The Impact of New-Type Urbanization Policy on Urban Green Total Factor Productivity: New Evidence from China

Abstract

China’s new-type urbanization has been instrumental in fostering high-quality economic and social progress. This research explores the impact and underlying mechanisms of the new-type urbanization policy on urban green total factor productivity (GTFP) by analyzing a sample of 165 Chinese prefecture-level cities from 2009 to 2020. Utilizing the difference-in-differences (DID) approach, the study uncovers several key findings: (1) compared to non-pilot areas, the new-type urbanization policy significantly boosts urban GTFP by 43.5%, supporting a shift towards more sustainable and eco-friendly urban growth; (2) the analysis of impact mechanisms indicates that these policies enhance urban GTFP through technological innovation and environmental regulation; and (3) the urban agglomeration model test reveals that being part of an urban agglomeration amplifies the effects of the new-type urbanization policy on GTFP. These results underscore the significance of these policies in driving sustainable and high-quality urban development in China, offering valuable insights for policymakers to formulate and implement effective urbanization strategies.

1. Introduction

As the world’s largest emerging economy, China’s urbanization process is unparalleled in size, speed, and scope. By the end of 2020, the urbanization rate had reached 63.89%, as reported by the Seventh National Census [1]. Rapid urbanization, fueled by investment growth, industrial concentration, and infrastructure construction, has significantly propelled China’s economic expansion [2]. However, traditional urbanization, dominated by extensive development, has led to expanding regional differences, environmental pollution, and resource depletion, causing uncontrolled urban expansion and degradation of land and water resources [3,4]. The energy- and pollution-intensive development mode has strained the resource and environmental capacity, severely limiting the potential for green city development [5,6]. Moreover, with China’s urbanization still accelerating and a significant gap remaining with developed countries, the rising urban population can inevitably increase total energy consumption, posing challenges to sustainable city development. This would also jeopardize the pledged zero net carbon emissions by 2060, as promised by Chinese policymakers, including Chairman Xi [7]. Given the above backdrop, exploring an urbanization path incorporating green and sustainable development principles is crucial for China’s economy and provides valuable insights into global city sustainability efforts.

China has introduced the National New-Type Urbanization Plan (2014–2020) (hereafter, the Plan) to address its development [8]. The new-type urbanization policy in China refers to a set of strategies and approaches introduced by the government in 2014 to promote sustainable and balanced urbanization. These policies address the challenges and imbalances associated with previous urbanization efforts. Some key features of the new-type urbanization policy include: (1) people-centered development focused on improving the quality of life for urban residents and ensuring their access to essential services like healthcare, education, and social security; (2) ecological and green development, with emphasis placed on environmental protection, resource conservation, and the development of eco-friendly urban infrastructure; (3) balanced rural-urban development aiming to bridge the urban-rural divide by investing in rural infrastructure, supporting rural industries, and promoting rural-urban integration; and (4) social integration and inclusiveness, with efforts made to address issues related to migrant workers, such as improving their access to public services, education, and healthcare [9]. Compared to urbanization policies before 2014, the new-type urbanization policy reflects a shift in priorities. These policies emphasize sustainable development, environmental protection, social equity, and the well-being of urban residents. The focus is on rapid urban expansion while creating livable, harmonious, and inclusive cities. China’s new urbanization strategy, launched in 2014 as a comprehensive pilot project, serves as a quasi-experimental setting for research purposes.

This study examines the effects of China’s new urbanization policy on urban green total factor productivity (hereafter, GTFP) and its mechanisms, which is crucial not only for China’s economic transformation and urban green development but also holds potential for global urban development. The results could interest other developing countries grappling with similar urban challenges (e.g., India, Indonesia, and Vietnam). Furthermore, the study fills a literature gap by analyzing China’s policy for new urbanization and its influence on urban GTFP. By utilizing the Difference-in-Differences (hereafter, DID) approach, we seek to offer a more robust evaluation of the impacts of pilot initiatives on urban GTFP, thereby providing valuable insights for policymakers worldwide.

The study is organized into six main sections. Following the introduction, Section 2 reviews the relevant literature, and Section 3 presents the research hypotheses. Section 4 meticulously describes the rigorous methodology and comprehensive data employed in the study, ensuring the reliability and validity of the research. Section 5 reports the empirical results, and Section 6 concludes the study and provides policy recommendations.

2. Literature Review

While conducive to rapid growth, the GDP-only assessment model has led to increasingly severe ecological and environmental problems. The 14th Five-Year Plan highlights accelerating the green transformation of China’s development model while promoting high-quality economic growth [10]. Since its introduction by Solow [11], the concept of total factor productivity (hereafter, TFP) has been widely adopted in academia to measure economic growth and development quality. However, TFP falls short in accounting for resource consumption and environmental pollution.

In contrast, GTFP, a comprehensive indicator that considers economic and environmental performances, aligns more closely with the principles of sustainable development, making it a crucial tool in empirical research. [12]. GTFP measures resource utilization efficacy and environmental performance in production, evaluating environmental considerations. Moreover, it assesses the ability of an economy to generate output while minimizing resource input and negative environmental impacts. Specifically, the calculation of GTFP involves measuring the productivity growth rate of factors of production while accounting for their energy consumption, emissions, and other environmental factors [13]. As the primary spatial carriers of economic development and population agglomeration, cities are also the primary sources of greenhouse gases and environmental pollution [14]. Therefore, balancing economic growth and environmental improvement and promoting sustainable economic development driven by urban GTFP growth is crucial to meeting people’s demands for a better quality of life [15].

Few studies have investigated the association between the new-type urbanization policy and GTFP. For instance, Li et al. [16] investigated the effect of policies on urban innovations using DID, Propensity Score Matching (hereafter, PSM)-DID, and other methods. Their findings show that the policy can significantly increase urban innovation capacity. Shang et al. [17] used provincial panel data and the improved entropy method to calculate the level of new-type urbanization. They found that new-type urbanization significantly influenced GTFP, positively or negatively impacting green technology progress (efficiency).

Recent studies have also begun to explore the policy effects of new-type urbanization on the quality of urban economic growth. For instance, Jiang et al. [18] concluded that traditional factors still predominantly drive the high-quality growth facilitated by new-type urbanization, with innovation and industrial structure playing relatively minor intermediary roles. Cheng et al. [19] constructed an urban-rural dual general equilibrium model incorporating environmental factors to theorize the impacts of urbanization on economic development and empirically tested the intermediary role of industrial structure upgrading. Gao et al. [20] explored the evolving coupling relationship between new-type urbanization and economic development, finding a shift from synergy to antagonism.

Additionally, the research on the green development aspect of new-type urbanization has focused on two main areas: (1) the effect on the environment and (2) the impact on energy utilization efficiency. Through the application of the DID method, Chen et al. [21] verified that new-type urbanization can significantly enhance the regional ecological environment. This is achieved through industrial structure transformation and upgrading, enterprise technological innovation, and the fortification of environmental regulations. Fan et al. [22] found that energy efficiency improvement mainly stems from innovation vitality, environmental improvement, and the gradient transfer of industrial structure. However, Feng et al. [23] argued that the new-type urbanization construction reduced urban energy utilization efficiency primarily because it failed to address the boundary between the government and the market appropriately. They suggested that government intervention should be appropriate while respecting market autonomy. Zhang et al. [24] employed the DID approach, kernel density function, and intermediary effect model to analyze the policy effect, spatial differentiation, and action path of the policies on urban land green usage efficiency.

The existing literature predominantly examines urbanization’s economic growth or green development aspects separately, with limited attention to the comprehensive impact of new-type urbanization on both areas. To address this research gap, our study employs GTFP as the critical variable capturing the interplay between economic growth, resource efficiency, and environmental sustainability. Furthermore, most literature focuses on GTFP at the provincial level, with a notable lack of city-level analysis. However, as the primary focus of new-type urbanization construction is at the city level, provincial-level analyses may introduce bias. Therefore, our study employs the DID method to explore the effects of the new-type urbanization policy on urban GTFP and its mechanisms. It aims to enrich understanding of the interplay between new-type urbanization and GTFP and provide insights for fostering green, high-quality urban development.

3. Hypotheses Development

The new type of urbanization is a scientific development path tailored to China’s national conditions. It complements and optimizes traditional urbanization by introducing higher and more systematic requirements for the quality of urban development. This form of urbanization refers to simple population aggregation and spatial expansion, encompassing comprehensive development across multiple dimensions, such as the economy, society, transportation, environment, and resources [9]. At its core, new-type urbanization prioritizes people-oriented development, emphasizing the coordinated progress of population, economy, land, and culture in space. It also places a strong emphasis on pursuing sustainable development goals like resource conservation, environmental friendliness, ecological diversity, and climate change adaptation [25]. Specifically, it aims to prioritize people’s quality of life and integrate rural migrants into urban society; promote green development and the economic and intensive use of resources; support coordinated urban development with urban agglomeration as the primary form; and bridge the urban-rural gap through shared infrastructure and equalized public services [9].

The primary manifestation of urbanization is the spatial migration and agglomeration of production factors. The impact of population and industrial agglomeration brought by urbanization on GTFP depends on the comprehensive effects of the agglomeration and crowding-out effects. Sensible population agglomeration can promote the continuous expansion of consumer demand, industrial scale, and regional high-quality development through knowledge spillover and the scale effect. However, the crowding-out effect caused by excessive agglomeration increases urban living costs and pressure, which is not conducive to regional economic development [26,27]. Drawing insights from the new economic geography framework, industrial agglomeration facilitates talent flow, knowledge spillover, and infrastructure sharing, reducing production and transaction costs and continuously improving production efficiency through scale effects. Yet, excessive industrial agglomeration may lead to “congestion” effects in terms of traffic congestion, resource supply, demand imbalance, and low management efficiency, all of which negatively impact GTFP [28].

On another note, the Plan regards ecological civilization and green and low-carbon construction as foundational to new-type urbanization. Saving resources and strengthening environmental protection can significantly promote GTFP growth. Furthermore, as urban residents’ income levels improve, their demands for quality of life and better public services increase, along with their awareness and practice of green consumption, which further enhances urban GTFP. In this study, we propose the following:

Hypothesis 1.

The new-type urbanization policy directly increases GTFP in cities.

Endogenous growth theory suggests that R&D capital and personnel inputs positively influence technological innovation [29]. In contrast, new economic geography theory posits that technological innovation and knowledge spillover require a reasonable spatial layout as carriers [30]. The accumulation of production factors induced by new-type urbanization provides talent and capital for technological innovation enhancement. Additionally, the information exchange network and infrastructure associated with urbanization offer channels for innovation diffusion [31]. On the one hand, the Plan encourages leveraging urban areas as innovation hubs to foster urban innovation and development by capitalizing on resource advantages. To elevate urbanization quality, pilot cities can boost innovation investment, promote an innovation-friendly environment, and energize societal innovation vitality. Technology innovation can effectively reduce energy consumption per unit output by improving resource allocation efficiency, thus improving GTFP [32].

Moreover, innovation’s primary agent, i.e., people and cities’ quality educational resources, contribute to human capital accumulation. New-type urbanization can promote human capital accumulation through the “labor attraction” and “human capital spillover” effects [33]. Moreover, urbanization can boost the spatial agglomeration of human and financial capital and the diffusion spillover effect of knowledge and technology. This, in turn, enhances the efficiency of production factor allocation, subsequently increasing urban GTFP. Nie et al. [34] utilized the quasi-natural experiment of innovative city construction to explore the green development externalities of innovative city construction. The authors used the DID model and found evidence of significant improvements in cities’ technological innovation capabilities and human capital levels. This, in turn, fosters GTFP improvement through enhanced resource allocation efficiency and the promotion of sustainable production technology. Therefore, we consider the technological effect as an intermediary mechanism through which new-type urbanization enhances GTFP.

The new development concept of urbanization has guided the reform of the original industrial structure, resulting in a “selection” effect that phases out extensive industries and accumulates high-tech industries, optimizing the industrial structure and enhancing market vitality [35]. With the expansion of urban labor demand and the urban-rural income gap, the rural labor force continues to migrate to cities where they are redeployed in urban and service industries, thus promoting their development. Additionally, from a demand perspective, the increasing consumption demand for entertainment among residents continuously propels the tertiary industry service sector. Compared to the industrial sector, the modern service industry, represented by transportation, finance, education, and entertainment, features low pollution emissions, low energy consumption, and a high resource allocation rate [36]. The shift in industrial structure enhances factor production efficiency and reduces energy consumption [37], benefiting GTFP improvement. Li [38] asserted that industrial structure transformation and upgrading facilitate regional GTFP enhancement, which supports the regional economy’s high-quality development. Thus, industrial structure improvement plays a more significant role than its rationalization. Given the above backdrop, upgrading industrial structures is considered an intermediary mechanism through which new types of urbanization boost GTFP.

While traditional urbanization blindly pursues expansive development, new-type urbanization pursues high-quality urban development. It emphasizes the importance of green development in achieving economic growth targets. Chen et al. [21] argued that the policy improves the regional ecological environment by reinforcing environmental regulation, which promotes energy savings and emission abatement through price and scale effects. The Plan advocates bolstering the environmental protection system by establishing an ecological civilization assessment mechanism, a compensation system, and a stringent environmental supervision framework.

Firstly, environmental regulations promote the improvement of GTFP via the “survival of the fittest” effect. They encourage labor and capital to flow towards the green sector while gradually phasing out high-pollution, high-energy-consuming industries. At the same time, as consumer environmental awareness rises, green products are more likely to gain consumer favor, whereas products failing to meet green standards face gradual elimination. Secondly, ecological regulation positively impacts GTFP through the “innovation compensation” effect. To comply with environmental regulations, enterprises innovate to abandon high-pollution, low-efficiency production modes, ultimately enhancing their production efficiency. Although production costs such as technological innovation and pollution control can be generated in this process, the delayed “innovation compensation” effect can offset the “cost” effect by improving environmental quality and increasing output [39]. Wen et al. [40] analyzed how environmental regulation affected GTFP with the DID approach, discovering a significant increase in GTFP within the region post-policy implementation, evidencing a time-lag effect. Given the above backdrop, environmental regulation is identified as an intermediary mechanism by which new-type urbanization impacts urban GTFP.

Drawing from existing theories, this study posits that the new-type urbanization policy can affect GTFP through (1) technical, (2) structural, and (3) environmental regulation effects. Accordingly, the influence of the new-type urbanization policy on GTFP can be analyzed from these three perspectives. Accordingly, we hypothesize as follows:

Hypothesis 2.

The new-type urbanization policy enhances urban GTFP via three pathways: elevating technological innovation, upgrading and optimizing the industrial structure, and intensifying environmental regulation efforts.

4. Research Design

4.1. Model Design

This study takes the new-type urbanization policy as a quasi-natural experiment based on the pilot areas in the Plan for New-Type Urbanization [41]. It uses the DID model to investigate the impact of the pilot policy on GTFP. The DID method is an effective research tool for assessing the effects of new policies by considering them as exogenous “natural experiments”. This approach is advantageous over other policy evaluation methods as it helps eliminate the common trends between the experimental and control groups, mitigating issues related to omitted variables and endogeneity in regression analysis [23].

We treat the initial batch of pilot areas for new-type urbanization announced in 2014 as the experimental group. To ensure research consistency and result accuracy, we focus on prefecture-level cities, excluding the four directly administered municipalities and cities where the first batch of pilot counties is located [42]. Subsequent batches of pilot regions unveiled in 2015 and 2016 and cities affected by administrative adjustments, such as Sansha, Haidong, and Chaohu, are also excluded from the analysis.

Accordingly, the final sample consists of 165 prefecture-level cities, shown in Figure 1, with the experimental group comprising 58 cities, including the first batch of pilot areas in two provinces, three planned cities, and 25 prefecture-level cities. The remaining 107 entities serve as the control group. Since the first batch of policies was decreed in December 2014, 2015 has been taken as the implementation point of the policies, and 2009–2020 has been selected as the research interval. The specific model (1) is set below.

$${GTFP}_{it}={\beta }_0+{\beta }_{1}DID+\sum {\gamma }_j{Controls}_{it}+{\mu }_i+{\delta }_t+{\epsilon }_{it}$$

(1)

where ${GTFP}_{it}$, represents the ith city’s green and high-quality development level in the period $t$. DID is the new-type urbanization pilot policy explanatory variable. ${Controls}_{it}$, ${\mu }_i$ and ${\delta }_t$ represent the set of six control variables, individual and time-fixed effects. ${\epsilon }_{it}$, represents the disturbance terms.

4.2. Variables

4.2.1. Explanatory Variable

Similar to Chen [43], this study calculates the city-level GTFP based on the DDF-GML (Directional Distance Function—Global Malmquist Luenberger) model. The DDF model encourages the expected output to expand to the production frontier and promotes the non-expected output to shrink to the minimization frontier, which conforms to the sustainable production process development concept. Therefore, we employ the DDF model with non-expected output to estimate the green efficiency of 165 Chinese cities between 2009 and 2020, combined with the GML index approach to measure these cities’ GTFP. The GML index represents the growth rate of GTFP. Specifically, we use the data from 2008 to 2020 to obtain the GML index from 2009 to 2020. Furthermore, we employ the GML index from 2008 as the base period data to receive the cumulative GTFP data from 2009 to 2020 through year-by-year multiplication [43]. The input and output variables involved in the measurement of GTFP are shown in

Table 1. GTFP input-output indicators.

Factor Categories	Indicators	Indicator Description
Input	Capital	The urban fixed capital stock is calculated using the perpetual inventory method [44], with 2003 as the base period.
	Labor	Utilize the year-end urban employment figures, encompassing the number of individuals employed by various units and those engaged in private and individual employment within cities.
	Energy	Energy data on coal, oil, and gas at the city level. We use the electricity consumption of the whole society as a proxy variable.
Expected output	Real GDP	The GDP index is converted to obtain gross domestic product (GDP) at constant prices, with 2003 as the base period.
Undesirable outputs	Pollutant discharge	Including urban industrial wastewater discharge, industrial sulfur dioxide discharge, and industrial dust discharge.

.

The base period for the capital calculation and GDP is set to 2003 for two reasons. First, as suggested by Zhang et al. [44] in their influential study on estimating China’s provincial capital stock, choosing an earlier base year helps to minimize the impact of potential errors in the initial capital stock estimate on the subsequent years’ calculations when using the perpetual inventory method. Second, complete data for all the cities in our study was only available starting in 2003, ensuring consistency and reliability in our calculations.

4.2.2. Core Explanatory Variables

We use a dummy variable that represents the comprehensive pilot policy variable for new-type urbanization, defined by the interaction term of Treat and T. Treat indicates whether a city is selected for the list of national new-type urbanization comprehensive pilot areas. Specifically, if selected, Treat = 1; otherwise, it equals 0. T signifies the temporal aspect of the pilot areas’ selection, acting as a virtual variable. Since the inaugural batch commenced their pilot programs before the end of 2014, 2015 serves as the demarcation point. Accordingly, before 2015, T = 0, whereas from 2015 onwards, T = 1.

4.2.3. Influencing Mechanism Variables

The first influencing mechanism variable, innovation capability, is assessed as in Nie et al. [34], who adopted the urban innovation index developed by the Industrial Development Research Center of Fudan University. The second variable is upgrading the industrial structure, encompassing rationalization and enhancement aspects. Rationalization is the dynamic adjustment process aiming for a more rational distribution of factors across various industries based on current technological levels. Specifically, we use the Thiel Index to evaluate urban industrial structure rationalization, which quantifies structural deviations in industrial output value and employment while reflecting the economic status of diverse industries [45]. As for industrial structure optimization, it entails regular shifts in the proportional representation of the primary, secondary, and tertiary sectors alongside labor productivity, measured as the product-weighted value of these proportions [46]. The third influencing mechanism variable is the intensity of environmental regulation. Similar to Li et al. [47], we use data on industrial wastewater discharge, sulfur dioxide emissions, and smoke (or particulate matter) emissions to calculate the intensity of the environmental regulations index.

4.2.4. Control Variables

In line with the relevant literature [13,34], we select the following six control variables: economic development level (Eco), proxied by urban per capita GDP; fiscal decentralization (Gov), represented by the local finance budget’s revenue-to-expenditure ratio; openness to the outside world (Fdi), expressed as the ratio of foreign capital utilization to GDP (with foreign capital converted to RMB based on the average annual exchange rate); human capital (Hc), denoted by the number of higher education enrollees per 10,000 population; financial development (Fin), calculated as the year-end ratio of total financial institution deposits and loans to GDP; and infrastructure (Infra), proxied by per capita urban road area.

Economic and social variable data come primarily from the China Urban Statistical Yearbooks. Due to the absence of city-level GDP and fixed asset price indexes, provincial-level indexes are obtained from the China Statistical Yearbooks. Missing data are addressed through linear interpolation and growth rate projection. Furthermore, we apply logarithmic transformations to all control variables to mitigate potential heteroscedasticity problems.

5. Results and Discussion

5.1. Benchmark Regression

The outcomes of the regression analysis regarding the impact of new urbanization pilot initiatives on urban GTFP are detailed in

Table 2. Baseline estimation results.

	(1)	(2)	(3)
	GTFP	GTFP	GTFP
DID	0.594 ***	0.435 ***	0.435 ***
	(0.048)	(0.056)	(0.155)
Constant	2.253 ***	0.349	0.349
	(0.438)	(0.649)	(0.762)
Control variables	Yes	Yes	Yes
Urban FEs	No	Yes	Yes
Year FEs	No	Yes	Yes
Cluster to city	No	No	Yes
Obs.	1980	1980	1980
R2	0.161	0.556	0.556

Note: *** denotes coefficient significance at 1% level, corresponding to p-value of < 0.001. Heteroscedasticity has a robust standard error in parentheses.

. In particular, column (1) displays the findings of the mixed OLS regression, which does not include urban and year-fixed effects (hereafter, FEs). The coefficient on the DID variable is positive (0.594) and significant at the 1% level (p < 0.01), suggesting a substantial positive influence of the policy on urban GTFP, holding all other factors constant.

Subsequently, column (2) showcases the estimated results from a fixed effects model, while column (3) presents the results when controlling for all FEs and utilizing cluster-robust standard error. Due to that, the estimated results in column (3) are regarded as more precise. We can observe that in columns (2) and (3), the core explanatory variable’s (DID) estimated coefficients are significant at the 1% level, positive, and of identical size (0.435). The low p values (p < 0.01) associated with the DID variable provide strong evidence to reject the null hypothesis and support our central hypothesis (Hypothesis 1) that the policy has a significantly positive impact on urban GTFP. The inferred outcomes indicate that, on average, GTFP experiences an increase of 43.5% following the implementation of the new-type urbanization policy, underscoring the policy’s favorable impact on urban GTFP.

The p-values for the constant term in columns (2) and (3) are high, suggesting that the estimated constant is not statistically significantly different from zero at conventional significance levels. However, this does not affect the interpretation of the estimated coefficient on the DID term. Summing up, the results in Table 2 offer robust empirical support for Hypothesis 1 and indicate that, as expected, the influence of the policy is highly significant (statistically and economically) and positive.

While our model adequately captures the effects of the pilot policies and hypothesized mechanisms, as indicated by the R-squared values of 0.556 in columns (2) and (3), other factors not accounted for in our analysis may contribute to the variation in urban GTFP, such as local government policies, regional economic conditions, or industry-specific characteristics. Thus, future research could explore these factors to offer a more complete grasp of the determinants of urban GTFP in the context of policies.

5.2. Parallel Trend Test

The parallel trend assumption is fundamental when employing the DID approach for policy evaluation. This assumption requires that GTFP in both the pilot and non-pilot regions exhibit a similar trend before implementing the new-type urbanization policy. We construct model (2) following Beck et al. [48] to verify whether this prerequisite is met.

$${GTFP}_{it}={\theta }_0+\sum _{t=2011}^{2018}{\theta }_1{Treat\times T}_t+{\gamma }_j\sum {Controls}_{it}+{\mu }_i+{\delta }_t+{\epsilon }_{it}$$

(2)

where taking 2015 (when the pilot list was released) as the base year, the estimated value of the multiple difference item from 2009 to 2020 is the time dummy variable for each year as the policy impact, for example, T = 1 in 2011 and T = 0 in other years. The remaining variables are the same as in model (1). Figure 2 illustrates that, before introducing the policy, the estimated coefficients of the multiple difference terms do not exhibit statistical significance. This finding confirms the lack of notable disparity in the GTFP between the pilot and non-pilot regions, thus meeting the parallel trend assumption. On the contrary, after implementing the pilot policy, the regression analysis reveals a significantly positive coefficient. Such findings imply that the GTFP in the pilot areas experienced a substantial increase relative to the non-pilot regions after the policy was implemented.

5.3. Robustness Test

5.3.1. Placebo Test

Next, we perform a placebo test to assess if the estimated results in previous sections were affected by other unobservable factors. Specifically, we generate the interaction terms of pseudo-virtual policies for analysis to assess the presence of a policy effect [18]. If the policy effect persisted, it would suggest that the baseline regression results were unreliable. The five specific test steps are as follows: (1) randomly selecting 58 prefecture-level cities from the sample of 165 cities as the treatment group, with the remaining cities as the control group; (2) randomly picking a year between 2009 and 2020 as the policy time point; (3) repeating the two groups of random sampling 500 times; (4) multiplying the obtained new Treat and T variables to obtain 500 DIDs; and (5) conducting regression analysis according to the benchmark regression model.

Figure 3 depicts the outcomes of 500 pseudo-policy regression analyses. The horizontal axis denotes the regression coefficients derived from 500 randomly generated pseudo-policy dummy variables, while the right-hand vertical axis shows their corresponding p-values. Each circle dot signifies the p-value associated with a specific pseudo-regression coefficient, and the solid line illustrates the kernel density distribution of these coefficients. Moreover, the red vertical line at the rightmost end of the graph indicates the estimated coefficient obtained from the genuine pilot policy.

We can observe that most pseudo-regression coefficients are concentrated between −0.2 and 0.2, i.e., smaller than the estimates of the core explanatory variable in Table 2. Additionally, the p-values of most coefficients are greater than 0.1. Therefore, Figure 3 suggests that the baseline estimated results (Table 2) remain relatively unaffected by other policies or unobservable factors, thus further supporting Hypothesis 1.

5.3.2. PSM-DID Test

In an ideal natural experiment, the pilot area selection is independent of individual characteristics and other factors, thus avoiding missing or endogenous variable bias using the DID method. In practice, choosing new-type urbanization pilot areas is usually deliberate. For instance, it considers initial conditions such as prefecture-level cities’ economic, infrastructure, and financial market development levels. Due to the different initial conditions between the treatment group and the control group, there may be selection bias issues. The solution involves comparing the GTFP when the pilot policy is implemented in the same prefecture-level city but in different periods. Each prefecture-level city can only choose whether to implement or not the pilot policy during the same time frame. However, if an entity decides to implement the pilot policy, data may be missing in the dependent variables of non-pilot areas. Propensity Score Matching (hereafter, PSM) can address selection bias by enhancing sample randomness and mitigating systematic differences in observable variables [49]. Therefore, this study adopts the PSM approach to match and screen the experimental and control groups, followed by differential estimation.

Given the substantial disparities in the initial conditions among the sample prefecture-level cities, the experimental and control groups could not satisfy the common trend assumption, a prerequisite for adequately applying the DID model. This implies that, in the absence of pilot policies for new-type urbanization, there was little difference in GTFP over time between pilot and non-pilot cities. To ensure the results are based on comparable entities, the PSM procedure matches two prefecture-level cities with similar probabilities in the experimental and control groups, which helps alleviate the common trend problem.

An essential prerequisite for the PSM is a sizeable common support domain where the propensity score values of the experimental and control groups overlap substantially. Thus, we employ control variables as covariates in PSM matching. The results obtained using nearest-neighbor matching indicate that 1966 samples are within the standard value range, with only 14 remaining unmatched. Consequently, the sample loss is relatively small, and the influence on the estimation bias is insignificant.

Furthermore, a balance test is employed to assess the balance of variables after sample matching. Following Xie et al. [50], the logit regression results pre- and post-matching are compared. A decrease in the estimated coefficient’s value of each matching variable translates into statistical insignificance post- matching, suggesting no systematic bias in the matching terms between the experimental and control groups [50].

Table 3. Balance test results of control variables before and after the propensity score.

Variables	Matches	Experimental Group Mean	Control Group Mean	Bias	t-Value	p-Value
Gov	Pre-match	0.504	0.788	58.100	12.220	0.000 ***
	Post-match	0.504	0.480	5.000	0.850	0.395
Pgdp	Pre-match	11.061	10.802	43.100	9.340	0.000 ***
	Post-match	11.061	11.000	10.200	1.900	0.057 *
Fdi	Pre-match	3.162	4.343	81.800	16.740	0.000 ***
	Post-match	3.162	3.222	4.200	0.860	0.388
Hc	Pre-match	6.011	5.646	38.200	7.820	0.000 ***
	Post-match	10.949	10.852	2.000	0.410	0.679
Fin	Pre-match	1.488	1.396	19.200	3.980	0.000 ***
	Post-match	1.488	1.505	3.700	0.690	0.493
Infra	Pre-match	2.587	2.312	51.000	10.760	0.000 ***
	Post-match	2.587	2.570	3.300	0.600	0.547

Note: * and *** denote coefficient significance at 10% and 1% levels, corresponding to p-values of < 0.01 and < 0.001. Heteroscedasticity has a robust standard error in parentheses.

reveals that the absolute values of standardized deviations of covariates after matching are below 10%, and the mean differences of covariates before and after matching change from statistically significant to insignificant. This indicates that the mean values of covariates are evenly distributed post-matching, suggesting no systematic deviation between the two groups of matching variables, thus meeting the balance test requirements.

The matching effect is further assessed based on the experimental and control groups’ propensity score probability distribution density functions in Figure 4 and Figure 5. We can observe that the probability density distributions of the propensity scores for the treatment and control groups (Figure 5) exhibited more remarkable similarity than their pre-matching distributions (Figure 4). This observation confirms the efficacy of the proposed matching approach.

Column 1 of

Table 4. Test results of PSM-DID.

	(1)	(2)
	PSM Results	Change Sample Interval
DID	0.432 ***	0.390 ***
	(0.155)	(0.146)
Constant	4.677	0.998
	(5.157)	(0.701)
Control variable	Yes	Yes
Urban FEs	Yes	Yes
Year FEs	Yes	Yes
Cluster to city	Yes	Yes
Obs.	1966	1815
R2	0.557	0.541

Note: *** denotes coefficient significance at 1% level, corresponding to p-values of < 0.001. Heteroscedasticity has a robust standard error in parentheses.

exhibits that the coefficient on the DID estimator remains statistically significant and positive. Such a finding suggests that the pilot policy has a robust effect on enhancing urban GTFP to a certain degree, which adds another layer of robustness and confirms Hypothesis 1.

5.3.3. Change in the Sample Interval

The COVID-19 pandemic began in late 2019 and has substantially affected China’s economic landscape [51]. To account for the potential influence of the pandemic on the GTFP of cities, we conduct a robustness check by excluding 2020 from our analysis. This allows us to isolate the impact of the pilot policies on urban GTFP. The corresponding results in column (2) of Table 4 reveal a highly statistically significant and positive estimated coefficient on the DID term. This finding additionally strengthens the robustness of our baseline regression results and reinforces the validity of Hypothesis 1.

5.3.4. Adjustment of Time Window Width

To identify whether the effect of pilot policies on improving urban GTFP varies across different periods, this study further evaluates the robustness of the results by altering the time window width before and after the promulgation of the policies. Specifically, taking 2015 as the benchmark time node, samples of 1 to 5 years before and after were selected for the regression analyses, as shown in columns (1) to (5) in

Table 5. Adjusting the time window width.

	(1)	(2)	(3)	(4)	(5)
	One Year	Two Years	Three Years	Four Years	Five Years
DID	0.140 ***	0.239 ***	0.309 ***	0.393 ***	0.431 ***
	(0.051)	(0.090)	(0.114)	(0.143)	(0.154)
Constant	3.162	3.949	2.146	7.653	1.276
	(5.902)	(7.177)	(3.622)	(5.512)	(5.096)
Control variables	Yes	Yes	Yes	Yes	Yes
Urban FEs	Yes	Yes	Yes	Yes	Yes
Year FEs	Yes	Yes	Yes	Yes	Yes
Cluster to city	Yes	Yes	Yes	Yes	Yes
Obs.	495	825	1155	1485	1815
R2	0.848	0.704	0.652	0.612	0.588

Note: *** denotes coefficient significance at 1% level, corresponding to p-value of < 0.001. Heteroscedasticity has a robust standard error in parentheses.

. We can observe that regardless of changes in the time window width, the estimated coefficients on the DID variable remain significant and positive. This, in turn, corroborates the robustness of the results from benchmark regression and Hypothesis 1.

5.4. Influence Path Test

The empirical analysis presented above demonstrates that the pilot policies are conducive to improving urban GTFP, validating Hypothesis 1. To further test the internal transmission mechanism of this effect (Hypothesis 2), this study adopts a different approach from the traditional intermediary effect model, following Jiang [52].

Specifically, most studies use the stepwise regression method to test the intermediary mechanism [17,53]. However, this method may have problems with endogeneity, omitted variables, and a high correlation between explanatory and mechanism variables [52]. To mitigate these issues, we adopt [52]’s approach and focus on the policy (DID) and the intermediary ($M_{it}$) variables as per model (3) presented below.

$$M_{it}={\alpha }_0+\beta DID+\sum {\omega }_j{Controls}_{it}+{\mu }_i+{\delta }_t+{\epsilon }_{it}$$

(3)

where $M_{it}$ denotes a set of intermediary variables, including technological innovation (Innov), industrial structure upgrading (Es), industrial structure rationalization (Tl), and environmental regulation (Er). The coefficient β in model (3) represents the effect of the pilot policies on the intermediary variable. The remaining variables are consistent with model (1). If the coefficient $\beta$ is statistically significant, the intermediation effect holds.

To test Hypothesis 2, we first regress each intermediary variable ($M_{it}$) on the policy variable (DID) and control variables using model (3). This step allows for examining the impact of the policy on each intermediary variable separately. Next, we refer to the existing literature to determine the direction of each intermediary variable’s influence on GTFP. Suppose prior research established that an intermediary variable positively affected GTFP, and our analysis using model (3) showed that the pilot policies significantly influenced the intermediary variable. In that case, we could infer that the intermediary variable served as a transmission channel for the impact of the pilot policies on GTFP, as per Hypothesis 2.

Column (1) of

Table 6. Mechanism test of the impact of the new-type urbanization policy on urban GTFP.

	(1)	(2)	(3)	(4)
	Innov	Es	Tl	Er
DID	0.354 ***	0.034	0.039	0.015 **
	(0.067)	(0.030)	(0.114)	(0.007)
Constant	1.039 ***	0.526 ***	2.250 ***	0.103 ***
	(0.367)	(0.191)	(0.667)	(0.028)
Control variables	Yes	Yes	Yes	Yes
Urban FEs	Yes	Yes	Yes	Yes
Year FEs	Yes	Yes	Yes	Yes
Cluster to city	Yes	Yes	Yes	Yes
Obs.	1320	1980	1980	1980
R2	0.978	0.718	0.761	0.762

Note: ** and *** denote coefficient significance at 5%, and 1% levels, corresponding to p-values of < 0.05, and < 0.001. Heteroscedasticity has a robust standard error in parentheses.

shows that, compared with non-pilot areas, the new-type urbanization policy can significantly enhance the technological level of cities (Innov). Increased technological innovation, in turn, can reduce energy consumption per unit output and increase GTFP by improving resource allocation efficiency. On the contrary, columns (2) and (3) of Table 6 reveal that the (DID) estimated coefficients are statistically insignificant. This suggests that there is no empirical evidence that, in the short term, the new-type urbanization pilot policy has a discernible impact on the industrial structure’s upgrading (Es) and rationalization (Tl). It might be that even though the policies aim to optimize the urban industrial structure and encourage the transformation of polluting enterprises, upgrading the industrial structure is a process that requires time. Given the above backdrop, in the future, sustained efforts can be made to leverage these policies to optimize industrial structures further. The statistically significant (at the 5% level) estimated coefficient on the DID term in column (4) shows that the new-type urbanization policy significantly enhances the intensity of urban environmental regulations (Er) compared with non-pilot areas. This, in turn, means that environmental regulation can improve urban GTFP through the “survival of the fittest” effect and the “innovation compensation” effect.

5.5. Heterogeneity Tests

5.5.1. Heterogeneity Test of Urban Agglomerations

The Plan proposed optimizing urbanization’s spatial layout through urban agglomerations as the leading platform. Moreover, the 19th CPC National Congress report emphasized the construction of urbanization development patterns with urban agglomeration as the main field of action [54]. Therefore, in-depth research on the effectiveness of new-type urbanization construction under the leadership of the urban agglomeration model is warranted. Similar to Ouyang et al. [55], we selected the ten most representative urban agglomerations for heterogeneity analysis, namely the Beijing Tianjin Hebei, Liaozhong South, Yangtze River Delta, Strait West Coast, Yangtze River Middle Reaches, Shandong Peninsula, Zhongyuan, Pearl River Delta, Sichuan Chongqing, and Guanzhong urban agglomerations. Concurrently, the effectiveness of new-type urbanization construction in Chinese cities outside urban agglomerations is also examined.

Table 7. Heterogeneity analysis of urban agglomerations.

	(1)	(2)
	Within Agglomerations	Outside Agglomerations
DID	0.788 **	0.133
	(0.296)	(0.145)
Constant	1.923	0.196
	(1.934)	(0.775)
Control variables	Yes	Yes
Urban FEs	Yes	Yes
Year FEs	Yes	Yes
Cluster to city	Yes	Yes
Obs.	648	1332
R2	0.574	0.564

Note: ** denotes coefficient significance at 5% level, corresponding to p-value of < 0.05. Heteroscedasticity has a robust standard error in parentheses.

shows that under the urban agglomeration model’s dominance in Column (1), the coefficient on the DID term is substantially larger (0.788) than in baseline regression (Table 2), indicating that the policy has a more significant impact on GTFP within urban agglomerations. Such findings suggest that urban agglomerations are conducive to the free flow of resource elements, expanding urban development space, releasing development potential, and significantly improving resource utilization efficiency and economic and social development levels. Thus, they are advantageous to the green and high-quality development of cities.

5.5.2. Testing the Heterogeneity of Urban Characteristics

The above results show that the pilot policies significantly improve urban GTFP. However, each city has different resource endowments. Therefore, it is necessary to analyze whether the pilot policies impact urban GTFP and whether the enhancement effect changes under heterogeneous city characteristics. Accordingly, we test for the heterogeneity of three significant factors of urban development: human capital (Hc), financial development (Fin), and infrastructure (Infra) on the impact of new-type policies on urban GTFP. Specifically, we rerun the baseline model (1) on six subsamples, each based on the above- and below-average values of the three factors.

Table 8. Testing the heterogeneity of urban characteristics.

	(1)	(2)	(3)	(4)	(5)	(6)
	Low Hc	High Hc	Low Fin	High Fin	Low Infra	High Infra
DID	0.102	0.768 ***	0.084	0.728 ***	0.143	0.546 **
	(0.094)	(0.232)	(0.129)	(0.251)	(0.157)	(0.235)
Constant	0.418	1.118	0.235	0.884	0.378	0.474
	(0.914)	(1.401)	(1.004)	(1.074)	(0.821)	(1.533)
Control variables	Yes	Yes	Yes	Yes	Yes	Yes
Urban FEs	Yes	Yes	Yes	Yes	Yes	Yes
Year FEs	Yes	Yes	Yes	Yes	Yes	Yes
Cluster to city	Yes	Yes	Yes	Yes	Yes	Yes
Obs.	996	984	996	984	996	984
R2	0.560	0.582	0.598	0.558	0.525	0.574

Note: ** and *** denote coefficient significance at 5%, and 1% levels, corresponding to p-values of < 0.05, and < 0.001. Heteroscedasticity has a robust standard error in parentheses.

shows that only for samples of cities with above-average levels of (Hc), (Fin), and (Infra), the new-type policy has a statistically significant positive effect on improving urban GTFP. Such findings indicate that human capital, financial development, and infrastructure are all essential resource factors for the new-type urbanization to support the improvement of Chinese cities’ GTFP. Regarding human capital, the labor force with higher education levels has higher learning ability, which means higher innovation ability and environmental awareness conducive to improving GTFP. Furthermore, the development of the financial system can provide vital financial support for the construction of new-type urbanization. Consequently, in cities with high financial development, new-type urbanization significantly promotes GTFP. Additionally, well-developed infrastructure provides a more potent environment for the positive effect on urban GTFP.

5.6. International Comparison

Many nations have initiated policies focused on developing Smart Cities. For instance, in 2009, IBM formally presented the “Smart Planet” concept to the US government, advocating for investment in advanced intelligent information infrastructure [56]. The Singaporean government was among the first to implement a national strategy for becoming a “smart nation”, launching the “Smart Country 2015” and the “Smart State 2025” plans in 2006 and 2016 [57]. In 2016, Japanese policymakers introduced the “Science and Technology Basic Plan” to create a knowledgeable society [58]. Such initiatives have paved the way for significant advancements in developing smart cities among high-income economies.

However, compared to developing countries, the advancement of smart cities in developed nations is relatively straightforward due to their access to advanced technology, ample resources, and efficient urban planning strategies [59,60]. Moreover, the above-listed initiatives could not be directly applied in China due to the distinct political, cultural, social, economic, and natural environments compared to other countries [61].

Given the above backdrop, the new-type urbanization policy is a distinct set of initiatives to address the challenges of rapid urbanization in China, such as environmental degradation, resource depletion, and social inequality. These policies have been designed and implemented based on China’s unique socio-economic conditions and development goals. For instance, the new-type policies integrate smart technologies into urban planning and development but significantly emphasize equitable growth, reducing the urban-rural divide, improving public services, and ensuring that technological advancements benefit all urban residents. However, in “Western” cities, deploying advanced technologies like the Internet of Things (IoT), AI, and big data often prioritizes optimizing urban operations over broader social goals [62].

Another unique aspect of the Chinese approach is its emphasis on social equity, ensuring that urbanization benefits all citizens and reduces disparities. In contrast, many smart city initiatives in developed states are criticized for neglecting social equity and giving rise to gentrification, the digital divide, and unequal access to smart technologies, further exacerbating social inequalities [63,64]. Furthermore, the Chinese government plays a pivotal role in planning and implementing urbanization policies with a strong top-down, centralized approach. This, in turn, ensures coordinated development; however, it sometimes limits local innovation and flexibility. This contrasts with many Western countries (e.g., the UK and the US) with decentralized governance models, encouraging public-private partnerships and community involvement [65]. This approach can foster innovation and responsiveness to local needs but may result in uneven development and fragmented policy implementation.

6. Conclusions and Policy Implications

This research uses a difference-in-differences (DID) methodology and panel data on 165 Chinese prefecture-level cities to empirically examine the impact of the new-type urbanization policy on urban green total factor productivity (GTFP). Specifically, we focus on the list of pilot regions included in the national new-type urbanization comprehensive pilot program.

The three main findings are as follows: Compared to cities not covered by the policy, introducing new-type urbanization pilot initiatives significantly enhances the GTFP of the prefecture-level participating cities. Second, the path analysis indicates that new-type urbanization can enhance urban GTFP by fostering technological innovation and intensifying environmental regulation. However, the results show that the pilot policies do not substantially influence the optimization and improvement of the industrial structure. Consequently, those two transmission channels do not appear to be a significant pathway through which the policies impact urban GTFP. Third, heterogeneity tests reveal that urban agglomerations are essential for new-type urbanization efficacy. That is, the impact of the new-type policy is more substantial in cities that are part of urban agglomerations. Furthermore, the analysis indicates that implementing new-type urbanization pilot initiatives in cities with more abundant human capital, financial resources, and infrastructure is conducive to the positive effect of policies on GTFP.

The study’s results yield several policy recommendations. First, local governments should promote the new-type urbanization policies in a coordinated effort from pilot regions toward nationwide coverage. At the same time, regions should learn from one another’s experiences to maximize the beneficial effects on their respective GTFPs. Meanwhile, promotion strategies should be tailored to local conditions, and industries should be planned and developed based on comparative advantages.

Second, we advocate exploring multidimensional approaches by the decision-makers for new-type urbanization construction to enhance urban GTFP. For instance, the future development of new-type urbanization could prioritize promoting technological innovation and strengthening the intensity of environmental regulation. Regarding technological innovation, government encouragement can be a driving force supporting urban green development. However, implementing effective environmental regulations also requires strong public participation and support, which calls for public awareness of green consumption to promote cities’ green and high-quality growth. Specifically, policymakers should consider implementing targeted educational campaigns, incentives for eco-friendly practices, and partnerships with community organizations to foster a culture of environmental stewardship among citizens.

Third, our findings call for continuous promotion of the development mode with urban agglomerations as the primary form to leverage the critical role of economies of scale. For instance, strengthening the cooperation mechanism among cities within agglomerations, realizing the integration of public services, infrastructure, and markets, and optimizing the structure of urban agglomerations to enhance GTFP should be critical considerations in the government’s strategic planning initiatives. Likewise, policymakers had better focus on developing comprehensive regional plans that identify critical areas for collaboration, establish shared goals and performance metrics, and create frameworks for resource sharing and joint decision-making. Furthermore, our results could interest other developing countries with similar urban problems.

As for the limitation, this study focuses on a specific period, and its findings may not be generalizable to other periods or scenarios. For instance, economic, social, and political shocks could influence the relationship between the new-type urbanization policy and GTFP. As such, the results should be interpreted within the context of the studied time frame, and follow-up empirical analyses are warranted to examine the long-run effects and robustness of conclusions under different temporal and socio-economic conditions.