Innovative City Construction and Urban Environmental Performance: Empirical Evidence from China

Abstract

Environmental performance is a key issue that relates to the sustainable development of the economy and the environment. Innovation-driven approaches are fundamental in improving environmental performance; however, innovation activities come with uncertainties and require supportive policies from the government. This study utilizes the implementation of the Innovation City Pilot (ICP) policy in China as a quasi-natural experiment. It employs a progressive Difference-in-Differences (DID) model using panel data from 283 Chinese cities during the period of 2005–2019 to evaluate the impact of the ICP policy on Urban Environmental Performance (UEP) and its underlying mechanisms. The empirical results indicate that (1) the ICP policy significantly promotes the improvement of UEP and robustness analyses further support this conclusion; (2) compared to cities in the central and western regions, resource-dependent cities, and higher administrative level cities, the ICP policy is more beneficial for enhancing UEP in eastern, non-resource-dependent, and lower administrative level cities; (3) mechanism tests suggest that the ICP policy facilitates UEP improvement by leveraging technological innovation, upgrading industrial structure, and optimizing resource allocation; (4) the ICP policy not only benefits the UEP enhancement in local cities but also promotes UEP improvement in neighboring cities through spatial spillover effects. This study provides evidence and insights from China, contributing to the global implementation of innovation-driven development strategies for sustainable urban economic and environmental development.

1. Introduction

The deterioration of global air quality and the increasing constraints on resources pose severe threats to human habitat. Countries worldwide are actively implementing measures to address these challenges and strive to achieve sustainable development. Sustainable development entails the harmonious and long-term advancement of society, the economy, the population, resources, and the environment. Environmental performance, a key indicator for assessing the coordinated development of the economy and the environment, aims to achieve high-quality economic growth at a minimal environmental cost and forms a critical component of sustainable development. Historically, China’s extensive economic development model, while fostering rapid economic growth, has also led to a range of environmental issues. According to the 2022 Environmental Performance Index (EPI) report, China scored 28.40, ranking 160th out of 180 evaluated countries [1]. This indicates significant room for improvement in China’s environmental performance, especially considering its status as the world’s second-largest economy. Enhancing China’s environmental performance is crucial for achieving global sustainable development. It is worth noting that in the past decade, China has exerted tremendous efforts to enhance environmental performance and realize sustainable development. The 5th plenary session of the 19th Central Committee of the Communist Party of China explicitly emphasized the need to expedite green transformation and development, with the primary objective of promoting sustained and robust economic growth.

Cities serve as major sources of energy consumption and pollution emissions [2], but they also serve as key drivers of innovation activities. Promoting urban innovation to drive economic growth and achieve coordinated development between economic growth and environmental quality is a fundamental approach to enhancing environmental performance [3,4]. However, innovation is characterized by significant investments, long cycles, and high risks [5], particularly in the field of technological innovation. Therefore, it is necessary to formulate relevant innovation support policies [6]. For instance, the United States and Germany have respectively published three editions of the “U.S. Innovation Strategy” and four editions of the “German High-Tech Strategy” [7,8]. These initiatives aim to foster the development of national innovation-driven cities [4]. The United Kingdom has launched the “Tech City” initiative to support the “Innovation City” strategy [9,10]. In 2008, China introduced the national strategy of innovation-driven development and proposed the establishment of national innovation-driven cities, with Shenzhen being selected as the first pilot city for this initiative. As of the end of 2018, China had established 78 pilot cities for innovation, including 76 cities at the prefecture level or above, and 2 county-level cities. Among them, from 2009 to 2013, a total of 60 cities were approved as pilot cities for innovation. In 2018, an additional 17 pilot cities were added, including Jilin City and Xuzhou City. In addition, in April 2010, the Ministry of Science and Technology issued the “Guiding Opinions on Further Promoting Pilot Work for Innovation-Driven Cities”. This document provided an evaluation indicator system for innovation-driven cities from six aspects: innovation investment, corporate innovation, technology commercialization, high-tech industries, technology benefiting the public, and innovation environment. In December 2016, the National Development and Reform Commission and the Ministry of Science and Technology jointly issued the “Guidelines for Building Innovation-Driven Cities”, further revising the indicator system. The assessment indicators, key tasks, policy support, and environmental performance of innovation pilot cities are deeply interconnected. So, do the policies supporting innovation-driven development, such as the innovative city pilot (ICP) policies, truly promote the improvement of urban environmental performance (UEP) in China? If so, what are the specific mechanisms, and do heterogeneity and policy spillovers exist? How can we maximize the promotion mechanisms of ICP policies? Exploring these questions is of great theoretical and practical significance for improving China’s modern innovation system and better utilizing innovation policies for sustainable economic and social development.

Currently, research on environmental performance primarily focuses on two aspects: measurement methods and influencing factors. In terms of measurement methods, the preferred approach is the use of composite indicator systems to measure environmental performance [11,12,13], such as the Environmental Performance Index (EPI) released by Yale University in 2022. However, the indicator selection in this method is somewhat subjective. Alternatively, some scholars measure environmental performance by the ratio of pollution emissions to GDP output [14]. However, this method does not consider the comprehensive impact of other production factors on environmental performance. Additionally, a portion of the research adopts Data Envelopment Analysis (DEA) to measure environmental performance [15,16,17,18]. This method takes into account inputs, ideal outputs, and undesirable outputs (pollutants), which can avoid the subjectivity of indicator selection and comprehensively consider other production factors [19]. Regarding influencing factors, according to the Environmental Kuznets Curve (EKC) hypothesis [20], economic growth, population size, and production technology are direct driving factors. Additionally, various urban infrastructure [21], energy efficiency [22], industrial agglomeration [23], and a range of environmental regulation measures [24,25] are crucial influencing factors. Furthermore, some studies have highlighted the role of innovation, particularly the impact of green technological innovation on environmental performance [26,27]. However, the role of innovation policies in influencing environmental performance, particularly ICP policies, has been largely overlooked.

Schumpeter [28] first proposed “innovation theory”, which views innovation as a “creative destruction” that disrupts the existing economic and market structures through the introduction of new products, technologies, and market openings to achieve dynamic high-quality growth. Since Schumpeter’s concept of innovation, the understanding, classification, and measurement of innovation have been continuously improved [29]. Chesbrough [30] introduced the concept of open innovation, and Hobcraft [31] utilized the Three Horizons Framework to articulate innovation. Developed countries, represented by the OECD, have released the Oslo Manual and the Frascati Manual, which define and classify innovation activities, standardizing the statistics of innovation activities [29]. Research on the relationship between innovation and environmental performance has primarily been conducted from the perspective of technological innovation. The uncertainty of innovation activities means that innovation does not inherently possess “green” attributes [32]. The research on the relationship between the two has formed three viewpoints: promotion theory [33], inhibition theory [34], and non-linear theory [35]. Innovation not only includes the five aspects proposed by Schumpeter but also encompasses institutional innovation and other innovative environments [29]. Innovative cities aim to eliminate the negative externalities of innovation activities by creating a favorable innovative environment. Therefore, some studies have explored the characteristics and evaluation indicators of innovative cities [36,37], the connotation of innovative cities [38], implementation paths [38], and development models [39]. Furthermore, many studies have assessed the economic and environmental impacts of ICP policies, including their effects on energy efficiency [5], urban innovation capabilities [40], ecological efficiency [4], and industrial structure [41]. However, further research is needed to determine whether innovative cities possess “green” attributes.

Based on this, the study first uses the EBM-DEA model to measure UEP, effectively avoiding the problem of underestimating efficiency values in traditional DEA. Secondly, the national innovative city pilot (ICP) policy is treated as a quasi-natural experiment, and the Difference-in-Differences (DID) model with two-way fixed effects is employed to accurately identify the impact of the ICP policy on UEP, effectively addressing endogeneity issues that existed in previous models. Additionally, this study conducts robustness analysis using the PSM-DID model to mitigate sample self-selection bias. Finally, this study not only reveals the transmission mechanisms of the ICP policy’s impact on UEP from the aspects of technological innovation, industrial structure, and resource location, but also elucidates the heterogeneity of the ICP policy’s influence on UEP from the perspectives of geographic location, resource endowment, and urban administrative level. Furthermore, this study explores the spatial spillover effects of the ICP policy on UEP. By enriching the research on the relationship between innovation and environmental performance through the analysis of innovation pilot policies, this study provides Chinese experiences and evidence for implementing a global innovation-driven development strategy and driving sustainable economic and environmental development in cities.

2. The Pilot Cities and Direct Drive Mechanism

2.1. The Pilot Cities

Considering the research period and sample, this study ultimately obtained 63 innovative pilot cities, while the remaining 220 cities were non-pilot cities. The spatial distribution of innovative pilot cities is shown in Figure 1. It can be observed that there are 38 cities in the eastern region, 14 cities in the central region, and 11 cities in the western region, accounting for 60.32%, 22.22%, and 17.46% of the total number of innovative pilot cities, respectively. The eastern region serves as the focus and core area for the innovative pilot city policies, and the spatial pattern of pilot cities aligns with the overall deployment of the innovation-driven development strategy, which aims to coordinate the layout of the eastern, central, and western regions.

2.2. Direct Drive Mechanism

The ICP policy mainly drives UEP through three action mechanisms: technological innovation, industrial structure upgrading, and resource allocation [4,40,42].

One important purpose for implementing an ICP policy is to increase the support of city innovation and promote the innovation elements agglomeration, which can accelerate the transformation of urban production mode from traditional extensive mode to green intensive mode, thus improving the UEP [43]. The ICP policy can promote the technological innovation of cities via three aspects, namely the strategic guidance effect, talent agglomeration effect, and economic agglomeration effect. First, an ICP policy can effectively promote city technological innovation through the strategic leading effect of the government [4]. Concerning “double externalities” in technological innovation, enterprises lack incentives to make technological innovation decisions [44]. Thus, innovation-oriented city pilot policies can enable the government to compensate for the risks of enterprises’ technological innovation through research and development subsidies [45]. Furthermore, it gives full play to the leading role of the government in the enterprises’ technological innovation and improve its level [46,47]. Second, the ICP policy is beneficial to technological innovation through the agglomeration effect of innovative talents [4]. Innovative talents primarily refer to individuals engaged in research and experimental development (R&D) activities. The agglomeration of innovative talents is one of the most critical elements of city innovation and an important carrier of city technological innovation. Innovative city pilot policy introduces domestic and foreign high-end talent and cultivates innovative talent. Furthermore, talent agglomeration can provide sufficient intellectual support for technological innovation [40]. Third, the ICP policy can promote the technological innovation development of cities through the agglomeration effect of an innovative economy. Innovation elements agglomeration brought about by implementing the ICP policy will accelerate the transformation of innovation achievements and form new economic growth poles. Apparently, the implementation of the ICP policy is conducive to technological innovation. In terms of the impact of technology innovation on UEP, early studies have revealed that technology innovation can exert a positive impact on UEP. The more advanced the technology, the “greener” the economic development [48]. For one thing, technological innovation at the production end helps to improve the efficiency of energy use [49], and reduce the emissions of pollutants in the production process, thus forming a pollutant prevention system and improving the urban environmental quality. Moreover, the innovation of environmental governance technology can improve the treatment efficiency of pollutants, promote the recycling of waste, and thus reduce the final discharge of pollutants [50]. In addition, technological innovation is also conducive to the use of clean energy, cleaner production [51], and the R & D and promotion of the pollution treatment equipment, thus improving the UEP. In sum, the ICP policy can help improve the UEP by promoting technology innovation.

In addition to technology innovation, industrial structure upgrading is another conduction mechanism in the causal connections between the ICP policy and UEP. The ICP policy is conducive to promoting industrial structure upgrading through the human capital effect and industrial cluster effect, thus promoting UEP. First, the ICP policy can promote the optimization of urban industrial structures through the human capital effect [52]. An innovative pilot city policy provides knowledge and technical personnel with high-quality jobs [53]. They create socioeconomic value to promote the land and other production factors flowing between regions and industries, and promote the industrial structure change speed, thus realizing industrial transformation and upgrade [41]. Second, the ICP policy can promote the optimization of urban industrial structures through the industrial cluster effect [54]. In the construction of innovative cities, key resources such as capital, workforce, technology, and equipment from the government are conducive to building new technology industries, forming new industrial clusters, promoting the rational urban industrial structure, and continuous transformation and upgrading [55]. As for the influence of industrial structure on UEP, a vast literature clarifies that the upgrading of industrial structure is an important driving factor for the improvement of urban environment [56,57]. The adjustment of industrial structure can bring about a reasonable distribution of resources among industries, and make full use of resources and reduce environmental pollution [58]. In addition, as the traditional industrial industry is gradually replaced by the technology- and knowledge-intensive industries, the pollution emissions of traditional industry will also be reduced, which will help improve urban environmental performance. Therefore, it is reasonable to believe that the ICP policy can help improve the UEP by promoting industrial structure upgrading.

The possible third conduction mechanism of the ICP policy affecting UEP is promoting resource allocation. First, the implementation of the ICP policy relies on the interconnection of super-complex innovation chains formed by several innovation factors. It realizes the optimal allocation of production factors among different regions, industries, and enterprises, therefore promoting synergy among production factors [59]. Second, the ICP policy is conducive to the inflow of innovative capital and innovative talents, thus helping reduce the misallocation of resources through the siphoning effect of investment and the aggregation effect of talents. Previous studies have proved that a mismatch of resources and distortion of the factor market inhibits the improvement of energy utilization efficiency [60], hinders the optimization of industrial structure, and aggravates environmental pollution [61]. Therefore, the implementation of ICP policies can improve urban environmental performance by mitigating element mismatch and improving resource allocation efficiency.

2.3. Spatial Spillover Mechanism

The ICP policy stresses the need to strengthen the open and sharing of all kinds of innovation resources within and between cities and speed up the spillover of innovation results [62]. This indicates that the ICP policy can form super-complex innovation chains to connect through characteristics of super networks. This will generate spatial connections through the flow of material resources and production factors, thus forming strong spatial correlation effects, which will also affect the UEP of surrounding cities [63,64].

From the perspective of the coordinated development of the regional ecological environment, the ICP policy not only plays the role of innovation radiation leading but also takes into account the role of pollution prevention and control “demonstration area”. This is a regional model to achieve the goal of a “win-win” economy and environment [65]. Scientific and technological innovation is the technological path to promote resource saving and control pollution emissions [45]. On the spatial scale, the ICP policy is conducive to regional financial accumulation and the talent inflow. This can promote innovation investment and scientific and technological activities, producing spatial spillover effects on the coupled development of an economy–environment system in adjacent regions [66], which is conducive to building a good ecological environment and market environment in neighboring units, effectively reducing the resource utilization of enterprises with high-pollution emissions and low-production efficiency, and promoting more high-quality resources to match clean enterprises with high-production efficiency, thus improving the UEP.

The above theoretical analysis is synthesized in Figure 2. According to the theoretical analysis, the following research hypotheses are constructed:

Hypothesis 1.

The implementation of ICP policies is conducive to the improvement of urban environmental performance.

Hypothesis 2.

Technology innovation, industrial structure upgrading, and resource allocation are three action mechanisms that the ICP policy affects UEP in China.

Hypothesis 3.

There is a spatial spillover characteristic in the improvement effect of the ICP policy on UEP.

3. Methods and Data

3.1. Methods

3.1.1. DID Model

As a government innovation policy implemented in batches, the ICP policy is regarded as a quasi-natural experiment in this study. Combined with the above theoretical analysis, this study constructs a progressive DID model to accurately identify the impact of ICP policies on UEP, as shown in Model (1):

$$UE{P}_{it}={\alpha }_0+{\alpha }_1{ \mathit{Policy} }_{it}+\phi X_{it}+{\mu }_i+v_t+{\epsilon }_{it}$$

(1)

where Policy is the variable of ICP. X_it is the control variable affecting the UEP. ${\alpha }_0$ is the constant term, and ${\alpha }_1$ and $\phi$ are the effects of Policy and control variables on UEP, respectively. ${\mu }_i$ is the individual fixed effect, $v_t$ is the time fixed effect, and ${\epsilon }_{it}$ represents the random interference term.

3.1.2. Panel Mediation Effect Model

To further investigate the action mechanisms of the ICP policy on UEP, by referring to the study of Baron and Kenny [67], we set up the mediating effect model and construct Models (2) and (3) successively based on Model (1):

$$ME{D}_{it}={\beta }_0+{\beta }_1{ \mathit{Policy}}_{it}+\phi X_{it}+{\mu }_i+v_t+{\epsilon }_{it}$$

(2)

$$UE{P}_{it}={\gamma }_0+{\gamma }_1{ \mathit{Policy}}_{it}+{\gamma }_{2}ME{D}_{it}+\phi X_{it}+{\mu }_i+v_t+{\epsilon }_{it}$$

(3)

where MED is the mediating variable, β₁ is the effect of ICP on the mediating variable, and γ₂ is the effect of the mediating variable on the UEP. The β₁ and γ₂ coefficients should be significant if the MED is an effective intermediary variable. If the γ₁ coefficient is significantly weaker than α₁ in the baseline Model (1), it indicates a partial mediation effect. If γ₁ is not significant, it indicates a complete mediation effect. The meanings of other variables in Models (2) and (3) are consistent with Model (1).

3.2. Variables Description and Data Sources

3.2.1. Variables Selection

(1) Explained variable. The evaluation system of UEP includes input, expected output, and unexpected output, as shown in

Table 1. Index system of UEP.

Indicators	Variable	Variable Description	References
Input	Land factor input	Urban land area (km2)	[24,70]
	Labor factor input	Number of employed persons in urban secondary and tertiary industries (10,000 people)
	Capital factor input	Urban capital stock (10,000 yuan)
	Energy factor input	Urban total social electricity consumption (10,000 kw·h)
Expected output	Ecological and environmental benefit output	Park green space of urban district (hm2)	[71]
		Urban district built-up area green coverage rate (%)
	Economic benefit output	Added value of urban secondary and tertiary industries (10,000 yuan)
	Social benefit output	Public financial revenue (10,000 yuan)
Unexpected output	Pollutant discharge	Total urban industrial wastewater discharge (10,000 tons)	[72]
		Total SO2 emission from urban industry (10,000 tons)
		Urban industrial smoke (powder) dust emission total (10,000 tons)

. The input index involves production factors such as land, energy, labor, and capital [18]. Among them, the urban capital stock is obtained from the urban fixed asset investment using the perpetual inventory method with a depreciation rate of 10.96% [68]. Output indicators include expected output and unexpected output [24]. Considering the availability of data and heavy pollution of industry, urban industrial pollutant emissions are used to measure the undesired output. Combined with the EBM model proposed by Tone and Tsutsui [69], this study uses a non-oriented super-efficiency EBM model with a variable return to scale to measure the UEP. The equations are as follows:

$${\gamma }^{*}=\mathit{min} \frac{\theta -{\epsilon }_x\left(\frac{1}{{{\displaystyle \sum }}_{i=1}^m{\omega }_i^{-}}\right){{\displaystyle \sum }}_{i=1}^m\frac{{\omega }_i^{-}{s}_i^{-}}{x_{ik}}}{\phi +{\epsilon }_y\left(\frac{1}{{{\displaystyle \sum }}_{r=1}^l{\omega }_r^{+}}\right){{\displaystyle \sum }}_{r=1}^l\frac{{\omega }_r^{+}{s}_r^{+}}{y_{rk}}+{\epsilon }_b\left(\frac{1}{{{\displaystyle \sum }}_{t=1}^p{\omega }_t^{b-}}\right){{\displaystyle \sum }}_{t=1}^p\frac{{\omega }_t^{b-}{s}_t^{b-}}{b_{tk}}}$$

(4)

$$\begin{array}{c}st.\{\begin{array}{c}{\displaystyle {\displaystyle \sum }_{j=1}^n}{x}_{ij}{\lambda }_j+s_i^{-}=\theta x_{ik},i=1,\cdots ,m\\ {\displaystyle {\displaystyle \sum }_{\begin{array}{c}j=1\\ j\ne k\end{array}}^n}{y}_{rj}{\lambda }_j-s_r^{+}=\phi y_{rk},r=1,\cdots ,l\\ {\displaystyle {\displaystyle \sum }_{\begin{array}{c}t=1\\ j\ne k\end{array}}^n}{b}_{ij}{\lambda }_j+s_t^{b-}=\phi b_{tk},t=1,\cdots ,p\end{array}\\ {\displaystyle {\displaystyle \sum }_{j=1}^n}{\lambda }_j=1,{\lambda }_j\ge 0,s_i^{-},s_r^{+},s_t^{b-}\ge 0,\theta \le 1,\phi \ge 1\\ \end{array}$$

(5)

In the above context, ${\gamma }^{*}$ represents the optimal efficiency value of the EBM model. $\theta$, ${\epsilon }_x$, ${\omega }_i^{-}$, $S_i^{-}$, k, $x_{ik}$, $y_{rk}$, $b_{tk}$, and ${\lambda }_j$ refer to the planning parameters of the radial component, key parameters for the conversion of radial and non-radial relaxation conditions, relative importance of input indicators, relaxation amount of the i-th input factor, evaluated decision-making unit (DMU), i-th input of the k-th DMU, $\gamma$-th desired output of the k-th DMU, t-th undesired output of the k-th DMU, and the linear combination coefficient of the DMU. In this case, ${\epsilon }_x$ satisfies the condition 0 < ${\epsilon }_x$ < 1, and ${\displaystyle \sum }_{i=1}^m{\omega }_i^{-}=1\left({\omega }_i^{-}\ge 0\right)$. $s_r^{+}$, $s_t^{b-}$, ${\omega }_r^{+}$, ${\omega }_t^{b-}$ represent the relaxation amounts of the $\gamma$-th desired output, t-th undesired output, weight of the $\gamma$-th desired output indicator, and weight of the t-th undesired output indicator, respectively. m and l denote the number of input types and output types, respectively.

(2) Core explanatory variable. According to the pilot cities of the ICP policy and the approval time of the policy pilot, the cities approved for the ICP policy pilot are defined as the experimental group, and the cities not approved are defined as the control group. Policy_it = pilot × time. pilot equals to 1 if the city is in the experimental group, and 0 otherwise. In the current year and subsequent years when the pilot city is approved, time is set as 1, and otherwise, the value is 0. The list of innovative cities is from the official website of the Ministry of Science and Technology. In this study, Changji, Shihezi, and other cities whose data are seriously missing in the experimental group are removed. Finally, 63 cities in the experimental group and 220 cities in the control group were obtained.

(3) Mechanism variables. This study adopts the number of urban patent authorization to characterize technological innovation (INNOV) [73]. Industrial structure upgrading is measured based on the ratio of value added of the tertiary industry to the value added of the secondary industry. Urban total factor productivity is used to represent the allocation of urban resources.

(4) Control variable. It mainly includes the following aspects:

Environmental regulation (ER): the entropy method is used to comprehensively calculate the intensity of environmental regulation by selecting the removal rate of sulfur dioxide, the removal rate of industrial smoke (powder) dust, and the comprehensive utilization rate of industrial solid waste [74].

Transport infrastructure (INFRA): transport infrastructure is measured by the per capita road area of urban districts with 10,000 inhabitants [75].

Opening to the outside world (OPEN): the proportion of annual actual foreign investment (converted into RMB according to the average RMB exchange rate of the year) in the GDP of the urban districts is used to represent the opening-up level [76].

Government intervention (GOV): government intervention is represented by the proportion of education and science expenditure in total fiscal expenditure except for fiscal expenditure [77].

Industrial agglomeration (AGG): this study adopts manufacturing agglomeration to represent industrial agglomeration [78].

Marketization (MAKE): this study adopts the proportion of urban individuals and private enterprises in total employment to represent marketization [79].

Financial Development (FIN): this study adopts the ratio of outstanding loans to GDP at the end of the year to represent financial development [80].

3.2.2. Data Sources

In this study, panel data of 283 cities in China from 2005 to 2019 were selected as research samples. The original data for the explained variable and control variables comes from The China Urban Statistical Yearbook and the EPS data platform (http://olap.epsnet.com.cn, accessed on 2 April 2023) from 2006 to 2020, and some missing values are supplemented by the interpolation method (

Table 2. Descriptive statistics.

	Variable	Mean	S.D.	Min.	Max.
Explained variable	UEP	0.370	0.390	0.000	1.286
Mechanism variable	INNOV	6.770	16.310	0.000	195.600
	INDUSTR	0.910	0.480	0.090	5.170
	RESOUR	1.050	0.400	0.080	16.430
Core explanatory variable	Policy	0.110	0.320	0.000	1.000
Control variables	ER	0.590	0.210	0.060	0.990
	INFRA	4.420	5.840	0.000	73.040
	OPEN	1.920	2.030	0.000	15.320
	GOV	0.800	0.050	0.500	0.980
	AGG	0.890	0.560	0.020	3.690
	MAKE	0.470	0.140	0.000	0.940
	FIN	0.870	0.550	0.080	9.620
	URB	45.520	18.910	7.670	100.000
	AGDP	10.340	0.840	7.800	13.190

).

4. Empirical Results

4.1. Spatiotemporal Characteristics of UEP

Figure 3 presents the spatial characteristics of UEP in China during 2005–2019. Overall, the mean values for UEP increased significantly during the study period, and in 2005, 2010, 2015, and 2019, the mean values were 0.307, 0.359, 0.367, and 0.434, respectively. The proportions of cities above the mean values for UEP in 2005, 2010, 2015, and 2019 were 31.45%, 35.34%, 36.40%, and 36.75%, respectively. The proportion of cities above the mean values for UEP showed an increasing trend from 2005 to 2019. There is evident heterogeneity in the spatial distribution of UEP. Cities with UEP > 0.7 are scattered in China. In general, the UEP in China increased during the study period, with only a few units (<0.2) mainly distributed in the central region. It was found that from 2005 to 2019, China’s urban environment performance showed a trend of increasing improvement.

4.2. Parallel Trend Test and Dynamic Test

To evaluate the implementation effect of the innovation city pilot policy, the hypothesis that the experimental group and the control group have a parallel trend before the policy implementation should be satisfied. After the implementation of the policy, this parallel trend is broken, and cities in the experimental group and control group will show different trends in the level of UEP. Simultaneously, there may be dynamic changes such as time delay or attenuation in the implementation of policy effects. Given that the data sample period for this study is from 2005 to 2019, and the ICP policy was first implemented in Shenzhen in 2008, we selected two cycles in advance and 11 cycles behind for testing.

The event study method used by Beck et al. [81] was utilized, for reference, to test whether the UEP meets the parallel trend hypothesis and the dynamic change of policy implementation. The test results are shown in Figure 4. It can be concluded that the estimated coefficient of Policy has no significant difference from 0 in the two periods before the implementation of the ICP policy and is not significant. This indicates that the parallel trend hypothesis is satisfied. Concurrently, the estimated coefficient of Policy is not significant in the current period and is significant at the 5% level in the first year after the implementation of the ICP policy (year = 2009). This indicates that the ICP has a certain lag effect on the UEP, and the lag effect of about one year is consistent with practical experience. Simultaneously, the estimated coefficient of Policy shows an increasing trend of fluctuation after the implementation of the ICP policy, and is significant at least at the 5% level. This indicates that the marginal effect of ICP on the UEP in the sample periods showed an increasing process, and this trend remained consistent until 2019.

4.3. Baseline Regression Results

This study uses the two-way fixed effect model to estimate the impact of the ICP policy on UEP. Column (1) is the estimated result without adding any control variables. It can be found that the estimated coefficient of Policy on UEP is 0.1038, which is significant at the 1% level (

Table 3. Baseline regression results.

Variable	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)
Policy	0.1038 ***	0.1058 ***	0.0960 ***	0.0976 ***	0.0932 ***	0.0913 ***	0.0887 ***	0.0886 ***	0.0906 ***	0.0867 ***
	(6.047)	(5.893)	(5.638)	(5.441)	(4.520)	(4.628)	(5.327)	(5.309)	(5.007)	(4.888)
ER		0.1197	0.1193	0.1187	0.1191	0.1247	0.1279	0.1277	0.1358	0.1388
		(1.594)	(1.592)	(1.591)	(1.601)	(1.613)	(1.563)	(1.559)	(1.559)	(1.589)
INFRA			0.0127 ***	0.0128 ***	0.0125 ***	0.0125 ***	0.0125 ***	0.0125 ***	0.0123 ***	0.0121 ***
			(5.059)	(5.015)	(4.700)	(4.699)	(4.691)	(4.669)	(4.663)	(4.532)
OPEN				0.0017	0.0021	0.0018	0.0020	0.0021	0.0013	0.0008
				(0.721)	(0.948)	(0.843)	(0.893)	(0.906)	(0.610)	(0.388)
GOV					−0.2286	−0.2317	−0.2206	−0.2216	−0.1830	−0.2326
					(−1.033)	(−1.053)	(−0.944)	(−0.944)	(−0.712)	(−0.926)
AGG						−0.0348	−0.0457	−0.0453	−0.0473	−0.0483
						(−1.496)	(−1.155)	(−1.140)	(−1.156)	(−1.178)
MAKE							−0.0804	−0.0796	−0.0746	−0.0699
							(−0.611)	(−0.602)	(−0.585)	(−0.551)
FIN								0.0033	0.0077	0.0209
								(0.237)	(0.579)	(1.558)
URB									0.0047	0.0042
									(1.319)	(1.204)
AGDP										0.1231 ***
										(3.944)
Constant	0.3563 ***	0.2853 ***	0.2305 ***	0.2271 ***	0.4117 *	0.4423 **	0.4791 ***	0.4765 ***	0.2241	−0.9987 *
	(85.481)	(6.789)	(4.703)	(4.439)	(1.903)	(2.197)	(3.042)	(3.057)	(0.702)	(−1.802)
Individual fixed effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Time fixed effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
N	4245	4245	4245	4245	4245	4245	4245	4245	4245	4245
R2	0.3207	0.3219	0.3251	0.3251	0.3253	0.3260	0.3262	0.3262	0.3320	0.3336
R2_a	0.2696	0.2707	0.2739	0.2738	0.2738	0.2743	0.2744	0.2742	0.2802	0.2818
F	36.5638	23.1417	21.3863	16.2958	16.3948	13.6192	14.4365	12.6369	10.8514	12.4198

Notes: *, **, and *** represent significance at 10%, 5%, and 1% levels, respectively; t values are in parentheses.

). When all control variables are further added, the estimated coefficient is slightly reduced to 0.0867, which is still significant at the 1% level. It is confirmed that the ICP policy can significantly promote UEP. In other words, this finding confirms the validity of research hypothesis 1. Specifically, when other conditions remain unchanged, compared with non-pilot cities, the UEP of innovative pilot cities increases by 0.0867 on average. According to the control variables in column (10), the transportation infrastructure and economic development level can significantly promote the improvement of UEP, while other factors have no significant impact on UEP.

4.4. Robustness Analysis

To test the robustness of the ICP policy promoting UEP, this study conducts many other robust analyses, such as excluding other policy pilot tests and placebo tests.

4.4.1. Exclude Other Policy Pilots

Although we have assessed the environmental effects of the ICP policy, the presence of omitted variables inevitably introduces endogeneity issues in the model. While innovative city construction is underway in China, there are also other pilot projects being implemented, and these policies may also affect urban environmental performance. Thus, the presence of these policies could act as omitted variables in the model. To address this potential endogeneity issue and obtain the net effect of the ICP policy on UEP, we incorporate the smart city pilot policy, low-carbon city pilot policy, civilized city construction policy, and entrepreneurial city construction pilot policy into the baseline regression and re-estimate the original model. Column (1) in

Table 4. Regression results excluding other policy pilots.

Variable	(1)	(2)	(3)	(4)
Policy	0.0857 ***	0.0879 ***	0.0757 ***	0.0676 ***
	(4.922)	(4.481)	(3.441)	(4.036)
Smart city construction	0.0167	0.0161	0.0127	0.0106
	(1.307)	(1.297)	(1.043)	(0.951)
Low-carbon city construction		−0.0214	−0.0176	−0.0176
		(−0.678)	(−0.477)	(−0.477)
Civilized city construction			0.0410 ***	0.0389 ***
			(3.854)	(3.709)
Entrepreneurial city construction				0.0401
				(1.024)
Control variables	Yes	Yes	Yes	Yes
Constant	−0.9800 *	−0.9993 *	−1.2455 *	−1.2674 *
	(−1.790)	(−1.757)	(−1.887)	(−1.868)
Individual fixed effect	Yes	Yes	Yes	Yes
Time fixed effect	Yes	Yes	Yes	Yes
N	4245	4245	3962	3962
R2	0.3337	0.3338	0.3383	0.3386
R2_a	0.2817	0.2817	0.2825	0.2826
F-value	11.3606	10.6649	8.4430	7.9545

Notes: * and *** represent significance at 10% and 1% levels, respectively; t values are in parentheses.

shows the regression results of incorporating smart city construction policies. The results show that the ICP policy significantly promotes the improvement of UEP. Furthermore, when low-carbon city construction policies, civilized city construction policies, and entrepreneurial city construction policies are added, the ICP policy still significantly promotes the improvement of UEP with a promotion coefficient of 0.0676. Compared to column (10) in Table 3, the regression coefficient decreased by 0.0191.

4.4.2. Placebo Test

To further exclude the improvement effect of other random factors on UEP, this study refers to the practice of Chen et al. [82] to randomly select the same number of cities as the control group and construct a “virtual” Policy variable according to the annual number of pilot cities. The baseline model was used to repeat 500 and 1000 regressions, respectively. Figure 5 plots the kernel density distribution of the estimated coefficients of policy for simulating 500 and 1000 times, respectively. Conclusively, the mean values for the two simulated regression coefficients are −0.0005178 and 0.0001621, respectively. These values are very close to 0 compared with the baseline regression coefficient of 0.0867. This indicates that the baseline regression coefficient is larger than most of the simulated values and can be regarded as an extreme value. Furthermore, the estimated result of the baseline regression can be understood as a large probability event, indicating that the improvement effect of the ICP policy on UEP is almost not affected by other policies or random factors.

4.4.3. Other Robustness Tests

Other robustness tests are shown in

Table 5. Regression results of other robustness tests.

Variable	PSM-DID	Traditional DID	A Test to Eliminate Samples	Re-Estimate UEP by SBM Model
	(1)	(2)	(3)	(4)
Policy	0.0828 ***	0.4188 *	0.0515 ***	0.0903 ***
	(4.417)	(1.834)	(3.453)	(3.045)
Control variables	Yes	Yes	Yes	Yes
Constant	−1.0639 **	−1.4099 *	−1.2542 *	−0.3532
	(−1.966)	(−1.886)	(−1.934)	(−0.451)
Individual fixed	Yes	Yes	Yes	Yes
Time fixed	Yes	Yes	Yes	Yes
N	4155	3315	3960	4245
R2	0.3252	0.3130	0.3227	0.1744
R2_a	0.2716	0.2584	0.2697	0.1103
F	10.9974	5.6883	8.2790	8.5044

Notes: *, **, and *** represent significance at 10%, 5%, and 1% levels, respectively; t values are in parentheses.

. Firstly, the successful implementation of innovative city construction in a particular region in China is influenced by various factors, and the selection of pilot projects itself may be influenced by factors such as a city’s economic development level, financial capacity, and technological capabilities, thus introducing potential selection bias in the pilot projects themselves. Consequently, the presence of selection bias gives rise to endogeneity issues in the Difference-in-Differences (DID) model. In order to address selection bias and ensure as much similarity as possible between the experimental and control group cities in terms of their characteristics, we employ the propensity score matching (PSM)-DID method to more accurately evaluate the impact of the ICP policy on UEP. This study takes control variables as matching covariables and uses Logit regression and 1:1 nearest neighbor matching to obtain the propensity score. On this basis, the PSM-DID model is used to re-examine the benchmark model. The regression results are shown in column (1). Subsequently, this study carries out the test of traditional DID based on a single time point. In this study, 2008 is used as the implementation year of the innovation-oriented city pilot policy. Meanwhile, to avoid the interference of the sample of pilot cities after 2008 on the regression results, this study excludes the pilot cities after 2008. On this basis, the traditional DID method is used to re-examine the benchmark model. The estimation results are presented in column (2). Third, this study carries out the test of eliminating samples. Considering the particularity of Beijing, Tianjin, Shanghai, and Chongqing in terms of administrative level and economic scale, the four municipalities directly under the Central Government and the sub-provincial cities are removed from the research sample for the robustness test. The estimation result is shown in column (3). Finally, the SBM model is used to measure the UEP again, and the DID model is used to re-test the benchmark model. Column (4) reports this estimation result.

From Table 5, it can be seen that after a series of robustness tests, the estimated coefficients of the core explanatory variable remain significantly positive at the 1% level, indicating that the implementation of the ICP policy can significantly improve UEP, and thus, the research result is reliable.

4.5. Action Mechanism and Heterogeneity Analysis

4.5.1. Action Mechanism Analysis

Benchmark regression results confirm that the ICP policy contributes to the improvement of UEP. Furthermore, does the ICP policy promote the improvement of UEP through technological innovation, industrial structure upgrading, and resource allocation? To answer this question, this study examines the three action mechanisms according to the panel mediation effect model set above.

In column (1) of

Table 6. Regression results of action mechanism.

Variable	(1)	(2)	(3)	(4)	(5)	(6)
Policy	9.3840 ***	0.0659 ***	0.1017 ***	0.0817 ***	0.0294 *	0.0773 ***
	(9.311)	(3.864)	(6.095)	(4.780)	(1.740)	(6.099)
INNOV		0.0022 ***
		(5.298)
INDUSTR				0.0496 *
				(1.926)
RESOUR						0.5557 **
						(2.238)
Control variables	Yes	Yes	Yes	Yes	Yes	Yes
Constant	102.4767 ***	−1.2258 **	6.8975 ***	−1.3410 **	0.9010 ***	−0.5840
	(6.638)	(−2.187)	(17.768)	(−2.112)	(6.537)	(−1.192)
Individual fixed effect	Yes	Yes	Yes	Yes	Yes	Yes
Time fixed effect	Yes	Yes	Yes	Yes	Yes	Yes
N	4245	4245	4245	4245	4245	4245
R2	0.8069	0.3353	0.8712	0.3341	0.0947	0.6293
R2_a	0.7919	0.2834	0.8612	0.2821	0.0248	0.6004
F	26.3841	13.9440	45.7548	11.7755	1.4278	10.6726

Notes: The explained variable of column (2), (4), and (6) is UEP. The explained variables of column (1), (3), and column (5) are INNOV, INDUSTR, and RESOUR, respectively; *, **, and *** represent significance at 10%, 5%, and 1% levels, respectively; t values are in parentheses.

, the estimated coefficient of Policy is significantly positive, indicating that ICP significantly promotes urban technology innovation. The estimated coefficient of INNOV in column (2) is 0.0022 (p = 0.01). The estimated coefficient of Policy is 0.0659 (p = 0.01), which is lower than the estimated coefficient of Policy in column (10) of the benchmark regression result in Table 3, indicating that technological innovation plays a partial mediating effect role in the causal connections between the ICP policy and UEP. In column (3), the estimated coefficient of Policy is significantly positive, indicating that the ICP policy significantly promotes the upgrading of urban industrial structure. The estimated coefficient of INDUSTR in column (4) is 0.0496 (p = 0.01), and the estimated coefficient of Policy is 0.0817 (p = 0.01), which is 5.76% lower than the estimated coefficient of Policy in column (10) of the benchmark regression results in Table 3, indicating that industrial structure upgrading has some mediating effect. Therefore, the ICP policy can promote the improvement of UEP through promoting industrial structure upgrading. In column (5), the estimated coefficient of Policy is significantly positive, indicating that the ICP policy significantly promotes the allocation of urban resources. The estimated coefficient of RESOUR in column (6) is significantly positive, and the estimated coefficient of Policy is 0.0773, which is 10.84% lower than the estimated coefficient of Policy in column (10) of the benchmark regression result in Table 3, indicating that there is a partial mediation effect of resource allocation. Therefore, the ICP policy promotes the improvement of UEP through resource allocation.

The above estimation results show that technological innovation, industrial structure upgrading, and resource allocation are the three effective mechanisms of the ICP policy to promote the improvement of urban environmental performance, and they all play a part of the intermediary role. These findings indicate that hypothesis 2 is valid. At the same time, other mechanisms of action remain to be studied.

4.5.2. Impact Heterogeneity Analysis

Due to the evident differences in geographical location, resource endowment, and administrative level among cities, the implementation effects of the ICP policy are also different. Column (1) in

Table 7. Heterogeneity regression results.

Variable	Geographical Location		Resource Endowment		Administrative Level of Cities
	(1)	(2)	(3)	(4)	(5)	(6)
Policy	0.0981 **	0.0541 ***	0.0575 ***	0.0966 ***	0.0449	0.0525 ***
	(2.483)	(3.261)	(3.702)	(3.369)	(1.576)	(3.051)
Control variables	Yes	Yes	Yes	Yes	Yes	Yes
Constant	−1.4281	−1.0616 **	−0.8831 **	−0.7842	−0.3976	−1.2344 *
	(−1.348)	(−2.481)	(−2.234)	(−0.722)	(−0.626)	(−1.807)
Individual fixed	Yes	Yes	Yes	Yes	Yes	Yes
Time fixed	Yes	Yes	Yes	Yes	Yes	Yes
N	1785	2460	1710	2535	525	3720
R2	0.2105	0.6507	0.6778	0.2724	0.7521	0.3165
R2_a	0.1422	0.6220	0.6497	0.2128	0.7213	0.2628
F	7.1864	10.8051	8.7099	6.2703	5.2409	6.9489

Notes: *, **, and *** represent significance at 10%, 5%, and 1% levels, respectively; t values are in parentheses.

is the regression result of eastern cities, and column (2) is the regression result of central and western cities. Predominantly, the ICP policy plays a stronger role in improving UEP in eastern cities than in central and western cities. Column (3) of Table 7 is the regression result of resource-based cities, and column (4) is the regression result of non-resource-based cities. Particularly, the improvement effect of the ICP policy on the UEP of resource-based cities is weaker than that of non-resource-based cities. Column (5) of Table 7 is the regression result of cities with high administrative levels, and column (6) is the regression result of cities with low administrative levels. Consequently, the improvement effect of the ICP policy on UEP of high administrative level cities is obviously weaker than that of low administrative level cities.

4.6. Spatial Spillover Effect

This study further explores the spatial spillover effect of the pilot policy in innovation cities because the spatial spillover of the pilot policy is of great significance for enhancing the regional innovation cooperation centered on the pilot city and improving the regional UEP. The SDM-DID is constructed based on the benchmark model, and the total marginal effect is decomposed into a direct effect and an indirect effect by a partial differential method in this study. This is carried out to better capture and explain the marginal effect of the Policy variable in SDM-DID. Before estimating the spatial econometric model, the construction of the spatial weight matrix W is the key to spatial econometric analysis. In this study, the adjacency spatial weight matrix (W1), distance space weight matrix (W2), economic space weight matrix (W3), and economic distance spatial weight matrix (W4) are constructed, respectively. W1 is 0 for diagonal elements, 1 for adjacent off-diagonal elements, and 0 for non-adjacent elements. The diagonal elements of W2 and W3 are 0, and the off-diagonal elements are the reciprocal of the distance between cities and the reciprocal of the absolute difference in per capita GDP. W4 is a linear combination of W2 and W3 (W4 = 0.5W2 + 0.5W3). After the LM test, Wald test, and LR test, the spatial Durbin model is selected for the model set. In this study, the maximum likelihood method is used to estimate it, and the estimation results are shown in

Table 8. Spatial spillover effect test.

Effect	W1	W2	W3	W4
LR_Direct	0.0722 ***	0.0748 ***	0.0341 *	0.0530 **
	(2.680)	(3.163)	(1.747)	(2.397)
LR_Indirect	0.1637 ***	0.7226 ***	0.2541 ***	0.9149 ***
	(3.321)	(3.055)	(3.161)	(3.808)
LR_Total	0.2360 ***	0.7975 ***	0.2883 ***	0.9679 ***
	(4.388)	(3.437)	(3.601)	(3.955)

Notes: *, **, and *** represent significance at 10%, 5%, and 1% levels, respectively; t values are in parentheses.

.

Table 8 implies that under the four different spatial weight matrices, the direct and indirect effects show that the ICP policy can not only significantly promote the UEP of the local unit but also drive the improvement of the UEP of surrounding units. This shows that the pilot policy of innovative cities has a significant policy spillover effect, and the pilot policy is effective in improving the UEP. Moreover, with the promotion of the successful experience of the pilot city, the radiation capacity of the pilot city is gradually enhanced, and the innovative elements and innovative achievements are constantly spreading to the surrounding units. Thus, it promotes the improvement of UEP in related units. It was found that research hypothesis 3 is valid.

5. Discussion

5.1. Interpretation of Findings

This study employs the progressive DID model to evaluate and test the effect and mechanism of the ICP policy on UEP in 283 Chinese cities. We found that the ICP policy exerts a significant positive effect on the improvement of UEP, showing a one-year lag and a gradually increasing trend. The possible reason is that the assessment indicators, main tasks, and driving modes of China’s ICP policy have a deep internal correlation with the UEP [3]. Specifically, the concept of green development contains the evaluation index of the ICP policy. The assessment indicators of building an innovative city index system include indicators highlighting green development, such as comprehensive energy consumption per ten thousand yuan of GDP. In addition, the ICP policy aims to achieve sustainable social development as the main task. Finally, the innovation-driven model meets the requirements of promoting the UEP. The reason for its delayed impact may be that the ICP was not mature at the early stage of policy implementation, which failed, to some extent, to promote the improvement of UEP. In addition, the ICP policy promotes the improvement of UEP through technological innovation, while technological research and development are characterized by large capital investment and a long return cycle [83]. This may lead to a certain lag in the policy effect reflected in UEP. The reason for the increase in the influence fluctuation may be that with the continuous promotion of the pilot policy of innovation-oriented cities, the number of pilot cities keeps increasing.

Evidently, the ICP policy can improve the UEP through technological innovation, industrial structure upgrading, and resource allocation. One of the important purposes of the implementation of the ICP policy is to increase the support of urban innovation and promote the agglomeration of innovation elements, which will promote the technological innovation of the city through three aspects: government strategic guidance effect, talent agglomeration effect, and economic agglomeration effect. The ICP policy is conducive to upgrading the industrial structure of cities through the human capital effect and industrial cluster effect; thus, promoting the improvement of UEP. The ICP policy is conducive to promoting the allocation of urban resources through promoting factor coordination, and reducing factor mismatch, thus enhancing the UEP.

Compared with central and western cities, eastern cities have location advantages in economic foundation, scientific and technological innovation capacity, industrial supporting systems, and factor flow mechanisms, which are more conducive to ICP’s role in promoting UEP. Resource-based cities, which are dominated by low-technology mining and processing, tend to rely on natural resources and have insufficient demand for innovative resource elements [84]. As a result, the implementation effect of innovative pilot policies in resource-based cities is not as good as that in non-resource-based cities. Cities with high administrative levels already have obvious agglomeration advantages of innovative resource factors and high production efficiency by virtue of grade advantages, and the implementation effects of pilot policies for innovative cities have not been fully manifested. Compared with cities with low administrative levels, cities with high administrative levels always have greater autonomy in innovation policy formulation and can implement more practical innovation policies according to their own urban conditions. The content of these policies often overlaps with that of the ICP policy, resulting in a weakened marginal contribution of the ICP policy. However, the innovation level and UEP of low administrative level cities are low, and the ICP policy can greatly promote the improvement of UEP. With the promotion of the successful experience of pilot cities, the radiation capacity of pilot cities is gradually enhanced, and innovative elements and achievements are constantly spreading to surrounding cities, which promotes the improvement of UEP for related cities [63].

On the one hand, from the perspectives of statistics and econometrics, these findings provide evidence that China’s innovative city construction may make a significant contribution to improving urban environmental performance. On the other hand, they also offer practical experiences for the deepening of sustainable development theory and its application in China. According to the report by the United Nations World Commission on Environment and Development [85], the essence of sustainable development lies in the ability of present generations to meet their own needs without compromising the ability of future generations to meet their needs. Furthermore, according to economist Pezzey [86], sustainable development can be understood as a development path that ensures the well-being of residents does not decline. The research in this study demonstrates that the implementation of innovative city construction policies in China serves as a critical pathway for driving green urban development. It represents an important development path driven by innovation to improve the dual benefits of residents’ economic and environmental well-being. These findings provide significant empirical evidence for the concept of innovation-driven sustainable development.

5.2. Policy Implications

In this study, we find that the ICP policy can effectively promote the improvement of UEP. Thus, it is necessary to summarize and extract the work experience of the existing pilot city and promote it.

Firstly, to ensure the promotion of the ICP policy in a reasonable and orderly manner and give full play to the leading and demonstration role of the innovative city, local governments need to innovate the system of environmental regulation and green development, encourage innovative practices oriented toward cleaner production, design reasonable and effective mechanisms for policy evaluation, supervision, adjustment, and improvement, and appropriately adopt exit mechanisms for cities with poor pilot results.

Secondly, it is necessary to pay full attention to the synergistic effect of multi-dimensional mechanisms to improve the UEP. It is suggested to effectively utilize the advantages of the national innovative city pilot policy, optimize the talent introduction policy, establish a scientific and technological mechanism to guarantee enterprise innovation, innovate the industrial upgrading environment, and provide effective external support for the multi-dimensional promotion mechanism of the ICP policy.

Thirdly, the formulation of innovation policies needs to combine the characteristics of cities and implement differentiated innovation policies following local characteristics and development needs. Cooperation and communication between eastern, central, and western cities, and resource-based and non-resource-based cities need to be strengthened. Cities with high administrative levels should consider the promotion effect of multiple policy factors on UEP, realize the effective docking of innovative city pilots and “civilized city” and other pilots, and strive to form a new green development model with overlapping, complementary, and collaborative policies. It is suggested to rationally optimize the spatial layout and effectively utilize the “demonstration first, radiation-driven” effect of the ICP policy. In addition, the ICP policy can be a decentralized pilot in multiple regions, combined with the characteristics of the city, to achieve multi-regional and full coverage for pilot cities. Moreover, neighboring cities should actively absorb the policy effects generated by innovative cities, form a cooperative mechanism of industrial coordination, technological cooperation, and pollution control, and promote the deep integration of pollution reduction.

5.3. Limitations and Future Directions

Although this study provides some reference and inspiration for future scholars to analyze the influencing mechanism of the ICP policy on UEP, it also has some limitations. This study only uses urban macro indicators such as UEP to evaluate the implementation effect of the ICP policy, which can be further expanded from micro-enterprises and other indicators in the future. In addition, this study only focuses on the impact of China’s innovative city pilot policy on UEP. In the future, whether innovation-driven policy can have an impact on UEP in other developing countries will be further studied.

6. Conclusions

In this study, the ICP policy and UEP are incorporated into the same analytical framework for the first time, which extends and enriches the research on the influencing factors of UEP and the effectiveness of the ICP policy. Combined with theoretical analysis, this study uses the panel data of 283 cities in China from 2005 to 2019 to empirically investigate the impact of the ICP policy on UEP by adopting a progressive DID model. Several conclusions can be drawn from this study.

Firstly, we found that the innovative city pilot policies can help significantly improve urban environmental performance in China, and the promoting effect shows a lag of about one year and increased fluctuation trend.

Secondly, the innovative city pilot policies can improve the urban environmental performance through technological innovation, industrial structure upgrading, and resource allocation.

Thirdly, there is evident heterogeneity in the impact of the ICP policy on UEP. Compared with central and western cities, non-resource-based cities, and cities with high administrative levels, the construction of innovative cities can promote the improvement of UEP in eastern cities, resource-based cities, and cities with low administrative levels.

Finally, further analysis shows that the innovative city pilot policies not only can significantly promote the improvement of UEP in the individual city but also promote the improvement of UEP in surrounding cities, presenting a significant policy spillover effect.