{i}\mathrm{q}}\times \mathrm{l}\mathrm{c}\mathrm{e}\mathrm{l}+\mathsf{\eta }\sum _{\mathrm{i}\ne \mathrm{q}}{\mathrm{W}}_{\mathrm{i}\mathrm{q}}\times \mathrm{X} \\ +{\mathrm{u}}_{\mathrm{i}}+{\mathsf{\lambda }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i},\mathrm{t}}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em} \end{gathered}$$

(10)

where $\mathsf{\eta }$ is the exogenous spatial interaction effect of the model, characterising the extent to which the explanatory variable X of other neighboring provinces q affects province i. $\mathsf{\rho }$ is the spatial autoregressive coefficient characterizing the degree of spatial dependence of employment skill structure, i.e., the degree to which the employment skill structure of the labor force in province i is influenced by the surrounding region q. $\mathsf{\theta }$₁ indicates the degree of spatial dependence of the low-carbon economy. Other symbols retain the same meaning as in (9). ${\mathrm{W}}_{\mathrm{i}\mathrm{q}}$ denotes the weight value of the corresponding province in the economic distance matrix.

This paper selects the geographic distance matrix as the primary analytical tool, mainly based on the following considerations: firstly, the low carbon economy indicator system involved in the study already contains economic variables such as GDP, whilst the utilization of the economic distance matrix may precipitate multiple covariances among the explanatory variables; secondly, the 0–1 adjacency matrix only captures the correlation of geographically neighboring regions, disregarding possible interactions among non-neighboring but spatially neighboring regions. The formula for the geographic distance matrix is as follows, where ${\mathrm{d}}_{\mathrm{i}\mathrm{j}}$ denotes the distance between region i and region j’s provincial capital.

$${\mathrm{W}}_{\mathrm{i}\mathrm{j}}\left\{\begin{array}{c}{\displaystyle \frac{1}{{\mathrm{d}}_{\mathrm{i}\mathrm{j}}}}, \mathrm{i}\ne \mathrm{j}\\ 0, \mathrm{i}=\mathrm{j}\end{array}\right.$$

(11)

3.2. Variables

3.2.1. Explanatory Variable

The explanatory variable in this study constituted the employment skill structure (ssoe), defined as the proportion of workers with different skill levels in the labor market. Following the extended human-capital framework, we treat skill as a multidimensional construct that is proxied, yet not fully captured, by educational attainment; actual skill also embodies cognitive ability, non-cognitive traits, and on-the-job training that remain unobserved in the data [28]. This study classified individuals based on years of education. According to the China Human Capital Report 2024, the average years of education among China’s labor force population reached 10.88 years in 2022, with per capita educational attainment at junior high school or higher. To better reflect the current skill distribution in the labor market, the employment skill structure was measured as the ratio of high-skilled workers to medium- and low-skilled workers. High-skilled workers were defined as those possessing a bachelor’s degree or higher, medium-skilled workers as those with a college degree (including higher vocational education), and low-skilled workers as those with a high school diploma or less (including secondary vocational education).

3.2.2. Core Explanatory Variable

The core explanatory variable was the low-carbon economy (lcel). Owing to global warming, it is now widely acknowledged that developing a low-carbon economy proves necessary. A low-carbon economy endeavors to enhance economic output with reduced energy, pollution, and emissions, whilst diminishing natural resource utilization and mitigating environmental pollution. This paper developed an index system to evaluate low-carbon economic development levels based on prior literature, adhering to the principles of comprehensiveness, systematicness, and hierarchy. The system encompasses four dimensions: low-carbon output, low-carbon consumption, low-carbon resources, and low-carbon environment, as detailed in

Table 2. Indicator system for the level of development of the low-carbon economy.

Dimension	Indicator	Measurement Methodology (Units)	Indicator Properties
Low-carbon output	Carbon productivity	Ratio of regional GDP to CO2 emissions (ten thousand CNY/tons CO2)	Positive
	Carbon emission elasticity	Ratio of growth rate of carbon emissions to growth rate of gross regional product (%)	Positive
	Energy consumption per unit of GDP	Ratio of total energy consumption to gross regional product (tons of standard coal per million CNY)	Negative
Low-carbon consumption	Carbon emissions per capita	Ratio of CO2 emissions to resident population (tons CO2/person)	Negative
	Energy consumption per capita	Ratio of total energy consumption to resident population (tons of standard coal/person)	Negative
	Public transport vehicles per 10,000 population	Ratio of public transport numbers to resident population (Vehicles per 10,000 people)	Positive
Low-carbon resources	Share of coal in total energy consumption	Ratio of coal consumption to total energy consumption (%)	Negative
	Percentage of forest cover	Ratio of forest area to total land area (%)	Positive
	Parkland area per capita	Ratio of parkland area to resident population in built-up areas (square meters/person)	Positive
Low-carbon environment	Sulfur dioxide emissions per unit of GDP	Ratio of sulfur dioxide emissions to gross regional product (tons SO2/ten thousand CNY)	Negative
	Comprehensive industrial solid waste utilization rate	Comprehensive use of industrial solid waste as a ratio of solid waste generation (%)	Positive

. These dimensions capture the low-carbon economy’s core elements and key areas, offering a systematic approach to reflect and assess its state. Low-carbon output indicates the level of low-carbon technology, focusing on energy efficiency and carbon emission intensity in the production process; low-carbon consumption represents consumption patterns, involving consumer behavior and habit changes; low-carbon resources pertain to the endowment, development, and utilization of low-carbon resources; and low-carbon environment signifies the synergy between carbon emission reduction and environmental protection. Following this index system’s construction, the comprehensive score was determined by objectively assigning weights to low-carbon economic development indicators using the entropy method.

3.2.3. Control Variable

The main control variables in this paper are as follows. First, industrial structure: according to the ‘Gatti-Clark Theorem’ and the ‘Kuznets Rule’, employment skill structure changes in the same direction as industrial structure [20]. Second, green innovation: green innovation can alleviate carbon emissions reduction pressure and promote low-carbon transition, thus augmenting demand for highly skilled labor employment [29]. Third, government financial support: by providing social security systems and optimizing living and working conditions, sufficient financial assistance can precipitate logical changes in employment skill structure [30]. Fourth, consumption level: escalating consumption levels promote domestic product and service quality optimization and upgrading, which consequently affect employment skill structure [31]. Fifth, technological progress: through factor substitution effects and other mechanisms, technological advancement influences employment skill structure [32]. Sixth, import and export: the impact of exports and imports upon employment skill structure proves complex, with some scholars contending that international trade growth, coupled with information and communications technology advances, has enabled firms in developed countries to shift offshorable jobs to developing countries at lower labor costs. These shifts have led to employment polarization in developed economies such as the United States, which has been exacerbated by declining demand for middle-skill occupations relative to high-skill and low-skill occupations [33]. Seventh, wage level: Marshall’s equilibrium wage theory states that labor income should equal labor’s marginal product value; consequently, high-skilled workers typically receive higher skill premium, while rising regional labor wages encourage high-skilled worker inflow, thereby increasing high-skilled labor’s share [34,35]. Eighth, fixed capital stock: social fixed assets expansion assists in optimizing industrial structure and improving resource utilization, both of which impact employment skill structure [36]. Overall, the variables in this paper are selected as shown in

Table 3. Variable symbols and calculations.

Categories	Variables	Symbols	Variable Calculation Method	Unit
Explanatory variable	Employment skill structure	ssoe	Ratio of high-skilled labor force employment to medium and low-skilled labor force employment	%
Core explanatory variable	Low-carbon economy	lcel	The synthesis measure was obtained by entropy value method
Control variables	Industrial structure	isp	Ratio of value added in tertiary sector to value added in secondary sector	%
	Green innovation	lj	Ratio of the number of green patents filed annually at the provincial level to the total number of patents filed	%
	Government financial support	gfs	Ratio of local government general fiscal expenditures to regional GDP	%
	Consumption level	xf	Ratio of per capita consumption expenditure to disposable income of urban households	%
	Technological progress	kj	Ratio of internal expenditure on R&D to regional GDP by region	%
	Import and export	jck	Ratio of total exports and imports to gross regional product	%
	Wage level	wage	Average wage of employed workers	CNY
	Fixed capital stock	zbcl	Use of the perpetual inventory method to obtain capital stock by province for 2006–2021	Billions CNY

.

3.3. Data Sources

The data utilized in this study were downloaded from the 2023 releases (retrieved between September and December 2023) of the China Statistical Yearbook, China Energy Statistical Yearbook, China Environmental Statistical Yearbook, China Carbon Accounting Database, China Labor Statistical Yearbook, China Population and Employment Statistical Yearbook, and China Research Data Service Platform. To ensure reliability and consistency, the following collection standards were applied: (1) only official or officially audited sources were employed; (2) all series were required to cover the complete 2006–2021 period; (3) Tibet, Hong Kong, Macao, and Taiwan were excluded owing to missing or non-comparable data; (4) wages and fixed-capital stock were log-transformed to diminish scale differences; (5) continuous variables were winsorized at the 1st and 99th percentiles to mitigate outlier influence. Consequently, the final panel comprises 480 observations (30 provinces × 16 years) spanning 2006–2021.

4. Empirical Results and Discussion

4.1. Descriptive Statistical Analysis

There exists significant regional imbalance in China’s low-carbon economy development, with substantial disparities amongst provinces. As demonstrated in

Table 4. Descriptive statistical analysis.

Variables	Obs	Mean	Std. Dev.	Min	Max
Low-carbon economy	480	0.331	0.098	0.115	0.729
Employment skill structure	480	0.093	0.107	0.008	0.779
Green innovation	480	0.059	0.019	0.020	0.118
Industrial structure	480	1.130	0.654	0.500	5.297
Technological progress	480	0.017	0.012	0.002	0.094
Consumption level	480	0.765	0.107	0.512	1.101
Government financial support	480	0.231	0.099	0.083	0.643
Import and export	480	0.294	0.349	0.008	1.800
Fixed capital stock	480	10.30	0.925	7.488	12.25
Wage level	480	10.89	0.709	9.640	13.78

, this imbalance manifests primarily in the numerical distribution of low-carbon economy development levels, where the maximum value reaches 0.729 whilst the minimum stands at only 0.115, exhibiting obvious polarization characteristics. Despite the imbalanced overall development, the national average level of low-carbon economy has attained 0.331, indicating that China’s low-carbon economy remains in a stage of rapid development with broad scope for future improvement. Similar to low-carbon economy development, China’s employment skill structure also possesses significant regional disparities. In terms of data performance, there exists a substantial gap between the maximum and minimum values of employment skill structure, whilst the average level stands at merely 0.093, representing a relatively low development level, reflecting that current employment skill structure optimization still confronts considerable challenges. Further analysis of control variables’ distribution characteristics reveals obvious gaps between the maximum and minimum values of variables such as green innovation, industrial structure, and government financial support. These data features collectively demonstrate that China’s provinces present multidimensional unbalanced development trends in low-carbon economy, employment skill structure, and related influencing factors. Based on the aforementioned analysis, conducting in-depth provincial-level research on low-carbon economy’s impact upon employment skill structure proves of great practical necessity. Through provincial-scale research, we can more accurately comprehend the differences and patterns in low-carbon economy development and employment skill structure evolution across different regions, providing robust data support and theoretical foundation for formulating more targeted and effective regional development policies.

4.2. Results of the Analysis of the Direct Impact of the Low-Carbon Economy on the Employment Skill Structure

The research results support hypothesis 1, namely that low-carbon economy development directly promotes employment skill structure optimization and upgrading.

Table 5. Results of the analysis of the direct impact of the low-carbon economy on the employment skill structure.

Variable	Employment Skill Structure
Low-carbon economy	0.099 ** (2.08)
Industrial structure	0.067 *** (5.93)
Green Innovation	0.411 *** (3.46)
Government financial support	−0.075 * (−1.86)
Consumption level	−0.144 *** (−4.11)
Technological progress	−1.398 * (−1.86)
Import and export	−0.107 *** (−4.78)
Wage level	−0.006 (−1.14)
Fixed capital stock	−0.028 *** (−2.88)
Constant	0.495 *** (4.30)
R-squared	0.952
Regional fixed effects	Yes
Time fixed effect	Yes

Note: ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively, while t-values are represented by the data in parenthesis.

presents the findings from analyzing the direct effects of low-carbon economy on employment skill structure. The low-carbon economy exhibited a coefficient of 0.099, indicating a significant and positive impact on the employment skill structure at the 5% significance level. Specifically, a 1% increase in low-carbon economy was associated with a 9.9% increase in the employment skill structure. Low-carbon economic growth has precipitated increased demand for highly qualified laborers. These workers are required for various tasks, such as developing appropriate low-carbon policies and implementing low-carbon programs. On the other hand, traditional energy companies, confronting intense pressure to reduce carbon emissions, have been compelled to accelerate their transition to low-carbon energy. To achieve low-carbon transformation, these industries have invested more substantially in environmental protection technology research and development while hiring more highly educated workers due to cost and substitution effects. Consequently, the employment skill structure has been optimized and upgraded.

Regarding control variables’ influence upon employment skill structure, industrial structure and green innovation demonstrate positive effects, and import and export exhibit negative effects, which aligns with previous studies’ findings; government financial support, fixed capital stock and wage level exert negative effects upon employment skill structure, yet the coefficient remains small (weak effect). Government financial support favoring investment in low-skill areas, fixed capital stock concentration in labor-intensive industries, and regional low wage tendencies leading to “low-end competition” in employment constitute the principal causes of the weak negative effect, respectively. Technological progress, conversely, exerts significant negative impact upon employment skill structure, the core of which stems from technological progress’s “low-skill bias” and “structural mismatch” during low-carbon economic transition: initially, certain low-carbon technologies are characterized by labor-intensive features, such as renewable energy infrastructure, simple energy-saving retrofits and low-carbon agricultural technology improvements, which rely more substantially upon low-skilled labor, whilst their popularization increases low-skilled job proportions. Subsequently, if technological progress concentrates in low-skill-intensive industries’ low-carbon transformation, such as traditional manufacturing, it may eliminate medium-skilled jobs whilst retaining low-skilled positions through efficiency enhancement, thereby strengthening the negative effect. Moreover, following new technology introduction during low-carbon transformation, enterprises encounter difficulty matching high-skilled labor in the short term, temporarily relying upon low-skilled labor with simple training for employment, whilst the model fails to capture long-term skill upgrading dynamics, resulting in significant negative coefficients.

Given the long-term and phased nature of the evolution of a low-carbon economy, this paper conducts cross-sectional tests at four pivotal nodes in 2006, 2010, 2015, and 2021, with results presented in

Table 6. A cross-sectional analysis of the direct impact of low carbon economy on employment skill structure.

Variable	2006	2010	2015	2021
lcel	0.0859	0.0212	−0.249 ***	−0.201
	(0.102)	(0.0744)	(0.0711)	(0.180)
Control variable	Yes	Yes	Yes	Yes
Regional fixed effects	Yes	Yes	Yes	Yes
Time fixed effect	Yes	Yes	Yes	Yes
_Constant	−0.937 *	−1.412 ***	−2.215	−0.697
	(0.491)	(0.394)	(1.436)	(1.154)
R-squared	0.894	0.952	0.932	0.935

Note: *** and * denote significance at the 1% and 10% levels, respectively, while t-values are represented by the data in parenthesis.

. Contrary to the prior expectation that ‘the low-carbon economy continuously enhances employment skills structure,’ the effect of low-carbon variables upon employment skills structure exhibits significant temporal reversal, from initially neutral to negative: in 2006 and 2010, the coefficients were 0.086 and 0.021, respectively, neither attaining statistical significance, indicating that low-carbon transformation’s promotional effect upon employment skills structure had not yet materialized during the transition’s early stages. After 2015, the coefficient plummeted to −0.249 (p < 0.05), and although it slightly rebounded to −0.201 in 2021, it remained significantly negative, indicating that as policies deepened, the low-carbon economy actually weakened the substitutability between high- and low-skilled labor. This result suggests that low-carbon technological upgrades may have strengthened the synergistic demand for both high-skilled and low-skilled labor through a skill complementarity mechanism, rather than a simple substitution relationship.

This divergence between short-term and long-term outcomes fundamentally reflects the dynamic interplay between the ‘short-term adjustment costs’ and ‘long-term upgrade dividends’ of the low-carbon transition. This can be analyzed in depth from three perspectives: (1) The positive effects observed in the panel data from 2006 to 2021 stem from the synergistic evolution of ‘technology-industry-skills’. From 2006 to 2021, China’s low-carbon transition underwent a phase transition from ‘passive emissions reduction’ to ‘active innovation.’ Technological level: The number of green patent authorizations increased from 14,000 in 2006 to over 340,000 in 2021. Breakthroughs in core technologies such as new energy and carbon capture have created a significant demand for high-skilled positions (e.g., R&D engineers, carbon management consultants), driving an annual increase of 0.8 percentage points in the proportion of high-skilled labor. Industrial level: A positive feedback loop has been established between ‘industrial upgrading’ and ‘skill demand.’ Policy level: From the ‘energy conservation and emission reduction’ focus of the 11th Five-Year Plan to the ‘dual carbon goals’ of the 14th Five-Year Plan, policy priorities have shifted from end-of-pipe governance to source innovation. The vocational training system has gradually adapted to low-carbon skill demands (e.g., ‘carbon management consultants’ were included in the new occupational directory starting in 2019), significantly improving the long-term alignment between skill supply and demand. The cumulative effect of these factors has made the positive impact of the low-carbon economy on the employment skills structure evident in the panel data from 2006 to 2021, confirming the long-term pattern of ‘low-carbon economy driving the optimization of the employment skills structure.’ (2) Negative cross-sectional effect in 2015: As the ‘policy breakthrough year’ for low-carbon transformation, several landmark policies were implemented simultaneously (such as the launch of the national carbon market pilot program and the focus on green manufacturing in ‘Made in China 2025’), which significantly suppressed the skill upgrading effect due to short-term adjustment costs. Meanwhile, high-carbon industries (such as steel and coal) accelerated capacity reduction due to stricter environmental standards, forcing laid-off workers to shift to low-skill jobs in construction and services, thereby increasing the proportion of low-skill employment; Meanwhile, the new energy sector was still in its infancy, and the increase in high-skill positions was insufficient to offset the passive transfer of low-skill labor. (3) Cross-sectional negative effects in 2021: In 2021, the ‘post-pandemic recovery’ and ‘accelerated carbon neutrality’ factors led to a short-term mismatch between skill supply and demand. Post-pandemic economic recovery focused on ‘stabilising employment,’ with 70% of fiscal stimulus funds flowing into infrastructure construction. Such projects created a large number of low-skilled jobs in the short term, resulting in a year-on-year increase of 1.2 percentage points in the proportion of low-skilled employment. Meanwhile, high-skilled jobs in the low-carbon industry (such as new energy research and development) lagged behind low-skilled jobs in terms of recovery speed due to long project cycles and slow investment returns. The first round of transformation following the introduction of the ‘dual carbon’ goals in 2021 primarily focused on ‘energy substitution’ (e.g., transitioning from coal-fired power to wind and solar power). This transformation led to a surge in demand for low-skilled labor in sectors such as photovoltaic panel installation and wind turbine erection (approximately 800,000 jobs), while high-skilled roles in system integration and intelligent maintenance accounted for less than 20% of the total, resulting in a short-term skill structure characterized by ‘increased quantity but low quality.’ It is evident that the negative effects observed in 2021 do not negate the long-term trend but rather result from short-term factors in a unique environment.

4.3. Robustness and Endogeneity Tests

This research employed the method of replacing the primary explanatory factors and replacing the dependent variable for robustness testing to further demonstrate the aforementioned findings’ robustness. The coefficient of variation method, which compares indicators’ mean values to their initial values to measure weights, was utilized to re-measure the low-carbon economy. The coefficient of variation method constitutes a completely data-driven objective weighting technique. It calculates the coefficient of variation by dividing each indicator’s standard deviation by its mean, and then normalizes it to directly generate weights: the greater an indicator’s variation across samples, the stronger its discriminative power, and the higher the weight assigned. Using this method to construct an indicator system can automatically suppress redundant or highly homogeneous indicators and highlight core variables with significant differences, thereby avoiding subjective biases in expert scoring, eliminating the need for additional standardization steps, and allowing weights to be dynamically adjusted as data are updated, making the evaluation system both simple and effective. Compared to the entropy method, which is based on the degree of dispersion between indicators, this method showed minor differences in the indicators’ weights. When testing the robustness of the impact of the low-carbon economy on the structure of employment skills by using alternative explanatory variables, consider using the ratio of the number of employed persons at each level of education to the total number of employed persons as a substitute, and test using indicators of the proportion of highly educated persons and the proportion of persons with medium to low levels of education. As evidenced by

Table 7. Robustness and endogeneity tests.

	Robustness Test			Endogeneity Test
Variable	Employment Skill Structure	gl	zd	Low-Carbon Economy	Employment Skill Structure
Low-carbon economy (entropy value method)					1.843 *** (3.13)
Low-carbon economy (coefficient of variation method)	0.123 ** (2.18)	0.061 *** (2.85)	−0.064 *** (−2.99)
Public transport vehicles per 10,000 population				0.009 *** (3.41)
Constant	0.488 *** (4.24)	0.351 *** (5.34)	0.595 *** (8.90)	−0.006 (−0.07)	−0.122 (−0.74)
Control variable	Yes	Yes	Yes	Yes	Yes
Regional fixed effects	Yes	Yes	Yes	Yes	Yes
Time fixed effect	Yes	Yes	Yes	Yes	Yes
R-squared	0.952	0.966	0.966	0.654
F value				11.714
Kleibergen–Paap rk Wald F statistic				14.987

Note: *** and ** denote significance at the 1% and 5% levels, respectively, while t-values are represented by the data in parenthesis.

, after applying the coefficient of variation method, the low-carbon economy’s optimizing effect upon employment skills structure improved compared to previously, indicating that this method more fully captures the low-carbon economy’s pull effect upon high-skilled job demand by amplifying indicators’ weight with strong discriminatory power. Simultaneously, the coefficient remains significant at the 5% significance level, indicating that the improved effect is not statistical noise, further validating Hypothesis 1’s robustness (a low-carbon economy promotes the upgrading of the employment skills structure). At the same time, the results indicated that the low-carbon economy has a significant promotional effect on the employment of highly skilled personnel and a significant inhibitory effect on the employment of medium- and low-skilled personnel, thereby optimizing the employment skill structure. As shown in the Appendix A, we also conducted other robustness tests, and the results were still significant.

Omitted factors, reverse causation, and other endogeneity issues may influence results’ accuracy and scientific validity when analyzing how the low-carbon economy has affected employment skill structure. To address this, this study used instrumental variables to investigate endogeneity. The instrumental variables method constitutes an important approach for addressing endogeneity issues, with its core steps primarily comprising three aspects. First, identifying endogeneity issues: when correlation exists between the low-carbon economy and the error term (such as omitted variables, bidirectional causality, or measurement errors), ordinary least squares (OLS) estimation results will be biased, necessitating instrumental variables’ introduction to resolve the issue. Second, selecting appropriate instrumental variables. These variables must satisfy two core conditions: first, correlation, meaning they must be highly correlated with the endogenous explanatory variables and capable of influencing the endogenous variables; second, exogeneity, meaning they must be uncorrelated with the error term and only influence explained variables through endogenous variables, without any direct effect. Next is model estimation, with the most commonly used method being two-stage least squares (2SLS): in the first stage, the endogenous explanatory variable is regressed using the instrumental variable to obtain the endogenous variable’s fitted value, thereby filtering out the endogeneity component; in the second stage, the fitted value from the first stage replaces the original endogenous variable, and the explained variable is regressed to obtain consistent estimates. The advantages of the instrumental variable method are significant. First, it effectively addresses endogeneity issues. When the low-carbon economy is correlated with the error term, leading to biased OLS estimates, the instrumental variable method introduces exogenous instrumental variables to obtain consistent parameter estimates. Second, it suits complex causal relationship scenarios, especially when bidirectional causality or omitted critical variables exist. Instrumental variables can isolate the exogenous changes in explanatory variables, thereby identifying true causal effects. Based on the instrumental variable selection principle, which requires correlation with the endogenous explanatory variable and exclusivity from the error term, this study selected the number of public transport vehicles per 10,000 population as an instrumental variable. A two-stage least squares regression was then performed. To verify the validity of the number of public transport vehicles per 10,000 people as an instrumental variable, we conducted correlation tests and weak instrumental variable tests. First, the coefficient of the instrumental variable was 0.009 and was significant at the 1% level, indicating that the number of public transport vehicles per 10,000 people was significantly correlated with a low-carbon economy. Second, the F-value was 11.714, indicating that it passed the weak instrumental variable test (F-value > 10). Finally, we referenced the K-P rk F statistic. Since 14.99 > 8.96 (at the 15% level), we reject the null hypothesis H0: there is a weak instrumental variable. The results, presented in Table 7, indicated that even after adding instrumental variables, the low-carbon economy still exerted a positive impact upon employment skill structure. The regression coefficient was higher than that obtained from the direct regression in the previous section. This suggested that the low-carbon economy’s influence upon employment skill structure became more pronounced when endogeneity issues were properly addressed.

4.4. Analysis of the Indirect Impact of the Low-Carbon Economy on the Employment Skill Structure

Earlier theoretical analyses indicated that three principal indirect transmission pathways existed through which the low-carbon economy affected employment skills structure: industrial restructuring, the innovation effect resulting from green innovation, and the factor substitution effect brought about by investment in pollution control. An empirical analysis would be performed to quantify each transmission pathway’s unique contribution to employment skill structure optimization, thereby further confirming the aforementioned theoretical assumptions.

4.4.1. Model Fit Testing

This research utilized the structural equation mediation analysis approach to evaluate the low-carbon economy’s indirect effects upon employment skill structure. Structural equation modeling was employed for this analysis, as it was considered more precise and reliable for examining indirect effects. This method could effectively manage the intricate causal relationships between the low-carbon economy and the employment skill structure while accounting for measurement errors in both.

In AMOS 28, the basic impact paths were defined as ‘low-carbon economy → industrial structure → employment skill structure’, ‘low-carbon economy → green innovation → employment skill structure’, and ‘low-carbon economy → investment in pollution control → employment skill structure’. Combined with model adaptation and data considerations, the basic model was adjusted, as shown in Figure 1. Rectangles denote observed variables, circles represent residuals, arrows indicate influence direction, and numbers correspond to standardized path coefficients.

As illustrated in Figure 1, the low-carbon economy indirectly affected employment skill structure through four pathways: ‘low-carbon economy → industrial structure → employment skill structure’, ‘low-carbon economy → green innovation → employment skill structure’, ‘low-carbon economy → investment in pollution control → employment skill structure’, and ‘low-carbon economy → green innovation → industrial structure →investment in pollution →employment skill structure’. Based on the original hypothesis that green innovation influenced industrial structure and indirectly enhanced energy efficiency while optimizing highly polluting industries’ industrial structure, the updated model more realistically and persuasively captures these relationships.

The model’s fit was evaluated to ensure the modified model was appropriate for the data. As displayed in

Table 8. Adaptation results for structural equation modeling.

Evaluation Indicators	Fitted Value	Criteria or Thresholds for Adaptation
Ratio of cardinal degrees of freedom CMIN/DF	0.102	<3
The chi-square value significance probability value P	0.750	>0.05
Root Mean Square of Approximation Error RMSEA	0.000	<0.05
Goodness-of-fit index GFI	1.000	>0.90
Adjusted goodness-of-fit index AGFI	0.999	>0.90
Comparative Fit Index CFI	1.000	>0.90
Normative Fit Index NFI	1.000	>0.90
Tucker–Lewis fitting index TLI	1.008	>0.90
Incremental Fit Index IFI	1.000	>0.90

, the results indicated an excellent fit, confirming that the modified model was suitable for analyzing indirect effects. We recognized the importance of transparency regarding potential suprasaturation, and to rigorously assess this, we conducted two core checks. Theoretical and Empirical Validation Against Suprasaturation: First, all model paths originate from the theoretical framework. Second, Parameter-to-Sample Ratio: The model includes 14 free parameters with a sample size of 480 (ratio ≈ 1:34). For suprasaturated models, this ratio often approaches 1:5 or lower; our ratio falls well below this threshold, indicating minimal risk of over—parameterization driven by sample size constraints. In addition, we also tried other model paths. Although the fitting index is not perfect, the explanatory power of the model is indeed poor, which does not conform to the theoretical and practical significance.

4.4.2. Analysis of the Results of the Indirect Impact of the Low-Carbon Economy on the Employment Skill Structure

As shown in

Table 9. Results of the analysis of the indirect impact of the low-carbon economy on the employment skill structure.

Action Paths	Estimated Value	p-Value	Contribution Rate
Low-carbon economy → Green Innovation → Employment Skill Structure	0.020	0.002	3.70%
Low-carbon economy → Industrial Structure → Employment Skill Structure	0.333	0.000	62.4%
Low-carbon economy → Investment in Pollution Control → Employment Skill Structure	0.040	0.000	7.50%
low-carbon economy → green innovation → industrial structure →investment in pollution →employment skill structure	−0.006	0.002	−1.2%
Low-carbon economy →employment skill structure	0.135	0.000	27.6%

, low-carbon economy positively impacted employment skill structure by enhancing industrial structure, green innovation, and investments in pollution control. The pathway “low-carbon economy → industrial structure → employment skill structure” contributed most substantially, accounting for 62.4% of the low-carbon economy’s indirect impact upon employment skill structure. In contrast, the pathway “low-carbon economy → green innovation → employment skill structure” possessed the smallest contribution, comprising merely 3.7% of the total indirect effect. These results suggest that, at this stage, the low-carbon economy primarily supports employment skill structure by upgrading industrial structure, while green innovation’s effect upon employment skill structure remains relatively limited. Two primary reasons account for the limited impact of the low-carbon economy on the employment skill structure through green innovation. Firstly, green technologies require time to progress from research and development to tangible outcomes. Significant demand for highly skilled labor only emerges when these innovations are genuinely implemented in production processes. Secondly, due to elevated costs, some green innovation technologies may possess limited application scope. Only when these technologies are widely adopted will they create substantial numbers of associated positions, thereby exerting stronger influence upon employment skill structure.

Contrary to the original hypothesis, this study discovered that the low-carbon economy’s impact upon employment skills structure follows a ‘negative chain pathway’: a low-carbon economy significantly promotes green innovation, which consequently drives industrial structure evolution towards high-end development. However, industrial structure upgrading significantly suppresses pollution control investment, ultimately exerting negative impact upon employment skills structure optimization. This finding suggests that green technology-driven industrial upgrading may, in the short term, reduce demand for end-of-pipe pollution control in traditional high-polluting industries through a ‘substitution effect,’ leading governments and businesses to curtail environmental protection investments and thereby suppress high-skilled green job expansion such as environmental engineers and carbon managers. In other words, without policy interventions to support pollution control investments in emerging industries, low-carbon transition may fall into an ‘industrial upgrading trap,’ where green innovation enhances industrial levels yet indirectly weakens pollution control investments’ employment-creation function. Therefore, policy-making must establish dynamic equilibrium between promoting green innovation and establishing environmental investment mechanisms to ensure that industrial upgrading and pollution control investments evolve concurrently, thereby achieving sustainable employment skills structure optimization.

4.5. Analysis on Spatial Influence of Low-Carbon Economy on Employment Skill Structure

4.5.1. Spatial Correlation Test

Before conducting spatial regression analysis, it proves necessary to examine whether spatial correlation exists in the employment skill structure. The global Moran index is utilized to measure the degree of agglomeration of employment skill structure in each province, which can reflect the spatial autocorrelation trend of different employment structures in the whole region. Its value interval is [−1, 1]: when the employment skill structure’s Moran index falls within the range of 0–1, it signifies that the employment skill structure throughout the entire region is positively correlated and exhibits agglomeration tendencies; if the employment skill structure’s Moran index takes values between −1–0, employment skill structure throughout the entire region is negatively correlated and demonstrates disaggregation states; and when the Moran index value equals 0, it indicates that no spatial correlation exists in provincial employment skill structure.

This paper examines the employment skill structure’s spatial correlation from 2006 to 2021 based on the geographic distance matrix, with results shown in

Table 10. Global Moran Index for the employment skills structure in 2006–2021.

Year	Moran Index	Z Statistic	p-Value
2006	0.025	1.953	0.025
2007	0.032	2.193	0.014
2008	0.019	1.732	0.042
2009	0.022	1.843	0.033
2010	0.050	2.895	0.002
2011	0.040	2.872	0.002
2012	0.038	2.803	0.003
2013	0.066	3.626	0.000
2014	0.036	2.546	0.005
2015	0.075	3.431	0.000
2016	0.067	3.197	0.001
2017	0.062	3.088	0.001
2018	0.077	3.568	0.000
2019	0.075	3.522	0.000
2020	0.062	3.228	0.001
2021	0.071	3.406	0.000

. The results demonstrate that China’s employment skill structure’s global Moran’s I (Moran’s Index) in 2006–2021 exhibits significant spatial correlation, with the Moran’s Index fluctuating from 0.025 in 2006 to 0.071 in 2021, whilst the overall significance level continues increasing over time. This indicates that when regional employment skill structure is upgraded, neighboring regions’ employment skill structure also tends towards upgrading, forming high-high agglomeration.

4.5.2. Results of Spatial Regression

As shown in

Table 11. Spatial regression results of low-carbon economy on employment skill structure.

Variable	Regional Impact	Space Lag Term
Low-carbon economy	0.018 ** (2.58)	0.145 *** (3.09)
Industrial structure	0.075 *** (9.10)	0.211 *** (2.73)
Green Innovation	0.512 *** (4.01)	4.358 *** (3.97)
Government financial support	−0.051 (−1.01)	0.177 (0.39)
Consumption level	−0.150 *** (−5.17)	−0.201 (−0.81)
Technological progress	−1.161 *** (−2.86)	−2.303 (−0.74)
Import and export	−0.106 *** (−8.15)	0.011 (0.10)
Wage level	−0.005 (−0.86)	−0.037 (−0.92)
Fixed capital stock	−0.011 (−0.88)	0.154 (1.40)
Spatial autocorrelation coefficient	−0.901 *** (−4.10)
Individual effects		0.000 *** (15.10)
Observed value	480	480
R2	0.156	0.156

Note: z-statistics in parentheses, *** and ** indicate significant at the 1% and 5% levels, respectively.

, the coefficient of low-carbon economy development’s influence upon regional employment skills structure is 0.018 and proves significant at the 5% level, indicating that local low-carbon economic development promotes employment skill structure optimization and upgrading. Meanwhile, the spatial lag term’s coefficient is 0.145 and proves significant at the 1% level, indicating that neighboring regions’ low-carbon economy exerts positive spillover effects upon local skill structure. This result may stem from deepening inter-regional green technology R&D and industrial chain division of labor; for instance, new energy industry cluster formation in neighboring regions may indirectly promote local high-skill job growth through technology diffusion or talent sharing mechanisms. The spatial autocorrelation coefficient of −0.901 proves significant at the 1% level, indicating significant negative dependence exists regarding employment skills structure between regions. This phenomenon may originate from high-skilled labor’s “siphon effect,” whereby economically developed regions attract neighboring talent through high-quality employment opportunities, precipitating skill resource depletion in neighboring regions.

Based on the aforementioned analysis, the spatial Durbin model’s total effect undergoes further decomposition and division into direct and indirect effects. The direct effect denotes the degree of influence exerted by a region’s low-carbon economy upon employment skill structure, whilst the indirect effect signifies neighboring regions’ low-carbon economy influence upon regional employment skills structure through spatial linkages, i.e., the spatial spillover effect. The total effect constitutes the summation of direct and indirect effects, representing the comprehensive impact of the low-carbon economy on employment skills structure, with spatial effects decomposition results presented in

Table 12. Results of the spatial effect decomposition of the low-carbon economy on the employment skills structure.

Variable	Spatial Direct Effect	Spatial Indirect Effects	Gross Spatial Effects
Low-carbon economy	0.014 ** (2.00)	0.074 *** (2.77)	0.088 *** (3.22)
Industrial structure	0.070 *** (9.73)	0.078 ** (2.08)	0.148 *** (3.73)
Green Innovation	0.402 *** (3.27)	2.263 *** (3.22)	2.666 *** (3.80)
Government financial support	−0.057 (−1.31)	0.112 (0.46)	0.054 (0.21)
Consumption level	−0.148 *** (−5.42)	−0.033 (−0.23)	−0.181 (−1.24)
Technological progress	−1.104 *** (−2.70)	−0.777 (−0.46)	−1.881 (−1.12)
Import and export	−0.109 *** (−8.52)	0.060 (1.10)	−0.049 (−0.83)
Wage level	−0.004 (−0.74)	−0.016 (−0.71)	−0.020 (−0.90)
Fixed capital stock	−0.015 (−1.36)	0.092 (1.52)	0.077 (1.18)
Observed value	480	480	480
R2	0.156	0.156	0.156

Note: z-statistics in parentheses, *** and ** indicate significant at the 1% and 5% levels, respectively.

.

In this study concerning low-carbon economy’s impact on employment skills structure, the direct, indirect and total effects of low carbon economy are significant at the 5%, 1% and 1% levels, respectively. Particularly noteworthy is that the indirect effect and total effect exhibit the highest degree of significance among these three effects. The direct effect coefficient of low carbon economy upon employment skills structure displays a positive value, indicating that the development of the low carbon economy can exert favorable influence upon local employment skills structure. Through continuous regional low-carbon economic progress, related industries undergo constant upgrades while high-skilled labor increases, consequently leading to regional employment skill structure optimization and enhancement. Concerning the indirect effect, the low-carbon economy coefficient also proves positive. This result demonstrates evident positive spatial spillover effects when low-carbon economy affects employment skills structure, i.e., whereby neighboring regions’ prosperous low-carbon economic development extends beyond their own regional employment skills structure, simultaneously exercising positive promotional effects upon regional employment skills structure through spatial correlation. For example, throughout neighboring regions’ low-carbon economic development processes, advanced low-carbon technologies and innovative concepts can be cultivated, subsequently radiating to the region through technological diffusion, talent flow and other means, driving related industry development in the region, and ultimately promoting regional employment skills structure development towards superior advancement.

5. Conclusions and Recommendations

5.1. Conclusions

Previous studies have demonstrated that the low-carbon economy can influence employment skill structure through both direct and indirect mechanisms. Our research confirms that the low-carbon economy assumes a significant role in optimizing employment skill structure, thereby contributing to sustainable development’s dual pillars: environmental sustainability and inclusive economic growth. Specifically, the low-carbon economy directly promotes employment skill structure improvement, establishing the foundation or sustainable workforce development that supports long-term low-carbon transitions. Indirectly, it enhances employment skill structure through industrial restructuring, innovation effects, and factor substitution effects, all constituting core pathways for achieving the United Nations Sustainable Development Goals (SDGs), particularly SDG 8 (decent work and economic growth) and SDG 9 (industry, innovation, and infrastructure). By facilitating industrial restructuring, the low-carbon economy amplifies demand for high-skilled workers. It simultaneously elevates green innovation levels, further refining employment skill structure through innovation effects. Additionally, low-carbon economic growth increases pollution control investment, optimizing employment skill structure through factor substitution effects.

Our analysis reveals that the industrial restructuring pathway is the primary means by which the low-carbon economy contributes to employment skill structure. This pathway accounts for 54.6% of the overall indirect effect and represents the most substantial contribution among the four identified transmission mechanisms. The findings indicate that industrial restructuring serves as a major driving force behind employment skill structure improvement and optimization within a low-carbon economy, with profound implications for sustainable development by ensuring that low-carbon transition generates quality employment opportunities. Low-carbon industrial structure transformation has facilitated economic development model shifts and significantly altered labor market demand structure. This has yielded sustained increases in highly skilled labor requirements, which positively impacts employment skill structure optimization and upgrading, ultimately supporting China’s dual-carbon goals while advancing the broader sustainable development agenda.

The impact of a low-carbon economy upon employment skills structure manifests spatial spillover effects. The low-carbon economy in neighboring regions exhibits positive spillover effects on the local skill structure, and significant negative dependence exists between regions in terms of employment skill structure. This results from the ‘brain drain effect’ of highly skilled labor, whereby economically developed regions attract talent from surrounding areas through high-quality jobs, consequently causing skill resource depletion in neighboring regions.

5.2. Recommendations

We recommend the following: Accelerate low-carbon economic development and employment skill upgrading in central and western regions to narrow regional disparities. China’s low-carbon economy and employment skills structure manifest a spatial distribution pattern extending from eastern regions towards central and western regions. Formulate relevant policies to facilitate underdeveloped central and western regions’ transformation from a ‘low-skill, high-carbon emission’ model to ‘medium-to-high-skill, low-carbon emission’ model. Strengthen regional cooperation and coordinated development, encouraging eastern coastal regions to collaborate with central and western regions in the field of low-carbon economy. Through such cooperation, the technological and financial advantages of eastern regions’ technological and financial advantages can complement central and western regions’ resource and labor advantages, collectively promoting low-carbon economic development and employment skills structure upgrading.

Promote green innovation. A low-carbon economy can enhance green innovation levels, thereby optimizing employment skills structure. Governments should develop and implement targeted policies while establishing comprehensive incentive mechanisms to fully support research and development activities, which serve as the technical cornerstone for achieving long-term sustainable development goals such as climate action (SDG 13) and sustainable industrialization (SDG 9). Specifically, empirical evidence from Table 5 highlights the critical role of green innovation: a 1% increase in green innovation drives a 41.1% improvement in the employment skill structure. However, our analysis reveals that this potential remains underutilized under current conditions, the path coefficient of low-carbon economy → green innovation → employment skill structure is only 3.7%, indicating that green innovation is not yet fully leveraged to optimize skills. To address this gap, governments should prioritize scaling green innovation by incentivizing green patent development. For actionable targets, we calculate that increasing the employment skill structure by 5% (a moderate policy goal) requires a 0.12% increase in green innovation (derived from the 41.1% elasticity: ΔSkill Structure = 0.411 × ΔGreen Innovation). Given the current average green innovation level (5.9% of total patents), this translates to an annual target of 6.61% green patent share. To achieve this, we propose establishing a special green patent subsidy fund equivalent to 0.5–1% of provincial GDP annually. This fiscal commitment aligns with the elasticity of green innovation on skill structure and ensures that R&D resources directly translate to high-skilled job creation. For example, a 0.5% GDP allocation to green patents could drive a 0.21% annual increase in skill structure (0.411 × 0.5%).

Drive the upgrading and transformation of traditional industries. A low-carbon economy contributes significantly to employment skills structure optimization through industrial structure improvement. Therefore, government departments should fully leverage policy guidance, strengthen traditional industry support and regulation, and promote their transition towards low-carbon development. First, based on the elasticity of 0.067 for industrial structure on the employment skill structure and the 62.4% contribution rate of the “low-carbon economy → industrial structure → employment skill structure” path, a two-stage transformation logic centered on “low-carbon industrial structure → high-end skill structure” should be followed. In the short term, focus on phasing out high-carbon production capacity. Over the next three years, aim to reduce the employment share of high-carbon industries from 35% to 25% by shutting down or renovating backward production capacity in sectors like steel and coal, thus freeing up space for skill upgrading. For the long-term skill structure adaptation, develop targeted training programs. Launch specialized training in “carbon management” and “clean production operation and maintenance” to meet the skill demands of industrial structure optimization. Strive to increase the annual share of high-skilled labor by 2 percentage points, matching the skill requirements of the mediation path in Table 9. To ensure the implementation, set quantifiable policy goals. If the aim is to drive a 3% upgrade in the employment skill structure in the medium term, industrial structure optimization of approximately 44.78% is needed.

Increase educational investment and strengthen labor skill training. Currently, China’s economy is undergoing a high-quality transformation and development phase, but the proportion of low-skilled labor in the labor market is relatively high, which proves unconducive to employment structure optimization and upgrading across different industries. On the other hand, low labor skill levels can inhibit technological innovation. Therefore, throughout forthcoming low-carbon economic development, the government should allocate greater resources to education for enhancing workers’ cultural literacy and thereby improving workforce skill levels. Additionally, the government should increase subsidies to create favorable market environments, encourage clean technology research and development, provide financial support for technological innovation, enhance the market’s capacity to absorb labor, and promote further structure optimization and upgrading across industries, ultimately achieving China’s low-carbon economic transformation and development. Scale up vocational education investment for low-carbon talent cultivation by increasing annual public vocational education expenditure to 6% of provincial GDP (from the current 4% average) by 2030, with 60% of incremental funding earmarked for green skills training. Launch a National “Low-Carbon Skills Upgrade Program” for Existing Workers. Prioritize 5 million low-skilled workers in high-emission sectors (coal mining, traditional manufacturing) for upskilling by 2030, with a focus on retraining 2 million workers for renewable energy operations. Upskilling 3 million workers in energy efficiency services Provide full wage replacement (80% of monthly income) during training (up to 6 months) and a CNY 3000 completion bonus. Tax and Subsidy Policies: Expand the R&D expense super-deduction for green tech firms from 100% to 120% of qualified expenditures, with an additional 5% deduction for companies that train ≥10% of their workforce annually in low-carbon skills.

6. Limitations and Future Improvements

Although this study has yielded certain findings regarding the exploration of the low carbon economy’s influence upon employment skills structure, several limitations persist owing to research conditions and objective factors.

First, the impact effects of public health emergencies remain insufficiently incorporated. COVID-19, constituting a major global emergency, has exerted a multidimensional impact on China’s economic and social development: on the one hand, the transformation pace of low-carbon industries (e.g., the progress of construction of new energy projects, and the cycle of research and development of green technologies) has experienced staged interference, whilst the labor market’s telecommuting popularization trend and employment pattern flexibility have accelerated. On the other hand, the epidemic’s restriction on cross-regional labor mobility and asymmetric recovery of some industries (e.g., high-carbon traditional manufacturing and low-carbon service industries) may have altered the transmission path of skills demand in the low-carbon economy. This study fails to systematically incorporate epidemic shock’s dynamic influences owing to data availability limitations and time-series coverage restrictions, consequently producing deficiencies in explanatory capacity regarding employment skill structure during the special period.

Second, limitations exist in the acquisition and measurement of segmented industry data. Low-carbon economy’s impact upon employment skill structure exhibits significant industry heterogeneity, but the current study faces two data constraint categories: first, insufficient micro-data concerning subsector skill demand; second, the employment statistical system of some emerging low-carbon fields (such as carbon accounting, new energy operation and maintenance) remains imperfect, and data timeliness disparities and statistical caliber differences impede accurate quantification of skill structure change magnitudes across different industries. This constrains, to a certain extent, the refined application of research findings at industry level.

Future research can be advanced in two ways: first, integrating epidemic impact time-series data and constructing multi-stage dynamic analysis models; second, expanding data source channels and supplementing micro-skill demand information through enterprise research and industry reports integration.