Differences in the effects of China’s new energy vehicle industry policies on market growth from the perspective of policy mix

Abstract

Regarding the characteristics of policy mix and its effectiveness, no complete consensus currently exists. This paper divides the new energy vehicle (NEV) industry policies issued by the Chinese government from 2009 to 2018 into two types: the supply-side and demand-side policy mix. From the three dimensions of comprehensiveness, consistency and balance, we applied the state space method to construct a time-varying parameter model and analyzed the differences in the overall and dynamic impact of the NEV policy mix characteristics on market growth. The results show that the three characteristics differ in their overall impact: comprehensiveness has the largest impact, followed by consistency, and balance has the smallest impact. The dynamic impact of the comprehensiveness and consistency characteristics of the two types of policy mix fluctuates greatly and presents clear differences in the stage effect. The dynamic impact of the balance characteristic is relatively stable, but this characteristic has little dynamic impact on market growth. The NEV industry policy design should fully consider the characteristics of the policy mix and the differences of this policy mix in terms of market development stage and region.

Keywords

New energy vehicle policy mix supply-side policy demand-side policy market growth

Introduction

New energy vehicles (NEVs) have the externalities of quasi-public products in terms of environmental protection and resource conservation. NEVs are currently facing immature commercial conditions, such as high prices, unstable performance and imperfect infrastructure facilities.^1,2 Government industrial policy support is a necessary condition for the growth and development of NEVs as quasi-public products and represents a common practice in countries around the world.³ or Since 2009, China has issued a series of supporting policies for the development of the NEV industry. With the increase in the number of policies, the policy mix system, which is composed of industrial policies with different subjects and goals, has become increasingly complicated.⁶ Different combinations of policy tools and their interactions will show different policy mix characteristics and will affect comprehensive effectiveness and the achievement of policy objectives.⁷

At this stage, does the model of China’s NEV industry policy mix effectively promote NEV market growth? What is the overall and dynamic effect of policy mix on market growth? What are the key points and directions for the optimization of China’s NEV industry policy mix model? To answer these questions, several aspects are worth considering. First, current studies generally believe that an effective policy mix model requires certain characteristics, such as comprehensiveness, consistency and balance,^8,9 and various characteristics within the policy mix system may differ in policy effects. Second, the policy effects of different policy mix tools, such as those on the supply side and on the demand side, may also differ.^10,11 Third, studies on the effectiveness of current policy mix mainly focus on public policies and evaluation in innovation or sustainable transitions,^12,13 and there are few studies on the characteristics of policy mix as well as the functional differences of these characteristics in the market growth of the NEV industry. Therefore, this paper intends to study the development of China’s NEV policy mix system and its impact on market growth through policy quantification, policy mix characteristics measurement and impact analysis, which includes both a theoretical exploration of current research limitations and realistic guidance for the development of China’s NEV industry.

The remainder of this paper is structured as follows. The next section is the literature review. The Research design section introduces the research design. The Empirical results and analysis section presents our empirical results and analyses of the differences in the overall and dynamic impacts of the NEV policy mix characteristics on market growth, and the Conclusions section concludes the paper with a discussion on research and policy implications.

Literature review

Policy mix

Policies usually need to be combined in a policy mix that uses different policy tools to solve problems for policy implementation.^14,15 The concept of policy mix first appeared in the economic policy literature in the 1960s, which was mainly used to analyse the interaction between monetary and fiscal policies. Later, the policy mix concept was gradually introduced into the study of public and innovation policy.¹⁶ Generally, a policy mix is a set of interacting policy tools.^17,18 With the increase in the number of various types of policies, policy systems have become increasingly more complicated, and the interaction between policies from different policy subjects with different policy objectives and measures is increasing.¹² Therefore, policies need to comprehensively use different policy tools in a policy mix to solve problems in the policy system.^19,20

There are many studies on the concepts, theories, effects and impacts of policy mix, and these studies can be approximately divided into two categories. The first category consists of theoretical and conceptual analyses. The OECD²¹ proposed a policy mix analysis framework that includes policy areas, principles, strategic tasks and tools. Flanagan et al.¹⁶ reconstructed the concept of a policy mix for innovation and emphasized the impact of a policy mix on innovation and the dynamics of policy interactions. On this basis, Rogge and Reichard²² proposed a more comprehensive concept of a policy mix, and pointed out that a policy mix should have the ‘4C’ characteristics: consistency, coherence, credibility and comprehensiveness. The second category consists of case analyses and empirical research that analyse the status of the policy mix and its impact. For example, Guerzoni and Raiter¹⁰ used the propensity score matching method to evaluate the policy mix application of 5,238 companies in 27 EU countries. Scordato et al.²³ used a case study to analyse the policy mix for the sustainability transition of the pulp and paper industry in Sweden.

Based on these studies, Reichardt and Rogge⁸ pointed out that the description of the characteristics of a policy mix is the premise of policy mix evaluation. Many scholars have noticed the important role of policy mix characteristics, such as comprehensiveness or consistency.^24,25 However, there is no consensus in either theoretical research or practical applications on the characteristics and effectiveness of policy mix. From the perspective of theoretical research, most scholars believe that the comprehensiveness, consistency and balance characteristics of the policy mix are important prerequisites for policy effectiveness. For example, Sovacool²⁶ believes that the comprehensiveness characteristic of the policy tool mix positively affects the implementation of renewable energy policies and energy efficiency. Costantini et al.⁹ found through empirical research that the comprehensiveness and internal balance of the policy mix positively affects technological innovation, while non-conformity and discreteness inhibit technological innovation. Uyarra et al.²⁷ noted that the complexity and inconsistency of the innovation policy mix hinder the efficiency of private sector investment and low-carbon innovation. However, some scholars have found that unbalances or differentials in the policy mix also play an important role in the specific market environment. For example, Ma et al.²⁸ believed that differentiated policy arrangements were necessary to promote the growth of the NEV industry and market. Xiong and Chen²⁹ noted that it is necessary to fully consider the unbalanced NEV industry policy and that the policy orientation of the supply side and the demand side should be adapted to local conditions with different emphases. Schmidt and Sewerin³⁰ proposed that the balance feature should be used as a new measure of policy mix characteristics. From the perspective of practical application, the policy of ‘letting some people get rich first’, which achieved great success in the early stage of China’s reform and opening up, is a typical unbalanced policy. The various types of development zones and test zones established at the present stage in China are essentially the specific practices of differential and unbalanced policy tools.

Clearly, in most cases, comprehensive, consistent and balanced policy mix models are considered to be ideal, but non-comprehensive, inconsistent or unbalanced policy mix models are not necessarily non-ideal in all cases. To accurately grasp the current characteristics of China’s NEV industry policy mix and its relationship with the growth of the NEV market, we collected NEV industry policy texts promulgated by the relevant departments of the Chinese central government from 2009–2018. According to the three dimensions of comprehensiveness, consistency and balance, this paper examines the model and characteristics of China’s NEV policy mix, and further analyses the differences in the impact of these characteristics on NEV market growth to provide a reference for further improving the precision of China’s NEV industry policy.

China’s NEV policy mix model and market growth

After reviewing the relevant policies on the development of China’s NEV industry, we find that since the introduction of ‘The Notice on the Pilot Work of Demonstration and Promotion of Energy-saving and NEVs’ policy in 2009, the Chinese government has adopted a variety of policy tools related to different aspects to form a policy mix model from the supply and demand sides to promote the development of the NEV industry.^31,32 Specifically, the supply-side policy mix mainly focuses on improving the quality and efficiency of the NEV consumer market through public resource investment and optimized allocation. The supply-side policy mix includes mainly measures such as the construction of NEV consumption and use infrastructure (charging piles), corporate financial support, market demonstration organizations, and standardized supervision of NEV consumption and transactions. The demand-side policy mix mainly focuses on stimulating the enthusiasm and purchasing power of NEV consumption by guiding end-consumers and providing a driving force for the maturity of commercialization conditions of the NEV consumption market. The demand-side policy mix includes mainly measures such as consumer direct purchase subsidies and corporate tax deductions, priority rights for NEV driving, government procurement and other measures.²⁹

Driven by strong policies, the scale of China’s NEV market has grown significantly over the past 10 years (Figure 1). As of the end of 2018, the global total sales of NEVs has exceeded 5.5 million, of which China has accounted for more than 53%. China has become the world’s largest NEV country and market. The large-scale growth of China’s NEV market is clearly inseparable from industrial policies.³³ However, due to the late start of the NEV market in China, the immature market mechanism and differences in consumer groups, some policies have gradually exposed certain problems and deficiencies during implementation.³⁴ With the further growth of China’s NEV market, related policies are currently undergoing adjustment, such as the withdrawal of consumer direct subsidies on the demand side.³⁵ Therefore, in this context, it is of great practical significance to study the characteristics and evolution rules of the two types of policy mix on the supply and demand sides of China’s NEV industry, so as to explore the dynamic relationship between the development of policy mix characteristics and market growth and determine how to maximize the effectiveness of a policy mix.

Figure 1.

Growth of China’s NEV market (Data source: China Automobile Industry Association).

Research design

Research samples and data sources

Research samples. Policy tools are the foundation of a policy mix.³⁶ This paper collected policies from the official website of the Chinese National Government Ministry, the ‘Energy Conservation and New Energy Vehicle Yearbook (2009-2018)’, the ‘China New Energy Automobile Industry Development Report (2009-2018)’ and other materials. The 113 NEV industry policy texts issued in 2009-2018 were used as research samples and classified into supply-side and demand-side policy mix. The supply-side policy mix mainly includes four types of specific policy tools: infrastructure construction, financial support, demonstration organizations and regulatory improvement. The demand-side policy mix mainly includes four types of specific policy tools: purchase subsidies, tax deductions, driving rights and government procurement.^29,37

Data sources. To maximize the effectiveness of government policies and promote market scale growth, in 2009, China established the first batch of 25 cities as pilot cities for NEVs, and the number of pilot cities has subsequently increased to 88.³⁸ Considering that China’s first batch of 25 NEV pilot promotion cities has experienced the whole process of China’s NEV industry policy implementation, the data for these cities are complete and have typical representative significance. Therefore, in this paper, the first batch of 25 NEV pilot promotion cities were selected as data sources. The specific research data collecting from 2009 to 2018, come mainly from the ‘Energy Conservation and New Energy Vehicle Yearbook’, ‘China Statistical Yearbook’, ‘China Economic and Social Development Statistics’ website and ‘incoPat patent database’. The specific policy statistics and market growth are shown in Figure 2.

Figure 2.

The quantitative statistics on the NEV supply-side and demand-side policies and market growth (the first batch of 25 pilot promotion cities).

Quantification and measurement of policies and policy mix characteristics

Policy quantification and statistics

Policy quantification refers to the quantification of the content of policy tool texts and their influence. In current research, Libecap³⁹ first used the Law Change Index to quantify the texts of various legal policies related to mineral property rights. Cools et al.⁴⁰ quantified the transport policy measures (such as traffic commuting, energy taxation and parking fees) from the four perspectives of ‘hard’, ‘soft’, ‘push’ and ‘pull’. According to the specific policy environment in China, Peng et al.⁴¹ proposed a specific quantitative process to refine and score policies in terms of three dimensions: policy power, policy measures and policy goals. This process also provided an effective blueprint for the subsequent quantitative and statistical research of policy tool texts. Zhao et al.⁴² studied the selection of the best travel mode of the last mile by using multi-attribute data mixed decision-making. The multiple weighting and grade assignment methods used by the authors also provide the basis for quantifying and calculating relevant policy texts.

Referring to current research results and specific methods, this paper also quantifies the NEV policy texts in terms of the three dimensions. The policy power reflects the size of the legal rights of the policy text, which is generally determined by the level of the policy release department. We refer to the Procedure Ordinance of Rules and Regulations Formulation by the State Council in China and the specific methods of policy quantification, and then assign the policy power 1–5 points. The policy measures refer to the methods and means used by the government to achieve the stated purpose when formulating and implementing policies. The policy goals are the objectives to be achieved by a policy. According to the degree of detail of the measures and objectives embodied in the policy text, the policy measures and the policy goals are assigned 1–5 points.⁴¹ The specific method is shown in Table 1.

Table 1.

Quantitative dimensions and standards of policy.

Quantitative dimensions	Assignment	Quantitative standards
Policy power	5	Legal documents promulgated by the National People’s Congress and its Standing Committee.
	4	Regulations and directives promulgated by the State Council; orders of various ministries.
	3	Interim regulations and regulations, plans, decisions, opinions, methods and standards promulgated by the State Council; regulations and decisions promulgated by various ministries.
	2	Opinions, methods, plans, guidelines, interim regulations, rules, conditions and standards issued by various ministries.
	1	Notice, announcement and planning.
Policy measures	5	Detailed policy implementation measures, related assessments, as well as inspection and supervision methods have been clearly formulated. Specific implementation plans have been issued. The objectives are clear; the routes are clear; the regulations are detailed; the evaluation is complete.
	3	The route and requirements of various policy measures, as well as assessments and proposed supervision and inspection methods have been implemented, but specific implementation measures or rules have not been formulated.
	1	Only some of the evaluation criteria for 5 and 3 points are mentioned. The policy measures are unclear and can only be used as macro guidance.
Policy goals	5	Quantitative standards for policy objectives are clearly formulated, and the goals and effects to be achieved in terms of market size growth, corporate innovation output, consumer purchase and use are clearly specified.
	3	Relevant policy goals are developed but lack corresponding inspection standards.
	1	Only some of the evaluation criteria for 5 and 3 points are mentioned. The policy goals are unclear and can only be used as a macro vision.

Note: The quantitative criteria of 4 points and 2 points in the evaluation dimension of the policy measures and policy goals are between 5 points and 3 points and between 3 points and 1 point, respectively.

Generally, the policies issued by higher-level institutions have greater legal effects but are more macroscopic. Hence, the scores of the policy power are higher, but the scores of the policy measures and policy goals are relatively lower. In contrast, the opposite is true for policies promulgated by lower-level institutions. Therefore, this study comprehensively evaluates and analyses policies according to the three dimensions of policy power, policy measures and policy goals, which can compensate for the shortcomings of single indicators in reflecting the effectiveness of policy content and improve the content validity of the policy texts.

According to the evaluation criteria and scoring criteria of the policies in Table 1, to ensure the feasibility and accuracy of policy quantification, relevant experts were hired to score policies in multiple rounds and by multiple groups. The average scoring results of the rounds were selected as the final evaluation scores for policy effectiveness. Therefore, the composite scores for each policy tool issued each year can be expressed as follows:

{T S}_{i} = \sum_{j = 1}^{N} {p p}_{i, j} \times {p m}_{i, j} \times {p g}_{i, j} i \in [2009, 2018]

(1)

where i is the year; N is the total number of policies promulgated in year i; j is the j-th policy promulgated in year i; pp_i,j indicates the policy power score of the j-th policy; pm_i,j indicates the policy measures score of the j-th policy; pg_i,j indicates the policy goals score of the j-th policy; and TS_i is the composite score of the various policy tools promulgated in year i.

Considering the adjustment, abolition or overlap of some NEV industry policies, we adjusted these policies. The effective cumulative scores of NEV-related policies for each year are NTS_i:

{NTS}_{i} = {NTS}_{i - 1} + {T S}_{i} - \sum_{j = 1}^{N^{*}} {p p}_{i, j} \times {p m}_{i, j} \times {p g}_{i, j} i \in [2009, 2018]

(2)

where N* represents the total number of policies that were abolished in year i or overlapped with those issued in that year.

Analysis dimensions and method for evaluating policy mix characteristics

Regarding the analysis dimensions and methods used to evaluate policy mix characteristics, Reichardt and Rogge⁸ used a case study to determine that policy mix characteristics analysis can be divided into comprehensiveness, consistency, credibility and stability dimensions. For researching quantitative policy mix, Costantini et al.⁹ creatively proposed a method that streamlines the number and concept of policy mix characteristics. Xu and Li¹¹ further modified the quantitative system of policy mix characteristics to consist of the three characteristic dimension: comprehensiveness, consistency and balance. Guo et al.⁴³ used this method to verify the impact of policy mix characteristics on the formation of China’s leading market for NEVs. The research results indicated that there is no consensus regarding the quantitative evaluation system of the policy mix characteristics, and research on the policy mix characteristics of NEV policies is still in the initial stage. Therefore, using the research results of the above scholars, this paper evaluates and analyses the mix characteristics of China’s NEV industry policy tools in terms of the three characteristic dimensions of comprehensiveness, consistency and balance. Such an evaluation and analysis has important exploration significance.

Comprehensiveness refers to the breadth and detail of the policy tools in a policy mix. The comprehensiveness of the policy mix can be indicated by eigenvalues (POL_Compre), which are determined by calculating the effective cumulative scores for each of the different policies:

{POL}_{Compr e_{i}} = \sum_{k = 1}^{r} {NTS}_{i}^{k} i \in [2009, 2018]

(3)

where k is the policy tool and r is the total number of tools in a certain type of policy. The number of tools in both the supply-side and demand-side policy mix in this study is 4. The parameter i is the year, and NTS_i represents the effective cumulative score of the NEV-related policies for each year.

Consistency refers to the degree of coordination between policy tools. The consistency of the policy mix is assessed through the interaction between the policy tools. This paper calculates the average vector angle cosine of the policy text according to the algorithm of the vector cosine value and then calculates the consistency characteristic value (POL_Consis):

\begin{array}{l} {POL}_{Consi s_{i}} = (\sum_{k = 1}^{r} {NTS}_{i}^{k}) \times \frac{\sum_{m = 1}^{N} \sum_{n = 1}^{N} \cos (X^{m}, X^{n})}{N \times (N - 1) / 2} i \in [2009, 2018], \\ C O S (X^{m}, X^{n}) = \frac{\sum_{k = 1}^{r} (x^{m k} \times x^{n k})}{\sqrt{\sum_{k = 1}^{r} {(x^{m k})}^{2}} \times \sqrt{\sum_{k = 1}^{r} {(x^{n k})}^{2}}} \end{array}

(4)

where m and n represent any two of the N policies promulgated in the i-th year and m ≠ n. The vector X^m = (x^m¹,x^m2…,x ^mr) is established according to the tool scores in each type of policy, and cos(X^m,Xⁿ) is the vector cosine of the two policies. The higher the mean value of the cosine of the angle between the policies is, the higher the similarity between the same types of policies promulgated in the i-th year is and the greater the overall consistency is.

Balance refers to the strength and development trend of different policies within the policy tool mix. The balance of the policy mix is first calculated by calculating the correlation index Rel between two different policy tools:

Re l = {[\frac{| {NTS}_{s} - {NTS}_{t} |}{\sqrt{{NTS}_{s} + {NTS}_{t}}}]}^{- 1}

(5)

where s and t are different tools in a certain type of policy and s ≠ dic. Rel represents the correlation between the two tools. Based on formula (5), the standard deviation of the correlation index is used to measure the balance characteristic value (POL_Bal).

Variable selection and model setting

Variable selection

In this paper, the number of NEV promotions and applications (market sales) in the first batch of 25 NEV pilot promotion cities is a dependent variable. Considering the lagging effect of policy mix and other control variables on market sales, this model lags the market sales variable (Y) by one year. Based on the research of Rogge and Schleich,⁷ Costantini et al.,⁹ and Xu and Li,¹¹ the independent variables consist of the comprehensiveness (POL_Compre), consistency (POL_Consis), and balance (POL_Bal) of the policy mix. Based on the research of Ma et al.²⁸ and Tan et al.,⁴⁴ the NEV market, region and other related indicators (including industry technology (IT), infrastructure level (IL), market consumption (MC) and traffic carrying capacity (TCC) variables) are introduced into the model as control variables. Related variables, data sources and algorithms are shown in Table 2.

Table 2.

Related variables and explanations.

Variable types	Variable name	Variable definition	Main data sources and algorithms
Dependent variable	Y	Number of pilot applications in the pilot promotion cities (market sales)	Energy Saving and New Energy Vehicle Yearbook
Independent variables	POL_Compre	The comprehensiveness characteristic of the policy mix	Energy Saving and New Energy Vehicle Yearbook, formula (3)
	POL_Consis	The consistency characteristic of the policy mix	Energy Saving and New Energy Vehicle Yearbook, formula (4)
	POL_Bal	The balance characteristic of the policy mix	Energy Saving and New Energy Vehicle Yearbook, formula (5)
Control variables	IT	Total number of industry patent applications	The incoPat patent database
	IL	Total number of public charging piles built in the city	Energy Saving and New Energy Vehicle Yearbook
	MC	Per capita disposable income of urban residents	China Economic and Social Development Statistics Database
	TCC	Urban road traffic density	China Economic and Social Development Statistics Database

Model setting

To study the dynamic impact of policy mix characteristics on market growth, a variable parameter model needs to be selected for analysis. Based on the selection of existing variables and the nature of the dependent variable, this paper constructs the basic measurement model as follows:

\begin{array}{l} {LnY}_{i + 1} = α_{1} {POL}_{Compr e_{i}} + α_{2} {POL}_{Consi s_{i}} \\ + α_{3} {POL}_{B a l_{i}} + β_{1} {I T}_{i} + β_{2} {I L}_{i} + β_{3} {M C}_{i} + β_{4} {TCC}_{i} + μ_{i} \end{array}

(6)

In this model, i is the year, LnY_i+1 is the logarithm of the dependent variable (market sales), and the other variables are defined in Table 2. Then, equation (6) is further modified, and the state space method⁴⁵ is used to construct the time-varying parameter model, which consists of the measurement equation and state equation:

\begin{array}{l} {LnY}_{i + 1} = c + α_{1 i} {POL}_{Compr e_{i}} + α_{2 i} {POL}_{Consi s_{i}} \\ + α_{3 i} {POL}_{B a l_{i}} + β_{1 i} {I T}_{i} + β_{2 i} {I L}_{i} + β_{3 i} {M C}_{i} + β_{4 i} {TCC}_{i} + μ_{i} \end{array}

(7)

{\begin{cases} α_{1 i} = φ_{1} α_{1 (i - 1)} + η_{1 i} \\ α_{2 i} = φ_{2} α_{2 (i - 1)} + η_{2 i} \\ α_{3 i} = φ_{3} α_{3 (i - 1)} + η_{3 i} \\ β_{1 i} = φ_{4} β_{1 (i - 1)} + η_{4 i} \\ β_{2 i} = φ_{5} β_{2 (i - 1)} + η_{5 i} \\ β_{3 i} = φ_{6} β_{3 (i - 1)} + η_{6 i} \\ β_{4 i} = φ_{7} β_{4 (i - 1)} + η_{7 i} \end{cases}

(8)

where formula (7) is the measurement equation. This equation indicates the relationship between the following: the three types of characteristic dimensions in the two policy mix models, some factors, such as IT, IL, MC and TCC; and the growth level of the NEV market. The parameters α₁–α₃ and β₁–β₄ represent the elastic values of the growth level of the NEV market for each independent and control variable at each time point, which are called state variables. The term μ_i is the disturbance term, which follows a normal distribution with a mean of 0 and a constant variance. Equation (8) is the state equation, which describes the generation of the state variables α₁-α₃ and β₁-β₄ in the time dimension. We can use the Kalman filtering algorithm to determine the estimated values of the state variables α₁-α₃ and β₁-β₄ and obtain the dynamic impact of each variable on the growth of the NEV market and their differences.⁴⁵

Empirical results and analysis

Model testing

To avoid false regression (or pseudo-regression) caused by non-stationary time series data, the unit root test and co-integration relationship test are first conducted.

(1) Unit root test. The augmented Dickey-Fuller (ADF) unit root test method was used to test the stationary of each time series. The constant term and the trend term were evaluated according to the sequence trend graph, and the lag order was evaluated according to the Schwarz information criterion. The test results are shown in Table 3.

Table 3.

Unit root test results.

Variables	The supply-side policy mix				T he demand-side policy mix
Variables	Inspection types (c,t,k)	ADF test value	P value	Test result	Inspection types (c,t,k)	ADF test value	P value	Test result
Y	(c,t,2)	–1.7938	0.6173	Non-stationary	(c,t,2)	–1.7938	0.6173	Non-stationary
POL_Compre	(c,t,2)	–2.2605	0.4041	Non-stationary	(0,0,2)	–2.7176	0.0144**	Stationary
POL_Consis	(c,t,0)	–1.0530	0.8755	Non-stationary	(c,2,2)	–4.1624	0.0657*	Stationary
POL_Bal	(c,t,1)	–2.8781	0.2251	Non-stationary	(c,0,0)	–2.3870	0.1695	Non-stationary
IT	(c,t,1)	–0.3761	0.9574	Non-stationary	(c,t,1)	–0.3761	0.9574	Non-stationary
TL	(0,0,1)	–2.8400	0.0109**	Stationary	(0,0,1)	–2.8400	0.0109**	Stationary
MC	(c,0,0)	–1.0726	0.6753	Non-stationary	(c,0,0)	–1.0726	0.6753	Non-stationary
TCC	(0,0,1)	1.0147	0.8999	Non-stationary	(0,0,1)	1.0147	0.8999	Non-stationary
D(Y)	(0,0,0)	–1.1697	0.2008	Non-stationary	(0,0,0)	–1.1697	0.2008	Non-stationary
D(POL_Compre)	(c,0,1)	–2.0574	0.2616	Non-stationary	(0,0,2)	–2.4456	0.0285**	Stationary
D(POL_Consis)	(c,0,2)	–1.4655	0.4809	Non-stationary	(c,t,2)	–4.8563	0.0485**	Stationary
D(POL_Bal)	(0,0,0)	–2.6881	0.0142**	Stationary	(0,0,1)	–2.8337	0.0119**	Stationary
D(IT)	(c,t,2)	–2.5251	0.3226	Non-stationary	(c,t,2)	–2.5251	0.3226	Non-stationary
D(IL)	(0,0,1)	–2.3864	0.0256**	Stationary	(0,0,1)	–2.3864	0.0256**	Stationary
D(MC)	(c,0,2)	–2.1155	0.2435	Non-stationary	(c,0,2)	–2.1155	0.2435	Non-stationary
D(TCC)	(c,0,0)	–2.3531	0.1796	Non-stationary	(c,0,0)	–2.3531	0.1796	Non-stationary
D(Y,2)	(0,0,1)	–2.4297	0.0251**	Stationary	(0,0,1)	–2.4297	0.0251**	Stationary
D(POL_Compre,2)	(0,0,0)	–1.8330	0.0671*	Stationary	(c,t,1)	–7.9588	0.0062***	Stationary
D(POL_Consis,2)	(c,0,1)	–6.1562	0.0181**	Stationary	(0,0,2)	–2.9057	0.0133**	Stationary
D(POL_Bal,2)	(0,0,1)	–3.2283	0.0071***	Stationary	(0,0,2)	–3.8589	0.0029***	Stationary
D(IT,2)	(0,0,0)	–2.2655	0.0319**	Stationary	(0,0,0)	–2.2655	0.0319**	Stationary
D(IL,2)	(0,0,2)	–1.9283	0.0597*	Stationary	(0,0,2)	–1.9283	0.0597*	Stationary
D(MC,2)	(0,0,1)	–3.6254	0.0041***	Stationary	(0,0,1)	–3.6254	0.0041***	Stationary
D(TCC,2)	(0,0,0)	–3.8668	0.0023***	Stationary	(0,0,0)	–3.8668	0.0023***	Stationary

Note: *, **, *** indicate that the null hypothesis is rejected at the levels of 10%, 5%, and 1%, respectively. D represents the first-order difference, and D(2) represents the second-order difference. (c,t,k) represents the test types for whether the unit root test contains the constant term, time trend term and lag order, respectively.

The test results show that the original sequence and the first-order difference sequence of the selected variables are mostly non-stationary. However, in the second-order difference sequence, the selected variables reject the null hypothesis at the 10% significance level; that is, the variables are all second-order single sequences. To avoid pseudo-regression, it is necessary to further perform a co-integration test on the sequence of variables.

(2) Co-integration test. The Engle-Granger (E-G) two-step method was used to co-integrate the variable sequence. First, the ordinary least squares (OLS) estimation is performed on each time series in the model. Then, whether the residual sequence is stationary is used to determine whether there is a co-integration relationship between the explanatory variables and the explained variable. The co-integration test results show that the R² values of the two policy mix models are 0.9957 and 0.9977, indicating that the fitness between the variables is higher and that all the P values of the estimated parameter t-test are less than 0.1. Furthermore, the residuals were subjected to the ADF test. The results are shown in Table 4. The residual sequences of the two policy mix models rejected the null hypothesis at the 1% significance level; that is, the residual sequence has no unit root and is a stationary sequence. Therefore, the variables in the two types of policy mix models have a long-term equilibrium relationship, and there is no pseudo-regression, thus indicating that the variables selected in this paper meet the requirements of the time-varying parameter model.

Table 4.

Unit root test results of the residual sequence.

Variables	Inspection types (c,t,k)	ADF test value	10% significance level	5% significance level	1% significance level	P value	Test result
The supply-side policy mix	(0,0,0)	–7.9613	–1.6001	–1.9882	–2.8473	0.0000	Stationary
The demand-side policy mix	(c,0,0)	–10.6731	–2.7712	–3.2598	–4.4206	0.0000	Stationary

Analysis of the relationship between policy mix characteristics and market growth

Substituting relevant data into formulas (7) and (8), the Kalman filtering algorithm was used to obtain the overall impact regression results of the variable parameters in the model and the dynamic impact of the supply-side and demand-side policy mix characteristics on market growth. The specific results are shown in Table 5 and Figure 3.

Table 5.

The variable parameters’ overall impact regression results.

	The supply-side policy mix				The demand-side policy mix
Parameter	Coefficient	Std. Error	Z-Statistic	Prob．	Coefficient	Std. Error	Z-Statistic	Prob．
c (1)	0.0570	0.0333	1.7115	0.0870	0.0518	0.0115	4.4943	0.0000
c (2)	–6.3751	1.6823	–3.7894	0.0002	–7.3950	2.7276	–2.7111	0.0067
	Final State	Root MSE	Z-Statistic	Prob．	Final State	Root MSE	Z-Statistic	Prob．
α_1i	0.7891	0.1304	6.0503	0.0000	0.2974	0.0575	5.1747	0.0000
α_2i	0.5107	0.1542	3.3128	0.0009	–0.1808	0.0673	–2.6875	0.0072
α_3i	0.0519	0.0289	1..8001	0.0719	–0.0763	0.0119	–6.4137	0.0000
β_1i	0.1976	0.0958	2.0625	0.0392	–0.4197	0.0530	–7.9159	0.0000
β_2i	0.4651	0.1134	4.1015	0.0000	0.3450	0.0458	7.5346	0.0000
β_3i	0.8865	0.2954	3.0013	0.0027	0.0188	0.0214	0.8795	0.3792
β_4i	–0.8550	0.2610	–3.2757	0.0011	0.9674	0.0458	21.1210	0.0000

Figure 3.

The dynamic impact of the policy mix characteristics.

The descriptive analysis of the overall impact

The three characteristics in the supply-side policy mix have positive effects on the growth of the NEV market, but there are differences in the intensity of the impact. The data in Table 5 show that the coefficients of the comprehensiveness, consistency and balance characteristics in the supply-side policy mix are 0.7891 (P = 0.0000), 0.5107 (P = 0.0009) and 0.0519 (P = 0.0719). The results show that these three types of mix characteristics positively affect the growth of the NEV market. However, the comprehensiveness of the policy mix has the strongest impact, consistency has the second strongest impact, and balance has the smallest impact. Thus, concerning the comprehensiveness characteristic of the supply-side policy mix, in promoting the NEV market, the mix of various policy tools on the supply side can effectively promote the development of the NEV market and create a large-scale market growth effect. Infrastructure and regulation improvement tools are the most widely used policy tools in the supply-side policy mix. Among the 113 policy texts, 68 policies involve using these two types of tools, which provide a strong foundational guarantee and normative role for the development of the NEV market. Concerning the consistency characteristic, the current supply-side policy tools are well coordinated, especially since 2015, and the number of supply-side policy tools (Figure 2) has grown explosively. The scale and synergy effect of the supply-side policy mix began to appear, which played a powerful role in promoting NEV market growth. Concerning the balance characteristic, regulatory improvement and infrastructure tools are the main tools, supplemented by financial support and demonstration organization tools, so the use of policy tools in the supply-side policy mix is relatively balanced. Although the regression result of the balance characteristic coefficient is positive, the promotion effect on market growth is unclear.

The three characteristics in the demand-side policy mix affect NEV market growth differently, the comprehensiveness characteristic has a positive effect, but the consistency and balance characteristics have negative effects. The data in Table 5 show that the coefficients of the comprehensiveness, consistency and balance characteristics of the demand-side policy mix are 0.2974 (P = 0.0000), −0.1808 (P = 0.0072) and −0.0763 (P = 0.0000). Thus, the comprehensiveness characteristic of the demand-side policy mix can also be important in driving market growth. The overall number of demand-side policy tools is small (Figure 2), and purchase subsidies and tax reduction tools, which represent 33 items, are the most common. Purchase subsidies and tax reduction tools benefit mainly from direct price reductions for consumers, exemption of the vehicle purchase tax and vehicle and vessel tax, and exemption of the NEV key material consumption tax, thereby allowing NEV products to be bought at a preferential price to promote market growth. In terms of driving rights and government procurement tools, some cities with greater traffic pressure (such as Beijing, Shanghai, Guangzhou and Shenzhen) have implemented the priority rights of NEVs in terms of license plates and driving. In addition, government departments have stipulated a proportion of the NEVs for purchase and use. Therefore, the demand-side policy mix in these aspects has played a positive role in promoting the use of NEVs. The regression results of the consistency characteristic are significantly negative (thus indicating a negative impact on market growth) possibly because the demand-side policy mix is mainly based on the purchase subsidies policy. However, the purchase subsidies policy ranges from full subsidy to subsidy withdrawal and is about to enter the stage of complete subsidy withdrawal. The instability of the purchase subsidies policy has made it difficult for this policy to play a synergistic role in policy development, thereby having a negative impact on market growth to a certain extent. The regression result of the balance characteristic is significantly negative, but this characteristic less negatively affects market growth. This result indicates that in the demand-side policy mix, the mix of the purchase subsidies and tax reduction tools, supplemented by driving rights and government procurement tools, plays an unclear role in the balance characteristic.

The dynamic analysis of the staged impact

Concerning the changes in the elastic coefficients of the explanatory variables in Figure 3, the comprehensiveness and consistency characteristics in the supply-side and demand-side policy mix have greater stage volatility for the impact on market growth, while the dynamic impact of the balance characteristic is unclear in the stage of market growth.

The first stage of the growth of the NEV market (2009–2011): the three characteristics of the supply-side and demand-side policy mix little affect market growth possibly mainly because China’s NEV market is still in its infancy at this stage. The purpose of the policy mix is to initially establish a market, and the number of policies on both the supply and demand sides are small (Figure 2). The policy itself is relatively weak and focuses on the macro-guidance of the industry. The influence of the three characteristics of the supply-side and demand-side policy mix has not been fully demonstrated. Therefore, market growth is also relatively slow.

The second stage of the growth of the NEV market (2012–2014): the positive impact of the comprehensiveness characteristic in the demand-side policy mix is rising rapidly, and the impact in the supply-side policy mix is at a secondary level. The impact of the consistency characteristics of the two types of policy mix are relatively low, even negative. As shown in Figure 1, from the perspective of market growth, China’s NEV market has gradually entered the growth stage, and the market scale has begun to accelerate. The main reason for this acceleration may be that at this stage, the Chinese government has significantly increased the implementation of the demand-side policy mix represented by purchase subsidies and tax reduction tools, which quickly exerted a significant effect on stimulating the NEV consumer market in a short time. However, because this stage overemphasizes the role of direct subsidies in the demand-side policy mix and uses fewer supply-side policy tools, the impact of the comprehensiveness characteristic in the supply-side policy mix is not high. Moreover, there is a lack of linkage between the policy tools within the supply-side and demand-side policy mix at this stage, so the consistency of the two types of policy mix are relatively low, even negative.

The third stage of the growth of the NEV market (2015–2018): the positive impact of the comprehensiveness and consistency characteristics in the supply-side policy mix has reached the highest level. The positive impact of the comprehensiveness characteristic in the demand-side policy mix remains high, but the negative impact of the consistency characteristic still exists. The main reason may be that the incentive effect of the demand-side policy mix based on the purchase subsidies tool is difficult to sustain, and the implementation focus of China’s NEV policies has begun to enter the adjustment period. This phase focuses on both increasing policy support for infrastructure construction from the supply-side policy mix and further improving the regulatory and supervision system. In addition, the ‘dual-credit policy (the Parallel Administrative Measures for Passenger Vehicle Corporate Average Fuel Consumption and New Energy Vehicle Credits)’ was implemented, so the effectiveness of the comprehensiveness and consistency characteristics in the supply-side policy mix grows rapidly at this stage, and the promotion effect on the growth of market size are significantly enhanced. However, at this stage, due to the gradual withdrawal of subsidies from the demand-side policy mix, other policy tools, such as tax reductions and driving rights, still cannot compensate for the gap left by the subsidy withdrawal. Therefore, the negative impact of the consistency characteristic on market growth still exists.

The dynamic impact of the balance characteristic of the two types of policy mix is relatively stable, and the effect is still small. Thus, in the development of China’s NEV market, the balance characteristic brought by the implementation of the supply-side and demand-side policy mix does not significantly affect NEV market growth. In future NEV policy design, while maintaining the comprehensiveness and consistency of the policy mix, we can more prominently consider the unbalance of the policy mix and effectively promote the positive role of policy differentiation.

Conclusions

The exploratory and positive contributions of this paper has three main aspects. First, this paper attempts to introduce an evaluation of the effectiveness of the policy mix into the environment of NEV market growth. We analyse both the overall role of the three types of characteristics in market growth and the dynamic impact in three stages. This investigation can compensate for a deficiency of current policy mix studies, which have paid insufficient attention to market growth and have mainly focused on the overall impact.⁴³ Thus, this investigation is not only an effective exploration of the performance evaluation of the policy mix but also a breakthrough in the study of the factors affecting NEV market growth. Second, we expanded the current research conclusions.^9,11 Based on the distinction between supply-side and demand-side policies, the results of this paper confirm that the unbalance or differentiation of the policy mix can also play an important role in the specific market environment and development stages. This finding is of great significance for further improving the accuracy of China’s NEV industry policy. Finally, there are some quantitative studies on the characteristics of policy mix and their impact, but empirical research is still in the exploratory stage.^7,22 Concerning the three dimensions of comprehensiveness, consistency and balance, this paper applies the state space method to construct a time-varying parameter model, and the overall and dynamic impacts of the NEV policy mix characteristics on market growth are analyzed. We further explore the research of policy mix performance evaluation empirically and also provide an effective reference for market development and policy implementation of NEVs in China and other countries. The main conclusions and policy implications of this study are as follows:

The comprehensiveness, consistency and balance characteristics of the supply-side and demand-side policy mix differ in overall impact: comprehensiveness has the largest impact, followed by consistency, and balance has the smallest impact. Therefore, to further improve the accuracy of China’s NEV industry policies, on the one hand, we must further promote the diversification of the policy mix and maximize the positive impact of the comprehensiveness characteristic of the supply-side and demand-side policy mix to provide continuous policy support for NEV market growth. On the other hand, we must pay attention to the coordination and interaction between policies, improve the consistency of the policy mix, and optimize the combination of policy tools on both the supply and demand sides to avoid market barriers caused by policy gaps.

The dynamic effects of the comprehensiveness and consistency characteristics in the supply-side and demand-side policy mix are highly volatile and show significant differences across stages. The dynamic impact of the balance characteristic in the two types of policy mix is relatively stable and has little effect on market growth. Therefore, while maintaining the effective comprehensiveness and consistency of the policy mix, China’s NEV industry policies should also highlight the unbalance of the policy mix and effectively promote the positive role of differentiation. On the one hand, we must pay attention to the differences of the policy mix across stages. At the present stage, the NEV policy mix is characterized mainly by the supply-side policy mix, with the demand-side policy mix as the auxiliary mix. This configuration of the NEV policy mix leads to problems, such as the small effect of the balance characteristic in the two types of policy mix and the negative effect of the consistency characteristic in the demand-side policy mix. In future policy design, we must first strengthen the infrastructure construction and core technology guarantee for the development of China’s NEV market through the supply-side policy mix and focus on grasping the major market opportunities facilitated by the implementation of the dual-credit policy. These methods can consolidate the dominant position of leading manufacturers through marketization, achieve the survival of the strong in the industry, and finally achieve the long-term healthy development of the whole NEV industry. In addition, the content will be enriched and the new demand-side policy mix will be formed. In the context of the gradual withdrawal of the direct subsidy tools, we can fully liberalize the NEV unique rights in terms of driving priority and supporting policies, such as parking fees, charging prices and road tolls. Additionally, explore a new combination of tools, such as tax incentives and government procurement will also be helpful.

On the other hand, we must pay attention to the regional differences in the policy mix. There are significant differences in the endowment factors for NEV market development in the different regions of China, such as across cities that limit and cities that do not limit the number of licenses. To further improve the usefulness of NEV policies and narrow regional market differences, it is necessary to seek a true connection between the policy mix and regional market growth. For license-limiting cities, we can continue to maintain special policies, such as free special licenses and driving rights, to maintain the steady growth of the NEV market. For non-license-limiting cities, it is necessary to accelerate the development of basic industries, such as power batteries and charging piles, and make public demonstrations and promotions to attract the active participation of potential consumers. Overall, the policy mix should be tailored to local conditions to further improve the matching of policy tools and regional market growth and to improve the execution of the policy mix.

There are still some limitations in this paper. Due to the short development of China’s NEV market, the available policies and related data are limited, and there are other factors affecting China’s NEV market growth. In the future, with the continuous development of the market and policies, and the increase in the number of years that data are available, the relationship between the policy mix and market growth can be further explored.

Footnotes

Declaration of conflicting interests

The authors declare no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was funded by the National Natural Science Foundation of China (No. 71874208).

ORCID iD

Shufeng Qin

Yongqing Xiong holds a PhD in Management and is working as a professor in School of Business at Central South University, China. His Research focuses broadly on the emerging industries and innovation strategies.

Shufeng Qin is a PhD candidate in School of Business at Central South University, China. His research focuses on the emerging industries and policy evaluation.

References

Yuan

Liu

Zuo

The development of new energy vehicles for a sustainable future: a review. Renewable Sustainable Energy Rev 2015; 42: 298–305.

Liu

Hao

Cheng

, et al. Critical issues of energy efficient and new energy vehicles development in China. Energy Policy 2018; 115: 92–97.

Sierzchula

Bakker

Maat

, et al. The influence of financial incentives and other socio-economic factors on electric vehicle adoption. Energy Policy 2014; 68: 183–194.

Langbroek

JHM

Franklin

Susilo

YO.

The effect of policy incentives on electric vehicle adoption. Energy Policy 2016; 94: 94–103.

Lim

Kim

Lee

Forecasting the future scale of vehicle to grid technology for electric vehicles and its economic value as future electric energy source: the case of South Korea. Energy Environ. Epub ahead of print 7 January 2020. DOI: 10.1177/0958305X19898283.

From government to market and from producer to consumer: transition of policy mix towards clean mobility in China. Energy Policy 2016; 96: 328–340.

Rogge

Schleich

Do policy mix characteristics matter for low-carbon innovation? A survey-based exploration for renewable power generation technologies in Germany. Res Policy 2018; 47: 1639–1654.

Reichardt

Rogge

How the policy mix impacts innovation: findings from company case studies on offshore wind in Germany. Environ Innov Soc Trans 2016; 18: 62–81.

Costantini

Crespi

Palma

Characterizing the policy mix and its impact on eco-innovation: a patent analysis of energy-efficient technologies. Res Policy 2017; 46: 799–819.

10.

Guerzoni

Raiteri

Demand-side vs. supply-side technology policies: hidden treatment and new empirical evidence on the policy mix. Res Policy 2015; 44: 726–747.

11.

CY.

S&T policy mix characteristics and their impact on industrial innovation in China. Stud Sci Sci 2017; 35: 45–53.

12.

Kern

Rogge

Howlett

Policy mixes for sustainability transitions: new approaches and insights through bridging innovation and policy studies. Res Policy 2019; 48: 103832.

13.

Falcone

Lopolito

Sica

Instrument mix for energy transition: a method for policy formulation. Technol Forecast Soc Change 2019; 148: 119706.

14.

Howlett

Del

RP.

The parameters of policy portfolios: verticality and horizontality in design spaces and their consequences for policy mix formulation. Environ Plann C Gov Policy 2015; 33: 1233–1245.

15.

Wang

Zou

Zheng

, et al. How will different types of industry policies and their mixes affect the innovation performance of wind power enterprises? Based on dual perspectives of regional innovation environment and enterprise ownership. J Environ Manage 2019; 251: 109586.

16.

Flanagan

Uyarra

Laranja

Reconceptualising the ‘policy mix’ for innovation. Res Policy 2011; 40: 702–713.

17.

Borr016

Edquist

The choice of innovation policy instruments. Technol Forecast Soc Change 2013; 80: 1513–1522.

18.

Schader

Lampkin

Mulle

, et al. The role of multi-target policy instruments in Agri-environmental policy mixes. J Environ Manage 2014; 145: 180–190.

19.

Kivimaa

Kern

Creative destruction or mere niche support? Innovation policy mixes for sustainability transitions. Res Policy 2016; 45: 205–217.

20.

Trencher

van der Heijden

Instrument interactions and relationships in policy mixes: achieving complementarity in building energy efficiency policies in New York, Sydney and Tokyo. Energy Res Soc Sci 2019; 54: 34–45.

21.

OECD. The Innovation policy mix. OECD Science, Technology and Industry Outlook 2010. Paris: OECD, 2010.

22.

Rogge

Reichardt

Policy mixes for sustainability transitions: an extended concept and framework for analysis. Res Policy 2016; 45: 1620–1635.

23.

Scordato

Klitkou

Tartiu

, et al. Policy mixes for the sustainability transition of the pulp and paper industry in Sweden. J Clean Prod 2018; 183: 1216–1227.

24.

Gawel

Strunz

Lehmann

A public choice view on the climate and energy policy mix in the EU – how do the emissions trading scheme and support for renewable energies interact?

Energy Policy 2014; 64: 175–182.

25.

Cantner

Graf

Herrmann

, et al. Inventor networks in renewable energies: the influence of the policy mix in Germany. Res Policy 2016; 45: 1165–1184.

26.

Sovacool

BK.

The importance of comprehensiveness in renewable electricity and energy-efficiency policy. Energy Policy 2009; 37: 1529–1541.

27.

Uyarra

Shapira

Harding

Low carbon innovation and enterprise growth in the UK: challenges of a place-blind policy mix. Technol Forecast Soc Change 2016; 103: 264–272.

28.

Fan

Feng

LY.

An evaluation of government incentives for new energy vehicles in China focusing on vehicle purchasing restrictions. Energy Policy 2017; 110: 609–618.

29.

Xiong

Chen

ML.

Study of the new energy vehicles demand market’s cultivated policy orientation: supply side or demand side. China population. Resour Environ 2016; 26: 129–137.

30.

Schmidt

Sewerin

Measuring the temporal dynamics of policy mixes – an empirical analysis of renewable energy policy mixes’ balance and design features in nine countries. Res Policy 2019; 48: 103557.

31.

Gong

Wang

New energy vehicles in China: policies, demonstration, and progress. Mitig Adapt Strateg Glob Change 2013; 18: 207–228.

32.

Zhang

Qin

China’s new energy vehicle policies: evolution, comparison and recommendation. Transp Res Part A Policy Pract 2018; 110: 57–72.

33.

Zhang

Liu

Analysis of new energy vehicles industry policy in china’s cities from the perspective of policy instruments. Energy Procedia 2016; 104: 437–442.

34.

Zhang

Bai

Incentive policies from 2006 to 2016 and new energy vehicle adoption in 2010-2020 in China. Renewable Sustainable Energy Rev 2017; 70: 24–43.

35.

Wang

Tang

Zhang

, et al. How to face the challenges caused by the abolishment of subsidies for electric vehicles in China? Energy 2019; 166: 359–372.

36.

Corradini

Costantini

Markandya

, et al. A dynamic assessment of instrument interaction and timing alternatives in the EU low-carbon policy mix design. Energy Policy 2018; 120: 73–84.

37.

Xiong

Wang

LY.

Policy cognition of potential consumers of new energy vehicles and its sensitivity to purchase willingness. J Clean Prod 2020; 261: 121032.

38.

Qiu

Zhou

Sun

HC.

Assessing the effectiveness of city-level electric vehicle policies in China. Energy Policy 2019; 130: 22–31.

39.

Libecap

GD.

Economic variables and the development of the law: the case of Western mineral rights. J Econ Hist 1978; 38: 338–362.

40.

Cools

Brijs

Tormans

, et al. Optimizing the implementation of policy measures through social acceptance segmentation. Transp Policy 2012; 22: 80–87.

41.

Peng

Zhong

Sun

WX.

Policy measurement, policy co-evolution and economic performance: an empirical study based on innovation policy. Manag World 2008; 9: 25–36.

42.

Zhao

Yang

Liang

, et al. Last-mile travel mode choice: data-mining hybrid with multiple attribute decision making. Sustainability 2019; 11: 6733.

43.

Guo

Tao

ZG.

An analysis of the impact of policy mix on leading markets by taking the new energy vehicle industry as an example. Sci Res Manage 2018; 39: 30–36.

44.

Tan

Tang

Lin

Policy impact of new energy vehicles promotion on air quality in Chinese cities. Energy Policy 2018; 118: 33–40.

45.

Harvey

AC.

Forecasting, structural time series models and the Kalman filter. Cambridge: Cambridge University Press, 1990.