The Effects of a Cross-Border Freight Railway Project on Chinese Cities’ Green Innovation Intensity: A Quasi-Natural Experiment Based on the Expansion of the China Railway Express

Abstract

This paper intends to clarify whether launching a cross-border freight railway project will cause adverse impacts on cities’ green innovation intensity. Based on the construction and operation of the China Railway Express, the impact mechanism of this cross-border freight railway project on cities’ green innovation is theoretically analyzed. Through using the data of 284 prefecture-level cities in China from 2017 to 2019, the effects of the China Railway Express on green innovation intensity are empirically tested by adopting the differences-in-differences method. The results show the following: 1. launching the China Railway Express project has a direct negative impact on cities’ green innovation intensity; 2. the China Railway Express has indirect positive impacts on green innovation intensity through increasing the foreign trade volume of cities as well as reducing the level of industrial specialization, and has an indirect negative impact on green innovation intensity through reducing the strictness of environmental regulations; 3. the radius of the impact coverage of the China Railway Express is 150 km, and its impact on surrounding cities is funnel-shaped; 4. the China Railway Express will cause a negative impact on green innovation intensity in large cities, cities with lower pollution emissions, and cities of Eastern China and Western China. Taking these findings into account, some policy implications are proposed.

1. Introduction

To further expand the scale of foreign trade, the Chinese government has launched a special cross-border freight railway project named the China Railway Express. The China Railway Express (CRE), which links the Chinese and European market, is the most important transport infrastructure serving ‘the Belt and Road’ initiative. Since the first part of the CRE project launched in Chongqing in 2011, 55 Chinese prefecture-level cities had launched the CRE project by the end of 2019. Among them, 12 express lines have achieved regular operation. Figure 1 shows the number of trains operated under the China Railway Express project. From 2011 to 2019, both outbound trips and inbound trips increased sharply, and the proportion of inbound trips even reached 44.98% in 2019. During the pandemic of COVID-19, the performance of the China Railway Express was quite significant in maintaining the stability of the global supply chain and pandemic control: total trips of the China Railway Express in 2020 increased by 50% from a year earlier to 12,406; 9.39 million pieces of anti-epidemic prevention materials were transported to Europe and Western Asia through the China Railway Express according to the data reported by the Ministry of Commerce of China, making it a veritable channel of life.

The expansion of the China Railway Express has caused a significant impact on regional economic growth. From the perspective of influencing the spatial allocation of trade, the China Railway Express has reshaped the traditional economic model of coastal ports, improved the comparative advantages of China’s inland cities in foreign trade, and led to the gradual realization of ‘two-way circulation’ in the East–West integrated global value chain [1,2]. Relying on the convenience, safety, and cheapness of the CRE, the scale of China’s foreign trade has increased substantially. In 2020, the volume of China’s foreign trade increased by 7.62 times compared with 2001. The total import and export volume between China and Europe in 2020 was 9.29 times that in 2001. From the perspective of stimulating economic growth, the CRE project broadened the trade channels connecting the European market and East Asian market, reduced competition in the domestic market and transportation costs of enterprises participating in international trade, and intensified the stimulating effect of foreign trade on economic growth as a result [3,4,5,6]. According to data from the National Bureau of Statistics of China, the average annual economic growth rate of cities that launched the CRE project was about 2 percentage points higher than that of cities not involved in the CRE project from 2011 to 2019.

At present, technological innovation has become a considerable driving force of economic growth. In order to clarify the deep reasons why the CRE project can stimulate the economic growth of the launching city, many scholars have studied the impact of the CRE project on technological innovation. Wang and Bu (2019) empirically verified that the CRE project promoted enterprise innovation by stimulating exports: enterprises participating in international trade through the China Railway Express may be more willing to increase innovation investments [7]. Li et al. (2020) tested the impact of the China Railway Express on technological innovation using data of Chinese prefecture-level cities and found that the CRE project promoted cities’ innovation, but this promotion of innovation was affected by governments’ behavior [8]. From the perspective of stimulating total factor productivity, the China Railway Express changed the conditions for domestic industrial agglomeration, further optimized the spatial distribution of industries, reduced the information search cost of enterprises, and improved total factor productivity of the cities within 200 km of the departure stations [9,10,11,12].

However, China is not only confronted with the problem of how to stimulate economic growth but also the ecological risks that have gradually appeared in the process of development. With the improvement of China’s urbanization and industrialization, compound environmental risks occur frequently in the process of the modernized transformation of Chinese society [13]. As a spontaneous attempt to expand the scale of foreign trade by Chinese cities, the CRE project also poses risks to the ecological environment [14]. Rather than only focus on the effect of the CRE project on stimulating economic growth, excessive attention should be paid to its impact on environmental protection. Green innovation refers to new methods and technologies that can help avoid environmental damage, and the intensity of green innovation reflects the ability of a region to control pollution and its willingness to follow the concept of ecological development [15]. Therefore, figuring out the impact of the CRE project on green innovation is more valuable for policy guidance than studying its impact on pollution emissions. Current studies have discussed how domestic transport infrastructure construction will affect green innovation. Li and Liu (2021) found that the construction of the high-speed railway in China improved the efficiency of factors’ cross-regional flow and thus promoted the green innovation of cities [16]. Xia and Tang (2018) constructed a spatial econometric model and then tested the impact of China’s transport infrastructure construction on green innovation efficiency and its spatial spillover effects [17]. They found that China’s domestic transport infrastructure construction promoted green innovation efficiency to some extent and caused a positive spillover effect on surrounding cities. However, the impact of the CRE project on green innovation is different from that of domestic railway lines: on the one hand, operation of the China Railway Express relies on established railway lines, which makes the CRE project more like a policy practice than infrastructure construction; on the other hand, factors such as customs clearance efficiency and cultural distance between China and countries along CRE routes will affect the operation of the international railway lines [18,19]. Therefore, it is necessary to theoretically and empirically analyze whether and how the China Railway Express affects cities’ green innovation. (The steps for launching the CRE project are as follows: first, a bilateral trade agreement should be signed; second, profits of the line will be assessed by China National Railway Group Limited; third, China National Railway Group Limited will negotiate with local governments to obtain government support for the operation of the CRE.)

First, this article theoretically analyzes the impact mechanism of the China Railway Express on green innovation. Then, a DID method is adopted to empirically test whether and how the China Railway Express affects cities’ green innovation based on the data of 284 Chinese prefecture-level cities. Finally, this article discusses the heterogeneity of the impact the China Railway Express has on cities’ green growth in detail and its spillover effect on surrounding cities.

2. Theoretical Hypotheses and Mechanism Analysis

In this section, we analyze theoretically how the launch of the CRE project affects cities’ green innovation intensity. Figure 2 reports the specific impact mechanism.

The China Railway Express directly affects green innovation. Cooke et al. (2004) modified the innovation system theory, stating that regional technological innovation intensity depends on both endowments and incentives of technological innovation. Endowments refer to regional reserves of technology, capital, and talent [20]. Incentives refer to the economic benefits generated by technological innovation, such as the improvement of production efficiency or government subsidies for R&D activities. This paper also analyzes the direct impact of the CRE project on cities’ green innovation intensity from these two aspects. From the perspective of green innovation endowments, launching the CRE project will enhance the innovation endowments of cities, which is conducive to green innovation. Firstly, the China Railway Express has broadened the trade channel between China and Europe, accelerated the flow of technology and promoted innovation cooperation between cities, and intensified the positive technology spillover effect [21]. Secondly, according to Gibbons and Machine (2005), the China Railway Express has optimized the allocation of human capital and R&D resources among Chinese cities [22]. Since most of the cities running the CRE project are inland industrial cities, this project provides new opportunities for the development of such cities, which in turn attracts the inflow of talent and capital, leading to a better match between innovation resources and the development goals of such cities. On the other hand, launching the CRE project will not be conducive to green innovation from the perspective of incentives for green innovation. Firstly, the China Railway Express promoted the upgrading of industries in launching cities and reduced energy consumption and pollution, thereby making green innovation less urgent in such cities.

Table 1. Major types of export goods transported through regular freight trains.

City	Launch Year	Major Types of Export Goods in Launch Year	Major Types of Export Goods in 2019
Chongqing	2011	Laptop products	Electronic products and accessories, mechanical products, vehicles and auto accessories, medicine, food, and daily necessities
Changsha	2012	Tea and porcelain	Mechanical and electronic products, tea, clothing, food, porcelain, steel, and auto accessories
Wuhan	2012	Electronic products and vehicles and auto accessories	Optical products, electronic products, and vehicles and auto accessories
Zhengzhou	2013	Tires, clothing, sports goods, and handicrafts	Mechanical and electronic products, clothing, cold-chain products, and medical devices
Suzhou	2013	Clothing and small commodities	LCDs, source boards, mechanical and electronic products, communication equipment, clothing, and small commodities
Chengdu	2013	Electrical appliances and electronic products	Vehicles and auto accessories, mechanical and electronic products, and red wine
Yingkou	2014	Food	Electronic products and auto accessories
Jinhua	2014	Handicrafts, drinks, and toys	Handicrafts, drinks, toys, luggage and bags, daily chemicals, and electronic products
Xiamen	2015	Building materials, daily necessities, and electronic products	Auto accessories and mechanical and electronic products
Qingdao	2015	Food and auto accessories	Textile products and mechanical and electronic products
Kunming	2015	Coffee and agricultural products	Coffee, agricultural products, and vehicle and auto accessories
Dalian	2016	Mechanical products, textile products, and auto accessories	Mechanical products, textile products, and auto accessories

Data sources: www.yidaiyilu.gov.cn/numlistpc.htm (accessed on 26 August 2022); www.nra.gov.cn/xwzx/ (accessed on 26 August 2022).

shows the changes in commodity types exported from cities providing regular freight train services from the launch year to 2019: scientific and technological products exported through the CRE in most cities increased substantially, and the upgrading of export goods reflects the upgrading of regional industrial structure. Secondly, the CRE project sharply expanded the market scale of industries not belonging to green industries, resulting in technological innovations to increase the production capacity of other industries and thereby crowding out green innovation [16,23]. The direct goal of launching the CRE project is to stimulate foreign trade in Chinese cities, so that most attention toward innovation has been geared toward improving the production efficiency and market acceptance of products. When the investment of capital and human resources is geared toward these innovation activities, the intensity of green innovation is adversely affected. According to the data of the China National Intellectual Property Administration, in 2019 green invention patents granted in CRE operation cities accounted for 3.7% of all the granted patents in those cities, whereas this proportion was 4.5% in cities that had not launched the China Railway Express. In 2007, this proportion in the two types of cities was relatively similar, accounting for 7.9% and 7.8%, respectively. This indicates that the CRE project could crowd out green patents.

Hypothesis 1.

Launching the CRE project has a net direct negative impact on cities’ green innovation intensity.

The China Railway Express indirectly affects green innovation through promoting the scale of cities’ foreign trade. On the one hand, launching the CRE project reduces transportation costs and risks and encourages international trade between China and European countries. Compared with air transportation, the cost of the CRE is much cheaper. Taking the Wuhan–Hamburg line as an example, the freight of a 40HQ container in 2020 was about USD 8000, and enterprises can also receive subsidies up to 30% of the total freight from the Wuhan government. Compared with shipping, the CRE has no advantage in costs, but it is much faster and safer. Fixed procedures and lines mean that the CRE can effectively avoid occurrences similar to the Suez Canal blockage [24]. On the other hand, CRE operating cities can expand the scale of foreign trade by encouraging more enterprises to participate in international trade, thereby making higher profits [25]. As for the impact of exports on green innovation, Dietzenbacher and Mukhopadhyay (2007) argued that some types of export goods could influence resource distribution in green innovation: if a destination country sets strict environmental protection requirements for foreign goods or has a strong demand for products of green industries, then this will stimulate cities’ green innovation; if foreign countries have larger demand for traditional products, it will only promote technological innovation to increase production capacity [26]. Ye and Yang (2019) verified that importing advanced production equipment and technologies can increase endowments for green innovation [27]. Since the CRE mainly stimulates trade between China and European countries, we can expect that it will have an indirect positive effect on cities’ green innovation: European countries have relatively stricter environmental protection standards and have a larger demand for new energy utilization and pollution control equipment. At the same time, the concept of environmental production in European countries can be introduced into Chinese cities through import.

The China Railway Express indirectly affects green innovation through changing the intensity of environmental regulations in cities. The intensity of regional environmental regulations represents the environmental pressure imposed by the governments on enterprises, which is an important force to stimulate green innovation. A higher intensity of environmental regulation means that it is more urgent for enterprises to invest in green innovation. In other words, environmental regulation forces enterprises to improve production techniques, reduce pollution emissions, or transform themselves into green industries. In theory, the CRE will have two opposite impacts on regional environmental regulation: on the one hand, operation of the CRE will accelerate the economic growth of cities through foreign trade, so governments can move their attention from economic growth to environmental protection [28]; on the other hand, at the initial stage of the CRE operation, a large amount of fiscal expenditure will be occupied in CRE operation subsidies or foreign trade stimulus policies, crowding out environmental protection expenditures, which may reduce the intensity of environmental regulations [29]. Chinese cities, especially small- and medium-sized cities, are facing a serious government debt crisis, making their financial funds not abundant. In 2017, the debt-to-GDP ratio of Chinese cities was 311.04% [30]. The stimulating effect of the CRE on economic growth of cities has not fundamentally solved the financial difficulties of small- and medium-sized cities. Therefore, it can be expected that launching the CRE project will reduce the intensity of environmental regulations and thus have an indirect negative impact on green innovation.

The China Railway Express indirectly affects green innovation through adjusting the degree of industrial specialization. From the above analysis, operation of the CRE can change the conditions for industrial agglomeration, for it has a stimulus effect on foreign trade. With the relocation of industries, the industrial specialization of cities is constantly changing: industries in small- and medium-sized cities tend to migrate to large cities as the CRE reduces the ‘market crowding effect’, leaving only industries that rely on local resource endowments or policy support to stay in those cities and foster specialized agglomeration in those cities. For large cities, the level of specialization may decline due to the influx of industries [9]. At present, most of cities that launched the CRE project are regional communication hubs and at a relatively higher hierarchy. Therefore, operation of the CRE will cause a ‘siphon effect’ on industries of surrounding cities and improve the degree of industrial diversification. As the relocation cost of small-sized enterprises is lower, industrial diversification of CRE operation cities is mainly driven by the relocation of small-sized enterprises. According to the website of the National Bureau of Statistics of China, the ratio of the average fixed assets of enterprises in cities operating the CRE to the average fixed assets of enterprises in cities not operating the CRE was 1.302 in 2007, which fell to 1.089 in 2019, indicating that the gap in enterprise size between the two types of cities is narrowing. These data prove that there is a correlation between operating the CRE and the influx of small-sized enterprises. The impact of industrial specialization on green innovation is also diverse, which is affected by factors such as types of enterprises, R&D efficiency, information spillover speed, and the objectives of clustered organizations [31,32]. Industrial specialization driven by the relocation of small-sized enterprises is not conducive to green growth: small-sized enterprises often have a flexible organization mode and more efficient creation system, making them more adaptable to green industries which have high requirements for innovation ability, whereas CRE operation cities can benefit from the reduction in industrial specialization. On the other hand, a decline in the level of specialization means a more complete industrial chain for big cities, which is conducive to establishing a more complete innovation chain. Therefore, it can be expected that launching the CRE project will reduce cities’ industrial specialization level and thus have an indirect positive impact on green innovation.

Hypothesis 2.

Launching the CRE project has an indirect positive impact on green innovation intensity through stimulating foreign trade and reducing industrial specialization, whereas it has an indirect negative impact on green innovation through reducing the intensity of environmental regulations.

3. Research Design and Data Analysis

3.1. Research Design

We take launching the CRE project as a quasi-natural experiment. To test Hypothesis 1, we constructed a set of econometric models:

$${\mathit{GINVG}}_{i,t}={\beta }_0+{\beta }_1{\mathit{TREAT}}_i\times {\mathit{POST}}_t+{\beta }_2{\mathit{TREAT}}_i+{\beta }_3{\mathit{POST}}_t+\mathit{\beta }{\mathit{Controls}}_{i,t}+{\epsilon }_{i,t}$$

(1)

$${\mathit{GINVG}}_{i,t}={\beta }_0+{\beta }_1{\mathit{TREAT}}_i\times {\mathit{POST}}_t+\mathit{\beta }{\mathit{Controls}}_{i,t}+{\mu }_i+{\gamma }_t+{\epsilon }_{i,t}$$

(2)

In Equation (1), GINVG represents the green innovation intensity of cities. TREAT represents whether a city has launched the CRE project and POST represents the year a city was influenced by the CRE project. TREAT × POST represents the experimental variable of launching the CRE project. Controls refers to a vector composed of control variables such as public financial expenditure, actual utilization of foreign capital, urbanization, natural resources, proportion of manufacturing, and financial support; i and t are the labels of city units and years, respectively; and ԑ_i,t is random disturbance. Since the launching time of the CRE project is not same in each city, we constructed Equation (2) to deal with the possible mistreat problem. In Equation (2), μ indicates city fixed effects, γ indicates year fixed effects, and the meanings of other variables are consistent with Equation (1). To test Hypothesis 2, we further constructed a set of econometric models:

$${\mathit{MED}}_{i,t}={\beta }_0+{\beta }_1{\mathit{TREAT}}_i\times {\mathit{POST}}_t+\mathit{\beta }{\mathit{Controls}}_{i,t}+{\mu }_i+{\gamma }_t+{\epsilon }_{i,t}$$

(3)

$${\mathit{GINVG}}_{i,t}={\beta }_0+{\beta }_1{\mathit{TREAT}}_i\times {\mathit{POST}}_t+{\beta }_2{\mathit{MED}}_{i,t}+\mathit{\beta }{\mathit{Controls}}_{i,t}+{\mu }_i+{\gamma }_t+{\epsilon }_{i,t}$$

(4)

The meanings of GINVG, TREAT, POST, and Controls in Equation (3) are the same as in Equation (1). MED represents mediation variables. To test Hypothesis 2, MED represents the foreign trade scale, the intensity of environmental regulations, and the level of industrial specialization separately. If the significance of β₁ and β₂ in Equation (4) cannot be refused, and β₁ in Equation (3) is significant at the same time, then we can prove that the CRE project will have indirect effects on cities’ green innovation intensity.

3.2. Data Analysis

This research uses the data of 284 prefecture-level cities in China from 2007 to 2019. The data of cities revoked or merged into other cities are excluded. Green patent data are from the website of the China National Intellectual Property Administration, the CNRDS database, and Wind. The CRE launch data are from the website of the Belt and Road Portal, the website of China National Railway Group Limited, and the website of the Ministry of Commerce of the People’s Republic of China. The longitude and latitude data of prefecture-level cities are from WebAPI of Amap. Export and import data are from Statistical Bulletins of each city from 2007 to 2019. Urbanization is estimated by NPP-VIIRS night light data. The data of other control variables and mediating variables are from China City Statistical Yearbooks (2008–2020), China Statistical Yearbooks (2008–2020), and the website of the National Bureau of Statistics.

3.2.1. Dependent Variable: Green Innovation Intensity

Green innovation intensity is measured by green invention patents granted in each city. Since statistical agencies at the city level do not directly report the amount of green invention patents granted, a specific method is adopted to calculate this data according to the IPC Green Inventory presented by the World Intellectual Property Organization in 2010: (https://www.wipo.int/classifications/ipc/green-inventory/home, accessed on 12 March 2023). First, we select invention patents granted belonging to Waste Management, Nuclear Power, Eco-Transportation, Energy Saving, Alternative Energy Production, Pesticide Substitutes, and Carbon Emissions Permit Trade as green invention patents granted from all of the granted invention patents of listed companies according to the IPC classification standards. Second, we aggregate the number of patents according to the location of companies as the green innovation level of cities. The green innovation level is expressed as GINVG. Figure 3 shows the changes in green invention patents granted in China, indicating that green innovation in China moved into a bottleneck period after 2017. The total number of green invention patents granted in China shows a trend of increase year by year, but the grow speed slows down after 2017. The proportion of green invention patents granted in the total patents granted also shows an overall upward trend before 2017, but the proportion decreases substantially after 2017.

3.2.2. Experimental Variable: Launch of China Railway Express

Considering that the CRE serves ‘the Belt and Road’ initiative, irregular operation can have great impacts on cities. Therefore, the 55 cities that had a CRE departure station in 2019 are defined as the treated group, and the remaining cities are assigned to the control group. This dummy variable is expressed as TREAT. The first year of launching is taken as the watershed to sort the value of the dummy variable POST. The value of POST before the first year of launch is 0 or the value is 1. Figure 1 shows that number of CRE launch cities is growing. Figure 4 details the spatial distribution of CRE launch cities: the number of CRE launch cities in inland areas and coastal areas is almost the same, and there is no obvious spatial agglomeration in cities that have regular freight trains in the CRE. Whether in eastern coastal areas or inland areas, cities that launched CRE project tend to be relatively better developed, and this characteristic is even more obvious among cities that have regular freight trains in the CRE (these cities are often provincial capitals in inland areas).

3.2.3. Mediating Variables

The total amount of imports and exports of Chinese prefecture-level cities are used to measure foreign trade scale, and they are expressed as IMP and EX, respectively. Referring to the method of Caldwell and Vogel (1996), the intensity of environmental regulations can be measured by the reciprocals of pollution emissions per unit of GDP, in which pollution emissions refer to sulfur dioxide emission, industrial soot emission, and industrial wastewater emission [33]. Environmental regulation is expressed as ENREG. Referring to the method of Li and Song (2008), we calculate each city’s relative-specialization index to measure the industrial specialization level of Chinese cities, which is RSI in the article [34]. Since statistical agencies of prefecture-level cities do not directly report the total output value by industry, we use the employment by industry as the substitute indicator to calculate the relative scale of each industry of the city.

3.2.4. Control Variables

According to related research, there are seven important factors that can change the intensity of green innovation. Therefore, we added those variables in the model to control such influences. Public financial expenditure (FAN) is measured by the proportion of general public budget expenditure in the real regional GDP. The real utilization level of foreign capital (FDI) is measured by the real foreign direct investment in the real regional GDP of prefecture-level cities. Urbanization is composed of two indicators: the scope of urbanization (CDS) is measured by the coverage of night lights and the degree of urbanization (CDD) is measured by the brightness of night lights. Finally, we controlled for the impacts of the external environment on industrial development. The first is natural resources (NR), which is measured by the proportion of employment in agriculture, forestry, animal husbandry, sideline fisheries, and extractive industries in total employment. The second is industrialization level (MAN), which is measured by the proportion of manufacturing employment in total employment. The last is financial support (FANJR). Restricted by the availability of the data, we used the year-end loan balance per enterprise to measure FANJR.

Table 2. Calculation method of control variables.

Control Variable	Abbreviation	Calculation Method
Public financial expenditure	FAN	the ratio of general public budget expenditure in cities’ real GDP
Scope of urbanization	CDS	coverage of night lights
Degree of urbanization	CDD	brightness of night lights
Utilization of foreign capital	FDI	ratio of real foreign direct investments in cities’ real GDP
Natural resources	NR	proportion of employment of agriculture, forestry, animal husbandry, side fisheries, and mining industries in the total employment of the city
Proportion of manufacturing	MAN	proportion of employment in manufacturing
Financial support	FANJR	the year-end loan balance per enterprise in each city

reports the calculation method of all control variables and

Table 3. Descriptive statistics.

Variable	Obs	Mean	Std. Dev.	Min	Med	Max
GINVG	3692	366.855	1197.978	0	61	21,197
TREAT	3692	0.194	0.395	0	0	1
POST	3692	0.338	0.473	0	0	1
EX	3692	689,949	2,343,922	0	80,997.970	30,600,000
IMP	3692	570,193	2,587,179	0	37,000	36,600,000
ENREG	3690	0.836	4.578	0.006	0.429	237.230
RSI	3692	3.989	4.593	1.253	2.430	65.064
FAN	3692	0.218	0.235	0.043	0.164	6.041
FDI	3692	0.003	0.004	0	0.002	0.115
CDS	3692	0.123	0.164	0.002	0.061	0.954
CDD	3692	0.887	1.852	0.004	0.301	20.815
NR	3692	0.081	0.117	0	0.033	0.780
MAN	3692	0.242	0.139	0.006	0.217	0.907
FANJR	3692	24,144.400	43,980.990	0	12,878.730	947,936.500

reports the descriptive characteristics of all data.

4. Empirical Test Results

4.1. Benchmark Results and Robust Test Results

4.1.1. Benchmark Results

Table 4. Benchmark results.

Variables	Model (1)	Model (2)
	LnGINVG	LnGINVG
POST × TREAT	−0.172 **	−0.119 **
	(−2.48)	(−2.31)
POST	0.329 ***
	(8.64)
TREAT	0.870 ***
	(5.70)
FAN	0.211 ***	−0.049
	(3.33)	(−0.88)
FDI	16.770 ***	6.886 **
	(3.97)	(2.13)
CDS	5.588 ***	1.748 ***
	(18.44)	(6.72)
CDD	−0.061 **	−0.023
	(−2.23)	(−1.05)
NR	−2.234 ***	1.349 ***
	(−7.01)	(4.50)
MAN	0.853 ***	0.209
	(3.96)	(1.16)
LnFANJR	0.676 ***	−0.067 **
	(27.29)	(−2.57)
Constant	−3.117 ***	3.005 ***
	(−12.32)	(12.34)
City and Year FE	NO	YES
R-squared	0.587	0.768
Observations	3692	3692

Note: t or z statistics in parentheses; *** p < 0.01, ** p < 0.05.

reports the test results of Hypothesis 1. To deal with heteroscedasticity, we performed logarithmic processing on GINVG and FANJR. The results of model (1) and model (2) are the estimated coefficients of Equations (1) and (2), respectively, which show that the net direct effect of launching the CRE project on cities’ green innovation intensity is negative. As analyzed above, CRE departure stations are often located in more developed cities. Such cities have formed R&D advantages over other cities by accumulating more scientific research institutions, R&D capital investments, and talent. After the launch of the CRE, the development of such cities was accelerated substantially by foreign trade, further expanding the comparative advantages of such cities in the field of R&D. Therefore, the decline in green innovation intensity in CRE operation cities is not caused by the destruction of original R&D endowments; on the contrary, it is more likely that R&D activities of other departments crowded out green innovation investments. On the one hand, this indicates that foreign trade of the green industry only accounts for a small part of the total foreign trade volume, possibly due to its relatively lower international competitiveness; on the other hand, this indicates that the urgency of pollution prevention and control may be not very high in CRE operation cities.

4.1.2. Robust Test Results

The prerequisite for testing the impact of the China Railway Express on green innovation intensity through the DID method is understanding if the change in green innovation intensity met common trend conditions before launching the CRE project. A graphical analysis method is only applicable to policies that occur in one period. Since the launch time of the CRE is different for each city, we tested this common trend hypothesis by setting false launch years. We constructed the false dummy variable POSTF to substitute POST by setting false launch years, which are 5 years before the real launch year, 3 years before the real launch year, 1 year before the real launch year, 1 year after the real launch year, and 2 years after the real launch year. Then, coefficients of Equation (2) were re-estimated using the new data, and the corresponding results are reported in

Table 5. Common trend test by constructing false launch year.

Variables	(1)	(2)	(3)	(4)	(5)
	LnGINVG
	5 Years before	3 Years before	1 Year before	1 Year after	2 Years after
POSTF × TREAT	−0.068	−0.098	−0.119	−0.129 **	−0.115 *
	(−1.36)	(−1.41)	(−1.61)	(−2.24)	(−1.65)
Controls	YES	YES	YES	YES	YES
City and Year FE	YES	YES	YES	YES	YES
R-squared	0.767	0.767	0.767	0.768	0.767
Observations	3692	3692	3692	3692	3692

Note: t statistics in parentheses; ** p < 0.05, * p < 0.1.

. The coefficients of POSTF × TREAT are not significant in model (1), model (2), and model (3), indicating that the development of green innovation intensity in Chinese cities met the common trend before the launch of CRE project. The coefficients in model (4) and model (5) show that after launching the CRE project, the development trend of the green innovation intensity of cities was significantly changed.

The net effect of the China Railway Express on cities’ green innovation intensity may be affected by other unobservable factors. In order to eliminate the interference of unobservable factors on the conclusion of this paper, we conducted a placebo test by constructing a false policy dummy variable (TREATF). To make the regression results more convincing, we randomly sampled 500 false treated groups in all cities and again estimated the coefficients of Equation (2). On this basis, we drew the kernel density diagram of all coefficients of experimental variables after randomization (Figure 5). Figure 5 shows that the estimated coefficients are unbiased and follow standard normal distribution. Then, we concluded that the impact of the China Railway Express on green innovation intensity is not affected by other unobservable factors.

In addition, we checked the robustness of the results reported in Table 4 with five methods, and the corresponding results are reported in

Table 6. Additional robustness test results.

Variables	(1)	(2)	(3)	(4)	(5)
	LnGINVG	LnGINVG	LnGINVGR	LnGINVG	LnGINVG
POST × TREATN	−0.096 *
	(−1.76)
POST × TREATA		−0.188 **
		(−2.05)
POST × TREAT			−0.088 *	−0.126 **	−5.211 **
			(−1.71)	(−2.03)	(−2.53)
FAN	−0.091	−0.048	−0.033	−0.082 *	−0.334 **
	(−0.41)	(−0.87)	(−0.52)	(−1.67)	(−2.11)
FDI	6.781	7.504 **	−9.706 ***	4.368	−3.756
	(1.39)	(2.32)	(−2.65)	(1.49)	(−0.49)
CDS	1.860 ***	1.731 ***	1.427 ***	1.524 ***	1.599 ***
	(3.73)	(6.64)	(4.85)	(5.16)	(3.10)
CDD	−0.069 **	−0.023	−0.016	−0.071 ***	0.307 **
	(−2.16)	(−1.04)	(−0.65)	(−2.72)	(2.18)
NR	1.473 ***	1.333 ***	0.621 *	1.009 ***	2.180 ***
	(2.92)	(4.44)	(1.83)	(3.46)	(3.21)
MAN	−0.84	0.184	−0.359 *	−0.144	−0.113
	(−0.26)	(1.02)	(−1.76)	(−0.85)	(−0.30)
LnFANJR	−0.371 ***	−0.070 ***	0.197 ***	−0.039	−0.077
	(−5.80)	(−2.69)	(6.67)	(−1.57)	(−1.50)
Constant	6.447 ***	3.041 ***	−5.205 ***	2.944 ***	2.973 ***
	(10.99)	(12.47)	(−18.87)	(12.61)	(6.20)
Method	PSM-DID	Treat cities with regular freight trains	Change dependent variable	Remove data of 2018 and 2019	Instrumental variable regression
City and Year FE	YES	YES	YES	YES	YES
R-squared	0.815	0.767	0.660	0.807
Observations	950	3692	3692	3124	3692

Note: t statistics in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1.

. First, we used the PSM-DID method to re-estimate the coefficients of Equation (2). There might be sample selection bias in this quasi-natural experiment. Therefore, a propensity score matching method was adopted to re-match the treated group with control group: step one is calculating the propensity score using the logit model; step two is using the nearest neighbor matching method to perform the re-match. Figure 6 reports that the match is of high quality. On this basis, we constructed a new treated group TREATN and re-estimated the coefficients of Equation (2), showing that the China Railway Express has a negative effect on green innovation intensity. Second, we took cities that provide regular freight train service as the treated group (TREATA) for a robustness test. It can be seen that regular freight train service is still not conducive to green innovation. Comparing the results of model (2) in Table 4, we can find that regular freight train service causes more negative impacts on green innovation. Third, we replaced the total number of green invention patents granted in cities with green invention patents granted per enterprise of cities as the dependent variable (GINVGR). The coefficient of the experimental variable in model (3) is significantly negative. Fourth, we excluded the data of 2018 and 2019. Before 2018, the China Railway Express refers to the China–Europe freight train service, which mainly connected the markets of China with Europe and West Asia. However, after 2018, China National Railway Group Limited revised the definition of the China Railway Express, including routes terminating in Southeast Asia and Central Asia in the China Railway Express. The coefficient of the experimental variable in model (4) is significantly negative, indicating that even the ‘real’ China–Europe express has a negative impact on the green innovation of launching cities. Last, we used instrumental variable regression to test the robustness. Cities along the ancient ‘Silk Road’ are regarded as the instrumental variable of TREAT [35]: the routes of the CRE are similar to the ancient ‘Silk Road’. The CRE serves ‘the Belt and Road’ initiative, but the ancient ‘Silk Road’ does not directly affect the level of cities’ green innovation. The coefficient of the experimental variable in model (5) is significantly negative. According to the above analysis, the China Railway Express has a negative impact on green innovation intensity. This conclusion passed the robustness test.

4.2. Mediating Effect Test Results

Table 7. Mediating effect test results.

Panel A: Mediating effects through stimulating foreign trade
Variables	(1)	(2)	(3)	(4)
	LnEX	LnGINVG	LnIMP	LnGINVG
POST × TREAT	0.167 ***	−0.139 ***	0.177 *	−0.123 **
	(3.06)	(−2.70)	(1.88)	(−2.38)
LnEX		0.116 ***
		(7.21)
LnIMP				0.021 **
				(2.24)
Controls	YES	YES	YES	YES
City and Year FE	YES	YES	YES	YES
R-squared	0.332	0.771	0.160	0.768
Observations	3692	3692	3692	3692
Panel B: Mediating effects through changing environmental regulations and industrial structure
Variables	(1)	(2)	(3)	(4)
	ENREG	LnGINVG	RSI	LnGINVG
POST × TREAT	−0.294 *	−0.119 *	−0.590 ***	−0.136 ***
	(−1.92)	(−2.31)	(−2.98)	(−2.65)
ENREG		0.002 *
		(1.67)
RSI				−0.029 ***
				(−6.44)
Controls	YES	YES	YES	YES
City and Year FE	YES	YES	YES	YES
R-squared	0.013	0.768	0.152	0.771
Observations	3692	3692	3692	3692

Note: t statistics in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1.

reports the test results of Hypothesis 2. To avoid the influence of heteroscedasticity, we logarithmized EX and IMP. The results of model (1) and model (2) in Panel A show that the China Railway Express indirectly promoted green innovation intensity through increasing the scale of cities’ exports. According to a theoretical analysis, there are two major reasons for this indirect positive impact. First, there was an increase in the demand for China’s green industrial products in the international market, especially the European market. Second, European countries implemented strict environmental protection standards for foreign commodities, meaning that even ordinary commodities need to meet environmental protection requirements. Therefore, Chinese export enterprises began to shift the focus of innovation to green innovation stimulated by the demand of the international market. The results of model (3) and model (4) in Panel A show that the China Railway Express indirectly promoted green innovation intensity through increasing cities’ import scale. The results indicate that the CRE project has expanded the import channel between China and European countries, has introduced environmental production technologies and equipment and eco-production concepts to Chinese cities, and has improved external conditions for green innovation in CRE operation cities and strengthened the internal determination of achieving green development in these cities. We can further infer that the CRE project has encouraged the development of domestic green industries such as new energy production and pollution prevention from the perspective of stimulating imports.

The results of model (1) and model (2) in Panel B show that the China Railway Express has an indirect negative effect on green innovation intensity through reducing the intensity of environmental regulations. The results indicate that the current operation of the CRE has failed to improve the government’s pollution control capacity. On the contrary, it might have crowded out environmental protection expenditures, resulting in a decline in the intensity of environmental regulations. The positive coefficient of ENREG shows that part of the driving force of green innovation in Chinese cities comes from the government’s environmental regulation pressure. The results of model (3) and model (4) in Panel B show that the China Railway Express indirectly promoted green innovation intensity through reducing cities’ industrial specialization level. The negative coefficient of the experimental variable in model (3) indicates that the China Railway Express has improved the convenience of international trade, formed a siphon effect on industries of surrounding cities, and fostered the diversified agglomeration of industries in the cities. The negative coefficient of RSI in model (4) indicates that industrial diversification has strengthened the ability of cities’ green innovation. Although the scale of enterprises agglomerated in CRE operating cities was relatively small, such enterprises are more willing to invest in green innovation. At the same time, with the rise of industrial diversification, the information spillover effect among industries could be magnified, creating better external conditions for green innovation.

4.3. Further Discussions on Spillover Effect and Heterogeneity Analysis

4.3.1. Spillover Effect of CRE Project

The impact of the CRE project may not be limited to the city where a departure station is located as it fosters agglomeration and an industrial adjusting effect on the region according to the theoretical analysis. Taking the spherical distance between a city and the nearest departure station as the standard, we constructed a set of dummy variables to identify the coverage of effect that the CRE has on green innovation intensity. We took 50 km as the changing threshold, then constructed five new dummy variables to identify cities within 50 km, 100 km, 150 km, 200 km, and 250 km of spherical distance from the departure station. For example, we used TREAT_ 50 to identify cities within 50 km of spherical distance from the nearest departure station. Figure 7 shows the distance between cities and the nearest departure stations nationwide. In 2011, Chongqing and its surrounding cities might be affected by the CRE project. In 2015, the impact of the CRE project expanded to almost the entire eastern coastal area and inland provincial capital city groups. In 2019, almost all prefecture-level cities were located within 250 km of the nearest CRE departure stations.

Based on the new TREAT_Dist dummy variables, we again estimated the coefficients of Equation (2) and the corresponding results are reported in

Table 8. Spillover effect of CRE project.

Variables	(1)	(2)	(3)	(4)	(5)	(6)
	LnGINVG
	Dsit ≤ 0	Dsit ≤ 50	Dist ≤ 100	Dsit ≤ 150	Dist ≤ 200	Dist ≤ 250
POST × TREAT_Dsit	−0.119 **	−0.103 **	0.075 *	0.112 ***	0.063	0.045
	(−2.31)	(−2.12)	(1.82)	(2.59)	(1.27)	(0.73)
FAN	−0.049	−0.049	−0.037	−0.032	−0.037	−0.039
	(−0.88)	(−0.88)	(−0.68)	(−0.58)	(−0.67)	(−0.71)
FDI	6.886 **	6.959 **	7.382 **	7.286 **	7.208 **	7.059 **
	(2.13)	(2.15)	(2.28)	(2.25)	(2.23)	(2.18)
CDS	1.748 ***	1.757 ***	1.670 ***	1.632 ***	1.693 ***	1.723 ***
	(6.72)	(6.75)	(6.32)	(6.18)	(6.40)	(6.55)
CDD	−0.023	−0.024	−0.035	−0.033	−0.031	−0.031
	(−1.05)	(−1.08)	(−1.57)	(−1.48)	(−1.42)	(−1.41)
NR	1.349 ***	1.348 ***	1.313 ***	1.318 ***	1.311 ***	1.304 ***
	(4.50)	(4.50)	(4.38)	(4.40)	(4.37)	(4.32)
MAN	0.209	0.215	0.223	0.217	0.218	0.216
	(1.16)	(1.19)	(1.24)	(1.20)	(1.21)	(1.20)
LnFANJR	−0.067 **	−0.067 **	−0.068 ***	−0.066 **	−0.066 **	−0.066 **
	(−2.57)	(−2.56)	(−2.62)	(−2.55)	(−2.51)	(−2.51)
Constant	3.005 ***	3.002 ***	3.027 ***	3.011 ***	3.001 ***	3.000 ***
	(12.34)	(12.32)	(12.41)	(12.37)	(12.31)	(12.30)
City and Year FE	YES	YES	YES	YES	YES	YES
R-squared	0.768	0.768	0.768	0.768	0.767	0.767
Observations	3692	3692	3692	3692	3692	3692

Note: t statistics in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1.

. It shows that the radius of influencing coverage of the CRE project is 150 km: the CRE project has an adverse impact on green innovation of cities within 50 km from the departure stations, whereas it has a positive effect on green innovation of cities 50 km to 150 km away from the nearest departure stations. For cities more than 150 km away from the nearest departure station, the CRE project will not affect their green innovation. The change in coefficients of experimental variables indicates that the impact the CRE project has on the green innovation of the surrounding cities is funnel-shaped (Figure 8). According to the theoretical analysis, benchmark results, and mediating effect test results, there might be three major reasons for the funnel-shaped impact: first, the direct crowding-out effect on green innovation caused by CRE operation in launching cities is quite large; second, cities far away from the departure stations suffer from CRE operating cities’ siphon effect on industries, causing a higher possibility of such cities to expand the scale of traditional industries to achieve short-term economic growth and thereby improving the necessity of environmental protection and green innovation of such cities; third, cities closer to CRE departure stations have more incentive to expand foreign trade to achieve growth goals that take advantage of the CRE project, inducing higher relevant government expenditures to crowd out environmental protection expenditures and causing greater adverse impacts on cities’ green innovation.

4.3.2. Heterogeneity of City Characteristics

Table 9. Heterogeneity analysis A.

Variables	(1)	(2)	(3)	(4)	(5)	(6)
	City Size		Relatively Specialized Manufacturing		Pollution Emission Intensity
	Large	Small	Yes	No	High	Low
	LnGINVG		LnGINVG		LnGINVG
POST × TREAT	−0.185 ***	0.066	−0.084	−0.118	0.173	−0.185 ***
	(−3.13)	(0.66)	(−1.13)	(−1.62)	(1.01)	(−3.23)
FAN	−0.785	−0.04	0.154	−0.021	−0.248 **	−0.001
	(−0.91)	(−0.65)	(0.94)	(−0.34)	(−2.50)	(−0.01)
FDI	4.948	7.692 **	−0.174	5.031	9.801	9.898 ***
	(0.69)	(2.05)	(−0.03)	(1.27)	(1.52)	(2.60)
CDS	1.514 ***	2.420 ***	1.566 ***	1.802 ***	1.781 ***	2.229 ***
	(4.36)	(3.80)	(4.64)	(3.17)	(3.28)	(7.08)
CDD	−0.004	−0.023	−0.044 **	0.019	−0.086 *	−0.016
	(−0.21)	(−0.23)	(−1.97)	(0.27)	(−1.83)	(−0.61)
NR	0.657	1.278 ***	3.789 ***	1.006 ***	1.173 *	1.735 ***
	(0.73)	(3.83)	(3.66)	(2.95)	(1.78)	(4.83)
MAN	0.524 *	−0.170	0.222	0.481	−0.477	0.163
	(1.95)	(−0.72)	(0.83)	(1.43)	(−1.58)	(0.64)
LnFANJR	−0.408 ***	−0.045	−0.323 ***	−0.038	−0.014	−0.062 *
	(−4.79)	(−1.57)	(−4.45)	(−1.32)	(−0.33)	(−1.93)
Constant	7.804 ***	2.224 ***	5.620 ***	2.476 ***	2.489 ***	3.199 ***
	(9.95)	(8.32)	(8.94)	(8.90)	(6.31)	(10.41)
City and Year FE	YES	YES	YES	YES	YES	YES
R-squared	0.812	0.744	0.823	0.728	0.784	0.701
Observations	1001	2691	1226	2466	1247	2445

Note: t-statistics in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1.

reports the heterogeneity of the impact the CRE project has on green innovation intensity from the perspective of city characteristics. Firstly, we divided cities into large cities and small-sized cities based on the average real GDP of all cities. The regression results of model (1) and model (2) show that launching the CRE project has a more obvious adverse impact on the green innovation intensity of large cities. Since large cities have better R&D endowments, CRE operation in large cities has induced a more significant crowding-out effect on green innovation compared to small-sized cities. The regression results of model (3) and model (4) show that the impact of the CRE project on green innovation intensity is not heterogeneous in cities with different levels of manufacturing specialization. The results also illustrate that the current operation of the CRE has not promoted the green transformation of manufacturing in Chinese cities. Lastly, the average value of total pollution emissions per unit of GDP is used as the standard to divide cities into cities with high pollution emission intensity and cities with low pollution emission intensity. The results reported in model (5) and model (6) show that launching the CRE project will cause a more significant negative effect on green innovation intensity in cities with low pollution emissions. The reason may be that cities with higher pollution emissions are confronted with greater pressure of pollution control and green industrial transformation whether launching the CRE project or not.

4.3.3. Heterogeneity of Cities’ Spatial Location

There exist great differences in development levels and development endowments among Eastern China, Central China, and Western China. We tested the impact of the CRE project on the green innovation intensity of cities located in different regions and analyzed the heterogeneity of the impact results from cities’ spatial locations. The corresponding results are reported in

Table 10. Heterogeneity analysis B.

Variables	(1)	(2)	(3)
	Eastern China	Central China	Western China
	LnGINVG	LnGINVG	LnGINVG
POST × TREAT	−0.104 *	0.001	−0.294 ***
	(−1.74)	(0.01)	(−2.69)
FAN	0.189	−0.067	0.123
	(1.29)	(−0.81)	(1.14)
FDI	−6.027	9.707 **	2.552
	(−1.06)	(2.14)	(0.17)
CDS	1.445 ***	4.308 ***	5.037 ***
	(5.59)	(4.63)	(2.89)
CDD	−0.025	0.008	−0.480 **
	(−1.40)	(0.06)	(−2.53)
NR	1.426 **	1.325 ***	1.166 **
	(2.28)	(2.59)	(2.31)
MAN	0.097	0.801 **	−0.847 *
	(0.51)	(2.10)	(−1.89)
LnFANJR	−0.250 ***	−0.008	−0.024
	(−4.96)	(−0.11)	(−0.67)
Constant	5.490 ***	1.962 ***	2.198 ***
	(12.20)	(3.16)	(6.24)
City and Year FE	YES	YES	YES
R-squared	0.862	0.723	0.766
Observations	1300	1300	1092

Note: t-statistics in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1.

. The results show that launching the CRE project has no significant impact on the green innovation intensity of cities located in Central China. The adverse impact of the CRE project on the green innovation intensity of cities located in Eastern China and Western China is significant, especially in Western China. The reason that the CRE project has no significant impact on the green innovation of cities in Central China may be that the high proportion of heavy industries, chemical industries, and resource industries in the region fosters strong willingness to achieve green transformation and pollution prevention in the region; therefore, the crowding-out effect on green innovation of the CRE project in Central China is not significant. The reason why the adverse impact of the CRE project on green innovation in western cities is stronger than eastern cities may be that scarce financial funds in western cities means that a higher proportion of financial expenditures will be occupied in support of the CRE project, leading to insufficient environmental protection expenditures and technological innovation stimulating expenditures.

5. Conclusions and Policy Implications

Based on the theorical and empirical analysis, we can conclude the following. The China Railway Express will cause a net direct adverse impact on the green innovation intensity of Chinese cities. The China Railway Express on the one hand has indirect positive impacts on green innovation intensity through increasing the import and export scale of cities as well as reducing the level of industrial specialization, and on the other hand has an indirect negative impact on green innovation intensity through reducing the strictness of environmental regulations. The radius of impact coverage of the China Railway Express project is 150 km, and its impact on surrounding cities is funnel-shaped, which means that the CRE project has a negative effect on the green innovation intensity of cities within 50 km of the departure stations, whereas it has a positive effect on the green innovation intensity of cities 50 km to 150 km away from the nearest departure stations. The impact of the CRE project is heterogeneous, as the green innovation intensity in large cities, cities with low pollution emission intensity, and cities located in Eastern China and Western China will have a more significant negative effect caused by the CRE project. To cope with the crowding-out effect of the CRE project on cities’ green innovation and promote the green industrial transformation of cities, a few policy implications are proposed:

(1)

Encourage green industrial transformation. There are two aspects of implications in green industrial transformation: the first is to encourage the development of green industries, including new energy production industries; the second is to encourage the green transformation of traditional industries to reduce energy consumption and pollution emissions. Green industries should be encouraged in CRE operating cities, especially in medium-sized and large-sized cities. Tax relief, financial support, and even land-transfer preferences can be given to enterprises of such industries. Enterprises should be encouraged to improve products’ quality to further expand the European market, stimulating investment in green industries due to increases in demand. The introduction of advanced production technologies and environmental protection and energy-saving equipment should be supported by giving tax subsidies and preferential customs clearance treatment to relevant enterprises. Energy-saving goals and pollution emission goals should be set as conditions for traditional chemical, resource, and heavy industries to apply for tax relief, tax reimbursement, and loans. Industries with high energy consumption and high pollution emissions in important ecological zones should be removed, transformed, or closed.

(2)

Strengthen environmental regulation. Since the higher intensity of environmental regulations can force enterprises to invest in green innovation, the environmental regulation of cities in Eastern and Western China should be strengthened. A subsidy ‘backslope’ should be implemented in CRE operating cities to reduce policy distortion on CRE operation and avoid the crowding-out effect of subsidies on environmental protection expenditures. Regular and irregular supervision should be implemented for enterprises to prevent increasing pollution emissions for excessive economic benefits. A strict accountability mechanism should be established to prevent local governments’ risk of ‘regulation capture’ in pursuit of short-term growth. A reward system should be established to encourage the public to participate in environmental supervision, creating public supervision to cover the periods and regions neglected by government supervision.

(3)

Improve the capability of green innovation. The promotion of regional green innovation capability is a necessary condition for the green transformation of industries. The CRE project has no adverse impact on regional green innovation endowments. The main reason for the reduction in regional green innovation capabilities is that R&D activities of other departments occupy too many resources of green innovation. R&D investments should be guided into the field of green innovation by subsidizing the costs of green innovation and giving tax relief to income generated by green innovation. The in-depth cooperation of enterprises with local scientific research institutions should be encouraged to give feedback regarding information and practical difficulties in a timely manner, such as conducting joint meetings, forums, and trainings. Research institutions in charge of governments should be encouraged to provide low-cost services to enterprises. Since the innovation ability of small-sized and medium-sized enterprises is restricted by capital and technology, large enterprises should be encouraged to participate in green innovation to forge diversified industrial clusters with stronger innovation abilities.