The Impact Path of New Energy Vehicle Promotion on Green Development—Empirical Research from the Provincial Level in China

Abstract

The new energy vehicle (NEV) industry has become one of the most important industries in China’s economic development. Based on the panel data of 27 provincial administrative regions in China from 2011 to 2022, combined with the random effect panel of the Tobit model and the Bootstrap method to test the multiple intermediary paths, this paper studies the impact of new energy vehicle promotion (NEVP) in China on regional green development, taking into account the intermediary effect and regional heterogeneity of NEVP on the green development level (GDL). The results show that NEVP significantly promotes the GDL. The mediating effect of NEVP to improve local-level green development through the digital economy level is significant in the eastern region, while in the central and western regions, it is not significant. NEVP can significantly promote the upgrading of regional industrial structure and the construction of transportation infrastructure in the eastern, central, and western regions so as to improve the local GDL.

1. Introduction

With the emergence of global warming and extreme weather, reducing carbon emissions has become an important global consensus [1]. China has begun to reduce carbon emissions in the transportation field by developing the NEV industry [2]. New energy vehicles mainly refer to the four-wheel vehicles using non-traditional fuels (bioethanol, liquid natural gas, biogas, and biodiesel), electric vehicles, battery electric vehicles, plug-in hybrid vehicles, and various hybrid types of these vehicles [3,4]; compared with traditional fuel vehicles, new energy vehicles (NEVs) show great potential in reducing air pollutant emissions [5], and their zero-exhaust emission is conducive to reducing global carbon emissions [6]. NEVP will help improve the low-carbon traffic efficiency of roads [7], promote the sustainable development of transportation [8,9], drive industrial upgrading [10], and push forward regional green development.

NEVP is an important step to realize the transformation of low-carbon transportation and reduce carbon intensity [11], but the sustainable development of NEVs is still hindered by various technical and policy challenges [12], such as high investment costs, limited mileage, and insufficient charging infrastructures. Therefore, whether NEVP can truly achieve the goal of green development is still a question worthy of discussion. This study attempts to answer the following questions:

(1)

What impact does NEVP have on green development?

(2)

What are the paths through which the impact of NEVP on green development can be realized?

(3)

Will the impact of NEVP on green development be affected by regional heterogeneity?

In order to solve the above problems, based on the data of 27 provinces in China from 2011 to 2022, we consider discussing the impact mechanism of NEVP on green development from the direct and indirect dimensions so as to provide a policy reference for high-quality economic development under the background of green and low-carbon transformation.

The innovations of this paper mainly include the following aspects: (1) Previous studies mainly used new energy vehicle sales [13], ownership [8,14] and whether to implement pilot policies for the promotion and application of new energy vehicles [15] as explanatory variables, while this paper uses the number of NEVP to study the impact of NEVP on green development. (2) In the past, scholars mainly focused on studying the impact of NEVP on carbon emission intensity [15] or energy efficiency [16]. This paper integrates NEVP and the GDL into a unified framework and deeply studies the impact of NEVP on the GDL. By examining three mediating variables, namely the digital economy level, industrial structure upgrading, and transportation infrastructure construction, the differences in the impact in different regions are explored. (3) Previous scholars have used greenhouse gas emissions to measure the green effects of NEVP [8]. This paper characterizes green development from four dimensions: economic benefits, green innovation, environmental governance, and green living.

The rest of this paper is as follows: Section 2 is a literature review. Section 3 puts forward the relevant assumptions. Section 4 gives the data source, variable description, and model construction. Section 5 analyzes the data and gives the results. Finally, according to the results of data analysis, the conclusions and policy recommendations are given in Section 6.

2. Literature Review

2.1. New Energy Vehicle Promotion

(1)

With the rapid development of China’s NEV industry, it is also actively exploring more factors affecting its development [17]. It mainly considers the impact of network embedding [18], technology interaction [19], carbon price [20], credit conditions [21], and other factors on NEVP. Many scholars have actively studied the NEV industry from the perspectives of environmental factors [22], residents’ health, traffic congestion, industrial foundation, etc. Lin and Wu [23] found that network externalities, price acceptability, government subsidies, vehicle performance, environmental problems, and other attitude factors have a significant impact on consumers’ willingness to buy electric vehicles. Pan et al. [24] evaluated the health benefits and monetary value of promoting new energy vehicles in Chongqing. Zhang et al. [25] found that the construction of charging piles can accelerate the penetration of NEVs, which will exacerbate traffic congestion.

(2)

NEVP is inseparable from government policy support [26]. Active government policies can play a key role in NEVP [27]. NEVP policies will directly promote the consumption and industrial upgrading of NEVs. Kuang and Wang [28] believed that the industrial policies of the Chinese government have greatly promoted the structural upgrading and optimization of the NEV industry and stimulated the improvement of the innovation level of enterprises. Zhu et al. [29] found that the unit carbon subsidy for electric vehicles promoted manufacturers’ profits and product demand. Zhao et al. [30] found that government subsidies can promote NEVP. However, excessive reliance on government subsidy policies may lead to insufficient endogenous power in the industry [31]. Therefore, China is gradually canceling the financial subsidies for the NEV industry [32] and shifting the incentive policy from policy guidance to market driven [33]. The market-driven mechanism is becoming an important internal driving force for the sustainable development of NEVs [34].

2.2. Green Development

(1)

There are different views on the definition of green development. Huang et al. [35] defined green development as the process of promoting the harmonious coexistence between man and nature, propelling the continuous accumulation of green assets, improving the environmental well-being of residents, and finally realizing the coordinated progress of the economic society and the ecological environment through the practice of “greening” and “Ecologicalization” within the framework of resource carrying capacity and ecological capacity. Zhu [36] proposed that green development is the recognition of the value of ecological nature; green should be used as the background of development, and development is also the basis of “greening”.

(2)

The measurement standard of green development has not been unified. The measurement mainly includes using the entropy weight method to build the index system to measure the regional green development level [37] and using the input–output model to take green development efficiency as the proxy variable of green development [38]. Che et al. [39] measured the efficiency of green development at the provincial level in China while considering the unexpected output. Hao and Zhu [40] divided the GDL into 6 first-class indicators, including resource utilization, environmental governance, and green life, and 44 s-class indicators, which were measured by the entropy method. Zhou et al. [41] measured the green development efficiency of Chinese cities by using the SBM model while considering the unexpected output and the spatial Markov chain. Zhao [42] divided the evaluation index of the urban GDL into three dimensions, namely ecological greening, economic greening, and social greening, and calculated it by the entropy method.

(3)

The realization path of green development has raised concerns. It mainly focuses on the significant role of industrial agglomeration [43], digital finance [2], environmental regulation [44], population migration [45], a new urbanization level [46], technological innovation [47], and other factors in promoting green development.

2.3. Impact of NEVP on Green Development

2.3.1. NEVP Influences the Effect of Energy Conservation and Emission Reduction

(1)

NEVP has an obvious positive effect on energy conservation and emission reduction [7,48]. Jochem et al. [49] believed that NEVP would help reduce CO₂ emissions. Trost et al. [50] found that the long-term promotion of new energy vehicles helps to further mitigate the greenhouse effect. Lane et al. [51] found that NEVs have certain green benefits to replace traditional fuel vehicles, but plug-in fuel cell vehicles are better than pure electric vehicles. Su et al. [52] analyzed the positive role of the transportation sector in reducing air pollution. Jiang et al. [53] evaluated NEVs and fuel vehicles from the perspectives of energy consumption, pollution emissions, and total driving cost and found that new energy vehicles were superior to traditional internal combustion engine vehicles in energy consumption and carbon emissions.

(2)

The environmental benefits of NEVs are controversial [54]. The high energy consumption and emissions of NEVs in the process of battery preparation make their carbon emissions higher than those of traditional internal combustion engine vehicles [55]; that is, NEVP only transfers emissions from the transportation field to the power generation sector and fundamentally still relies on coal, making it difficult to achieve true emission reduction effects [54].

2.3.2. NEVP Policies Affect Green Development

(1)

The NEVP policy has a certain impact on environmental improvement and the energy structure [56]. Tan et al. [57] found that the emission reduction effect of the subsidy policy was continuously enhanced with the deepening of the policy. Zhang et al. [58] found that the construction of a low-carbon transportation system significantly promoted the optimization of the energy consumption structure. Wang et al. [15] found that the promotion and application of NEV policies significantly reduced the carbon intensity of the cities.

(2)

The implementation effect of NEVP policy is by no means completely positive. Ruan and Liu [59] believed that the emission reduction effect of the subsidy policy for NEVs on mobile pollution sources showed nonlinear characteristics. With the expansion of the industrial scale, its emission reduction efficiency will gradually weaken. Guo and Wang [60] found that government subsidies have a two-way effect on emission reduction: the substitution effect reduces vehicle exhaust emissions, while the direct effect increases electricity emissions by stimulating additional demand. Because the substitution effect is weaker than the direct effect, it shows that subsidies are not an effective means of emission reduction.

(3)

Other impact paths: Sheldon and Dua [61] found that thanks to the cost advantage of new energy vehicles, it indirectly promotes the reduction in fuel consumption for fuel vehicles, and the improvement of the fuel economy has become a way for NEVs to reduce emissions. Road traffic congestion [62] and charging station congestion [63] are considered in the green development path.

2.4. Literature Summary

The existing research on NEVP mainly focuses on the relationship between it and carbon intensity [15], carbon emission reduction [11], and the impact of government policies on NEVP [31]. Few scholars explore the impact of NEVP on the detailed perspective of the green economy such as green development.

Existing studies have not reached a consensus on the relationship between NEVP and green development. On the one hand, NEVP is good or bad, and how to measure the green development, there are differences between the two research conclusions as separate elements. On the other hand, some scholars believe that the development of the NEV industry will exacerbate the negative impact of industrialization on the ecological environment. This dispute has led to the continuous discussion on whether NEVP hinders the realization of the “double carbon” goal, thus affecting green development. This paper constructs a green development evaluation index system from four dimensions, namely economic benefits, green innovation, environmental governance, and green living, to characterize the GDL, which fills the gap of existing research that only characterizes green development from a single dimension.

Table 1. Comparison of variable selection.

References	Explained Variable	Explanatory Variable	Mediating Variables
Wang et al., 2025 [15]	Carbon intensity	Promote and apply pilot policies for new energy vehicles	The upgrading of the transportation industry structure and innovation of green technology
Tan et al., 2018 [57]	Nitrogen dioxide concentration	Policy of “Ten Cities, Ten Thousand New Energy Vehicles Project”
Wang et al., 2022 [64]	Carbon intensity	Policy implementation of “Thousands of Vehicles, Ten of Cities” (TVTC)	The number of newly registered electric vehicles, trams, pure electric vehicles, hybrid buses, taxis, buses, and trams
Gu et al., 2022 [8]	Greenhouse gas emissions (GHGs)	New energy vehicle ownership	Fuel-fired car ownership, the number of end-of-life vehicles, and the traffic congestion index
Xie et al., 2021 [65]	Urban air pollution	NEV subsidy
Cai et al., 2025 [66]	Carbon emission	NEV subsidy policy
Xiong and Cheng, 2023 [16]	Energy efficiency	The total number of NEVs applied in each province	Inter-provincial electricity transmission and renewable energy development
This paper	Green development level (GDL)	NEVP	The digital economy level, industrial structure upgrading, and the transport infrastructure construction level

compares the differences in variable selection between the existing literature and this paper.

3. Research Assumptions

The path assumption of the impact of NEVP on the GDL is shown in Figure 1.

3.1. Direct Impact of NEVP on Green Development

Numerous studies have proved that NEVs can significantly reduce carbon emissions [67]. Driven by burning fossil fuels, traditional internal combustion engine vehicles emit a large amount of greenhouse gases, leading to global warming [30]. However, NEVs do not directly emit carbon dioxide or other greenhouse gases during driving, reducing emissions during the vehicle use phase [68], and if the power source of NEVs is clean energy, the carbon emissions during the whole lifecycle will be further reduced [58]. The most intuitive green effect of NEVs replacing traditional fuel vehicles is to reduce greenhouse gas emissions [50], especially carbon dioxide emissions [69]. NEVP can also improve the green innovation ability of automobile enterprises [70]. Therefore, NEVP directly contributes to reducing carbon emissions and promoting sustainable development [6], bringing positive green effects.

The direct impact of NEVP on green development is not only reflected in the impact on the regional market environment, especially the automotive market environment, but also on consumer behavior [29]. Research on the transformation of China’s policy portfolio shows that consumer-oriented incentives can meet the economic and psychological needs of consumers in the process of using NEVs [71]. NEVP not only promotes the industrial transformation of automobile manufacturers but also redounds the change in consumers’ purchase preferences, which results in the growth of sales or the actual use of NEVs and makes NEVP an indispensable part of cultivating emerging markets for NEVs [31].

Based on this discussion, we propose the following research hypothesis:

H1: 

NEVP plays a positive role in promoting the regional GDL.

3.2. Indirect Impact of NEVP on Green Development

Existing studies show that the impact of NEVP on green development is often not directly generated but indirectly caused by other factors.

3.2.1. Digital Economy Level

The digital economy level plays a key role in the formulation of NEVP policies by the government [26] and the technological innovation of NEVs by enterprises [41]. On the one hand, the development of the NEV industry will directly promote and facilitate the development of the digital economy. On the other hand, an NEVP policy can promote the development of the city in the direction of digitalization and intelligence by driving related supporting industries, transportation infrastructure upgrading, and other ways, thus boosting the development of the digital economy [11]. The improvement of the digital economy can promote high-quality economic development and bring the economic green effect [72].

3.2.2. Industrial Structure Upgrading

NEVP can promote the reform and transformation of China’s automotive industry [73], thus propelling the reconstruction of the entire industrial chain and supply chain [74] and stimulating the development of green and intelligent industrial chains [75]. NEVs have the potential to reconstruct the traditional energy innovation ecosystem [76], which makes the upgrading of the industrial structure force the green innovation and transformation upgrading of automobile enterprises [77], thus prompting the change in green development efficiency [78].

3.2.3. Transport Infrastructure Construction Level

NEVP needs well-developed charging infrastructure [79], which can effectively improve the convenience [80] and popularity [81] of NEVs. The technological progress in energy utilization brought about by the improvement of transportation infrastructure can promote the recycling of energy among industries, improve energy efficiency [82], reduce carbon emissions, and facilitate the sustainable development of the overall transportation system and the green development of cities [83].

Based on this discussion, we propose the following research hypotheses:

H2a: 

NEVP indirectly improves the GDL by improving the level of the digital economy.

H2b: 

NEVP indirectly improves the GDL through the upgrading of the industrial structure.

H2c: 

NEVP indirectly improves the GDL by improving the level of transportation infrastructure construction.

4. Research Methods

4.1. Data Sources

The provincial data rely on a unified standard and are relatively systematic and complete, making them easy to compare and analyze between different provinces. However, new energy vehicles in China mainly lie in medium and large-sized cities; therefore, the provincial data largely represent the data of medium and large-sized cities; open city-year data are available in medium and large-sized cities, while other open city-year data lack availability. Therefore, this paper selects the panel datasets of 27 provincial administrative regions in China (except Xinjiang, Tibet, Ningxia, Qinghai, Hong Kong, Macao, and Taiwan) from 2011 to 2022 as samples and obtains 324 observations, all of which are consistent with the statistical caliber.

The data of the NEVP volume are from the yearbook of energy saving and new energy vehicles (2012–2022), and the data of the GDL are mainly from the China Statistical Yearbook, the China Energy Statistical Yearbook, the China Environmental Statistical Yearbook, the China Science and Technology Statistical Yearbook, the China Industrial Statistical Yearbook, and local statistical yearbooks and bulletins during the period of 2011–2022. All the data of the control variables and intermediate variables are from the China Statistical Yearbook and the local statistical yearbook during the period of 2011–2022. For some missing values, the interpolation method is used to supplement the data, and 1% quantile Winsorize tail reduction is applied to continuous variables to eliminate the impact of abnormal values. All regression results for the data analysis in this paper are obtained by using Stata18 version software.

4.2. Development Status of New Energy Vehicles in China

China’s NEV sales have gradually entered a stage of rapid development [33]. Since 2013, NEV sales have increased significantly, as shown in Figure 2. The research on NEV policies mainly focuses on different policy priorities [84], difficulties in policy implementation [85], and the rationality and effectiveness of the policy [86]. Some cities benefited from government subsidies [65], which stimulated a large number of green technology innovations in the production of NEVs, making NEVs play a vital role in diminishing carbon intensity [15], thus contributing to the improvement of the urban environment and sustainable development.

Different cities adopt different NEVP policies to solve the prominent problems such as urban traffic congestion and air pollution [66]. According to the division of the China Bureau of statistics, China is divided into four parts: the central region, the eastern region, the western region, and the northeast region. But the sample size of the northeast region (the three eastern provinces) is too small, and the comparability is poor. Therefore, this paper divides China into three regions for interregional analysis. The NEVP volume and growth rate in the three regions are shown in Figure 3. We found that for the NEVP volume, the eastern region increased almost exponentially from 2013 to 2015, and the promotion volume from 2017 to 2022 was significantly higher than that in the central and western regions; The central region has been growing steadily from 2011 to 2022. The overall situation of NEVs in the western region is similar to that in the central region, but there is also a big gap with the eastern and central regions, and the development of NEVP is still relatively backward. As for the growth rate, the growth rate in the eastern and central regions fluctuated significantly before 2016 and tended to be stable after 2016; The growth rate in the western region is relatively large, but it is still different from that in the eastern and central regions, and the overall growth rate is also lower than that in the eastern and central regions.

4.3. Green Development Level of Provincial Administrative Regions

In this paper, the entropy–TOPSIS method is used to measure and calculate the green development level comprehensive index of 27 provincial administrative regions in China to reflect the GDL (see

Table 2. The GDL in China’s 27 provincial-level administrative regions.

Province	2011	2012	2013	2014	2015	2016	2017	2018	2019	2020	2021	2022
Anhui	0.445	0.462	0.477	0.476	0.493	0.498	0.527	0.523	0.519	0.538	0.537	0.542
Beijing	0.489	0.513	0.538	0.571	0.573	0.587	0.609	0.584	0.576	0.556	0.571	0.579
Fujian	0.455	0.491	0.493	0.493	0.506	0.512	0.539	0.516	0.542	0.561	0.559	0.530
Gansu	0.334	0.353	0.359	0.372	0.394	0.406	0.444	0.437	0.449	0.475	0.456	0.401
Guangdong	0.497	0.517	0.535	0.528	0.555	0.546	0.578	0.588	0.592	0.599	0.606	0.615
Guangxi	0.420	0.425	0.438	0.428	0.445	0.457	0.451	0.454	0.466	0.480	0.474	0.487
Guizhou	0.354	0.394	0.409	0.410	0.440	0.444	0.438	0.458	0.477	0.492	0.496	0.492
Hainan	0.402	0.418	0.428	0.411	0.436	0.463	0.445	0.452	0.444	0.471	0.492	0.511
Hebei	0.390	0.426	0.440	0.445	0.484	0.493	0.509	0.508	0.512	0.529	0.502	0.413
Henan	0.407	0.410	0.416	0.423	0.441	0.469	0.474	0.475	0.471	0.498	0.486	0.499
Heilongjiang	0.356	0.364	0.399	0.399	0.426	0.435	0.435	0.432	0.459	0.470	0.462	0.473
Hubei	0.412	0.434	0.456	0.458	0.468	0.497	0.486	0.494	0.499	0.523	0.516	0.521
Hunan	0.422	0.433	0.441	0.451	0.469	0.493	0.491	0.505	0.518	0.533	0.527	0.534
Jilin	0.365	0.396	0.417	0.419	0.432	0.462	0.408	0.440	0.449	0.471	0.473	0.483
Jiangsu	0.504	0.518	0.537	0.540	0.562	0.566	0.580	0.575	0.581	0.607	0.603	0.605
Jiangxi	0.431	0.445	0.447	0.444	0.452	0.453	0.505	0.501	0.496	0.525	0.523	0.545
Liaoning	0.380	0.407	0.425	0.407	0.431	0.458	0.456	0.461	0.470	0.491	0.489	0.418
Inner Mongolia	0.406	0.420	0.436	0.457	0.478	0.482	0.483	0.473	0.477	0.494	0.492	0.510
Shandong	0.491	0.515	0.529	0.516	0.538	0.556	0.553	0.543	0.542	0.560	0.560	0.560
Shanxi	0.404	0.416	0.431	0.426	0.430	0.441	0.446	0.460	0.476	0.480	0.480	0.497
Shaanxi	0.434	0.452	0.466	0.470	0.485	0.492	0.476	0.479	0.480	0.500	0.490	0.508
Shanghai	0.416	0.452	0.457	0.462	0.489	0.510	0.529	0.508	0.505	0.517	0.512	0.523
Sichuan	0.403	0.414	0.420	0.413	0.431	0.448	0.446	0.456	0.460	0.480	0.466	0.479
Tianjin	0.482	0.501	0.513	0.500	0.518	0.540	0.531	0.487	0.534	0.522	0.515	0.510
Yunnan	0.370	0.417	0.422	0.437	0.432	0.432	0.467	0.453	0.455	0.456	0.472	0.489
Zhejiang	0.507	0.522	0.529	0.542	0.557	0.568	0.569	0.579	0.591	0.604	0.588	0.576
Chongqing	0.489	0.491	0.496	0.490	0.504	0.518	0.511	0.497	0.498	0.530	0.527	0.527

for details).

There are differences in the GDL in different regions. Figure 4 shows the annual average value and growth rate of the GDL in the eastern, central, and western regions of China. We can find that the green development situation in the eastern region is significantly better than that in the central and western regions. The GDL in the central and western regions is relatively close, and the overall level of the three regions is low. On the whole, the GDL presents the spatial distribution characteristics of decreasing from coastal to inland.

4.4. Variable Selection

Variables are defined in

Table 3. Definitions of the variables.

Variables	Definitions	Types
GDL	Green development level	Explained variable
PLNEV	Promotion level of new energy vehicles	Explanatory variable
FD	Fiscal decentralization	Control variable
FDI	Foreign direct investment	Control variable
DO	Degree of opening up	Control variable
ER	Environmental regulation	Control variable
DEL	Digital economy level	Mediating variable
ISU	Industrial structure upgrading	Mediating variable
TICL	Transport infrastructure construction level	Mediating variable

.

4.4.1. Explained Variable

The green development level (GDL) is the explained variable of this paper. This paper comprehensively considers the coordinated development of the green economy and traditional industries, the innovation ability, and the impact of residents’ lives; evaluates the GDL of each provincial administrative region; and selects 4 first-class indicators and 19 s-class indicators to build the evaluation index system of GDL from the perspectives of economic benefits, green innovation, environmental governance, and green life (see

Table 4. Green development level evaluation indicator system.

Primary Indicator	Secondary Indicators	Indicator Measurement Method	Unit	Indicator Direction
Economic benefits	Government financial capacity	Per capita local fiscal revenue	Yuan/person	+
	Economic development level	Per capita gross domestic product (GDP)	Yuan/person	+
	Socioeconomic contribution rate	Industrial added value/GDP	%	+
	Economic growth quality	The proportion of added value of the tertiary industry to the GDP	%	+
Green innovation	Green R&D funding investment	Expenditure on R&D and experimental development funds for industrial enterprises above a designated size/GDP	%	+
	Green R&D personnel intensity	Full-time equivalent R&D personnel of industrial enterprises above a designated size	Person/year	+
	Green innovation capability	Number of green invention patent authorizations	Number/year	+
	Efficiency of achievement conversion	Technology market turnover/GDP	%	+
Environmental governance	Pollution control	Proportion of local government environmental protection expenditure to general budget expenditure	%	+
	Cycle regeneration	Utilization rate of general industrial solid waste	%	+
	The total amount of industrial wastewater has decreased	(Total amount of industrial wastewater in the previous period-total amount of industrial wastewater in the current period)/total amount of industrial wastewater in the previous period	%	+
	Reduction in total industrial sulfur dioxide emissions	(Total amount of industrial sulfur dioxide in the previous period-total amount of industrial sulfur dioxide in the current period)/total amount of industrial sulfur dioxide in the previous period	%	+
	Reduction in total industrial smoke and dust	(Total amount of industrial smoke and dust in the previous period-total amount of industrial smoke and dust in the current period)/total amount of industrial smoke and dust in the previous period	%	+
Green life	Green travel	Number of standard urban public transportation vehicles/total population of the city	Standard platform/10,000 people	+
	Wastewater treatment level	Wastewater treatment capacity/total sewage discharge	%	+
	Harmless level of household waste disposal	Harmless treatment capacity of household waste/output of household waste	%	+
	Forest greening level	Forest area/total land area	%	+
	Park greening level	Urban park green area/total urban population	Square meters/person	+
	Road traffic level	Urban road area/urban population	Square meters	+

Notes: “+” represents a positive impact of secondary indicators on primary indicator.

for details).

4.4.2. Core Explanatory Variable, Control Variables, and Intermediary Variables

(1)

Core explanatory variable

The promotion level of new energy vehicle (PLNEV) is the core explanatory variable of this paper, which can usually be measured by NEV sales [13], NEV ownership [8,14], NEV sales per capita [87], etc. Referring to the practice of Xiong and Lin [88], this paper uses the newly increased NEVP volume in provinces (cities and autonomous regions) to replace the NEVP level in various regions and directly obtains the NEVP volume from the yearbook of energy saving and new energy vehicles for the period of 2011–2022. At the same time, in order to ensure the stability of the data, this variable is processed by natural logarithm.

(2)

Control variable

Hao and Zhu [40] conducted an empirical study on the influencing factors of green development. This paper selects four variables as control variables: fiscal decentralization, foreign direct investment, degree of opening up, and environmental regulation.

➀

Fiscal decentralization (FD)

Fiscal decentralization refers to how to reasonably divide the powers of fiscal revenue and expenditure management, fiscal decision-making, and resource allocation between the central government and local governments under the framework of the government system. Fiscal decentralization may also bring environmental pollution problems, which is not conducive to regional green development. Therefore, this paper uses fiscal decentralization as one of the control variables. The calculation formula of fiscal decentralization is as follows:

$$FD={\displaystyle \frac{FiscalexpenditurebyRegion}{Centralpercapitafiscalexpenditure}}$$

(1)

➁

Foreign direct investment (FDI)

Foreign direct investment is one of the main forms of modern capital internationalization, which carries the function of technology spillover. As technology introduction is mainly realized in the form of foreign direct investment, this paper selects the proportion of foreign direct investment in GDP in various regions to measure technology introduction.

$$FDI={\displaystyle \frac{Foreigndirectinvestment}{localGDP}}\times exchangerate$$

(2)

➂

Degree of opening up (DO)

The degree of opening up reflects the level of market openness of a region. The improvement of the degree of opening up can not only improve the level of local green technology and promote the efficiency of the green economy through the technology spillover effect but also lead to the transfer of foreign high pollution, low technology, and high energy consumption industries to the local, thus intensifying the large-scale development of local pollution industries. The calculation formula for the degree of opening up is as follows:

$$DO={\displaystyle \frac{Importandexportamountofgoods}{localGDP}}\times exchangerate$$

(3)

➃

Environmental regulation (ER)

The variable of environmental regulation indicates the intensity of environmental regulation. In this paper, the proportion of industrial pollution control investment in industrial added value is used to express that pollution emissions and governance directly determine the regional environmental quality. However, there may be an endogenous correlation between pollution control and pollutant emissions; that is, the higher the degree of environmental pollution, the higher the cost of pollution control; Accordingly, the more pollution control costs, the higher the degree of environmental pollution. Therefore, the increase in investment in environmental protection may correspond to a lower degree of industrial pollution in the region, but it may also correspond to higher levels of pollution. The calculation formula of environmental regulation intensity is as follows:

$$ER={\displaystyle \frac{Completedinvestmentinindustrialpollutioncontrol}{Industrialaddedvalue}}\times exchangerate$$

(4)

(3)

Mediating variable

➀

Digital economy level (DEL)

Referring to Zhao et al. [89], this paper measures the development level of the Internet by using indicators such as the Internet penetration rate and the number of Internet-related employees, etc., and uses the China digital inclusive finance index to represent the development level of digital finance. Finally, the principal component analysis method is used to calculate the digital economic comprehensive development index. The index system is shown in

Table 5. Indicator system for the comprehensive development of the digital economy.

Project	Index	Variables	Positive or Negative
Comprehensive development index of the digital economy	Internet penetration	Number of Internet users per 100 people	Positive
	Number of Internet-related employees	Proportion of computer services and software employees	Positive
	Internet-related output	Total telecom services per capita	Positive
	Number of mobile Internet users	Number of mobile phone users per 100 people	Positive
	Digital finance inclusive development	China digital inclusive finance index	Positive

.

➁

Industrial structure upgrading (ISU)

Industrial structure refers to the proportion and interrelationship of different industries in the gross national product. The upgrading of the industrial structure is a dynamic evolution process, which follows the change law from the primary industry to the secondary industry and then to the dominant position of the tertiary industry. The optimization and upgrading of the industrial structure can promote economic development, improve energy efficiency, and reduce the negative impact on the environment. The calculation formula of industrial structure upgrading is as follows:

$$IS={\displaystyle \frac{Addedvalueoftertiaryindustry}{Addedvalueofsecondaryindustry}}$$

(5)

➂

Transport infrastructure construction level (TICL)

This paper uses regional traffic density to measure the transport infrastructure construction level. The construction of transportation infrastructure will promote economic growth. The formation of the transportation network reduces the transportation cost, further reduces the factor endowment effect, and promotes trade exchanges, which leads to the improvement of the efficiency of factor allocation, makes enterprises produce economies of scale, and then affects the economic output. The calculation formula for the construction level of transportation infrastructure is as follows:

$$TICL={\displaystyle \frac{Regionaltrafficmileage}{Regionaladministrativearea}}$$

(6)

4.5. Model Construction

4.5.1. Tobit Model

Because the GDL value calculated by the entropy TOPSIS method in this paper has the characteristics of truncation, if OLS regression is used, the nonlinear disturbance term will be included in the error term, which may lead to biased estimation results. Therefore, the Tobit model is more suitable for this paper. Combined with H1, the lower limit and the upper limit are 0 and 1, respectively. The panel Tobit model is as follows:

$$G{DL}_{it}={\alpha }_0+{\alpha }_1{PLNEV}_{it}+{\sum }_{j=1}^n{\gamma }_{i}C{V}_{ijt}+{\epsilon }_{it}$$

(7)

Subscript $i$ refers to the region; $t$ refers to the time; $j$ refers to the number of control variables; $G{DL}_{it}$ refers to the GDL value of the province (municipality directly under the Central Government) $i$ in the period $t$; and ${PLNEV}_{it}$ refers to the NEV market promotion level of the province (municipality directly under the Central Government) $i$ in the period $t$. ${\epsilon }_{it}$ is a random disturbance term, and $C{V}_{ijt}$ represents a series of control variables.

4.5.2. Mediation Effect Model

This paper uses the stepwise regression method to study the impact of NEVP on urban green development and establishes the following intermediary effect model:

(1)

Digital economy level

$$DE{L}_{it}={\phi }_0+{\phi }_1{PLNEV}_{it}+{\sum }_{j=1}^n{\gamma }_{i}C{V}_{ijt}+{\epsilon }_{it}$$

(8)

$$G{DL}_{it}={\phi }_2+{\phi }_{3}DE{L}_{it}+{\phi }_4{PLNEV}_{it}+{\sum }_{j=1}^n{\gamma }_{i}C{V}_{ijt}+{\epsilon }_{it}$$

(9)

(2)

Industrial structure upgrading

$$I{SU}_{it}={\theta }_0+{\theta }_1{PLNEV}_{it}+{\sum }_{j=1}^n{\gamma }_{i}C{V}_{ijt}+{\epsilon }_{it}$$

(10)

$$G{DL}_{it}={\theta }_2+{\theta }_{3}I{SU}_{it}+{\theta }_4{PLNEV}_{it}+{\sum }_{j=1}^n{\gamma }_{i}C{V}_{ijt}+{\epsilon }_{it}$$

(11)

(3)

Transport infrastructure construction level

$$T{ICL}_{it}={\varphi }_0+{\phi }_1{PLNEV}_{it}+{\sum }_{j=1}^n{\gamma }_{i}C{V}_{ijt}+{\epsilon }_{it}$$

(12)

$$G{DL}_{it}={\phi }_2+{\phi }_{3}T{ICL}_{it}+{\phi }_4{PLNEV}_{it}+{\sum }_{j=1}^n{\gamma }_{i}C{V}_{ijt}+{\epsilon }_{it}$$

(13)

Taking the digital economy level path test as an example, the test steps of the mediation effect model are as follows: First, check whether ${\phi }_1$ and ${\phi }_3$ are significant, and if ${\phi }_1$ and ${\phi }_3$ are significant, the indirect effect is significant. Continue to test ${\phi }_4$; if it is significant, consider it to be part of the mediation effect. Then check whether ${\phi }_1\times {\phi }_3$ and ${\phi }_4$ are the same symbol; if they have the same symbol, we can obtain the proportion of part of the mediation effect ${\displaystyle \frac{{\phi }_1\times {\phi }_3}{{\phi }_4}}$. If at least one of ${\phi }_1$ or ${\phi }_3$ is not significant, use the bootstrap method to test ${\phi }_1\times {\phi }_3$; if it is significant, the sign test is performed, and the proportion of the mediation effect is obtained. If it is not significant, the indirect effect is not significant, and this path is not a mediation path. If ${\phi }_1\times {\phi }_3$ and ${\phi }_4$ are different signs on the premise of ${\phi }_4$ being significant, the masking effect is obtained.

5. Empirical Results and Analysis

5.1. Sample Descriptive Statistics

By sorting out the relevant data of 27 provinces, a descriptive statistical table as shown in

Table 6. Sample descriptive statistics.

Variables	Obs.	Mean	Std. Dev.	Min	Max
GDL	324	0.480	0.050	0.356	0.606
PLNEV	324	2.380	1.720	0.000	5.793
FD	324	6.500	2.500	3.812	14.586
FDI	324	0.020	0.020	0.001	0.105
DO	324	0.290	0.280	0.033	1.294
ER	324	0.000	0.000	0.000	0.011
DEL	324	0.600	0.070	0.485	0.850
ISU	324	1.390	0.770	0.638	5.022
TICL	324	1.040	0.470	0.143	2.125

is obtained. The maximum value of the explained variable GDL is 0.606; the minimum value is 0.356; the average value is 0.480, and the standard deviation is 0.05. It can be seen that there are some differences in green development in different regions. The maximum value of the explanatory variable PLNEV is 5.793; the minimum value is 0; the average value is 2.380; the standard deviation is 1.720; and the overall coefficient of variation is 0.720, indicating that the national new energy promotion level has great differences in different regions and years. For control variables and intermediate variables, FD, FDI, DO, ER, DEL, ISU, and TICL have no abnormal values and normal ranges.

5.2. Stability Test

LLC and IPS are used to test the stationarity of the core variables to ensure the stability of the panel data and avoid the pseudo-regression problem. When both methods are significant, it can be determined that the variable data is stable. The results in

Table 7. Unit root test.

Variables	LLC	IPS	Whether It Is Stable
GDL	−5.66 ***	−2.15 ***	Yes
PLNEV	−6.64 ***	−6.95 ***	Yes
FD	−5.93 ***	−5.23 ***	Yes
FDI	−6.17 ***	−2.53 ***	Yes
DO	−3.30 ***	−2.89 ***	Yes
ER	−9.27 ***	−3.57 ***	Yes
DEL	−9.25 ***	−2.38 ***	Yes
ISU	−2.52 ***	−5.20 ***	Yes
TCIL	−7.85 ***	−2.48 ***	Yes

Notes: * p < 0.10, ** p < 0.05, and *** p < 0.01.

show that the LLC test of all variables is significant at the significance level of 1%, indicating that the data are stable; The IPS test also supports this conclusion.

5.3. Basic Regression Analysis

The correlation analysis of variables is carried out to determine the degree of correlation between variables and the VIF test to determine whether there is multicollinearity between variables. The results of the correlation analysis and the VIF test are shown in

Table 8. Correlation analysis of the sample variables.

Variables	GDL	PLNEV	FD	FDI	DO	ER
GDL	1
PLNEV	0.68 ***	1
FD	0.22 ***	0.07	1
FDI	0.12 **	−0.09	0.40 ***	1
DO	0.47 ***	0.15 ***	0.65 ***	0.42 ***	1
ER	−0.37 ***	−0.43 ***	0.04	0.02	−0.22 ***	1

Notes: * p < 0.10, ** p < 0.05, and *** p < 0.01.

and

Table 9. VIF test results.

Variables	PLNEV	FD	FDI	DO	ER	DEL	ISU	TCIL
VIF	4.01	3.39	1.54	2.59	1.45	6.09	1.98	1.99

, respectively. The results of Table 9 show that there is no multicollinearity between variables.

Regarding the research on influencing factors, many scholars have conducted causal inference based on difference in differences (DID) and propensity score matching (PSM) to study the impact of NEVP policies [57,60,65]. However, there are differences between the types and timing of NEVP policies proposed by each province in this paper, so it is difficult to distinguish between the “treatment group” and the “control group”. Therefore, this paper does not consider using the DID and PSM methods for causal inference.

Due to the inability of fixed-effects Tobit models to find sufficient statistical measures for individual heterogeneity and the inconsistency of their estimation results, fixed effects are not suitable for panel Tobit model estimation. The LR test results indicate the presence of individual effects; therefore a random-effects panel Tobit model should be used for regression analysis. This paper conducts stepwise regression of the Tobit model using the Stata17.0 software, and the results are shown in

Table 10. Baseline regression results.

Variables	Model (1)	Model (2)	Model (3)	Model (4)	Model (5)
	GDL	GDL	GDL	GDL	GDL
PLNEV	0.017 ***	0.016 ***	0.016 ***	0.015 ***	0.015 ***
	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)
FD		0.008 ***	0.010 ***	0.010 ***	0.010 ***
		(0.002)	(0.002)	(0.002)	(0.002)
FDI			−0.189 *	−0.191 *	−0.175
			(0.108)	(0.107)	(0.108)
DO				−0.016	−0.019
				(0.015)	(0.015)
ER					−0.823
					(0.713)
Constant term	0.443 ***	0.394 ***	0.388 ***	0.389 ***	0.392 ***
	(0.007)	(0.014)	(0.015)	(0.015)	(0.015)
LR test	Pass	Pass	Pass	Pass	Pass
Model selection	Random effect	Random effect	Random effect	Random effect	Random effect
N	324	324	324	324	324

Notes: * p < 0.10, ** p < 0.05, and *** p < 0.01; values in parentheses denote the robust standard error for the coefficient.

. Model (1) only regresses the explanatory variable PLNEV, while models (2)–(5) gradually add the control variables FD, FDI, DO, ER, and regresses, respectively. The results show that with the introduction of control variables one by one, the significance and coefficient direction of the core explanatory variables remain stable, indicating that the model estimation results are robust.

From models (1)–(5) in Table 10, it can be found that the core explanatory variable PLNEV is significant at the significance level of 1%, and the regression coefficient values are greater than 0, indicating that it has a significant positive role in promoting green development. For the control variables, FD has a significant positive effect on the GDL, FDI, DO, and ER has no significant negative relationship with the GDL.

5.4. Multiple Intermediary Tests and Analyses

The panel Tobit model with random effects is used to estimate the stepwise regression model, and the nonparametric percentile bootstrap method with deviation correction is used for the intermediary test. Bootstrap sampling is set for 5000 times, and the confidence level is 95%. The results are shown in

Table 11. Stepwise regression results.

Variables		Digital Economy Level		Industrial Structure Upgrading		Transportation Infrastructure Construction Level
		DEL	GDL	ISU	GDL	TICL	GDL
Core explanatory variable	PLNEV	0.026 ***	0.003 **	0.075 ***	0.012 ***	0.048 ***	0.012 ***
		(0.001)	(0.001)	(0.008)	(0.001)	(0.004)	(0.001)
Mediating variables	DEL	-	0.420 ***	-	-	-	-
			(0.041)
	ISU	-	-	-	0.032 ***	-	-
					(0.006)
	TICL	-	-	-	-	-	0.050 ***
							(0.012)
Control variables	FD	0.013 ***	0.008 ***	0.070 ***	0.008 ***	0.017 *	0.009 ***
		(0.002)	(0.002)	(0.023)	(0.002)	(0.010)	(0.002)
	FDI	−0.606 ***	−0.010	−3.995 ***	−0.044	−0.327	−0.170
		(0.121)	(0.097)	(0.973)	(0.106)	(0.425)	(0.104)
	DO	−0.002	−0.013	−0.796 ***	−0.004	0.226 ***	−0.036 **
		(0.015)	(0.013)	(0.144)	(0.015)	(0.063)	(0.015)
	ER	−3.307 ***	0.615	−1.537 ***	−0.320	−1.072	−0.752
		(0.847)	(0.627)	(6.248)	(0.679)	(2.740)	(0.695)
Bootstrap test		Digital economy level		Industrial structure upgrading		Transportation infrastructure construction level
		upper bound	lower bound	upper bound	lower bound	upper bound	lower bound
Mediation path		0.0049	0.0108	0.0002	0.0019	0.0002	0.0028
Direct path		0.0088	0.0157	0.0166	0.0218	0.0153	0.0218

Notes: * p < 0.10, ** p < 0.05, and *** p < 0.01; values in parentheses denote the robust standard error for the coefficient; “-” represents a blank value.

.

It can be found from Table 11 that the regression coefficients of PLNEV to DEL and DEL to GDL, PLNEV to ISU and ISU to GDL, and PLNEV to TICL and TICL to GDL are significantly positive, indicating that the improvement of the NEVP level has significantly promoted the improvement of the digital economy level, industrial structure upgrading, and the transportation infrastructure construction level, thus significantly promoting green development. The mediating effect of the digital economy level, industrial structure upgrading, and the transportation infrastructure construction level are significant, which shows that there are three paths: “NEVP is improved → the level of digital economy is improved/the industrial structure is upgraded/the level of transportation infrastructure construction is improved → the regional GDL is improved”, so H2a, H2b, and H2c are all proved, and the mediating effect of each path is shown in Figure 5.

5.5. Regional Heterogeneity Analysis

China has a vast geographical space, and there are huge differences in economic development, the geographical environment, and resource endowment in different regions. The natural endowment, economic situation, and infrastructure construction in the eastern coastal areas of China have certain advantages over those in the central and western regions. Therefore, in order to identify the heterogeneity of different regions, China is divided into its eastern, central, and western regions. The eastern region includes 11 provinces and cities, including Beijing, Fujian, Guangdong, Hainan, Hebei, Jiangsu, Liaoning, Shandong, Shanghai, Tianjin, and Zhejiang. The central region includes nine provinces and cities including Chongqing, Anhui, Heilongjiang, Henan, Hubei, Hunan, Jiangxi, Jilin, and Shanxi. The western region includes seven provinces and cities (autonomous regions) including Gansu, Guangxi, Guizhou, Inner Mongolia, Shaanxi, Sichuan, and Yunnan. The regression results of the total effect in different regions is shown in

Table 12. Regression results for total effects in China’s eastern, central, and western regions.

Variables	Eastern Region	Central Region	Western Region
	GDL	GDL	GDL
PLNEV	0.013 ***	0.017 ***	0.011 ***
	(0.002)	(0.001)	(0.002)
FD	0.006 **	0.031 ***	0.012 ***
	(0.003)	(0.006)	(0.004)
FDI	−0.119	−0.254	−0.312
	(0.127)	(0.290)	(0.395)
DO	−0.040 **	−0.220 ***	−0.056
	(0.020)	(0.069)	(0.071)
ER	1.359	−0.696	−4.200 ***
	(1.062)	(1.339)	(1.177)
_cons	0.458 ***	0.289 ***	0.385 ***
	(0.024)	(0.030)	(0.033)
LR test	Pass	Pass	Pass
Model selection	Random effect	Random effect	Random effect
N	132	96	96

Notes: * p < 0.10, ** p < 0.05, and *** p < 0.01; values in parentheses denote the robust standard error for the coefficient.

.

The regression results in Table 12 show that NEVP in the eastern, central, and western regions has a significant positive effect on the GDL, and the central region has the largest regression coefficient, indicating that the promotion effect of this region on the GDL is more obvious, mainly because the state has increased its support for the central region in recent years, and its industrial structure optimization and transformation are also accelerating, which has a more obvious role in promoting green development. The reasons for the eastern region are because the NEV market has been relatively mature, the government’s support policies and market demand are relatively strong, and the incremental effect is relatively limited. The economic development in the western region is relatively backward, and its infrastructure construction and market acceptance are relatively insufficient. Due to the constraints of economic volume, industrial structure adjustment, and policy support, the effect of its promotion is not as good as that in the eastern and central regions.

To further explore the path differences in the impact of new energy vehicle market promotion on green development in different regions, the intermediary effect analysis was conducted by regions again. The regression results are shown in

Table 13. Stepwise regression results in China’s eastern, central, and western regions.

Regions	Variables	DEL	GDL	ISU	GDL	TICL	GDL
Eastern region	PLNEV	0.024 ***	0.008 ***	0.054 ***	0.012 ***	0.037 ***	0.010 ***
		(0.002)	(0.002)	(0.015)	(0.002)	(0.005)	(0.002)
	DEL		0.225 ***
			(0.072)
	ISU				0.014 *
					(0.008)
	TICL						0.081 ***
							(0.027)
Central region	PLNEV	0.026 ***	0.003	0.101 ***	0.012 ***	0.038 ***	0.014 ***
		(0.002)	(0.002)	(0.013)	(0.002)	(0.007)	(0.002)
	DEL		0.535 ***
			(0.053)
	ISU				0.048 ***
					(0.011)
	TICL						0.065 ***
							(0.014)
Western region	PLNEV	0.028 ***	0.000	0.068 ***	0.007 ***	0.057 ***	0.009 ***
		(0.002)	(0.002)	(0.015)	(0.002)	(0.007)	(0.002)
	DEL		0.421 ***
			(0.072)
	ISU				0.055 ***
					(0.011)
	TICL						0.033 ***
							(0.012)
	Control variable	Yes	Yes	Yes	Yes	Yes	Yes

Notes: * p < 0.10, ** p < 0.05, and *** p < 0.01; values in parentheses denote the robust standard error for the coefficient.

.

It can be seen from Table 13 that the impact of NEVP on the GDL through the path of industrial structure upgrading and transportation infrastructure construction is significantly positive in the eastern, central, and western regions, but there are certain regional differences in the mediating effect through the level of the digital economy. Specifically, the mediation path of the digital economy level in the eastern region is significant, while it is not significant in the central and western regions. This is because the digital infrastructure in the eastern region is better, the degree of digital transformation of enterprises is higher, and the integration ability of digital technology and green technology is stronger, so the path of promoting green development through the digital economy of new energy vehicles is more unblocked. However, the digital infrastructure in the central and western regions is relatively weak; the digitalization level of enterprises is low; and the penetration ability of the digital economy to traditional industries is slightly insufficient, so the path of the digital economy level is not significant.

Based on the results of Table 12 and Table 13, we find that for the eastern region, the mediating effect of the digital economy level accounts for 41.5% of the total effect, the mediating effect of industrial structure upgrading accounts for 5.8%, and the mediating effect of the transportation infrastructure construction level accounts for 2.8%. The mediating effect of the three paths on green development is significant, but the mediating effect of industrial structure upgrading and the transportation infrastructure construction level is relatively low. For the central region, the mediating effect of industrial structure upgrading accounted for 28.5% of the total effect, and the mediating effect of the transportation infrastructure construction level accounted for 14.5%. The mediating effect of industrial structure upgrading and the transportation infrastructure construction level on green development was significant and accounted for a relatively high proportion. For the western region, the mediating effect of industrial structure upgrading accounts for 34% of the total effect, while the mediating effect of transportation infrastructure construction level accounts for 17.1%. Due to the relative lag of economic development, the western region has greater potential for upgrading the industrial structure and improving the transportation infrastructure, and NEVP shows a more significant mediating effect in this region.

5.6. Robustness and Endogenous Test

5.6.1. Robustness Test

To avoid the deviation of measurement results caused by the lax selection of indicators, this paper selects green development efficiency as an alternative variable to further verify the impact of NEVP on the GDL; using ref. [39] for reference, this paper considers the SBM model with an unexpected output and calculated the green development efficiency (GE) of each province through the matlabr2021b software. The measurement of green development efficiency (GE) needs to include four types of indicators: non-resource input (labor and capital), resource input, expected output, and unexpected output. Specifically, the labor input index is represented by the number of employees at the end of the year. The capital input index estimates the capital stock based on the perpetual inventory method, extends the regional capital stock data to 2022, and converts it into a capital stock data proxy based on 2010. The resource input index is represented by the total energy consumption (standard coal) of each province. The expected output is represented by GDP (based on CPI in 2010). The unexpected output is the discharge of industrial waste gas, waste water, and solid waste, which is represented by the comprehensive index calculated by the entropy method. The test results of the random-effects model and the fixed-effects model are shown in

Table 14. Substitution of the explained variables of the random-effects model.

Variables	GE	DEL	GE	ISU	GE	TICL	GE
PLNEV	0.005 **	0.026 ***	0.006 *	0.075 ***	0.003	0.048 ***	0.001
	(0.002)	(0.001)	(0.004)	(0.008)	(0.003)	(0.004)	(0.003)
DEL			0.403 **
			(0.111)
ISU					0.023
					(0.015)
TICL							0.070 **
							(0.034)
Control variable		Yes	Yes	Yes	Yes	Yes	Yes

Notes: * p < 0.10, ** p < 0.05, and *** p < 0.01; values in parentheses denote the robust standard error for the coefficient.

and

Table 15. Substitution of the explained variables of the fixed-effects model.

Variables	GE	DEL	GE	ISU	GE	TICL	GE
PLNEV	0.006	0.002	0.012	0.001	0.006	0.018	0.005
	(0.009)	(0.001)	(0.009)	(0.030)	(0.009)	(0.013)	(0.009)
DEL			−2.662 **
			(1.035)
ISU					−0.112 *
					(0.055)
TICL							0.027
							(0.065)
Control variable		Yes	Yes	Yes	Yes	Yes	Yes

Notes: * p < 0.10, ** p < 0.05, and *** p < 0.01; values in parentheses denote the robust standard error for the coefficient.

, respectively. Comparing the regression results in Table 11 and Table 14, it can be found that the regression coefficient and significance are basically the same, so the model established in this paper is robust.

For the robustness test of the mediating effect, this paper also replaces the mediating variables. Specifically, the overall upgrading of the industrial structure is the agent of the upgrading of the industrial structure. We use the entropy method to re-measure the level of the digital economy. The natural logarithm of the regional freight volume is used as the proxy variable for the construction level of transportation infrastructure. The test results of the random-effects model and the fixed-effects model are shown in

Table 16. Substitution of the mediator variables of the random-effects model.

Variables	DEL_new	GDL	ISU_new	GDL	TICL_new	GDL
PLNEV	0.070 ***	0.003 ***	0.025 ***	0.008 ***	0.063 ***	0.013 ***
	(0.003)	(0.001)	(0.002)	(0.001)	(0.006)	(0.001)
DEL_new		0.159 ***
		(0.014)
ISU_new				0.253 ***
				(0.031)
TICL_new						0.029 ***
						(0.007)
Control variable	Yes	Yes	Yes	Yes	Yes	Yes

Notes: * p < 0.10, ** p < 0.05, and *** p < 0.01; values in parentheses denote the robust standard error for the coefficient.

and

Table 17. Substitution of the mediator variables of the fixed-effects model.

Variables	DEL_new	GDL	ISU_new	GDL	TICL_new	GDL
PLNEV	0.006 *	−0.003 **	0.010 ***	−0.004 **	0.029	−0.004 **
	(0.003)	(0.002)	(0.003)	(0.002)	(0.026)	(0.001)
DEL_new		−0.020
		(0.065)
ISU_new				0.059
				(0.049)
TICL_new						0.004
						(0.006)
Control variable	Yes	Yes	Yes	Yes	Yes	Yes

Notes: * p < 0.10, ** p < 0.05, and *** p < 0.01; values in parentheses denote the robust standard error for the coefficient.

, respectively. It shows that after replacing the intermediary variable and keeping the control variable unchanged, the stepwise regression effect coefficients are significant, and the positive and negative coefficient are consistent with Table 11, which confirms that NEVP has a significant impact on the GDL.

The promotion effect of NEVP on green development may have a time lag effect. Therefore, this paper also uses the lag period of the new energy vehicle promotion level (L. PLNEV) as the core explanatory variable and performs a regression analysis on the benchmark model to test the robustness of the direct effect mechanism. Finally, this paper also randomly eliminated the sample data of 2011, 2014, and 2020 to test the robustness of the model. The test results of the random-effects model and the fixed-effects model are shown in

Table 18. Lagged variables and randomized censored year results of the random-effects model.

Methods	Explanatory Variable Lag Phase I	Random Elimination Year
Variables	GDL	GDL
PLNEV		0.013 ***
		(0.001)
L.PLNEV	0.013 ***
	(0.001)
Control variable	Yes	Yes

Notes: * p < 0.10, ** p < 0.05, and *** p < 0.01; values in parentheses denote the robust standard error for the coefficient.

and

Table 19. Lagged variables and randomized censored year results of the fixed-effects model.

Methods	Explanatory Variable Lag Phase I	Random Elimination Year
Variables	GDL	GDL
PLNEV		−0.003
		(0.002)
L.PLNEV	−0.003 *
	(0.002)
Control variable	Yes	Yes

Notes: * p < 0.10, ** p < 0.05, and *** p < 0.01; values in parentheses denote the robust standard error for the coefficient.

, respectively. The results in Table 18 indicate that the improvement of NEVP can significantly improve the regional GDL.

5.6.2. Endogenous Test

NEVP can promote the GDL; at the same time, the improvement of the GDL makes the region provide support for NEVP in science and technology, industry, policy, market environments, and other aspects. Therefore, there may be a reverse endogenous relationship between NEVP and the GDL. In order to more accurately explore the impact of NEVP and the GDL, the instrument variable method was used for the endogenous test. Since there is no correlation issue between the residuals of lagged variables and current variables, therefore, the lag phase I (L.GDL) and lag phase II (L2.GDL) of the explained variable (GDL) are used as instrumental variables in this paper, and the two-stage least square method is used for regression analysis.

Table 20. Endogeneity test results.

Variables	Phase I: PLNEV		Phase II: GDL
PLNEV			0.042 ***	0.041 ***
			(0.004)	(0.005)
L.GDL	14.663 ***	13.827 ***
	(2.962)	(3.644)
L2.GDL	22.120 ***	16.794 ***
	(3.100)	(3.891)
Control variable	No	Yes	No	Yes
Hausman test	p< 0.01	p < 0.01
Unrecognizable inspection	41.33 ***	28.08 ***
Weak instrumental variable test	59.15 ***	61.02 ***
Over identification test	Pass	Pass
N	324	324	324	324

Notes: * p < 0.10, ** p < 0.05, and *** p < 0.01; values in parentheses denote the robust standard error for the coefficient.

shows the results of the endogeneity test. Firstly, the Hausman test rejected the original hypothesis at the 1% significance level, indicating that the model has endogenous problems. In the unrecognizable test, Kleibergen Paaprk LM statistic is significant at the 1% significance level, indicating the identifiability of instrumental variables. In addition, both the overidentification test and the weak instrumental variable test passed the test, indicating that the model does not have the problem of weak instrumental variables.

Comparing the endogeneity test results in Table 20 with the Tobit benchmark regression results in Table 11, we can find that the coefficients of the instrumental variables are significantly positive at the 1% level, and the direction of the coefficients of the core explanatory variables is consistent with the Tobit model, verifying the robustness of the model and indicating that selecting lagged variables as instrumental variables played a role in endogeneity testing.

6. Conclusions and Policy Suggestions

This paper considers the multiple mediating effects between NEVP, the upgrading of industrial structure, the level of the digital economy, the level of transportation infrastructure construction, and the GDL and uses provincial panel data to build a random effect panel Tobit model to empirically study the impact of NEVP on the GDL. The results show that NEVP significantly promotes the GDL in the region, and this significant impact is also transmitted through three indirect paths: industrial structure upgrading, the digital economy level, and transportation infrastructure construction. The regional heterogeneity analysis also verifies the positive promotion effect of NEVP on the GDL, but the transmission effect of the three paths on green development is different. The mediating effect of the digital economy level path in the eastern region is significant, while the mediating effect of the digital economy development level path in the central and western regions is not significant. NEVP in the eastern, central, and western regions can significantly promote the upgrading of regional industrial structure and the construction level of transportation infrastructure so as to enhance the local GDL. The policy recommendations of this paper are as follows:

(1)

Adhere to innovation-driven development, increase research and development support for core technologies of NEVs, promote green technology innovation, support data-driven innovation mode, optimize the efficiency of the transportation system, and reduce environmental pollution.

(2)

Optimize the industrial structure, promote the green transformation of traditional industries, support the transformation of the traditional automobile manufacturing industry to a high-tech industry, promote the green and low-carbon development of the industrial chain, and encourage the coordinated development of NEVs with related supporting industries and basic industries to form a green industrial cluster.

(3)

Improve the construction of transportation infrastructure, accelerate the construction and planning of roads and charging infrastructure, improve supporting facilities, promote the green transformation of the urban transportation system, and build an efficient and low-carbon green transportation system.

(4)

Maintain the NEVP level in the eastern region, improve the NEVP level in the central region, enhance the policy support for NEVs in the western region, and further promote regional green development to a greater extent.

This paper makes an attempt to study the impact of NEVP on the green development level of comprehensive indicators. However, the within-province heterogeneity is not taken into account because some open city-year data lack availability. Furthermore, the within-province heterogeneity will be researched if all open city-year data are provided. In the future, we can also consider taking public sentiment from social media data such as the public’s views and acceptance, consumers’ green consumption willingness, and subsidies for new energy vehicles as mediating variables to study the impact of NEVP on the GDL.