Introduction

The World Commission on Environment and Development defines sustainable development as “development that meets the needs of the present without compromising the ability of future generations to meet their needs.” The environmental issues arising from China’s economic development have received widespread attention from the international community. The Chinese government has taken a series of initiatives and has achieved significant results. However, China is still in a period of mid-industrialization, and its industries are highly energy intensive. The continued high demand for energy from industrialization has led to continuous environmental degradation, and the Chinese government’s management of the environment needs to continue to be strengthened. In the Global Environmental Performance Index 2022 report (See Appendix), China ranks 160th out of 180 countries and territories. In addition, the adoption of the Sustainable Development Goals (SDGs) has required the government to implement new environmental governance measures. As important sites for industrial development, industries in urban industrial parks emit pollutants such as sulfur dioxide and nitrogen oxides. Environmental pollution seriously endangers the lives of residents and hinders green economic development (Bakıcı et al. 2013; Sarma and Sunny 2017). The operation mode of traditional industrial parks is unsustainable. The construction of eco-industrial parks (EIPs) aims to achieve ecological environmental protection and develop a circular economy and has gradually become an important way to promote urban development.

The concept of EIPs was first transformed from the concept of “industrial symbiosis,” which is a new type of industrial organization based on the theory of circular economy and the principle of industrial ecology and is a gathering place of eco-industry and an important link in the construction of a circular economy. To stimulate the realization of the goal of building the industrial symbiosis system, the Chinese government officially promoted the national-level eco-industrial park (NEIP) pilot policy. China’s NEIP pilot policies were officially launched in 2008, and thus far, the construction of the first generation of eco-industrial parks has been completed. Unlike the construction model of EIPs in other countries, the NEIP serves as a comprehensive park that integrates industrial production and residential life. In addition, these industrial parks are also showing clear signs of urbanization due to the gradual domination of the economy by secondary and tertiary industries (Yu et al. 2015) and are becoming part of the cities in which they are located, influencing urban development.

The literature on NEIPs is dominated by case studies (Lambert and Boons 2002; Shi et al. 2010), qualitative analysis (Zhang et al. 2010), and evaluation of NEIP development (Bai et al. 2014; Shi et al. 2012). Most empirical studies on the impact of NEIP policies on urban development, on the other hand, have focused more on the direct impact of the NEIP on a single indicator such as the economy or the environment. Moreover, few scholars have considered the direct effects of the NEIP on pilot cities and the possible indirect effects on the surrounding areas. As the embodiment of the concept of industrial ecology and the circular economy, the NEIP has a significant impact on the economic and environmental development of its region. NEIP policies promote the economic development of pilot cities while improving the environmental situation by applying the principles of industrial ecology and technological innovation to achieve cleaner production design. Furthermore, the circular economy model adopted by NEIP enables a symbiotic integration of industries across cities, sharing resources and exchanging byproducts. Additionally, the establishment of NEIP brings a set of preferential policies that can attract high-quality enterprises. Therefore, NEIP policies can indirectly affect sustainable regional development by modifying intercity factor transfers that affect both regional economic and environmental development. To explore the impact of NEIP pilot policies on the co-development of urban economy and environment, this study aims to address two main focal points. Firstly, can the NEIP policies effectively contribute towards the co-development of economy and environment in the pilot cities? Secondly, considering the spatial impact of the NEIP policy, to what extent does it affect the economic and environmental growth of the neighboring cities and how far does its spatial influence extend into the surrounding areas? By answering these questions, we can gain valuable insight into the impact of NEIP pilot policies and their point-to-point rollout on urban sustainable development, thereby offering a fresh perspective to address the aforementioned concerns.

To answer the above questions, this study presents a comprehensive theoretical and empirical analysis of the impact of NEIP policies on the economic and environmental co-development of cities and regions using panel data of 271 prefecture-level cities in China from 2004 to 2018 using a staggered difference-in-differences (DID) and spatial staggered difference-in-differences (SDID) models. This study makes three main contributions and innovations. First, this study uses an SDID model based on the spatial Durbin model (SDM), which provides reliable empirical support for the theoretical assumptions under the most robust conditions possible, allowing us to control for the spatial effects established by NEIP policies and satisfy the stable unit treatment value assumption (SUTVA) of the DID model (Rubin 1974). Second, by setting different distance thresholds on the spatial weight matrix, this study precisely estimates the scope of the policy spatial effect of the national eco-industrial park pilot policy on urban eco-efficiency and investigates the spatial decay boundary of the NEIP on the eco-efficiency of the surrounding cities, which has not been done in previous articles studying the NEIP. Third, this study uses the number of NEIPs owned by each city to measure the policy intensity based on the interleaved SDID model, thus ensuring the accuracy and reliability of the conclusions and overcoming the shortcomings of other studies that focus only on whether NEIPs are built but ignore the actual policy intensity differences.

The remainder of this study is organized as follows: the “Literature review” section discusses the EIP- and NEIP-related literature review; the “Policy background and theoretical hypothesis” section returns to the institutional context of the NEIP and develops theoretical hypotheses; the “Estimation strategy” section establishes the model of the empirical analysis and proposes the study design; the “Empirical analysis” section presents empirical findings; the “Discussion” section discusses the results of empirical analysis; and the “Concluding remarks and policy implications” section summarizes the limitations and main conclusions and provides policy implications.

Literature Review

Some literature provides case studies of the sustainability effects of the NEIP. Using evidence from environmental engineering studies, Nie et al. (2022) found that NEIP policies depress the peak of the environmental Kuznets curve and reduce urban carbon intensity. Qian et al. (2022) similarly found that NEIP policies can reduce urban carbon emissions by improving energy use efficiency. In addition, Jung et al. (2012) conducted an logarithmic mean Divisia index (LMDI) analysis of industrial parks in Korea and found that the establishment of EIPs can significantly reduce energy intensity and hence carbon emission intensity. In a case study of Tianjin-TEDA Industrial Park, Yu et al. (2014) found that TEDA achieved a high-quality transition from a traditional industrial park to a NEIP through the promotion of NEIP policies in five areas: institutional activities, technological promotion, economic and financial factors, information activities, and corporate activities. Chen et al. (2017) found that NEIP policies attract interfirm alliances to relocate into NEIPs and improve industrial environmental performance. In addition to improving the industrial environment, the NEIP policy has been very significant in terms of co-benefits. Liu et al. (2018) concluded from a case study of the Dalian Economic Development Zone that NEIPs contributes not only to economic profitability and resource conservation but also to greenhouse gas (GHG) emission reduction and environmental services and that its co-benefits go far beyond simple direct economic benefits.

There are also some studies focusing on the environmental impact of NEIPs. Under the premise of building a resource-saving and environmentally friendly society, China’s NEIPs have made progress in clean production and environmental improvement (Zhang et al. 2010). For example, Fan and Fang (2020) found that NEIPs’ industrial symbiosis system was highly effective in improving the ecological environment. In their case study of Jiangyin Hi-Tech Park, they found that the energy-ecological footprint deficit and energy-ecological footprint intensity of the park were reduced by 16.75% and 16.74%, respectively, after the implementation of eco-industrial development. In addition, Tian et al. (2014) conducted a series of empirical analyses on the effects of NEIP policies on pollution emissions from industrial enterprises and found that NEIPs significantly reduced the COD (chemical oxygen demand) emissions of enterprises in industrial waste. Moreover, Sacirovic et al. (2019) found that NEIPs can reduce multiple types of pollution emissions and reduce water wastage while improving living conditions in urban areas and can increase the efficiency of the use of many natural resources. In fact, the indirect impact of NEIPs on the regional environment is even more significant. Song and Zhou (2021) investigated the effect of NEIPs on urban industrial SO2 emissions using panel data from 2003 to 2017 using a DID model. They found that industrial agglomeration due to improved efficiency in pollution control and energy use could reduce industrial sulfur dioxide (SO2) emissions by approximately 2.5%. Cao et al. (2022) used DID and SDID models and found that NEIPs can significantly reduce SO2 and carbon dioxide (CO2) emissions in the surrounding areas through technology spillover and demonstration effects.

There are also studies focusing on the economic impact of NEIPs. Gibbs and Deutz (2007) argue that NEIPs can create jobs, improve working conditions, and thus promote economic development. Liu et al. (2021) use NEIP policies as a quasi-natural experiment and find that NEIPs can promote total factor productivity in cities through structural upgrading, government competition, and economic agglomeration while finding that it significantly increases the green economic efficiency of cities in pilot cities. Similar to the indirect environmental effects of the NEIP, its indirect effects on the economy are also significant. Wu et al. (2021) calculated the number of urban green innovations in China from 2003 to 2018 and used NEIPs as a quasi-natural experiment. They found that NEIPs significantly improved the economic and environmental conditions of the cities in the pilot city due to the increase in foreign direct investment, urban research investment, and green technological innovation. Hu et al. (2021) found through a case study of the Beijing Economic Development Zone (BEDZ) that by optimizing the industrial structure, the BEDZ was able to achieve 186% economic growth from 2020 to 2025.

However, there are still some shortcomings in the existing studies, both in the econometric and ecological fields. The aim of these studies is to explore the relationship between NEIP and urban development. However, these studies have not explored the spatial effects of NEIP on urban development. Compared with traditional development zones, the EIP may serve as a key method for achieving ecological environmental protection and developing a circular economy development model Gibbs and Deutz (2007). NEIP policies aim to enhance the management of NEIPs and improve the efficiency of business operations and resource conversion by sharing the infrastructure of neighboring enterprises. Thus, NEIPs have an impact on regional economic and environmental development through intercity factor transfer, which in turn has a spatial impact on regional sustainable development. If the effects of these spatial effects are ignored, the robustness and credibility of the empirical findings are weakened. Some studies examine the mechanisms by which NEIPs affect individual cities at the economic or environmental level. However, the studies mentioned above do not consider the important role of elements such as human capital, infrastructure, and even government actions in sustainable development. Therefore, conducting research on the economy or the environment alone does not represent sustainable development. In addition, since the NEIP was established with the goal of achieving a win–win situation for economic, environmental, and social development, the use of a single indicator for the economy or the environment does not adequately measure the effect of the NEIP on the co-development of the economy and environment. Previous studies have tended to use NEIP as exogenous policy shocks to study a single impact on economic or environmental indicators. The NEIP policy aims to achieve green and high-quality development, which requires the analysis of comprehensive indicators of eco-industrial parks on the efficiency of urban economic and environmental development. Therefore, this study introduces the concept of ecological efficiency, which is a comprehensive indicator to measure the economy and environment.

Policy background and theoretical hypothesis

Background on NEIP policy

In order to facilitate the establishment of an industrial symbiosis system, the Chinese government officially inaugurated the NEIP pilot program in 2008. Despite the relatively brief duration of its implementation, both the central and local governments have introduced a range of policies, including laws and regulations pertaining to circular economy, ecological industry, energy efficiency, and emissions reduction. These policies have been implemented at the national level, and more inclusive local and regional policies have been formulated by integrating the national policies and regulations with the distinct natural resources and economic endowments of each region.

In accordance with the Regulations of the State Environmental Protection Administration of China on NEIP Declaration, Naming, and Management, the process of establishing an NEIP consists of six stages: declaration, creation, acceptance, naming, supervision, and management. The primary objective is to foster circular economic development in China, while simultaneously reducing the output of pollutants and waste and enhancing the utilization of resources and energy to improve the environmental quality of both urban and regional areas. To be eligible for NEIP status, an industrial park must possess functional leadership and institutional support, demonstrate a clear model of circular economy, and achieve a balanced development that benefits both society and the economy, while also being environmentally sustainable. Generally, the industrial parks that satisfy the aforementioned conditions are those with a certain foundation of development. In fact, most of the officially recognized NEIPs were originally economic development zones or high-tech zones. As of the end of 2016, China had established a total of 93 NEIPs, including 45 parks under construction and 48 approved to establish NEIPs, encompassing 29 provinces and autonomous regions throughout China. Currently, the construction of NEIPs is gradually being promoted. The Ministry of Ecology and Environment, Ministry of Commerce, and Ministry of Science and Technology (See Appendix) decided to approve 12 parks, including Shanghai City North High-Tech Service Park, as NEIPs to deepen construction.

Theoretical hypothesis

The NEIP and urban eco-efficiency

The economic and environmental co-development impact of the NEIP is not only within the park, but it may also affect the economic and environmental development of the whole city through the synergistic development of infrastructure. Yu et al. (2015) found that EIPs can develop into eco-cities when they can improve their environmental performance and promote the economic prosperity generated by their industrial sites. Moreover, the improvement of energy and resource efficiency caused by NEIPs is also key to achieving sustainable urban development. Yedla and Park (2017) found that NEIPs can improve decarbonization at the city level through eco-industrial networking constituted by the development of eco-industries linked to urban systems, which in turn leads to sustainable urban development.

NEIP policies can have an impact on the economic and environmental co-development of the surrounding cities. A certain study has discovered that NEIP policies possess a powerful spatial spillover effect, which encourages a reduction in carbon emissions within a radius of 450 km through the adoption of energy efficiency improvements. Furthermore, the spillover effects of industrial agglomeration induced by NEIPs have significant implications for the promotion of sustainable development (Yu et al. 2014). Specialized and diversified industrial agglomerations generate negative externalities to the environment through overcrowding and concentrated emission effects, and both include significant spatial spillover effects (Hong et al. 2020), which have a negative effect on the development of the surrounding areas. Furthermore, although the absorption of various factors by the establishment of NEIPs can have a positive external impact on the sustainable development of the pilot city, for its surrounding areas, the siphoning effect of NEIPs causes the loss of its production factors, which is also detrimental to the achievement of its sustainable development. Thus, Hypothesis 1 is proposed.

  • Hypothesis 1: NEIP policies have a catalytic effect on the eco-efficiency of the pilot cities but a negative spatial effect on the eco-efficiency of the surrounding cities.

The “crowding out” effect of heavily polluting enterprises

To achieve the goal of industrial symbiosis, the NEIP influences the sustainability of cities through measures such as setting emission reduction targets for enterprises in parks (Cao et al. 2022). Primarily, the NEIP facilitates the establishment of a mutually beneficial integration of various industries across different enterprises within a city as well as between cities. By doing so, the NEIP can effectively tackle environmental pollution while simultaneously promoting harmony with the environment, through the regeneration of industrial waste. In this regard, the waste generated in the upstream production process is transformed into raw material for the downstream production, thereby enabling the reuse of essential elements and improving the overall quality of the urban environment (Geng et al. 2016). In addition, government subsidies and tax incentives for NEIPs will attract enterprises or production factors to NEIPs (Zhao et al. 2021), and the spatial-level factor mobility caused by this phenomenon will constrain the economy and environment of the surrounding areas. Moreover, the government focuses on increasing policy research and key projects for NEIPs. Due to the rarity of NEIPs and the rigorous application process, the incentives provided are highly appealing. As a result, more capital inflows and enterprises are drawn to the park, which in turn promotes the economic growth and technological advancement of both the park and the city, thereby enhancing overall development. It is worth noting that most of the enterprises and elements that benefit from NEIPs are environmentally friendly or green high-technology enterprises. Most of the enterprises attracted to NEIPs are likely to have high technology and low pollutant emissions (Susur et al. 2019), and the presence of these enterprises can cause heavily polluting enterprises to be crowded out. Furthermore, owing to the development objectives and obligations of NEIPs, heavily polluting enterprises operating within NEIPs are mandated to fulfill the emission reduction targets stipulated by NEIP policies. Consequently, those heavily polluting enterprises that fail to conform to NEIPs’ clean production standards will be compelled to relocate to the surrounding areas due to NEIPs’ restrictions on enterprise pollution emissions. These areas, therefore, are more susceptible to pollutant emissions resulting from the relocation of heavily polluting enterprises that have been displaced, impeding the attainment of SDGs.

Technological innovation affects

NEIPs’ treatment or utilization of waste (i.e., solid waste and used energy) for local industries to achieve a win–win situation is a breakthrough (Dong et al. 2016), which cannot be supported by technological innovation. As the cradle of innovation, companies play an important role in this process. NEIPs have built a good platform for enterprise technology innovation. First, clusters of enterprises within NEIPs concentrate innovation resources and stimulate extensive exchanges between enterprises (Park and Lee 2004). In addition, NEIPs can accelerate the exchange and application of innovation by strengthening cooperation through exchanges with external markets, and this coupling has a significant impact on the technological innovation capacity of cities (Wu and Gao 2022). In addition, the tax incentives and subsidies provided by NEIPs to enterprises drive their innovation motivation and level of innovation. These tax incentives and subsidies can directly promote the innovation ability of enterprises and can drive high-quality talent to enter NEIP pilot cities. With this gathering of talent, human capital can interact with production factors inside and outside the park to generate knowledge spillover, accelerate the rate of innovation diffusion, and improve the overall innovation capacity of the city (Tang-wei et al. 2019), which in turn provides technological guarantees for sustainable urban development. According to the growth pole theory, innovation emerges first in regions where conditions are favorable (Benedek 2016). The preferential subsidies and priority development policies enjoyed by NEIPs strengthen the clustering of innovation factors and absorb human and physical capital from neighboring regions to a certain extent. As costs fall sharply and labor demand expands, innovation in cities requires greater returns through expanded production patterns (Banerjee et al. 2010). The unidirectional flow of innovative factors towards the central city may lead to a siphoning effect, which, in turn, could widen the innovation development gap between the NEIP and its neighboring regions. Such a situation implies that neighboring regions would have to forfeit their own innovation enhancement levels to facilitate the increase in innovation development momentum in the central city. Moreover, the siphoning effect of the pilot city would entail a continuous loss of human and physical capital of the cities that are adjacent to the NEIP pilot cities. This, in turn, would lead to a weakening of their innovation factors and thus impede the achievement of their Sustainable Development Goals. Thus, Hypothesis 2 is proposed.

  • Hypothesis 2: The spatial effect of NEIPs is achieved through the extrusion of heavily polluting enterprises with technological innovation clustering and siphoning.

The theoretical mechanism framework of this study is shown in Fig. 1.

Fig. 1
figure 1

Theoretical mechanism framework

Full size image

Estimation strategy

Econometric models

Previous studies have encountered potential endogeneity problems in establishing causal relationships between the economic and environmental impacts of NEIP. To address this issue and account for differences in the timing of NEIP establishment, this study employs a staggered DID model. The model places the NEIP pilot, a policy shock, in the same model at different times and treats the cities that start to have NEIP at a certain time as the treatment group and the others as the control group. This generates a double difference between the treatment and control groups, as well as between the “pre-treatment” and “post-treatment” groups, and effectively addresses endogeneity problems such as omitted variables and reverse causality. The staggered DID model is set up as Eq. (1):

$${\mathrm{lnEE}}_{it}=\alpha +\beta {\mathrm{DID}}_{it}+\gamma \sum {\mathrm{Control}}_{it}+{\delta }_{i}+{\mu }_{t}+{\varepsilon }_{it}$$
(1)

where subscripts i and t denote city and year, respectively; \(\mathrm{lnEE}\) represents eco-efficiency; DID represents the independent variable; \(\mathrm{Control}\) represents a vector of the control variables; \(\varepsilon\) is an error item; \(\alpha\), \(\beta\), \(\gamma\), \(\delta\), and \(\mu\) are the parameters to be estimated.

Nonetheless, according to the theoretical hypothesis outlined in the preceding section, there exists a possible spatial effect in the implementation of the NEIP. This spatial effect implies that the traditional DID model may not provide consistent estimates as it violates the stable unit treatment value assumption (SUTVA) of the DID model. Therefore, to ensure an accurate assessment of the impact of NEIP on regional development and account for potential spatial effects, this study employs a spatial staggered difference-in-differences (SDID) model for empirical analysis.

Furthermore, in spatial econometric models, spatial correlations can arise from dependent variables, independent variables, or error terms. Common models include the spatial lag model (SLM), spatial autoregressive model (SAR), and spatial Durbin model (SDM). The SDM is a more general model as it combines and extends the SLM and SEM models by adding corresponding constraints. Additionally, the SDM is a SAR model that incorporates spatial lag variables (Elhorst 2014). Compared to the traditional SDID model, a staggered SDID model that includes NEIP establishment can more accurately assess the impact of NEIP establishment on economic and environmental co-development. Therefore, this study uses Eq. (2), the SDM-based staggered SDID model, for the empirical analysis:

$${\mathrm{lnEE}}_{it}=\alpha +\rho W{\mathrm{lnEE}}_{it}+\omega W{\mathrm{DID}}_{it}+\beta {\mathrm{DID}}_{it}+\gamma \sum {X\mathrm{Control}}_{it}+\xi W\sum {\mathrm{Control}}_{it}+{\delta }_{i}+{\mu }_{t}+{\varepsilon }_{it}$$
(2)

where W is an element of the spatial weight matrix which describes the spatial proximities between cities and ρ, ω, and ξ are the parameters to be estimated, while the remaining variables and parameters to be estimated are the same as in Eq. (1).

Assessment of eco-efficiency

Eco-efficiency has become a consistent tool in the transition to sustainable development, and eco-efficiency indicators have been used for comparative research and decision-making tasks, as well as to provide a better evaluation of financial, environmental, and social performance (Caiado et al. 2017). Eco-efficiency is an improved measure of sustainability and is defined by the WBCSD (World Business Council for Sustainable Development) as meeting human needs and reducing ecological impacts and resource use on the basis of creating economic benefits (Ehrenfeld 2005). Eco-efficiency plays an important role in assessing sustainable development due to its correlation with environmental and economic performance. Compared to assessing the economic benefits of NEIP or changes in pollution emissions, calculating eco-efficiency considers economic, environmental, and external factors and provides a more comprehensive measure of regions’ sustainability. Moreover, eco-efficiency is closely related to economic and environment co-development, and its importance is evident in the concept of economic and environment co-development (Stergiou and Kounetas 2022). This is because the concept of eco-efficiency favors more realistic policy measures that can limit domestic economic activity to a lesser extent and lead to higher levels of sustainability (Kuosmanen and Kortelainen 2005).

The existing methods for measuring eco-efficiency mainly use the data envelopment analysis (DEA) method (Moutinho et al. 2020) as it solves the problem of inconsistent units of various resource consumption and pollution emissions in the eco-environmental impact item of eco-efficiency measurement indexes. In this study, a three-stage DEA model is used to measure urban eco-efficiency. Unlike the traditional DEA model, the three-stage DEA model can eliminate the bias in efficiency measurement caused by stochastic factors due to the management inefficiency term, resulting in more accurate eco-efficiency measurement (Fried et al. 2002). The measurement steps are shown in the Appendix.

Variable selection

Key independent variable

Starting in 2001, the Chinese government initiated the construction of EIPs. Based on this, this study compiled a total of 93 industrial parks that were approved to build NEIPs or have become NEIPs from 2001 to 2016 according to the Model List of National Eco-Industrial Zones. From this selection, this study identified the NEIPs that have been officially designated as quasi-natural experiments.

Control variables

Industrial structure (IUI)

The realignment of the industrial structure is a critical development goal for the economy, as it has a substantial influence on the harmony between economic development and the environment. Accordingly, this study employs the ratio of added values of the tertiary and secondary industries to quantify the industrial structure.

Retail sales of social consumer goods (con)

Consumer demand influences the urban environment by changing firms’ production preferences, which in turn has an impact on the assessment of eco-efficiency. This study uses the logarithm of retail sales of social consumer goods as the proxy variable to measure the retail sales of social consumer goods.

Urban green investment

Infrastructure directly or indirectly affects SDGs. Specifically, the level of infrastructure development can promote economic development and reduce environmental pollution. This study uses the logarithm of green space area (lngreen) and urban energy saving and environmental protection expenditure (greenexp) to measure the level of urban green development.

Foreign direct investment (FDI)

Foreign direct investment (FDI), as an indicator of regional openness, exerts a significant influence on both the economic and environmental development of a city. With regard to the impact of FDI on pollution emissions, two hypotheses have been put forth: the pollution paradise and the pollution halo hypotheses. The pollution paradise hypothesis postulates that FDI exacerbates regional environmental degradation by transferring polluting industries to the region. On the other hand, the pollution halo hypothesis suggests that FDI has the potential to reduce environmental pollution by introducing green technologies through spillover effects. Therefore, FDI plays a critical role in the co-development of economic and environmental factors. In this study, annual foreign direct investment in prefecture-level cities, converted using the current year’s exchange rate, is employed as the measure of FDI.

Population density (den)

The concentration of urban population and the size of a city can have an impact on the level of regional economic development and the amount of pollution emitted, which, in turn, can affect the calculation of eco-efficiency. Thus, this study employs the measure of population density, defined as the number of individuals per square kilometer, to gauge the impact of the agglomeration of the urban population and the size of the city.

Environmental regulation

Flexible environmental regulatory policies can significantly contribute not only to technological innovation but also to economic and environment co-development (Yuan and Zhang 2020). Moreover, as one of the ways in which the government manages the environmental behavior of companies, environmental regulation can effectively limit the harmful effects of economic activities on the environment (Afshar Jahanshahi and Brem 2018). Therefore, environmental regulations have an impact on economic and environment co-development. Previous studies have mostly used firm-level variables such as emission costs to measure the intensity of environmental regulations, but these data are difficult to obtain at the city level (Xie et al. 2017). Therefore, according to Xie et al. (2017), this study uses the rate of solid waste disposal (regulation) to measure environmental regulation.

Mechanism variables

The “crowding out” of heavily polluting enterprises

Serious environmental problems caused by heavily polluting enterprises have a great negative impact on urban eco-efficiency. In addition, the emission reduction of heavily polluting enterprises is also an important goal of the NEIP construction plan. This study obtains information on the number of heavily polluting enterprises contained in each city by collecting information on industries and enterprises of listed companies and screening and determining the geographical location of heavily polluting enterprises. This study uses the number of heavily polluting enterprises in each prefecture-level city as a proxy variable for heavily polluting enterprises substituted into Eq. (2) for regression.

Technological innovation

High-tech zones and economic development zones are the backbones of urban innovation, within innovation being the leading force and the motivation for development. Innovation is divided into substantive innovation to promote technological progress and gain competitive advantage and strategic innovation for the purpose of seeking other benefits and pursuing “speed” and “quantity” of innovation output through different innovation behaviors. The product of the substantive innovation orientation is often invention patents that are primarily inventive in nature, while the product of strategic innovation orientation is non-invention patents that are intended to receive ex post government subsidies and include rent-seeking behavior. As an extended form of the high-tech zone (Liu et al. 2022; Yang et al. 2022), the EIP continues the innovative role of the high-tech zone while incorporating the government’s ardent expectations for eco-efficiency. Therefore, this paper regresses urban green invention patent innovation as a proxy variable for green innovation capacity as the mechanism variable.

Identify assumptions and checks

The definitive assumption for the DID estimates is that if the NEIP policy is not adopted, then the pilot cities will follow the same time trend as the non-pilot cities. However, the main threat to this hypothesis is that the pilot cities were not randomly selected and may have resulted from some inherent differences between cities with and without NEIP. To address this issue, this study employs the approach used by Li et al. (2016) to address and identify this problem. Specifically, key factors for NEIP selection were identified, and trends in the differences in outcomes between cities with NEIP and cities without NEIP resulting from these determinants were controlled for after the NEIP policy was implemented. The goal of NEIP construction is to achieve “clean production and circular economy,” so this study chose the altitude of each city (altitude), the amount of investment in pollution control (thousand million yuan) (waste), the logarithm of public expenditure (pub), and whether it is an ecological civilization city (EC, 1 if it is, 0 if it is not) as the key selection criteria in the year before the policy was implemented. The results and descriptions of the identified assumptions and checks are performed in the Appendix (Table B).

The treatment and control groups are balanced after adding these variables, which is crucial for the subsequent empirical analysis.

Data description

Considering that the first NEIP approved by the Chinese Ministry of Environmental Protection (renamed as the Ministry of Ecology and Environment in 2018) was in 2008, we selected annual data for 271 prefecture-level cities from 2004 to 2018 for regression, which excluded cities with severe missing data. According to the NEIP Pilot List, a total of 30 cities including 93 approved NEIPs were selected as the treatment group for the empirical analysis, and the rest were the control group, with a total of 271 prefecture-level cities. The raw data we used in this study were obtained from authoritative databases and statistical yearbooks: where the industry structure data is from the CNRDS database, urban energy conservation and environmental protection expenditures are from the China Energy Statistical Yearbook, solid waste disposal volume and environmental pollution control investment amount are from the China Environmental Statistical Yearbook, and population density, green area, foreign direct investment, and public finance expenditures are from the China Urban Statistical Yearbook and the statistical yearbooks of each province. Some missing values are extrapolated through the interpolation method. To ensure the comparability of the data, we deflate them to 2004 constant prices. The descriptive statistics of the above variables are shown in Table 1.

Table 1 Data description
Full size table

Empirical analysis

Resolution of heterogeneous treatment effects

If the processing effects are not identical for all individuals, this is called a “heterogeneous treatment effect.” For example, in this study, the effect of NEIP construction on cities’ eco-efficiency is examined. Under the homogeneous treatment effect, the changes in NEIPs affecting urban eco-efficiency should be the same, but obviously, this is unrealistic. Although this study controls for two-way fixed effects and performs a series of robustness tests, heterogeneity in treatment effects causes the negative weighting of treatment effects. Despite the sum of all weights being 1, the presence of negative weights may lead to unrobust estimation results under two-way fixed effects. Therefore, according to De Chaisemartin and d’Haultfoeuille (2020), this study examines whether the two-way fixed effects are robust before starting the empirical analysis.

The results are shown in Fig. A in the Appendix, where all weights are positive among all 168 weights and the sum of the positive weights is 1. The metric is estimated to be approximately 0.259 under the heterogeneity treatment for the positive weight indicator. Therefore, the two-way fixed effects model used in this study can be considered robust under the heterogeneity treatment effect.

Parallel trend test

One of the requirements for applying the DID model is to satisfy the parallel trend assumption. That is, the changes in eco-efficiency of the treatment and control groups need to be consistent before the NEIP policy begins. Following Armitage (1995), this study uses the event study method based on Eq. (3) to test the parallel trend assumption:

$${\mathrm{ln\;EE}}_{it}=\alpha +\sum\nolimits_{k=1}^{K}{{F}_{k}\mathrm{DID}}_{i,t-k}+\sum\nolimits_{m=0}^{M}{{L}_{m}\mathrm{DID}}_{i,t+m}+\gamma \sum {\mathrm{Control}}_{it}+{\delta }_{i}+{\mu }_{t}+{\varepsilon }_{it}$$
(3)

where \({\mathrm{DID}}_{i,t-k}\) represents the event window dummy variables before NEIP policy implementation (k = 1, 2,…, K), which is set to test the treatment group and the control group before the promulgation of NEIP policy whether there is the same trend. A non-significant \({F}_{k}\) coefficient implies no significant systematic difference between the treatment and control groups prior to the NEIP policy implementation, supporting the hypothesis of parallel trends. \({\mathrm{DID}}_{i,t+m}\) represents the event window dummy variable after the implementation of NEIP policy implementation (m = 1, 2,…, M) year of the promulgation of NEIP policy, the post terms are set to identify the lasting impact of NEIP construction. Other variable settings are the same as Eq. (1). Because of the issue of sparse samples (refer to Table C in the Appendix), this study adjusts for the timing of policy implementation. Years with a relative time before policy implementation greater than 8 years are categorized as the eighth year before policy implementation, while years with a relative time after policy implementation greater than 7 years are categorized as the seventh year after policy implementation. Consequently, this study employs a parallel trend test covering 8 years before and 7 years after policy implementation.

The results depicted in Fig. 2 indicate that the assumption of consistent time-varying trends of eco-efficiency holds for both cities with and without NEIPs before the pilot policy’s implementation. Thus, the parallel trend assumption is reasonable, and the construction of NEIPs can be treated as a quasi-natural experiment for investigating its effects on eco-efficiency. Furthermore, following the construction of NEIPs, its impact on urban eco-efficiency gradually increases and reaches its peak in the sixth year.

Fig. 2
figure 2

Parallel trend test

Full size image

Benchmark regression results

According to Eq. (1), the regression results of the impact of NEIP construction on the local eco-efficiency are shown in Table 2 using OLS.

Table 2 Benchmark regression and robustness test results
Full size table

Both columns (1) and (2) show that NEIP has a significant positive effect on urban eco-efficiency. This study’s analysis is based on the regression results of column (2), i.e., Eq. (1) with the inclusion of control variables. Holding other conditions constant, the construction of NEIP leads to a 0.069 increase in urban eco-efficiency.

Considering the time lag effect of NEIP construction, this study includes a 1-year time lag for the independent variable in Eq. (1). The regression results in column (3) demonstrate that the policy continues to have a significant positive impact on urban eco-efficiency, with an improvement of approximately 0.06.

Robustness test

Placebo test

In Eq. (1), this study includes the control variables from the previous benchmark regression results, but it is uncertain whether unobservable factors still interfere with the improvement effect of NEIP construction on eco-efficiency. To ensure the robustness of the estimation results, this study employs a counterfactual framework and conducts a placebo test on the benchmark regression results. The specific steps of the placebo test are as follows. To obtain 500 false variables \({\mathrm{DID}}_{\mathrm{false}}\), we first extract the NEIP policy time and perform random sorting 500 times. Second, we replace the original variables DID with \({\mathrm{DID}}_{\mathrm{false}}\) to form 500 new samples while keeping the order of the other variables. Finally, we use 500 new samples to estimate the effect of \({\mathrm{DID}}_{\mathrm{false}}\). By using this method, we can determine whether the benchmark regression results are “chance findings” brought on by the effects of missing variables.

According to Fig. 3, the estimated coefficients followed a normal distribution, with all 500 regression coefficients estimated to the left of the actual coefficient value of 0.069 for the estimation result of Eq. (1), as expected in the placebo test. The P value, which was greater than 0.1 (not significant at the 10% level), indicates that the estimated result was not a chance finding. Thus, the results of the placebo test confirm that the increase in eco-efficiency resulting from NEIP construction is not due to other random or unobservable factors, ensuring the reliability of the benchmark regression results.

Fig. 3
figure 3

Placebo test

Full size image

Excluding the interference from other policies

Low-carbon city policy

The objective of the low-carbon city policy is to construct a sustainable development model and establish a low-carbon society as a prototype and example. The implementation of this policy aims to enhance the ecological environment of cities and promote economic growth. However, as a consequence, the estimation of urban eco-efficiency may be impacted, and the results obtained in the benchmark regression may be biased due to the influence of this policy.

National-level economic development zone and high-tech zone

Based on our statistics, 80% of the NEIPs were previously economic development zones and high-tech zones, which influence urban economic development and environmental pollution. Hence, the notable impact of NEIP construction on eco-efficiency may be affected.

To eliminate this concern, we add two dummy variables, LC and Former, which indicate whether LC policy and national-level economic development zone and high-tech zone have reached a city, as control variables in Eq. (1).

Column (4) in Table 2 shows the estimation result that controls the low-carbon city policy and National-level economic development zone and high-tech zone, respectively. The effects of the construction of NEIP are still significantly positive after controlling these policies. Thus, the result of benchmark regression remains robust.

PSM-DID

Although the control variables have been identified and checked in this study, a series of control variables have been chosen to ensure that the benchmark regression results are as robust as possible. However, since eco-efficiency is not completely exogenous, the government may set up pilot projects in eco-industrial parks with good eco-efficiency, which can lead to erroneous estimates of the regression results. According to Rosenbaum and Rubin (1983), this study uses propensity score matching (PSM) to perform robustness tests on the sample data. This study uses the propensity score and 1:1 nearest neighbor matching to match the treatment and control groups. The matched samples were brought into Eq. (1) as control variables for regression.

Table D in the Appendix shows the balance test results of each matching control variable. The self-selection bias of the samples was significantly reduced after the balance test. The absolute value of the standard deviation of all matching control variables was less than 20% after matching, and the t-statistic was not significant, so the balance test was passed. The regression results after PSM are reported in column (5) of Table 2, and the regression results show that the improvement of eco-efficiency by NEIP remains robust.

Spatial effect analysis

SUTVA is one of the assumptions of the DID model (Rubin 1974). In other words, exogenous shocks only impact certain individuals and do not affect others. Failing to meet this assumption would significantly compromise the reliability and robustness of the estimated results in Eq. (2). Given the regional relevance of China’s policies, the eco-efficiency of local cities and surrounding areas is affected by the NEIP policy, which can be achieved through economic “siphon” and “diffusion” effects as well as the dispersion of pollutants. To control for these spatial effects and satisfy the SUTVA of the DID model, this study utilizes a spatial DID model based on a spatial Durbin model (SDM).

Spatial weight matrix

To estimate the spatial effect, this study constructed a common geographic distance weight matrix (W1) and a spatial boundary matrix (W2). As the direct spatial effect of NEIP mainly involves the absorption effect of resources and production factors in other regions, the absorption intensity of resource factors may decrease with the expansion of intercity distance within a certain range. Therefore, this study selected the geographic distance matrix as one of the spatial weight matrices. Additionally, we used the common boundary matrix (W2) in regression to ensure the robustness of the geographic distance matrix results. The elements of \(W1\) and \(W2\) are shown in Eqs. (4) and (5):

$$W1=\left\{\begin{array}{l}\frac{1}{{D}_{ij}},\;\;\;\;\\ \;\;\;0 ,\end{array} \begin{array}{l}i\ne j\\ i=j\end{array}\right.$$
(4)
$$W2=\left\{\begin{array}{c}1,\\ 0,\end{array} \begin{array}{l}\mathrm{adjacent}\\ \mathrm{not\;adjacent}\end{array}\right.$$
(5)

where \({D}_{ij}\) represents the geographic distance between city i and city j, which is calculated from the latitude and longitude of each city, respectively. \(W2\) represents whether the border between cities i and j is contiguous. If it is adjacent, it is assigned a value of 1 and not adjacent to 0.

Spatial correlation test

This study performs global spatial correlation tests for eco-efficiency, where a positive (negative) spatial correlation is indicated by a global Moran’s index greater (smaller) than 0. Table 3 presents the global Moran index of urban eco-efficiency in China from 2004 to 2018, which is significantly positive at the 1% level in all spatial matrices.

Table 3 Global Moran’s index of eco-efficiency
Full size table

Spatial parallel trend test

Under the framework of linear conditional expectation, the spatial effect of the default NEIP policy on the treatment group and the control group is the same. Therefore, in order to accurately analyze the spatial effects of NEIP policy, this study follows the findings of Jia et al. (2021), adding the spatial lag term to Eq. (2) as Eq. (6) to test the parallel trend assumption of the spatial econometric model:

$${\mathrm{ln\;EE}}_{it}=\alpha \mathrm{^{\prime}}+\rho \mathrm{^{\prime}}W\mathrm{^{\prime}}{\mathrm{lnEE}}_{it}+{\sum\nolimits_{k=1}^{K}{F}_{k}\mathrm{DID}}_{i,t-k}+\sum\nolimits_{m=0}^{M}{{L}_{m}\mathrm{DID}}_{i,t+m}+\gamma \sum {\mathrm{Control}}_{it}+{\delta }_{i}+{\mu }_{t}+{\varepsilon }_{it}$$
(6)

where \(W\mathrm{^{\prime}}\) represents the spatial weight matrix and \({\alpha }^{^{\prime}}\) and \(\rho \mathrm{^{\prime}}\) are the parameters to be estimated; other variables are consistent with Eq. (2). This study uses W1 for spatial parallel trend test.

The results of the spatial parallel trend test are shown in Fig. 4. Before the NEIP policy was implemented, the spatial total effect on the treatment and control groups was not significantly different, and \({F}_{k}\) was not significant during all periods. Therefore, it can be assumed that the spatial parallel trend assumption is satisfied. In addition, after the implementation of NEIP, its total spatial effect on regional eco-efficiency had a significant negative effect, indicating a fluctuating downward trend.

Fig. 4
figure 4

Spatial parallel trend test

Full size image

The spatial effect of NEIP on eco-efficiency

Table 4 presents the estimation results of Eq. (2) using MLE, where columns (1) to (3) are the estimation results of the geographic distance matrix W1 and columns (4) to (6) are the estimation results of the boundary matrix W2. This study also reports the estimation of SAR, SLM, and LR test statistics for robustness in addition to the estimation results of the SDM. The independent variables for both matrices in the SLM and SAR are significantly positive at the 1% level. According to these findings and those of the LR test, the benchmark model’s spatial correlations are caused by the dependent variables, independent variables, and error terms, so the SDM cannot be degraded into the SLM and SAR. Therefore, this study turns to discuss the estimation results of columns (1) and (4).

Table 4 The regression results of Eq. (2)
Full size table

The results show that the coefficient of the spatial lagged term of the independent variable is negatively significant at the 1% level for the geographical distance matrix, indicating that the eco-efficiency of the local area has a significant negative effect on the eco-efficiency of the surrounding areas; that is, it demonstrates a siphoning effect. However, the spatial lagged term of the independent variable is not significant, and the estimated coefficients are negative, which may be because W2 is a boundary spatial matrix, which limits the geographic spillover range without infinite diffusion and effect accumulation, thus affecting the estimation of spatial spillover effects. However, due to the nonlinear character of the SDM data generation process, the spatial lag term coefficients of the NEIPs in Table 4 cannot be interpreted as partial effects. Therefore, partial derivatives of the SDM model need to be obtained to decompose its direct effects, indirect effects, and total effects.

Table 5 presents the direct, indirect, and total effect estimation results of Eq. (2), respectively. The geographic distance matrix estimation result is shown in column (1), and the common boundary matrix estimation result is shown in column (2). The estimation results of the direct effect show that the construction of the NEIP has a positive impact on local eco-efficiency. This study takes the results with W1, that is, column (1), as an example for discussion. The direct effect of the NEIP is 0.078 and statistically significant at the 1% level, indicating that the construction of the NEIP increases the local eco-efficiency by approximately 0.078. However, the indirect and total spatial effects are significantly negative at the 1% level. This implies that the NEIP has a siphoning effect on the eco-efficiency of the surrounding areas, due to the improvement of local urban eco-efficiency, ultimately leading to a decrease in the eco-efficiency of the whole region. Thus, Hypothesis 1 is verified.

Table 5 Direct, indirect, and total effects based on the results of columns (1) and (4) in Table 4
Full size table

Identification of policy effects

Number of NEIP

The previous section of this study identified the policy effects of NEIPs. However, as mentioned earlier, there are 93 NEIPs located in 30 cities, meaning that some cities may have more than one NEIP, and the policy effects may vary with the number of NEIPs. Due to the high concentration of enterprises, cities with multiple NEIPs may experience lower production and R&D (Research and Development) costs due to enhanced communication and mutual learning between enterprises. Additionally, cities with more NEIPs may experience a greater siphoning and snowballing effect on human and physical capital than cities with only one or no NEIP, resulting in a greater eco-efficiency improvement effect from NEIP construction.

Location of the NEIP

As the political and even economic centers of provinces and regions, provincial capitals and municipalities have a strong snowball effect and siphoning effect on the resources of the surrounding areas (Li et al. 2022). These cities are given priority to enjoy a series of preferences, such as preferential policies and financial subsidies, thus making them more attractive to various capitals from neighboring cities. In addition, these cities will transfer some heavily polluting enterprises to the surrounding areas to maintain the local ecological environment. For example, the Bohai Sea region has taken over many heavily polluting enterprises relocated from Beijing and Tianjin, causing serious damage to the local ecological environment. Thus, the siphoning effect of provincial capitals and municipalities directly under the central government on the eco-efficiency of the surrounding areas is more obvious.

To identify the policy effects of NEIP at a deep level, this paper constructs a spatial difference-in-difference-in-differences (SDDID) model (Eq. (7)) as follows:

$${\mathrm{lnEE}}_{it}={\alpha }^{^{\prime}}+\varphi {{W}_{i}\mathrm{lnEE}}_{it}+{\beta \mathrm{^{\prime}}\mathrm{DID}}_{it}+\beta {\mathrm{^{\prime}}\mathrm{^{\prime}}\mathrm{DID}}_{it}\times {PE}_{i}+\partial {\mathrm{^{\prime}}\mathrm{DID}}_{it}\times {\mathrm{PE}}_{i}+\partial {\mathrm{^{\prime}}{W}_{i}\mathrm{DID}}_{it}\times {\mathrm{PE}}_{i}+\gamma \mathrm{^{\prime}}\sum {\mathrm{Control}}_{it}+\gamma \mathrm{^{\prime}}\mathrm{^{\prime}}\sum {{W}_{i}\mathrm{Control}}_{it}+{\delta }_{i}+{\mu }_{t}+{\varepsilon }_{it}$$
(7)

where \({\alpha }^{^{\prime}}\), \(\varphi\), \(\beta \mathrm{^{\prime}}\), \(\beta \mathrm{^{\prime}}\mathrm{^{\prime}}\), \(\beta \mathrm{^{\prime}}\mathrm{^{\prime}}\mathrm{^{\prime}}\), \(\partial \mathrm{^{\prime}}\), \(\gamma \mathrm{^{\prime}}\), and \({\gamma }^{\mathrm{^{\prime}}\mathrm{^{\prime}}}\) are the parameters to be estimated and \({\mathrm{PE}}_{i}\) denotes the policy effect, including Number (number of NEIPs owned by cities) and Provcap (dummy variable set to 1 if it is a provincial capital city or a municipality directly under the central government, otherwise 0); the remaining variables are the same as in Eq. (2).

The regression results are shown in columns (1) to (4) of Table E in the Appendix. Columns (1) and (2) show the regression results for the number of NEIP as the independent variable. Columns (3) and (4) show the regression results for whether the NEIP is in the provincial capital city or the municipality directly under the central government, with all the linear term and secondary interaction terms controlled. In this study, we analyze the regression results of W1, i.e., columns (1) and (3).

The regression results show that the direct effects of Number × DID and Provcap × DID are both significantly positive at the 1% level, indicating that the greater the number of NEIPs, or if the NEIPs are in provincial capitals and municipalities, the greater they improve the eco-efficiency of the cities with NEIPs (0.094 and 0.076, respectively). In addition, the results for indirect effects were significantly negative at the 1% level, indicating that the more NEIPs a city has, the greater the negative effect of that city on the eco-efficiency of the surrounding area, which is approximately − 1.64, much larger than the results of the benchmark regression. If the NEIP is located in a provincial capital or municipality, its negative impact on the eco-efficiency of the surrounding area reaches 4.2. The total spatial effect is significantly negative at the 1% level, which indicates that the greater the number of NEIPs and the more NEIPs are located in provincial capitals and municipalities directly under the central government, the more obvious the siphoning effect is on the surrounding areas, and it is detrimental to the eco-efficiency of the whole region.

Spatial attenuation boundary

As the geographic distance between cities increases, the spatial correlation between each variable will continue to decrease and the policy effect of NEIP will continue to diminish. Therefore, the spatial effects of the NEIP policy may only be effective to a certain extent. Therefore, to deeply identify the spatial effects of NEIP policies, this study sets a threshold value for the spatial weight matrix W1, which is set to W3, as shown in Eq. (8):

$$W3=\left\{\begin{array}{l}\begin{array}{ccc}\frac{1}{{D}_{ij}},& i\ne j& {D}_{ij}>\mathrm{Dsc}\end{array}\\ \begin{array}{ccc}\;0,\;\;& i=j & {D}_{ij}\le \mathrm{Dsc}\end{array}\end{array}\right.$$
(8)

where \(\mathrm{Dsc}\) represents the distance threshold. This study sets the initial value of the distance threshold to 100 km and gradually increases from 100 to 800 km. By using W3, we can investigate the spatial effect of NEIP policy on cities within a certain distance. By bringing W3 into Eq. (2), we can calculate the spatial total effect with a 95% confidence interval.

The regression results are presented in Fig. 5. Similar to the benchmark regression results, the NEIP policy significantly increases local eco-efficiency. However, as distance increases, the negative spatial effect of NEIP gradually becomes apparent. This result is consistent with the regression results of Eq. (2), which show that NEIP has a siphoning effect on the ecological efficiency of the surrounding cities up to a distance of 100 km. However, the siphoning effect is limited to areas within 100 km, i.e., cities in close proximity to NEIP. Beyond 100 km, the spatial effect of NEIP is not significant.

Fig. 5
figure 5

Spatial attenuation effects

Full size image

Mechanism and heterogeneity analysis

Mechanism analysis

Table 6 reports the estimation results of the mediated effects models for the spatial weight matrices of W1 and W2. Columns (1) and (2) correspond to the estimation results based on technological innovation as the mechanism variable. Columns (3) and (4) correspond to the estimation results based on the number of heavily polluting enterprises as the mechanism variable. The regression results in column (1) show that the direct effect of NEIPs on urban technological innovation is significantly positive, indicating that NEIPs significantly promote local technological innovation. However, the indirect and total effects of NEIPs on technological innovation in the surrounding cities are significantly negative, indicating that NEIPs inhibit technological innovation in neighboring areas and even in the whole region. The regression result in column (3) shows that the implementation of NEIP policies has significantly reduced the number of local heavily polluting enterprises. In addition, the indirect effect of NEIPs on the number of heavily polluting enterprises in the surrounding area is significantly positive, but the spatial total effect is not significant. Thus, Hypothesis 2 is verified.

Table 6 Mechanism analysis results
Full size table

Heterogeneity analysis

China’s vast size results in significant variations in resource endowments, geographic conditions, economic development, and institutional arrangements among its cities. Economic policy support has predominantly targeted cities located in eastern China since the implementation of China’s reform and opening-up policy. Moreover, historical and geographical factors have granted better economic development prospects to eastern China than to central and western regions. The division of economic regions not only portrays the economic impact brought about by geographical location but also demonstrates the development process of China’s gradient economic policy. Referring to the regional divisions issued by the report The State Council issued the implementation opinions on several policies and measures for the development of western China, this study divides China into three regions: eastern, central, and western areas (Northeast China is dominated by heavy industry, so this study classifies Northeast China as Central China). According to the research of Almond et al. (Almond et al. 2009), there are differences between northern China and southern China in terms of air pollution due to temperature, humidity, heating policies in northern China, etc. The Qinling-Huaihe line serves not only as the 800 mm precipitation boundary in China but also as the 0 °C isotherm, thus providing a scientific basis for dividing the northern and southern regions of China.

To identify the above heterogeneity effects, this study constructs an SDDID model as follows:

$${\mathrm{lnEE}}_{it}={\alpha }^{^{\prime}}+\varphi {{W}_{i}\mathrm{lnEE}}_{it}+{\beta \mathrm{^{\prime}}\mathrm{DID}}_{it}+\beta {\mathrm{^{\prime}}\mathrm{^{\prime}}\mathrm{DID}}_{it}\times {\mathrm{HE}}_{i}+\partial {\mathrm{^{\prime}}\mathrm{DID}}_{it}\times {\mathrm{HE}}_{i}+\partial {\mathrm{^{\prime}}{W}_{i}\mathrm{DID}}_{it}\times {\mathrm{HE}}_{i}+\gamma \mathrm{^{\prime}}\sum {\mathrm{Control}}_{it}+\gamma \mathrm{^{\prime}}\mathrm{^{\prime}}\sum {{W}_{i}\mathrm{Control}}_{it}+{\delta }_{i}+{\mu }_{t}+{\varepsilon }_{it}$$
(9)

The regression results in Table F in the Appendix show that the NEIP has caused a significant increase in eco-efficiency for eastern and central China. The NEIP has improved eco-efficiency in central China even more than in eastern China. The regression results in column (4) show that the closer to the eastern region, the more significant the improvement in the direct effect of the NEIP on eco-efficiency. The indirect effect of East × DID and Lon × DID is significantly negative at the 1% level, which indicates that the closer to eastern China, the more significant the siphoning effect of the NEIP on the resources of the surrounding area. Notably, the indirect impact of NEIP on eco-efficiency is significantly positive in central China, suggesting a positive spatial spillover effect of NEIP on eco-efficiency in this region, thereby driving the eco-efficiency of the surrounding areas. In contrast, the impact of NEIP on eco-efficiency in western and northern China was not significant.

Discussion

Spatial effects of the NEIP on urban eco-efficiency

NEIP policies can significantly improve the eco-efficiency of the city where the NEIP is located and promote eco-efficiency. The NEIP uses the principles of industrial ecology to achieve economic progress from the consumption of limited resources and waste emissions based on meeting the requirements of clean production (Lambert and Boons 2002). The improvement of energy utilization reduces the emission of waste while reducing the production cost of the enterprise so that the enterprise has more production factors to invest in other eco-economic activities to achieve sustainable development. In addition, as an agglomeration of enterprises, the NEIP sets emission reduction targets for enterprises to reduce their pollution emissions in the production process and achieve pollution reduction at the city level (Song and Zhou 2021). Moreover, the government has implemented a series of tax incentives and innovation subsidies for enterprises in NEIPs to promote green innovation in the production industry (Wu and Gao 2022), which improved cities’ green innovation and economic efficiency. As a result, the implementation of NEIP emission reduction targets and government policies such as tax incentives have reduced local urban pollution emissions in parks while improving economic efficiency and promoting eco-efficiency (Abbas and Sağsan 2019). Thus, the NEIP contributes to local eco-efficiency. NEIPs need to take into account the economic benefits of enterprises while limiting their pollution emissions (Shi et al. 2010). Furthermore, NEIPs cannot allow enterprises to discharge uncontrolled pollution because of the need to meet their production targets (Sacirovic et al. 2019). Therefore, assessing eco-efficiency improvements as a result of NEIP policies is an important challenge for NEIP policy design.

Nevertheless, while NEIPs can promote the eco-efficiency of local cities, they have a negative impact on the eco-efficiency of the surrounding areas, which is incompatible with the cities around the NEIP’s Sustainable Development Goals. The NEIP policy has a “clustering” effect of innovative factors and a “crowding out” effect of heavily polluting enterprises on the eco-efficiency of the surrounding areas. As NEIPs promote the clustering of enterprises and thus externalities, scale effects, and clustering effects (Liu et al. 2022), enterprises within NEIPs reduce their energy consumption and improve their technical efficiency (Zhang et al. 2016). Therefore, enterprises in NEIPs are more likely to have sufficient capital to absorb and purchase human capital, physical capital, and other production factors from the surrounding areas. As the carrier of technological innovation and material production, the concentration of human capital and material capital can significantly improve the innovation capacity and production level of cities (Xu and Li 2020) and improve the eco-efficiency of the city where the NEIP is located. However, due to the loss of production factors, the economic development of the surrounding areas is bound to be hindered, and the lack of government policy support for these areas makes it difficult to implement some policies related to environmental protection. In addition, to accomplish environmental performance, major electronics and automotive enterprises in the NEIP tend to outsource the production of the lowest value-added, pollution-intensive intermediate components and materials outside of the NEIP (Shi et al. 2010), leading to a decline in eco-efficiency in the surrounding area. Therefore, NEIPs can hinder the eco-efficiency of the surrounding cities while increasing the eco-efficiency of local cities.

Discussion on the identification of policy effects

Through a deeper identification of policy effects, this study finds that the direct and siphoning effects of NEIPs are stronger in provincial capitals and municipalities directly under the central government for eco-efficiency. If a city has more than one NEIP, this situation is also the same. Li (2022) argues that provincial capitals and municipalities directly under the central government are significantly different from other cities in that they have a strong siphoning and snowball effect on the resources of cities in the same province and surrounding areas. In fact, as the political and economic center of the province, the provincial capital city has more financial resources for economic development and environmental protection due to its preferential policies and production factors, and all types of capital will always be transferred to the provincial capital city from low development areas. Therefore, compared to NEIPs located in other cities, NEIPs located in provincial capitals have a stronger ability to attract production factors. Therefore, NEIPs located in provincial capitals have more eco-efficiency gains and stronger siphoning effects. In addition, the greater the number of NEIPs a city has, the stronger its direct and siphoning effects on eco-efficiency. Due to the concentration of NEIPs, cities have more incentives to subsidize NEIPs, and the reduced distance between NEIPs will facilitate exchanges between NEIPs, reducing production and innovation costs (Park and Lee 2004). Therefore, its siphoning effect on physical capital and innovation factors is bound to be higher than that of cities with only one or no NEIP. Moreover, the spatial effects of NEIP policies only exist in areas where the NEIP is 100 km from the periphery. Most likely, considering the cost of relocation of enterprises and other reasons, NEIP can only attract enterprises or industries from neighboring cities to move into the park, and the distance of its siphoning effect is not far.

Discussion on the mechanism and heterogeneity analysis

This study finds that the spatial effect of the NEIP on eco-efficiency is realized through the crowding out of heavily polluting enterprises. The reduction targets set by the NEIP for enterprises lead to only two options for polluters: to move out of the NEIP or to achieve cleaner production through technological innovation and thus meet the reduction targets set by the NEIP (Zhu et al. 2010). However, some enterprises may be forced to move away from the NEIP and the city due to their own reasons, such as the shortage of production factors caused by the lack of capital, which makes it difficult for them to carry out technological innovation, increasing the number of polluting enterprises in the surrounding cities. In addition, due to the UN (the United Nations) requirements for cleaner production and the establishment of the SDGs, the Chinese government has continued to promote environmental regulations as well as emission reduction policies across the region (Wang and Shen 2016), and each NEIP consolidates heavily polluting enterprises, which in turn has led to the crowding out effects of heavily polluting enterprises in the NEIP. However, while the NEIP pushes heavily polluting enterprises out to neighboring cities, the overall number of them in the whole region does not increase as a result. This is precise because of the emergence of policies such as environmental regulations and carbon reduction targets (Shi et al. 2022) and the relentless efforts of the Chinese government to remediate heavily polluting industries, and the number of heavily polluting enterprises is decreasing (Lu 2021; Shi et al. 2022). As a result, the number of heavily polluting enterprises throughout the region has not increased. However, due to the absorption of production factors by the NEIP, surrounding cities may not be able to reduce emissions while taking into account economic benefits, leading to a decline in their ecological efficiency and contributing to the NEIP siphoning effect. Additionally, the spatial effect of NEIPs on eco-efficiency is realized through the technological innovation effect. First, according to Porter’s hypothesis, NEIP policies, as government-imposed environmental regulation policies for enterprises, will lead enterprises to actively participate in technological innovation to reduce production costs, enhance competitiveness, and produce innovative outcomes (Porter and Van der Linde 1995). Moreover, NEIPs enable the clustering of enterprises to reduce the cost of new technology output through the rapid exchange of innovation resources and synergy effects and increase the output of innovation resources of enterprises (Park and Lee 2004). The increased output of technological innovation in NEIPs also creates superior production conditions, thus attracting innovation factors from neighboring regions to cluster in NEIPs. Therefore, NEIPs, while enhancing the technological innovation capacity of local cities, reduce the technological innovation capacity of the surrounding cities through the absorption of talent and other capital, which in turn has a siphoning effect on eco-efficiency and hinders the achievement of Sustainable Development Goals in the surrounding cities. The direct promotion effect of NEIPs on urban eco-efficiency was more pronounced in eastern and central China, but there was no significant north‒south difference. The direct effect of NEIPs on the eco-efficiency of central China is higher than that of eastern China. Compared to eastern China, the industries in central China are mainly heavy industries. Thus, since the construction of the NEIP, which is based on the concept of clean production and industrial ecological symbiosis, the NEIP has had a significant effect on reducing the pollution of the original high-polluting industries, thus significantly increasing the eco-efficiency of central China. Additionally, eastern China maintains obvious advantages in technology and the market environment. The construction of the NEIP will further strengthen these advantages and enhance the eco-efficiency of eastern China while strengthening its siphoning effect on eco-efficiency. However, there was no significant difference in the effect of the NEIP on eco-efficiency between northern and southern China. The reason may be that the economic bases of southern and northern China are basically the same. Furthermore, the policy implementation efforts in southern and northern China are essentially the same. Therefore, there is no significant difference between the eco-efficiency of the NEIP for southern and northern China.

Concluding remarks and policy implications

Conclusions

Using a panel dataset of 271 Chinese cities at the city level from 2004 to 2018, this study is the first to adopt a staggered spatial difference-in-differences (SDID) model to investigate the effect and mechanism of the NEIP policies on eco-efficiency. In addition, this study conducts a heterogeneity test and thoroughly identifies the policy effect by employing a spatial difference-in-difference-in-differences (SDDID) model, spatial attenuation boundary, and event study method. The outcomes of the direct effects from the event study method indicate that the NEIP substantially and consistently enhances urban eco-efficiency and supports sustainable urban development. Conversely, the outcomes of total spatial effects reveal that the NEIP has a continuous negative impact on the eco-efficiency of the surrounding region and, therefore, is not conducive to sustainable regional development. The siphoning effect of the NEIP on eco-efficiency is particularly notable in cities with more than one NEIP or in provincial capitals and municipalities with NEIPs. Furthermore, the spatial effect of the eco-efficiency of the NEIP can span approximately 100 km, indicating that there is a negative impact on cities in the immediate vicinity of the NEIP (within 100 km). The impact of the NEIP on urban and regional eco-efficiency is primarily realized through the crowding out effect of heavily polluting enterprises and the technological innovation effect. Additionally, the direct effect of the NEIP on eco-efficiency in eastern and central China was significantly positive. However, the siphoning effect of the NEIP on the eco-efficiency of the surrounding areas is only present in eastern China, and this effect becomes more pronounced the closer the NEIP is to eastern China.

This study provides a new empirical explanation of the impact of NEIP construction on regional eco-efficiency through an in-depth empirical analysis. Although many studies have examined the impact of NEIP construction on urban economy or environment from different perspectives (Wu and Gao 2022), no existing study has investigated the impact of NEIP construction on the overall efficiency indicators of urban development. The insight that NEIP construction has a siphoning effect on the eco-efficiency of the surrounding cities within a certain range may provide a new contribution to the literature in this area. This study highlights the internal mechanism of the NEIP’s impact on regional eco-efficiency, providing researchers with a fresh perspective on the role of EIP in regional development. Furthermore, by focusing on the spatial effects and diffusion boundaries of EIP construction on regional economic green development, the analysis of this study may provide valuable experience and methodology for EIP construction in the Chinese context, as well as other countries globally.

Policy implications

Our findings provide insights into the integrated development of cities and regions through the construction of EIPs. The NEIP construction in China highlights how it promotes pilot city development and compresses the development of the surrounding areas. Therefore, for the rest of the world, governments should establish regional eco-industrial city clusters centered on pilot cities, adopt policies such as innovation incentives and tax preferences, and build the appropriate infrastructure. Governments should also encourage NEIPs to promote technological innovation and proliferation, as well as infrastructure co-building and sharing, to achieve cleaner production, meet emission reduction targets, and establish an early completion of the symbiotic system of industry and environment to achieve regional eco-efficiency. Moreover, to achieve sustainable regional development based on environmental protection, the government should construct differentiated EIP goals and support policies according to the industrial layout, environmental carrying capacity, and economic strength of each region, including capital cities and other cities, to promote the establishment of an industrial symbiosis system rich in regional characteristics while considering regional differences.

Limitations

This study analyzes the NEIP policy from a Chinese perspective, and while some implementation elements and policy recommendations can be applied to other countries, the specificity of China’s economy and environment limits the applicability of China’s NEIP construction experience to other countries. In future research, it is important to compare China’s NEIP to similar EIPs in other countries. Additionally, this study has a broad focus on the development of NEIP in China as a whole, making it difficult to identify the specific impact of NEIP on individual regions. Further research should focus on specific regions to provide case support for the development of EIPs in similar regions around the world.