Green Growth with Equity: How Low-Carbon City Pilot Policy Builds Coordinated Development Pathways in China

Abstract

In light of the challenge of imbalanced econometric growth, social equity, and environmental sustainability, the low-carbon city pilot policy offers a scientifically grounded framework for low-carbon development and a practical resolution to the urban evolution paradox. This article employs the Double Machine Learning methodology to examine the impact of the Low-Carbon City Pilot Policy (LCCP) on inclusive green growth (IGG) using data from prefecture-level cities from 2008 to 2021. Empirical evidence supports the assertion that the LCCP significantly contributes to the advancement of IGG in urban regions. Moreover, multiple robustness tests provide additional support for this conclusion. The mechanism tests demonstrate that the LCCP enhances IGG by fostering technological innovation, regional entrepreneurship, and labor structure upgrading. Furthermore, the heterogeneity study shows that the LCCP has a stronger impact on key environmentally protected cities, cities without an old industrial base, and cities without a resource-based economy than on other types of cities. We recommend fostering green innovation, building entrepreneurial ecosystems, and implementing differentiated strategies to establish a more effective low-carbon development strategy.

1. Introduction

The rapid expansion of urbanization, while driving global economic development, has generated severe environmental and social challenges, making the reconciliation of economic growth, social equity, and environmental sustainability a pressing worldwide issue. As the world’s largest developing economy, China exemplifies this tension. As depicted in Figure 1, while China’s per capita GDP demonstrated a steady rise from 2008 to 2021, its carbon emissions remained at persistently high levels throughout most of this period—illustrating the difficulty of decoupling economic growth from environmental pressure under conventional development pathways. This trajectory occurred against a backdrop of accelerated urbanization, where the urban population share surged from 20.16% (1981) to 59.58% (2018) [1]. With 57 Chinese cities ranked among the world’s 100 most polluted [2], it is clear that achieving growth without exacerbating environmental degradation and social inequality requires a new urban development model.

Figure 1. Trends of CO₂ Emissions and Per Capita GDP in China (2008–2021). Note: CO₂ emissions data are sourced from the China Emission Accounts and Datasets (CEADs). Per capita GDP data are obtained from the China Statistical Yearbook (National Bureau of Statistics, various years).

It is in this context that the paradigm of Inclusive Green Growth (IGG) gains critical relevance. Formally advanced at the 2012 United Nations Conference on Sustainable Development (Rio+20), IGG provides a normative framework that explicitly balances economic progress, social equity, and environmental protection [3]. This framework has been strongly endorsed by China’s national strategy. The 20th National Congress of the CPC explicitly called for harmonizing human activity with nature and promoting green development, aligning national policy directly with the IGG agenda [4]. The central question, therefore, is not whether China should pursue IGG, but how it can be effectively operationalized. This paper investigates China’s Low-Carbon City Pilot Policy (LCCP) as a primary mechanism for translating this national commitment into tangible urban outcomes.

Existing research suggests that the LCCP is one of the most significant measures China has taken to achieve a balance between economic development and environmental protection. After the implementation, pilot provinces and cities explored establishing capped carbon emission control systems, major project carbon emission evaluation systems, and low-carbon product standard certification systems. Following the introduction of the policy, the overall carbon intensity in the pilot areas decreased faster than the national average [5]. In terms of its social impact, some scholars argue that it promotes a shift in the employment structure towards high-skilled labor, leading to labor migration from the primary to the secondary and tertiary industries. This affects labor supply and relative wage levels, and eventually widens the urban–rural income gap [6]. However, the existence of an inversed U-shaped relationship between environmental management and income disparity has been pointed out by others, with the suggestion that overcoming this obstacle would result in an improvement in income inequality. Therefore, in this context, studying whether LCCP can achieve inclusive green growth has significant theoretical and practical implications for resolving current disputes and exploring sustainable development paths.

This study makes several contributions to the existing literature. Firstly, previous research has primarily focused on the individual effects of low-carbon city pilot policies on environmental indicators, such as carbon emission levels and air quality, and social indicators, such as the urban–rural income gap. This article broadens the scope of the research to include prefecture-level cities and above, combining inclusive and green growth. Using indicator evaluation methods and data envelopment analysis, it re-evaluates the inclusive green growth levels of prefecture-level cities. It discusses the impact of low-carbon city pilot policies on inclusive green growth from both theoretical and empirical perspectives, enriching the relevant research on policy effects and supplementing the influencing factors of inclusive green growth. Secondly, by using a DML model for analysis, the present article has the objective of reducing the bias arising from model misspecification. In addition, it eliminates bias through residual modeling. The consequence of these actions is the accurate identification of causal and mediation effects. And what it does is make it more accurate when it comes to identifying causal and mediation effects. Lastly, this article comprehensively examines the mechanisms by which low-carbon city pilot policies affect inclusive green growth from multiple angles, including technological innovation, entrepreneurial vitality, and labor structure upgrading. This provides empirical evidence for solving the short-term dilemmas of LCCP policies and achieving regional coordinated development and inclusive green growth goals. Fourthly, this article explores the heterogeneous impacts of low-carbon city pilot policies on inclusive green growth in different types of cities from the perspectives of environmental institutional foundations, industrial bases, and resource endowments.

The structure of this article is as follows. Section 2 of this text reviews the relevant literature. Section 3 of the text provides a review of LCCP in China, alongside the proposal of theoretical hypotheses. Section 4 of this text provides a comprehensive explanation of the methods, variables, data, and sources employed in this study. Section 5 presents the estimation results and the robustness tests. Section 6 of this study will analyze the intermediary mechanisms of the LCCP. The evaluation of the heterogeneity effects of LCCP is the purpose of Section 7. Section 8 of the text provides a summary and discussion of the policy recommendations.

2. Literature Review

IGG has been proposed as a novel approach to achieving sustainable development, with the objective of realizing economic growth, social equity and environmental protection [3]. In the following section, the consequences of the LCCP are delineated in relation to each of the three dimensions of inclusive green growth.

The first area of research concerns the economic effects of the LCCP. The implementation of LCCP has the potential to significantly promote technological innovation in pilot cities and neighboring cities. This is due to the enhancement of green total factor productivity, optimization of the industrial structure, and alleviation of financing constraints [7]. It is evident that the role of carbon emission reduction technology in promoting economic growth is becoming more and more significant [8]. The construction of low-carbon cities is regarded as a significant measure to enhance economic competitiveness. For instance, Hanna and Oliva found that a decrease in sulfur oxides (SO₂) led to an increase in working hours in a refinery in Mexico City [9]. Porter and Linde conducted case studies and determined that appropriate environmental regulation could generate “innovation compensation” for firms, including green technologies, processes, and products, which may enhance economic performance [10].

The second area of research concerns the social effects of the LCCP. An important way to achieve IGG is through full employment and income distribution. The LCCP has been demonstrated to facilitate a transition of labor from the primary sector to the secondary and tertiary sectors through the output effect and the factor substitution effect [11,12], which leads to changes in labor supply and demand and relative wage levels, and ultimately affects the urban–rural income gap [6]. Additionally, the Porter hypothesis [10] posits that the implementation of stringent environmental regulations, when meticulously designed, has the potential to catalyze innovation. This innovation, in turn, has the capacity to counterbalance the immediate private expenses associated with such regulations. Consequently, this innovation may ultimately contribute to an enhancement in a nation’s productivity [10]. According to Foa [13], the implementation of environmental regulations has the potential to engender distributive impacts, encompassing the promotion of enhanced gender equality and the generation of diverse economic and social advantages. Conversely, a contrary viewpoint is held by certain scholars. The imposition of environmental regulations has been demonstrated to engender costs and exert a detrimental effect on competitiveness. Furthermore, such regulations have been shown to occasion adverse socioeconomic effects, including the reduction in employment rates and welfare. As the distribution of income and the employment sector are pivotal factors in the context of welfare, the implementation of these regulations may result in the creation of employment opportunities for specific categories of workers within certain geographical areas, whilst concurrently generating job prospects for other categories of workers in disparate regions [14].

The third research area concerns the environmental impacts of LCCPs. A significant corpus of research has been dedicated to the exploration of the relationship between the development of low-carbon cities and green growth [15,16,17]. Furthermore, some scholars have explored the impact of the LCCP on environmental indicators such as eco-efficiency, carbon emission efficiency and air quality [5,18,19,20]. It has been demonstrated by extant research that the LCCP is capable of achieving environmental protection and emission reduction objectives through industrial restructuring and technological innovation [21,22,23]. Furthermore, a number of studies have indicated a negative correlation between China’s LCCP and carbon intensity. Significant regional heterogeneity has been observed in terms of policy effectiveness, including government support [16], infrastructural development [15], environmental information disclosure, and green innovation [24].

The green transition of enterprises serves as the foundation for a city’s sustainable development, and the LCCP inevitably impacts the production and operation of these enterprises. In addition to the examination of cities as research subjects, the study of the effects of LCCP on businesses represents a further crucial direction for research. Like the Porter hypothesis [10], the LCCP is believed to promote the green technological innovation in enterprises without setting specific targets [7,25]. Moreover, the heterogeneity of enterprises has been demonstrated to exert an influence on the performance of the LCCP. Yang [26] utilized a sample of Chinese A-share listed firms and ascertained that state-owned firms and firms situated in pilot zones with municipal secretaries who possess greater promotion incentives are more susceptible to the LCCP. In conclusion, the preponderance of scholarly opinion is that the LCCP has had a substantial and beneficial effect on China’s economic, social and environmental performance.

In light of the extant research, three gaps in the field should be noted: Firstly, as has previously been demonstrated by the extant literature, the LCCP has a significant impact on the economic and environmental outcomes of IGG. However, there is a paucity of attention paid to incorporating social inclusion within the policy effect analysis framework. IGG possesses inherent characteristics that allow for a comprehensive reflection of the economic, social, and environmental effects of the LCCP. Therefore, it is imperative to investigate the influence of the LCCP on IGG. Secondly, most of the existing research on IGG uses two main methods for measuring it, data envelopment analysis and the index system comprehensive evaluation method. The former focuses more on the issue of “Technical Efficiency”, while the latter is based on the core concept of IGG and constructs comprehensive evaluation indicators to measure it. This article argues that the comprehensive indicator evaluation method is better suited to reflect the true meaning of IGG. In addition, when using comprehensive indicators to evaluate the level of IGG at the city level, the relevant literature does not include the impact of carbon emissions. Therefore, based on the characteristics of the LCCP, it is necessary to consider carbon emissions when analyzing the impact of LCCP on IGG. Thirdly, contemporary scholars hypothesize that environmental regulation has the potential to influence IGG through a variety of macro-mechanisms, including technological innovation, industrial structure adjustment, and government expenditure. Compared with the existing studies, this article synthesizes the macro and micro, and analyzes the direct, indirect, and heterogeneous impacts that the LCCP may have on IGG from the directions of technological innovation, industrial structure upgrading, and SMEs entrepreneurial activeness, and improves the relevant theoretical and empirical studies. the relevant theoretical and empirical research.

3. Policy Background and Research Hypotheses

3.1. Policy Background

In response to the pressing challenge of global climate change and the international community’s collective expectation for environmental conservation, China has unequivocally recognized its global obligation to combat climate change and has committed to the ambitious targets of peaking carbon emissions by 2030 and achieving carbon neutrality by 2060. The rapid urbanization rate, particularly from approximately 36% in 2000 to over 60% in 2020, has transformed cities into major energy consumers and focal points for greenhouse gas emissions. Cities contribute to more than 70% of national energy demand and significant greenhouse gas emissions. To mitigate greenhouse gas emissions and achieve sustainable development, the National Development and Reform Commission (NDRC) announced the Notice on Conducting the Pilot Program of Low-Carbon Provinces and Cities (hereafter referred to as “Notice”) in 2010, initially designating five provinces and eight municipalities for the pilot program. The second cohort was launched in 2013, including Hainan province and 28 other cities. In 2017, the NDRC further expanded the program to include 45 cities (Districts and Counties), such as Wuhai City in the Inner Mongolia Autonomous Region, as part of the third batch of the LCCP. Figure 2 illustrates the distribution of pilot regions. The Notice mandates that all pilot regions integrate local industrial characteristics and development strategies, adhere to low-carbon principles in urban infrastructure planning and construction, and formulate and implement industrial, fiscal, taxation, and technology promotion policies to foster low-carbon development. This article aims to examine the impact of low-carbon city pilot policies on inclusive green growth, a concept that aligns closely with the goals of low-carbon city construction, and to investigate whether the LCCP can influence inclusive green growth and its mechanisms, thereby aiding policymakers in the future adjustment of strategies to achieve sustainable economic development.

Figure 2. Geographic distribution of pilot and non-pilot areas.

3.2. Research Hypotheses

Inclusive green growth fundamentally encapsulates principles of environmental sustainability paired with social equity, which is an organic integration of inclusive growth and green growth. The paradigm of low-carbon urban development heralds a novel trajectory for achieving inclusive and sustainable urban growth. On one hand, mandated investments in environmental governance under the LCCP are poised to reshape the urban–rural employment landscape due to the disparities in the urban–rural industrial structure. This phenomenon is anticipated to influence the income disparity between urban and rural inhabitants [27]. Although this dynamic could potentially widen the urban–rural income gap in the short term, in line with the Porter hypothesis [10], stringent environmental regulations are projected to augment the disposable income of citizens through technological innovation, thus narrowing the urban–rural income gap [28]. Additionally, the environmental regulations may precipitate a decline in executive remuneration while concurrently elevating the wages of the labor force, thereby reducing within-firm wage inequality [26]. From this perspective, the LCCP embodies inclusive growth. On the other hand, the LCCP has not only proved to be effective in reducing carbon emissions from transportation, electricity and heating [29], but also in transforming the low-carbon economy by stimulating industrial and technological advancements, promoting the clustering of new industries and services, upgrading energy-consuming industries, and increasing the rate of urbanization [30]. From this perspective, the LCCP embodies green growth. Accordingly, this article proposes the first hypothesis:

Hypothesis 1.

The Low-Carbon City Pilot Policy has a significant positive effect on inclusive green growth. This article asserts that the LCCP may affect the IGG through the following three channels (see Figure 3).

Figure 3. The influencing mechanism of LCCP on IGG.

The LCCP may enhance IGG by fostering technological innovation. The Porter hypothesis [10] indicates that rational environmental regulations can drive technological innovation and competitive gains. Firstly, the LCCP is expected to significantly influence technological innovation. On one hand, the LCCP encourages firms in the pilot regions to develop and implement green strategies, embrace green technologies, and create green products, enhancing their competitive edge [31]. On the other hand, all pilot regions have adopted policy tools to incentivize technological innovation for green progress in businesses [32], such as increasing financial support for low-carbon development, setting up financial specialization for low-carbon technology research, increasing financial support for low-carbon talent training, and accelerating the establishment of a carbon financial market. These policies direct businesses to focus their innovation efforts on green technology, directly boosting green innovation. Secondly, technological innovation can create jobs and improve income, allowing wider sharing of economic benefits [33]. Furthermore, technology innovation can boost social welfare and support inclusive growth by promoting green consumption [34,35]. Accordingly, this article proposes the second hypothesis:

Hypothesis 2.

The Low-Carbon City Pilot Policy can promote technological innovation and thus promote inclusive green growth.

LCCP can promote IGG by enhancing entrepreneurial activity. According to the Porter hypothesis [10], reasonable environmental regulation can be a driving force for corporate innovation, pushing companies to develop new green technologies and products, thereby creating new business opportunities. Firstly, LCCP encourages the combination of low-carbon development strategies with local traditional characteristic industries while vigorously developing strategic emerging industries and establishing low-carbon industrial systems, thereby creating more entrepreneurial opportunities. People tend to actively explore the entrepreneurial opportunities implied under new policies when policies change [36]. Secondly, policy changes send the message to entrepreneurs that entrepreneurship is feasible, and pilot cities enhance the legitimacy and value of related industries by strengthening the statistics and regulation of GHG emissions [37], which mobilizes potential entrepreneurs, especially environmental proponents, and increases the probability of entrepreneurship occurring. Thus, the LCCP significantly increased entrepreneurial activity in the pilot cities. Concurrently, enhanced entrepreneurial activity contributes to IGG in cities. Firstly, from the perspective of shared development, enhanced entrepreneurial activities provide more employment and development opportunities for various social groups [38]. By developing low-carbon industries and supporting green entrepreneurship, cities can attract more green investments, create more new jobs, enhance the financial well-being of the local populace and narrow the income gap. This model of economic growth not only benefits a wide range of social groups, but also enhances the overall qualitative aspect of life and societal well-being. The diversity and breadth of low-carbon entrepreneurship programs can better accommodate and absorb the participation of people from different backgrounds and with different skills, thereby promoting the common prosperity of society. Secondly, from the standpoint of green development, such entrepreneurial activities under the oversight of the pilot policy of low-carbon cities have the potential to promote the development of green and environmental protection industries. In the context of low-carbon city construction, start-ups will not only meet new environmental standards in the development process, but also explore and apply more green technologies to facilitate the development of sustainable industrial chains. Additionally, by supporting these entrepreneurial activities, the pilot low-carbon city policy can not only reduce dependence on traditional high-pollution industries, but also enhance its own sustainable development capability. This path of green development not only helps to improve the quality of the urban environment, but also boosts the overall competitiveness and attractiveness of the city. In summary, LCCP can promote green and shared development in cities by enhancing the entrepreneurial activity of the region, thus realizing IGG. In accordance with the preceding discussion, the present article puts forward the third hypothesis:

Hypothesis 3.

The Low-Carbon City Pilot Policy can increase entrepreneurship and thus promote inclusive green growth.

LCCP can increase IGG by promoting the labor structure upgrading. Human capital theory [39] suggests that investment in human capital (e.g., education, skills, health, etc.) is critical for economic growth and individual income. Improving the skills and knowledge of workers through education and training can increase labor productivity and economic competitiveness. Firstly, the LCCP explicitly proposed at the beginning of its implementation to cultivate composite, high-end, and specialized technological talents adapted to low-carbon development, and attracted high-end talents and research institutions by setting up a municipal low-carbon technology research financial specialization and supporting scientific research projects. During the implementation of the policy, the governments of the pilot cities take proactive measures to cultivate low-carbon skilled personnel to promote low-carbon transformation in their regions [40]. For example, Ankang City issued the Action Program for Low-Carbon City Pilot Work in Ankang City in 2018, which explicitly states that it is imperative to fortify the construction of a talent team for low-carbon development. A robust mechanism must be established for the identification, nurturing, cultivation, and assessment of talent. There is a necessity to prioritize the cultivation and introduction of high-caliber professionals specializing in energy management, conservation, and protection. The same applies to industrial energy-saving technology, among others, with all parties involved in this endeavor being industry, academia and research. The formulation of a comprehensive talent cultivation plan that integrates all these sectors is crucial for the city’s low-carbon development. Consequently, the implementation of the low-carbon city pilot policy has the potential to enhance the supply of high-skilled talents, thereby improving the city’s labor structure. Secondly, the enhancement of the labor structure has also been demonstrated to contribute to the improvement of IGG in the city. In the context of the low-carbon city pilot policy, local governments are responsible for organizing green and environmental protection publicity activities, as well as other public participation policies. The purpose of these activities is to encourage urban residents to participate more actively in the construction of low-carbon transformation and to pursue low-carbon skilled jobs. The ultimate objective of these initiatives is to promote the green development of the city. Conversely, it has been demonstrated that the enhancement of the labor structure has the capacity to diminish the skill premium [41], which can narrow the urban–rural income gap through human capital accumulation, skill mobility, and incentives for skill supply [42], thereby being conducive to the inclusive development of society. In accordance with the foregoing, the present article puts forward the fourth hypothesis:

Hypothesis 4.

The Low-Carbon City Pilot Policy can promote labor structure upgrading of pilot cities and thus promote inclusive green growth.

4. Methodology and Data

4.1. Econometric Model

4.1.1. Basic Model

The present article investigates the impact of LCCP on IGG. Indeed, IGG is a comprehensive measure of transformational urban growth, influenced by numerous factors in the economy and society. To ensure the accuracy of the estimation of policy effects, the impact of other factors on inclusive green growth in cities should be controlled as much as possible. It is challenging for conventional econometric models and methodologies to efficiently address the issue of high-dimensional data with limited sample size. This is primarily due to their vulnerability to the curse of dimensionality, multicollinearity, and constrained control of key covariates, which consequently leads to estimation bias. With regard to machine learning models, it has been demonstrated that they are capable of obtaining effective estimates in high dimensions, thus circumventing the so-called ‘curse of dimensionality’ issue. Nevertheless, machine learning models continue to exhibit a regularity bias problem that is not negligible. In light of the aforementioned findings, Chernozhukov et al. [43] proposed the Double/debiased machine learning (DML) model. This model incorporates numerous machine learning algorithms and regularization techniques to automatically screen pre-selected sets of high-dimensional control variables. The objective is to identify an effective set of control variables that exhibit high prediction accuracy. The DML model has two primary benefits. Firstly, it circumvents the ‘dimension curse’ that arises from redundancy in control variables. Secondly, it mitigates the estimation bias that arises from the limited number of main control variables. This is achieved by addressing the estimation bias problem [44,45].

In order to empirically analyze and test the effect of the LCCP on IGG, this article has constructed a partial linear regression (PLR) model based on the DML method for Hypothesis 1. The following details are provided:

Firstly, this article specifies the partially linear dual machine learning model as shown in Equations (1) and (2)

$${\mathit{IGG}}_{i,t}={\theta }_0{\mathit{LCCP}}_{i,t}+g\left(X_{i,t}\right)+U_{i,t},E\left(U_{i,t}|X_{i,t},{\mathit{LCCP}}_{i,t}\right)=0$$

(1)

${\mathit{IGG}}_{i,t}$ denotes inclusive green growth in city i in year t, which is a continuous composite index that quantifies the city-level performance of inclusive green growth, constructed following the multidimensional framework and entropy weighting methodology detailed in the Variable Description section. ${\mathit{LCCP}}_{i,t}$ is a dummy variable for city i’s low-carbon city pilot policies in year t (the city takes the value of 1 in the year of the LCCP, and 0 otherwise), and if ${\theta }_0$ is significantly positive, it indicates that the LCCP has a facilitating effect on the IGG. According to Hypothesis 1, the coefficient is expected to be significantly positive in this article; and $X_{i,t}$ is a multidimensional control variable for the city in year t, which affects the explanatory variables through the function $g$. The exact form of $g$ is unknown, and its estimator $g$ is obtained through machine learning models; $U_{i,t}$ is the error term with a conditional mean of 0 (0 < i < I, 0 < t < T).

If Equation (1) is estimated directly, the estimate ${\widehat{\theta }}_0$ is biased. This is due to the fact that, in high-dimensional or complex model settings, the machine learning model is required to introduce a regular term in order to reduce the dimensionality. This is intended to avoid excessive variance in the estimator, but it also introduces a regular bias in the function $g$, thereby making it difficult for ${\widehat{\theta }}_0$ to converge to ${\theta }_0$. The resolution of this issue can be achieved through the introduction of the auxiliary Equation (2).

$${\mathit{LCCP}}_{i,t}=m\left(X_{i,t}\right)+V_{i,t},E\left(V_{i,t}\mid X_{i,t}\right)=0$$

(2)

Firstly, this article estimates the function $m$ in Equation (2) by using a machine learning model to obtain the estimate $\widehat{m}$ and obtains the residual ${\widehat{V}}_{i,t}={\mathit{LCCP}}_{i,t}-\widehat{m}\left(X_{i,t}\right)$. Secondly, the present article considers ${\widehat{V}}_{i,t}$ as an instrumental variable for ${\mathit{LCCP}}_{i,t}$ and employs a machine learning model to estimate an estimate $\widehat{g}$ of the function $g$, which in turn computes an estimate ${\widehat{\theta }}_0$ of ${\widehat{\theta }}_0$. In this equation, $n$ denotes the total sample size.

$${\widehat{\theta }}_0={\left({\displaystyle \frac{1}{n}}\sum _{i\in l,t\in T} {\widehat{V}}_{i,t}{\mathit{LCCP}}_{i,t}\right)}^{-1}{\displaystyle \frac{1}{n}}\sum _{i\in l,t\in T} {\widehat{V}}_{i,t}\left({\mathit{IGG}}_{i,t+1}-\widehat{g}\left(X_{i,t}\right)\right)$$

(3)

4.1.2. Mediation Effects Model

In order to verify the influence of technological innovation as path mechanisms, the present article employs the DML model for regression estimation. This article follows the established procedure outlined in the seminal methodological work to test the mediating variable [46], and the specific research model is as follows:

$$\begin{array}{c}{\mathit{IGG}}_{i,t}={\theta }_0{\mathit{LCCP}}_{i,t}+g\left(X_{i,t}\right)+U_{i,t},E\left(U_{i,t}|X_{i,t},{\mathit{LCCP}}_{i,t}\right)=0\\ \end{array}$$

(4)

$$\begin{array}{c}{M}_{i,t}={\theta }_0{\mathit{LCCP}}_{i,t}+g\left(X_{i,t}\right)+U_{i,t},E\left(U_{i,t}|X_{i,t},{\mathit{LCCP}}_{i,t}\right)=0\\ \end{array}$$

(5)

$$\begin{array}{c}{\mathit{IGG}}_{i,t}={\theta }_0{\mathit{LCCP}}_{i,t}+{\theta }_1{M}_{i,t}+g\left(X_{i,t}\right)+U_{i,t},E\left(U_{i,t}|X_{i,t},{\mathit{LCCP}}_{i,t}\right)=0\\ \end{array}$$

(6)

$${\mathit{LCCP}}_{i,t}=m\left(X_{i,t}\right)+V_{i,t},E\left(V_{i,t}\mid X_{i,t}\right)=0$$

(7)

In Equation (5), $M_{i,t}$ is the mechanism variable. This article also uses the PLR model for estimation, and the other relevant control variables are consistent with the previous section. From Equations (4) to (6), the test steps for mediating effects are as follows: The initial step in the research process is to conduct a test that will ascertain whether the LCCP has the capacity to enhance IGG to a significant degree. The second step is to test whether the LCCP has the capacity to strengthen the mechanism variables to a significant degree. The third step is to test the effects of the LCCP and the mechanism variables on IGG.

4.2. Variable Description

4.2.1. Dependent Variable

Inclusive green growth (IGG): IGG refers to a development model that is both inclusive and green within the context of economic progress. Drawing on these authoritative studies [47,48,49], we quantify IGG across three interdependent pillars: Economic Development (ED), Social Equity (SE), and Green Development & Environmental Protection (GDEP). The selection of specific indicators under each pillar, detailed in Table 1, is not arbitrary but is rigorously based on two criteria: (1) their theoretical relevance to the core concepts of inclusive and green growth, and (2) their widespread adoption and validation in prior research. For instance, the SE pillar includes indicators like the urban–rural income gap and educational equity to directly capture the inclusive dimension, while the GDEP pillar employs indicators such as industrial sulfur dioxide emissions and public green space per capita to reflect the green dimension. This approach ensures our index is both theoretically sound and comparable with existing academic work.

Table 1. Calculation index of the explained variable.

Target Layer	Dimension Layer	Indicator Name	Nature	Unit	W_2008	W_2021
Inclusive green growth	Economic development	Per capita GDP	+	Yuan/person	0.056	0.106
		Ratio of fiscal revenue to GDP	+	%	0.037	0.058
		Share of secondary sector in GDP	−	%	0.008	0.030
		Share of tertiary sector in GDP	+	%	0.010	0.035
	Social equity	Registered urban unemployment rate	−	%	0.004	0.012
		Per capita disposable income of rural residents	+	Yuan/person	0.013	0.024
		Per capita disposable income of urban residents	+	Yuan/person	0.023	0.049
		Ratio of disposable income per capita of rural residents to urban residents	+	multiple	0.005	0.009
Inclusive green growth	Social equity	Proportion of social security expenditures to fiscal expenditures	+	%	0.299	0.125
		Public library books per 10,000 population	+	Volumes	0.261	0.140
		Hospital and health center beds per 10,000 population	+	Sheets	0.027	0.042
		Public transport vehicles per 10,000 people	+	Vehicles	0.083	0.126
	Green development and environmental protection	Sulfur dioxide emissions per capita	−	Tons	0.002	0.003
		Carbon dioxide emissions per capita	−	Tons	0.001	0.004
		Wastewater discharge per capita	−	Tons	0.001	0.002
		Industrial smoke and dust emissions per capita	−	Tons	0.004	0.004
		Public green space per capita in cities	+	Hectares/person	0.130	0.191
		Comprehensive utilization rate of general industrial solid waste	+	%	0.009	0.031
		Centralized treatment rate of sewage treatment plants	+	%	0.013	0.005
		Non-hazardous domestic waste disposal rate	+	%	0.014	0.003

To aggregate these indicators into a composite index, we employ the fixed-base extreme difference entropy methodology [50]. This method is chosen specifically to eliminate subjectivity in assigning weights. Unlike subjective weighting methods (e.g., Analytic Hierarchy Process), the entropy weight method determines weights objectively based on the variation in the data itself: An indicator with greater variation across cities and time receives a higher weight, as it is deemed to carry more information. This ensures that the structure of the data, rather than our personal judgment, drives the results. This approach is widely recognized for its objectivity in constructing comprehensive indices in economic and environmental research.

4.2.2. Independent Variables

Dummy variables for Low-Carbon City pilot policies: From 2010 to 2017, the National Development and Reform Commission (NDRC) established three batches of low-carbon pilot cities. Specifically, as the Low-Carbon City pilot policies were implemented in July 2010, November 2012, and January 2017, and considering that the effects of these policies may exhibit a lag [51], this article uses 2011, 2013, and 2018 as the base periods for policy implementation. The analysis covers up to nine years following the policy shock and up to thirteen years preceding the policy impacts.

4.2.3. Mechanism Variables

The present article aims to elucidate the mechanisms through which the LCCP impacts IGG via three pathways: The enhancement of urban innovation, the promotion of entrepreneurial vitality, and the encouragement of the upgrading of labor structure are of paramount importance.

Technological innovation: Innovation has been identified as a significant catalyst for high-quality development, particularly in urban areas where innovation capacity is a critical determinant of economic development quality. It is evident that technological innovation, amongst the various innovation activities that exist, has the most immediate impact on economic development quality. In an attempt to negate the impact of population size on the outcomes, this article employs the number of patent applications per capita and the number of green patent applications per capita to gauge the levels of technological innovation (Patent) and green technological innovation (GPatent) in each city.

Entrepreneurial Vitality (Entrep): The number of new private enterprises has been identified as an indicator of regional entrepreneurial activity [52]. The present article employs the National Enterprise Credit Information Publicity System as the data source to obtain the number of new private enterprises in each city from the Qixinbao Database, adopting an approach that has been successful in previous studies. The number of new private enterprises is then standardized using the labor force population aged 15–64. Furthermore, in order to account for variations in urban magnitude, the total population of the city at the year’s conclusion is utilized as the standardized base, and the number of new enterprises per 100 individuals is calculated to ascertain entrepreneurial vitality, denoted as Entrep.

Labor structure upgrading (HL): Labor structure upgrading is represented by the ratio of employed persons in high-skilled to low-skilled labor forces. Referring to existing studies [53], this article employs the scientific research, technical services, and geological survey industries to represent high-skill industries closely related to low-carbon industries, and the agriculture, forestry, animal husbandry, and fisheries industries to represent low-skill industries.

4.2.4. Control Variables

Referencing to relevant studies [54,55,56], this article controls the potential factors affecting IGG. (1) The level of educational development (Edu) is quantified by the proportion of education expenditure in GDP. (2) The level of scientific and technological development (Tec) is assessed by the proportion of scientific and technological expenditure in GDP. (3) The level of investment in infrastructure (Invest) is determined by the proportion of urban fixed asset investment in fiscal expenditure. (4) Internet penetration (Inter) is measured by the ratio of the number of international Internet users to the total population at the end of the year. (5) The level of financial development (Finan) is evaluated by the ratio of the balance of deposits and loans of financial institutions to regional GDP at the end of the year. (6) The level of industrialization (Indus) is gauged by the proportion of regional industrial value added to GDP. (7) The level of social consumption (Consump) is measured by the ratio of total retail sales of consumer goods to regional GDP. (8) Population density (Popden) is expressed as the ratio of the total urban population to the administrative area. (9) Economic density (Ecoden) is expressed as the ratio of city GDP to administrative area. (10) Government self-sufficiency (GS) is measured by the ratio of local general budget revenues to regional GDP, reflecting the capacity of a regional government to be self-sufficient. (11) The government’s intervention capacity (GI) is measured by the ratio of local general budget expenditures to regional GDP. (12) The level of urbanization (Urb) is measured by the ratio of the resident urban population to the total population of the area at the end of the year. (13) The level of human capital development (HC) is measured by the natural logarithm of the number of students enrolled in primary, general secondary, and general tertiary schools per 10,000 population. (14) The level of transport is measured by the natural logarithm of road passenger traffic per capita (RPT) and road freight traffic per capita (RFT). Furthermore, the article incorporates quadratic terms for each city variable in the regression analysis, with the objective of enhancing the precision of the fitted model. Concurrently, city and time fixed effects are incorporated in the form of individual and year dummy variables to circumvent the omission of information pertaining to the city and time dimensions.

4.3. Data Sources and Descriptive Statistics

This article employs panel data from 282 prefecture-level cities in China spanning the period from 2008 to 2021. It is selected for its methodological appropriateness in evaluating the Low-Carbon City Pilot policy. The starting point of 2008 accounts for preliminary low-carbon initiatives preceding the formal national pilot program, while the endpoint of 2021 is determined by data availability constraints. Specifically, the compilation and public release of official city-level data on key metrics—particularly those related to energy consumption and green growth—typically experience a lag of two to three years. Consequently, 2021 represents the most recent year for which a complete, consistent, and officially verified panel dataset can be constructed across all included cities. This data range ensures reliability and is consistent with the timeframe employed in influential prior studies [57,58], thereby facilitating meaningful comparison of our empirical findings with the existing body of literature. Due to the unavailability of data and the existence of economic and cultural disparities between regions, city data for Tibet, Hong Kong, Macao, and Taiwan have been excluded from the test group. The data, control variables, and mechanism variables associated with the construction of the inclusive green growth indicator system for prefecture-level cities are primarily sourced from the China Urban Statistical Yearbook and the official websites of each city’s statistical bureau. In order to address minor data gaps, this article integrates information from the Statistical Bulletin of National Economic and Social Development of each city and applies the linear interpolation method. Table 2 delineates the main variables and their descriptive statistics.

Table 2. Variables and descriptive statistics.

	Variable Symbol	Sample Number	Mean Value	Standard Deviation	Minimum Value	Maximum Value
Dependent variables	IGG	3948	3.182	1.305	0.544	13.263
	ED	3948	4.177	1.879	1.131	26.781
	SE	3948	3.173	1.546	0.170	19.002
	GDEP	3948	2.385	0.615	0.091	8.783
Core independent variable	LCCP	3948	0.277	0.447	0.000	1.000
Control variables	Edu	3948	0.178	0.041	0.000	0.377
	Tec	3948	0.016	0.016	0.001	0.207
	Invest	3948	4.811	2.113	0.000	17.168
	Inter	3948	2.593	12.620	0.000	355.180
	Finan	3948	2.394	1.190	0.560	21.302
	Indus	3948	0.371	0.167	0.000	2.849
	Consump	3948	0.373	0.110	0.000	1.013
	Popden	3948	5.735	0.939	0.000	7.882
	Ecoden	3948	0.308	0.753	0.001	15.356
	GS	3948	0.460	0.225	0.054	1.541
	GI	3948	0.192	0.103	0.043	1.485
	Urb	3948	0.550	0.156	0.174	1.000
	HC	3948	5.992	0.635	3.829	7.830
	RPT	3948	2.493	0.886	0.000	7.957
	RFT	3948	3.059	0.805	0.000	8.021
Mechanism variables	Patent	3948	3.921	7.286	0.000	74.436
	Gpatent	3948	0.466	0.938	0.000	10.972
	Entrep	3948	1.178	1.160	0.147	20.236
	HL	3948	0.103	0.348	0.000	7.399

5. Empirical Results and Discussion

This article employs the DML method to examine the impact of the LCCP on IGG using data from prefecture-level cities from 2008 to 2021. The structure of the empirical strategy is as follows. Section 5.1 explains the basic machine learning model setting. Section 5.2 presents the empirical analysis, firstly establishing the baseline causal effect of LCCP on IGG, and subsequently Section 5.3 subjects this finding to a series of robustness tests. Section 6 follows by investigating the underlying transmission mechanisms of technological innovation, entrepreneurial activity, and labor structure upgrading. Section 7 evaluates the heterogeneity of policy effects across different city types.

In this section, the methodology of model estimation is described in the initial stages. Subsequently, the results of the foundational analysis are presented for the purpose of examining the impact of the LCCP on IGG. Subsequent to this, the empirical results are subjected to a series of robustness checks. Finally, the causally identified effects of the low-carbon city pilot policies on urban inclusive green growth are elucidated.

5.1. Model Estimation Description

Regarding machine learning model selection, this article employs the random forest model as the estimation model within the DML method, except for the robustness test. Except for the primary term of the above 15 control variables, the article also incorporates the quadratic terms of the control variables to account for the nonlinear relationship between the explanatory variables and the control variables. For the in-sample and out-of-sample cross-validation method used in the DML method, k is set to 5. The parameter settings of other relevant estimation procedures, unless otherwise specified, are directly referenced from the practices of Chernozhukov et al. [43].

5.2. Effect of LCCP

The effectiveness of China’s policies has long been debated. Therefore, this paper evaluates the actual level of impact of LCCP on IGG through the DML model. To account for unobserved, time-invariant heterogeneity across cities and common time shocks, our model explicitly includes a full set of city fixed effects and year fixed effects as high-dimensional controls. The machine learning algorithms in the first stage of DML automatically handle the selection and regularization of these numerous dummy variables. The DML’s cross-fitting and debiasing procedures then ensure that the final estimate of the policy effect (θ₀) is robust to this high-dimensional conditioning, providing a consistent and asymptotically normal estimator without the need for pre-testing the correlation between unobserved effects and the regressors [43]. As illustrated in Table 3, the results of the estimation of low-carbon city pilot policies and inclusive green growth in cities are presented. Model (1) incorporates controls for city fixed effects, time fixed effects, and the primary terms of other city variables over the full sample interval. The regression coefficient of LCCP on IGG is positive and significant at the 5% level, indicating that the low-carbon city pilot policy significantly enhances the level of inclusive green growth in cities. As a development on model (1), model (2) incorporates the quadratic terms of urban variables. The regression coefficient remains significantly positive, with minimal change in value. In summary, the impact of low-carbon city pilot policies on inclusive green growth is significantly positive, regardless of the inclusion of high-dimensional control variables, thereby largely validating Hypothesis 1.

Table 3. Benchmark regression.

VARIABLES	IGG		ED	SE	GDEP
	(1)	(2)	(3)	(4)	(5)
LCCP	0.106 *** (0.032)	0.084 ** (0.033)	0.143 *** (0.044)	0.101 ** (0.042)	−0.037 * (0.021)
Constant	0.001 (0.007)	0.0004 (0.006)	0.003 (0.009)	0.001 (0.009)	0.001 (0.005)
The primary term for the control variable	YES	YES	YES	YES	YES
Quadratic term for the control variable	NO	YES	YES	YES	YES
Time fixed effect	YES	YES	YES	YES	YES
Urban fixed effect	YES	YES	YES	YES	YES
N	3948	3948	3948	3948	3948

Note: Results are from the Double Machine Learning (DML) estimator. Consistent with the reporting convention in DML applications [43,44] the table presents the debiased causal coefficient and its standard error. The validity of inference is ensured by the DML procedure itself, and goodness-of-fit metrics are not the focus of this methodology. Standard errors in parentheses. The following table is the same. * p < 0.1, ** p < 0.05, *** p < 0.01.

This article further examines the specific impacts of low-carbon city pilot policies on various dimensions of inclusive green growth, presenting the regression coefficients and significance of LCCP on ED, SE, and GDFP in models (3) to (5), respectively. Firstly, the effect of LCCP on both ED and SE is significantly positive, with the coefficient on ED being larger than that on SE. This disparity may stem from the fact that LCCP is typically accompanied by substantial economic incentives and investments, particularly in green technologies and infrastructure. To provide an illustration, investments in renewable energy projects, energy efficiency, and emission reduction technologies have been proven to contribute directly to economic growth. These investments frequently yield substantial economic returns that may be more immediate and perceptible than improvements in social equity. Moreover, the regression coefficient of LCCP on GDEP is significantly negative, suggesting a short-term adverse impact on urban green development. This paradoxical outcome may be attributed to the complex and nonlinear nature of the socio-technical transition that the LCCP instigates. Initially, the policy may trigger substantial investments in new green infrastructure and technologies, the construction and manufacturing phases of which can be resource-intensive and temporarily elevate environmental pressures—a phenomenon akin to the “investment displacement effect” in transition economics. Simultaneously, this finding resonates with the left segment of the Environmental Kuznets Curve [59], where the aggressive pursuit of low-carbon growth might initially amplify the scale effect of economic activity, thereby overwhelming the nascent technique effect of green innovation. Beyond these structural explanations, behavioral and systemic factors are also at play. The policy incentives, for instance, could inadvertently encourage symbolic compliance or “greenwashing” among some firms [60], thereby diluting the actual environmental benefits. Furthermore, when confronted with certain special circumstances, such as when local governments attract low-quality foreign direct investment, the intense regulatory focus and resources dedicated to carbon reduction could temporarily sideline enforcement and investment in other critical, non-climatic environmental domains, such as water pollution or biodiversity conservation, leading to a perceived decline in overall environmental performance [61]. Therefore, while counterintuitive, this short-term disincentive does not negate the policy’s long-term potential. Rather, it underscores the intricate dynamics and necessary trade-offs inherent in orchestrating a fundamental economic shift towards sustainability.

5.3. Robustness Tests

The robustness of the findings is demonstrated in Table 4 and Table 5 through four distinct methods. The dependent variable was replaced, incorporating province and time interaction fixed effects, excluding the impact of parallel policies, and resetting the DML Model. Each approach independently corroborates our primary conclusion, providing consistent evidence across different analytical techniques. The methodological consistency of the present study serves to reinforce the validity of the results obtained. These results indicate that pilot low-carbon city policies have a significant impact on inclusive green growth. The convergence of results across these robustness tests significantly enhances the credibility of our core conclusion.

Table 4. Basis robustness tests.

VARIABLES	IGG
	(1)	(2)	(3)	(4)	(5)
LCCP	0.0289 *** (0.010)	0.133 *** (0.039)	0.123 *** (0.035)	0.169 *** (0.036)	0.159 *** (0.036)
Constant	−0.002 (0.002)	−0.003 (0.008)	0.002 (0.008)	0.001 (0.008)	0.002 (0.008)
The primary term for the control variable	YES	YES	YES	YES	YES
Quadratic term for the control variable	YES	YES	YES	YES	YES
Time fixed effect	YES	YES	YES	YES	YES
Urban fixed effect	YES	YES	YES	YES	YES
Province–Time fixed Effects	NO	YES	NO	NO	NO
N	3948	3948	3948	3948	3948

*** p < 0.01.

Table 5. Dual machine learning robustness tests.

VARIABLES	IGG						Instrument
	(1)	(2)	(3)	(4)	(5)	(6)	(7)
LCCP	0.0907 ** (0.038)	0.055 ** (0.021)	0.889 *** (0.044)	0.159 *** (0.014)	0.135 *** (0.033)	0.145 *** (0.035)	1.940 *** (0.598)
Constant	−0.000 (0.006)	0.000 (0.008)	0.060 *** (0.019)		0.000 (0.007)	−0.000 (0.007)	−0.005 (0.011)
The primary term for the control variable	YES	YES	YES	YES	YES	YES	YES
Quadratic term for the control variable	YES	YES	YES	YES	YES	YES	YES
Time fixed effect	YES	YES	YES	YES	YES	YES	YES
Urban fixed effect	YES	YES	YES	YES	YES	YES	YES
Sample partition ratio	1:4				1:2	1:7	1:4
Machine learning algorithms	Lasso	Gradboost	SVM	Random forest
Regression model	PLR			IR	PLR
N	3948	3948	3948	3948	3948	3948	3948

** p < 0.05, *** p < 0.01.

5.3.1. Replacement of the Dependent Variable

This article employs the green development efficiency method to measure inclusive green growth, as outlined by Dong and Xu [62]. In considering the inclusiveness of economic development concurrently, this article has revised the green total factor productivity (GTFP), and the improved method has added the urban–rural income development gap that can characterize inclusiveness. It is therefore evident that the specific formula for inclusive green development is as follows:

$${IGG}_{it}=W_1\times {GTFP}_{it}+W_2\times {Gap}_{it}$$

(8)

where ${GTFP}_{it}$ and ${Gap}_{it}$ denote the green total factor productivity and the urban–rural income gap, respectively, and $W_1$ and $W_2$ denote the focus weights of the two in this article, respectively. Given the current stage of development, this article considers both equally important, so $W_1$ = $W_2$ = 0.5. GTFP is measured using the undesirable-SBM model, and Gap is measured by dividing the disposable income per capita of rural residents by the disposable income of urban residents.

The calculation of GTFP draws on existing research [63,64], and this article employs a triad of input indicators comprising labor, capital and energy, alongside a single output indicator in the form of GDP. The article goes on to identify industrial sulfur dioxide emissions, industrial soot emissions and industrial wastewater emissions as indicators of undesirable output.

The results presented in column (1) of Table 5 demonstrate that the impact of LCCP on IGG remains statistically significant at the 1% level, following the replacement of the dependent variable. This finding serves to indicate the robustness of the benchmark regression.

5.3.2. Incorporating Province and Time Interaction Fixed Effects

As provinces represent pivotal administrative nodes within the Chinese government’s governance structure, it is conceivable that they may exhibit distinct characteristics, including levels of economic development, policy environments, and cultural backgrounds. These factors have the potential to influence the explanatory variables. In addition, it is important to note that differing macroeconomic environments, policy changes, and other factors may have a significant impact on different periods. Consequently, this article incorporates province–time interaction fixed effects into the baseline regression, thereby ensuring the control of the impact of different provinces over time. The detailed regression results are displayed in column (2) of Table 5. The regression results indicate that, upon consideration of the correlation effect of disparate city characteristics within a single province, the impact of LCCP on IGG remains significantly positive at the 1% level. This finding corroborates the initial conclusion.

5.3.3. Excluding the Impact of Parallel Policies

Since 2010, the pilot policies of low-carbon cities have been launched one after another, and the related policies include the construction of Smart city and the pilot policy of carbon emissions trading, which were implemented in 2012 and 2013, respectively, according to which this article constructs the policy dummy variables of Smart city (Smartcity) and the pilot policy of carbon emissions trading (Carboncity) and adds them into control variables. The detailed regression results are shown in column (3) to column (5) of Table 5. Following the exclusion of the effects of the two parallel policies in the same period, the significance of the policy effect of the pilot low-carbon city policy remains unchanged. This serves to illustrate the robustness of the conclusions of this article.

5.3.4. Resetting the Double Machine Learning Model

To mitigate the bias in the dual machine learning model setup, this study examines the robustness of the findings by adjusting the sample partition ratio from 1:4 to 1:2 and 1:7. Additionally, different machine learning algorithms (Lasso), Gradient boosting (Gradboost), Support vector machines (SVM) are used instead of the previous random forest algorithm for prediction. Once again, the baseline regression based on double machine learning to construct the partially linear regression model (PLR) for analysis, the model form setting is not immune to subjectivity. The following part of this article employs double machine learning to construct a more general interactive regression model (IR) to explore the impact of the model setting on the conclusions of the article. The main regression and auxiliary regression changes are as follows.

$$\begin{array}{c}{\mathit{IGG}}_{i,t}=g\left({\mathit{LCCP}}_{i,t},X_{it}\right)+U_{it},E\left(U_{i,t}|X_{i,t},{\mathit{LCCP}}_{i,t}\right)=0 \\ {\mathit{LCCP}}_{i,t}=m\left(X_{it}\right)+V_{it},E\left(V_{i,t}\mid X_{i,t}\right)=0\end{array}$$

(9)

where the coefficient that explains the variable ${\mathit{LCCP}}_{i,t}$ is ${\widehat{\theta }}_0=E\left[g\left({\mathit{LCCP}}_{i,t}=1,X_{it}\right)-g\left({\mathit{LCCP}}_{i,t}=0,X_{it}\right)\right]$, and the other relevant variables are consistent with Equation (1).

Finally, despite the fact that this article has attempted to take into account as many factors affecting urban inclusive green growth as possible, due to limitations in the data available, there are inevitably omitted variables. Furthermore, regression analysis faces endogeneity problems, and the instrumental variable method can effectively alleviate the endogeneity problem. Accordingly, this article constructs a partially linear instrumental variable model for dual machine learning based on Chernozhukov et al. [43] with the following settings.

$$\begin{array}{c}{\mathit{IGG}}_{i,t}={\theta }_0{\mathit{LCCP}}_{i,t}+g\left(X_{it}\right)+U_{it}\\ { \mathit{Instrument} }_{it}=m\left(X_{it}\right)+V_{it}\end{array}$$

(10)

where ${ \mathit{Instrument} }_{it}$ is the instrumental variable for ${\mathit{LCCP}}_{i,t}$, which refers to the relevant research [65]. This instrumental variable utilizes the interaction term between the degree of urban terrain relief and the time trend term (IV), which satisfies the exogenous and correlation assumptions for instrumental variables.

Table 5 presents the regression results after resetting the dual machine learning model. It is evident that factors such as the sample partition ratio, the machine learning algorithm used, the model estimation form, and the endogeneity issue do not affect the conclusion that the LCCP promotes IGG in the city. These factors only change the policy effect’s magnitude to some extent, demonstrating the robustness of the original conclusions.

6. Mechanism Identification

The above discussion demonstrates that the LCCP has a significant positive impact on IGG. Understanding the mechanism through which this impact is transmitted and identifying the pathway of the impact is beneficial for better comprehending how the Low-Carbon City Pilot Policy can contribute to inclusive green growth in urban areas. This article will analyze the potential pathways of the policy’s effects in terms of technological innovation, entrepreneurship, and labor structure upgrading. To examine these mechanisms, this article adopts the causal mediation effect analysis of DML as outlined by Farbmacher et al. [66] with the specific test results presented in Table 6. Furthermore, this article continues to utilize the DML method to test the three transmission pathways, with the specific test results depicted in Table 7.

Table 6. The first result of mechanism tests.

VARIABLES	Total Effect	Disposal Group Direct Effect	Control Group Direct Effect	Disposal Group Indirect Effect	Control Group Indirect Effect
Patent	0.320 *** (0.030)	0.260 *** (0.020)	0.270 *** (0.020)	0.050 *** (0.010)	0.060 ** (0.030)
Gpatent	0.320 *** (0.030)	0.250 *** (0.020)	0.260 *** (0.020)	0.060 *** (0.010)	0.070 ** (0.030)
Entrep	0.320 *** (0.030)	0.250 *** (0.030)	0.290 *** (0.030)	0.040 *** (0.010)	0.070 ** (0.020)
HL	0.320 *** (0.030)	0.290 *** (0.030)	0.300 *** (0.030)	0.020 *** (0.010)	0.030 (0.020)

** p < 0.05, *** p < 0.01.

Table 7. The second result of mechanism tests.

VARIABLES	Patent	Gpatent	Entrep	HL	IGG
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
LCCP	0.732 *** (0.234)	0.074 ** (0.032)	0.161 *** (0.056)	0.059 *** (0.000)	0.103 *** (0.033)	0.107 *** (0.034)	0.110 *** (0.035)	0.116 *** (0.034)
Patent					0.028 *** (0.003)
Gpatent						0.229 *** (0.023)
Entrep							0.076 *** (0.025)
HL								0.115 *** (0.039)
Constant	−0.045 (0.043)	−0.004 (0.006)	0.001 (0.010)	0.006 (0.358)	0.003 (0.007)	0.003 (0.008)	0.002 (0.008)	0.002 (0.008)
The primary term for the control variable	YES	YES	YES	YES	YES	YES	YES	YES
Quadratic term for the control variable	YES	YES	YES	YES	YES	YES	YES	YES
Time fixed effect	YES	YES	YES	YES	YES	YES	YES	YES
Urban fixed effect	YES	YES	YES	YES	YES	YES	YES	YES
N	3948	3948	3948	3948	3948	3948	3948	3948

** p < 0.05, *** p < 0.01.

The second and third rows of Table 6 illustrate the estimation results with technological innovation (Patent) and green technological innovation (Gpatent) as mediating variables, respectively. The findings suggest that the indirect effects of technological innovation and green technological innovation in both the treatment and control groups are significantly positive. Moreover, the direct effects in the treatment and control groups remain positive following the exclusion of technological innovation pathways and the passing of the test of significance. This confirms that the LCCP can significantly contribute to IGG by enhancing innovation levels, thus validating Hypothesis 2. The LCCP is an organization that fosters the enhancement of policies and regulations. The purpose of this is to create a favorable institutional environment for technological and green technological innovation. Incentives are implemented by policymakers with the aim of encouraging enterprises and research institutions to invest in the development and application of green technologies. The incentives may take the form of tax breaks, subsidies and technology research and development funds. Consequently, the policy has been shown to have a dual impact: on the one hand, it has been demonstrated to drive technological advancement; on the other, it has been shown to promote inclusive economic growth. This has allowed a broader segment of society to benefit from the development of the green economy.

The fourth row of Table 6 presents the estimation results with entrepreneurial vitality (Entrep) as the mediating variable. The findings suggest that the indirect impact of the LCCP on IGG via entrepreneurial vitality is substantial and favorable in both the treatment and control groups. Furthermore, the direct impact of LCCP on urban IGG is also found to be positive. These findings lend support to Hypothesis 3, which suggests that LCCP can enhance IGG by fostering entrepreneurship. The LCCP is an organization that promotes the innovation and development of emerging enterprises in the low-carbon technology sector. It achieves this by providing support for green entrepreneurship.

The fifth row of Table 6 illustrates the estimation results with labor structure upgrading (HL) as the mediating variable. Overall, the impact of the indirect effect in the treatment group is significantly positive, suggesting that the LCCP can enhance IGG by promoting labor structure upgrading, thereby confirming Hypothesis 4. The LCCP fosters close collaboration between educational institutions and industries to advance labor capabilities. The policy encourages universities and vocational schools to establish partnerships with green enterprises to jointly develop curricula and training programs, and to provide internships and employment opportunities. This industry–academia collaboration model not only enhances the quality of education but also ensures that educational content aligns closely with market needs, thereby effectively improving the employability and adaptability of the workforce. Additionally, for the control group cities, the indirect effect of labor structure upgrading is greater than 0 but not significant. The LCCP typically includes specialized green skills training programs and educational courses that equip the workforce with knowledge and skills in areas such as renewable energy technologies, energy efficiency management, and environmental engineering. In cities without these policy supports, the workforce lacks access to these specialized skills, making it challenging to meet the demands of a green economy.

Moreover, the second test results in Table 7 indicate that the effects of LCCP on the three mechanism variables—technological innovation, entrepreneurial vitality, and labor structure upgrading—are all significantly positive, suggesting that LCCP is conducive to increasing the level of technological innovation, enhancing entrepreneurial activity, and promoting labor structure upgrading in the city. Through further testing, Models (4) to (6) are employed to reaffirm the role of the transmission mechanism of the mediating variables, thereby successfully verifying our hypothesis.

7. Heterogeneity Analysis

It has been verified above that the LCCP significantly improves IGG, but due to the uneven economic and social development across China’s regions, there are still notable differences among various pilot cities. Therefore, this article presents a further analysis of the heterogeneous effects on IGG in different pilot cities from the perspectives of environmental institutional foundations, industrial base, and resource endowments, which are closely related to low-carbon pilot cities.

It is evident that while certain cities may have well-established environmental protection mechanisms, others may be in the nascent stages of addressing environmental issues. This discrepancy has a direct impact on the effectiveness of low-carbon policies and the extent to which inclusive green growth can be achieved. In order to analyze the effect of heterogeneity in the environmental institutional foundations, this article employs a classification system based on the National Environmental Protection Eleventh Five-Year Plan issued by the State Council in 2007. This system categorizes Chinese cities into two groups: key environmental protection cities and non-key environmental protection cities. Secondly, the characteristics of industrial structure have been demonstrated to have a significant impact on urban ecological governance and economic growth. The high levels of energy consumption and emissions that are characteristic of production sectors dominated by heavy industry have the potential to impede progress towards green development. In order to examine the impact of LCCP on IGG under different industrial bases, this article is based on the National Old Industrial Base Adjustment and Rehabilitation Plan (2013–2022). The study sample is divided into old industrial base cities and non-old industrial base cities. It is evident that cities with abundant resources frequently depend on a single mode of production, which exerts a long-term lock-in effect on crude economic growth characterized by high inputs. This is not conducive to the process of industrial transformation and upgrading, and it results in low economic efficiency. This, in turn, has the potential to cause the city to fall into the trap of the “Resource Curse” [67]. Accordingly, the National Plan for Sustainable Development of Resource-Based Cities (2013–2020) stipulates the division of the study sample into two distinct categories: resource-based cities and non-resource-based cities. A subsequent analysis is then conducted to examine the differentiated impacts of LCCP from the perspective of resource endowment.

As shown in column (1) of Table 8, the LCCP exerts a more substantial influence on IGG in key environmental protection cities than in non-key environmental protection cities. It may be posited that this is due to the fact that key cities in the field of environmental protection generally possess a greater quantity of resources and a more robust capacity for the execution of policy, in addition to the provision of technical support and financial investment. Moreover, these cities have accumulated more experience and data on environmental governance during the planning cycle, providing a more reliable analytical basis for understanding the long-term impacts of policy effects.

Table 8. The result of heterogeneity tests.

VARIABLES	IGG
	(1)		(2)		(3)
	Environmental	Non-Environmental	Industrial	Non-Industrial	Resource	Non-Resource
LCCP	0.353 *** (0.0661)	0.0779 ** (0.0378)	0.111 * (0.0673)	0.230 *** (0.0450)	0.103 ** (0.0479)	0.197 *** (0.0494)
Constant	−0.00649 (0.0140)	0.00168 (0.00885)	−0.000661 (0.0122)	−0.00917 (0.0102)	0.000267 (0.0109)	−0.00445 (0.0106)
The primary term for the control variable	YES	YES	YES	YES	YES	YES
Quadratic term for the control variable	YES	YES	YES	YES	YES	YES
Time fixed effect	YES	YES	YES	YES	YES	YES
Urban fixed effect	YES	YES	YES	YES	YES	YES
N	1554	2394	1316	2632	1582	2366

* p < 0.1, ** p < 0.05, *** p < 0.01.

As demonstrated in column (2) of Table 8, the LCCP exhibits a more substantial promotion effect on IGG in non-old industrial base cities. This phenomenon may be attributed to the predominance of heavy industries and manufacturing in traditional industrial base cities. These sectors are characterized by high energy consumption, significant pollution levels, and considerable complexity and expense in terms of transformation. By contrast, non-old industrial base cities may be characterized by a preponderance of relatively light-polluting industries.

Column (3) of Table 8 shows that the LCCP contributes more significantly to IGG in non-resource-based cities than in resource-based cities. This may be because resource-based cities tend to achieve economic growth through factor inputs and expansion. This leads to stronger path dependence and a relatively homogeneous and consolidated development approach, which does not encourage technological advancement or improve energy efficiency. In contrast, non-resource-based cities experience faster development of the tertiary industry. This enables them to accelerate the transformation of urban development modes and achieve inclusive green growth with the help of digital technology [56].

8. Conclusions and Policy Suggestions

The worsening issues of global climate change and environmental pollution have become pressing challenges shared by all humankind. Governments and international organizations have implemented measures to reduce greenhouse gas emissions and promote sustainable development. As the largest emitter of greenhouse gases globally, China faces immense environmental pressure and international responsibility in terms of emission reduction. In this context, this article has developed the inclusive green growth level of 282 prefecture-level cities in China from 2008 to 2021 across the dimensions of economic development, social equity, green development, and environmental protection. It has empirically investigated the impacts of China’s LCCP on IGG through quasi-natural experiments using the DML method, supplemented by robustness tests, mechanism analyses, and heterogeneity analyses. These findings offer insights for achieving sustainable development. The specific findings of the study are summarized as follows.

Firstly, the empirical results indicate that China’s implementation of LCCP can significantly promote regional IGG, and this conclusion remains robust after tests such as substituting explanatory variables and applying machine learning methods. Secondly, this article has verified the mechanisms of technological innovation, entrepreneurial activity, and labor structure upgrading through mechanistic analyses. The LCCP encourages enterprises to engage in green technological innovation by increasing financial support for low-carbon development and research, enhancing support for training low-carbon talent, and accelerating the establishment of a carbon financial market. These policy changes signal to entrepreneurs that entrepreneurship is feasible, thereby mobilizing potential entrepreneurs and increasing entrepreneurial activity. The growing number of green patents and pool of low-carbon talent provide a solid foundation for inclusive green urban development. Finally, the impact of the LCCP on IGG varies significantly across cities. Heterogeneity analysis reveals that the LCCP has a greater impact on IGG in key environmental protection cities than in non-key environmental protection cities. For cities with an old industrial base, the LCCP has a greater impact on IGG than in cities without an old industrial base. For resource-based cities, the LCCP has a greater impact on IGG than in non-resource-based cities. Overall, the LCCP significantly impacts ecological governance and sustainable development in China.

Based on these conclusions, the article makes the following policy recommendations: Firstly, financial support for the LCCP should be enhanced, particularly investment in green technology innovation. Research shows that technological innovation, particularly green innovation, is an important mechanism through which the LCCP promotes IGG. The government could promote technological innovation and the green transformation of enterprises further by setting up special funds to support projects in low-carbon technology R&D, application and talent development. Secondly, the government should actively promote the development of an entrepreneurial ecosystem in the city, particularly with regard to low-carbon and green technologies. Through policy incentives and entrepreneurship support programs, potential entrepreneurs can be encouraged to engage in entrepreneurial activities. The government could support start-ups in achieving breakthroughs in low-carbon technological innovation and green development by providing entrepreneurship training, incubator support and tax incentives, among other multifaceted measures. This would promote inclusive green growth in the regional economy. Thirdly, the government should formulate differentiated low-carbon development strategies tailored to different types of cities. In non-key environmental protection cities, infrastructure development and environmental governance capabilities should be strengthened, overall environmental quality improved, green buildings promoted, and the use of renewable energy encouraged, with the aim of reducing pollutant emissions. For cities with an old industrial base, the government should enhance the technological transformation and upgrading of traditional industries, encourage enterprises to introduce low-carbon technologies, promote industrial restructuring, facilitate the development of new green industries and reduce the negative environmental impact of traditional industries. For resource-based cities, the government should prioritize improving resource utilization efficiency, encouraging resource recycling and the development of renewable resources, promoting the transformation of resource-based industries towards greater efficiency and reduced environmental impact, and mitigating environmental damage during resource extraction and utilization. Meanwhile, the government should support these cities in developing alternative industries, such as tourism and services, to achieve sustainable, diversified economic development.

While this study is grounded in the Chinese context, its findings offer valuable insights for urban climate policy in other countries, particularly emerging economies. Firstly, the success of the LCCP demonstrates the efficacy of a “pilot-then-spread” governance approach. Instead of implementing a high-risk, nationwide policy at once, policymakers can designate pilot cities to test and refine low-carbon strategies, creating a learning cycle that minimizes costs and maximizes effectiveness before broader rollout. Secondly, our mechanism analysis confirms that technological innovation and talent cultivation are universal drivers of the green transition. This underscores the need for all governments to integrate science, technology, and human capital development into the core of their climate action plans. Finally, the heterogeneous effects remind policymakers that “one-size-fits-all” strategies are ill-advised. The development of differentiated low-carbon strategies, tailored to a city’s industrial base, resource endowment, and development stage, is crucial for a just and effective transition. Therefore, the key lesson from China’s LCCP is not to replicate its specific measures, but to adopt its core principles: experimental governance, sustained investment in innovation, and policy flexibility.