1. Introduction
In recent years, with the rapid development of the economy, the air pollution has increased simultaneously in China [
1,
2]. The continuous improvement of air quality is an urgent requirement to achieve the ecological civilization development goals of carbon peak and carbon neutrality. Air pollution not only leads to multiple adverse health outcomes [
3,
4,
5], but also has a negative effect in economic growth [
6]. Improving air quality has become a significant task and a persistent issue [
1,
3,
7]. As a result, more and more strategies and tactics have been formulated and implemented in response to air pollution [
2]. Therefore, the improvement of air quality not only needs the support of government policies and the public’s awareness of environmental protection, but also depends on the technology innovation of manufacturing to reduce pollutant emissions [
7], because exhaust emission from manufacturing is an important factor that contributes to air pollution [
7]. China is one of the world’s largest emitters of pollutant emissions [
4]. To reduce pollutant emissions and improve air quality, China has implemented the policy portfolio about the technology innovation of manufacturing, including the automotive industry as an example in regard to the reduction in exhaust emissions and utilization of clean energy [
8].
Scientists have been dedicated to exploring innovation and new technology in automobiles and substituting new energy vehicles (NEVs) for traditional energy vehicles to improve air quality [
9]. Past studies have shown some evidence that the innovation of hybrid and plug-in NEVs can effectively alleviate CO
2 emissions [
9,
10,
11], and renewable energy technological innovation can reduce air nitrogen dioxide concentrations [
12].
Furthermore, previous studies have shown that the technology innovation for NEVs have triggered a massive growth in the production and purchase of NEVs, which can improve air quality by replacing traditional energy vehicles [
13,
14,
15].
However, this mediating role of the output of NEVs in improving air quality through technology innovation in NEVs has not been quantified. Our study is to fulfil this task.
Some studies show that vehicle and vessel tax (VVT) contributes to technological innovation [
12] and optimizes air quality [
12]. Some provincial governments in China have implemented a higher VVT to reduce the emissions of traditional energy consumption vehicles and support the NEV industry [
16,
17]. NEV companies in these regions may be incentivized to focus on the technology innovation of NEVs thereby expanding sales, which is conducive to improving the air quality [
18,
19]. Therefore, the impact that the technology innovation of NEVs has on air quality is more notable in regions with a higher VVT than those with a lower VVT.
To test the stationarity of the panel data next, based on the mediating effect models, our study analyzed that the technology innovation of NEVs (defined as the invention patent counts of each region) could enhance air quality through total output of NEVs (defined as the annual production quantity of NEVs in each region, unit: thousand). Then, we found that the impact of technological innovation of NEVs on air quality is stronger in high tax regions compared with low tax regions. This indicates that increasing VVT significantly enhances the positive effect of technology innovation in NEVs on air quality. In the analysis of the above relations, the number of invention patents for the NEV industry is used to quantify technology innovation, and the annual number of days of excellent air quality in each provincial administrative region is used to quantify this region’s air quality.
The previous studies only focus on some specific kinds of NEVs or treat all kinds of patents evenly in measuring new energy technology innovation [
9,
10,
16]. Up until now, to our best knowledge there have been no studies using all NEVs as a whole industry together with the output of NEVs and VVT to explore the relationship between technology innovation and air quality. In addition, it is generally accepted that invention patents, among patents for invention, patents for utility model and patents for design, have the strictest assessments requiring both prominent substantive features and notable progress [
20]. Therefore, we only single out invention patent data as a measurement of technology innovation rather than all patent data used by the previous literature. Our study contributes to the literature and public policy with empirical evidence in three aspects: (1) analyzes the direct relationship between innovation of NEVs as a whole industry and air quality for the first time; (2) explores the mechanisms of relationship between NEVs’ technology innovation and air quality, with the mediating role of output of NEVs and the moderating role of VVT; (3) affirms that governments should lay emphasis on levying tax appropriately to keep a balance between technology innovation of NEVs and air quality improvement.
This paper is organized as follows:
Section 2 discusses the role of technological innovation of NEVs in improving air quality, and develops three hypotheses to test;
Section 3 describes the data and model;
Section 4 reports and discusses empirical results; and
Section 5 concludes.
4. Empirical Results
This study attempts to quantify the impact that the technology innovation of NEVs has on air quality in China by the mediating role of the output of NEVs and the moderating role of VVT.
4.1. Descriptive Statistics
Table 1 is a description of statistics about the data. We summarize the panel data of 31 provincial administrative regions in mainland China during 2014–2019, such as the number of days that have excellent air quality status, the number of NEV invention patents per year, the population, the proportion of GDP in the secondary industry, the sum of VVT, the forest coverage rate, the output of NEVs and the VVT charging criteria in each provincial administrative region.
As shown by
Table 1, the indicators such as mean values, standard deviations (SD), minimums and maximums of each variable are listed. The variations of the data for most variables are very large. For example, the SD values of AQI and Patent are 60.700 and 70.660, respectively. Given that large deviation may bias our results, we used the natural logarithm of the value of invention patents of NEVs in our estimation.
Table 2 presents pairwise correlations between all variables in this research. All correlation coefficients are lower than 0.4. Pearson’s correlations of all variables are on the bottom left corner of the table and Spearman’s correlations on the top right corner.
To test the stationarity of the panel data, we performed the Breitung panel unit root tests according to Breitung (2000) [
35]. The null hypothesis is that a variable contains a unit root (i.e., it is not stationary). The estimated results for major interesting variable data are reported in
Table 3.
Table 3 includes level results of the unit root tests and all data series are stationary with high statistical significance in I (0). Therefore, we can soundly reject the null hypothesis of the panel unit root. This stationarity evidence probably also echoes the statement when the panel individual number N (N = 31 in this study) is larger than the time series number T (T = 6 in this study), the stationary testing is not required and usual panel data procedures are recommended [
36].
After the Breitung panel unit root tests, we performed the Granger causality tests. The results are reported in
Table 4. In the first line, we found that Patent can be regarded as the Granger cause of AQI. In the second line, the union significance of the coefficient of the test variable (AQI) is with the statistic chi-square of 0.066 and the relevant
p-value is 0.798. It indicates that AQI cannot be regarded as the Granger cause of Patent. To sum up, the evidence suggests that the technology innovation of NEVs represented by invention patent counts Granger causes the air quality improvement measured by AQI, but not vice versa.
4.2. Baseline Results
Before the baseline analysis, we carried out a further search to the invention patents in the NEVs industry, finding that more than 70% of all invention patents are directly related to the technology of air quality improvement. For example, one patent concerns the pollution prevention in mold making of new energy vehicles manufacturing. The remaining 30% of (at most) invention patents are not so relevant to air quality, which focus on the mechanistic designing of production efficiency improvement, the reduction in production cost of vehicle parts, etc.
Table 5 reports the panel fixed effect regression results for the relationship between the technology innovation effect of NEVs and AQI. The dependent variable is AQI, being measured as days of excellent air status. The independent variable is the technology innovation of NEVs, being measured as the number of invention patents of NEVs. The variable definitions are shown in
Table 1. Robust
t-statistics are reported in parentheses. The estimates of the two models are highly statistically significant, indicating that the value of invention patent counts is one of contributors for emissions reduction and environment protection. According to the Hausman test (chi-square = 10.440,
p = 0.001), we chose to use fixed effects regression models for data analysis. In column (1), we only include the interesting variable with yearly fixed effects and provincial fixed effects. The estimated coefficient of the interesting variable: Patent, is 1.102 and statistically significant at the 0.1% level, suggesting that the technology innovation of NEVs is positively correlated with excellent air quality days. In column (2), we add controls in our estimation. The coefficient on Patent is still positive and statistically significant at the 0.1% level. It suggests that the causal relationship between Patent and AQI is robust. Our results are basically similar to those of the previous studies [
37]. Overall, our results are consistent with Hypothesis 1 and support the positive association between the technology innovation of NEVs and AQI [
37,
38], indicating that the technology innovation of NEVs can enhance air quality in China.
4.3. Analysis of Mediator Effect
Mediator analysis is a set of statistical methods used to investigate whether a variable has an intermediary structure [
37]. According to the previous study [
39], mediation is tested with the causal step procedures (Baron and Kenny, 1986), in which the relationships among the independent variable, dependent variable and the mediator variable are analyzed as follows: (a) the dependent variable is regressed on the independent variable; (b) the mediator variable is regressed on the independent variable; (c) the dependent variable is regressed on both the independent variable and mediator variable. These regression equations can be written in Model 1, 2 and 3. We suggest that the output of NEVs is the mediator variable.
In Model 1, 2 and 3, we tested the mediating effect of the output of NEVs on the process where the technology innovation of NEVs influences air quality. The direct effect of technology innovation on air quality remains statistically significant after including the mediator variable in Model 1 (α = 1.034,
p < 0.001), which has been reported in the baseline results of
Table 5. In Model 2, Output is statistically significant and positively associated with Patent (α′ = 0.673,
p < 0.001), which indicates that the technology innovation of NEVs enhances the output of NEVs. In Model 3, there is a significant association in AQI with Patent and Output (α″ = 0.019 < α,
p < 0.001), which indicates the statistically significant mediating effect by the output of NEVs between technology innovation and air quality. The estimates of panel fixed effect regression are shown in
Table 6.
4.4. Analysis of Moderator Effect
Appendix A of this paper lists the indicators of vehicle and vessel tax, abbreviated as VVT in 31 provincial administrative regions (yuan/year) and the preferential policies about VVT in mainland China.
According to the VVT in each provincial administrative region, we select the tax rate standard of 1.0–1.6 L (the mainstream vehicle type that accounts for about 70 percent of total automobile sales in China [
39]) as the moderator variable. As shown in
Appendix A, this tax varies in each provincial administrative region, with an average of 335.5 yuan, a maximum of 420 yuan and minimum of 300 yuan.
Therefore, in order to better understand the role that VVT plays, the whole sample was divided into two groups (low tax group and high tax group) based on the tax rates of VVT. Between them, the low tax group is that the tax rates are less than the median (350 yuan/L), consisting of 17 provincial administrative regions. The other group is that the tax rates are more than 350 yuan/L, which includes 14 provincial administrative regions. We ran regressions of Model 1 for both groups. In the low tax group, only three control variable estimates were statistically significant while in the high tax group five variable estimates were statistically significant (including the independent variable).
The results of panel fixed effect regression analysis of both groups are shown in
Table 7.
In
Table 7, the estimates of VVT are not statistically significant in low tax regions (only three control variables significant with
p < 0.01). However, they become statistically significant in high tax regions with independent variables and five control variables are also statistically significant (
p < 0.001). In addition, we ran a seemingly unrelated regression estimation to examine the difference of significances between the above two groups, with results of difference = 0.908 and chi-square = 5.12, meaning the two groups are statistically significantly different. Therefore, we can conclude that VVT is a moderator variable between the technology innovation of NEVs and air quality. VVT can moderate the association between the technology innovation of NEVs and air quality in China.
4.5. Robustness Test
Lastly, we perform a robustness check by singling out the air-quality-related invention patents and regressing
AQI on the air-quality-related invention patents only. The results are reported in
Table 8. The difference between the results in
Table 5 and the results in
Table 8 is that
Patent in
Table 5 includes all invention patents of NEVs while Patent in
Table 8 includes only the invention patents directly related to air quality. Based on the results, we can see clearly that when including invention patents directly related to air quality, the significance of our variable estimates in regressions even rises, with higher absolute values of coefficients. This indicates that the invention patents closely correlate with emissions reduction and being environment friendliness.
5. Conclusions
Nowadays, the control of environmental pollution has become a hot and prioritized issue in China and abroad. In this paper, we used a panel fixed effect model to analyze the data from 31 provincial administrative regions in mainland China to test our three hypotheses: (1) the technology innovation of NEVs industry is positively correlated with air quality; (2) the technology innovation of NEVs industry correlates with air quality improvement with an impact path of NEVs’ output; and (3) VVT can enhance the positive relationship between the technology innovation of NEVs and air quality.
This study finds that these influencing factors improve the air quality statistically and economically significantly. The relevant policy implications that can be derived from this study are as follows.
For the automotive manufacturing industry, the companies in this industry should increase expenditures on research and development in pursuit of state-of-the-art technological innovation which addresses air quality and environmental friendliness. A massive replacement of traditional energy vehicles with NEVs having frontier technologies can be popularized to reduce pollution emissions. For relevant government agencies and departments at all levels, this study provides strong empirical evidence in China to appropriately increase VVT rates, because VVT can incentivize consumers to substitute traditional energy vehicles with NEVs. At the same time, central and local governments can also enact a series of preferential tax treatments to the vehicle manufacturers that have invested heavily in NEV research and development.