How Do Transportation Influencing Factors Affect Air Pollutants from Vehicles in China? Evidence from Threshold Effect

Abstract

In recent years, China has promoted a series of legal norms to reduce the environmental impact of air pollutants from vehicles. The three main vehicle emission species (carbon monoxide, hydrocarbons, nitrogen oxides) contribute significantly to air pollution. In this study, the emission factor method was used to estimate air pollutants from vehicles in 31 provinces from 2006 to 2016. The results show a trend of total vehicle carbon monoxide (CO) and hydrocarbons (HC) emissions decreasing with time; the vehicle nitrogen oxides (NOx) emission trend is divided into two stages: an upward trend between 2006 and 2012 and a downward trend after 2012. Based on a panel threshold, a regression method was used to divide the vehicle NOx and CO emissions in China into four emission zones: low emissions, medium emissions, high emissions, and extra-high emissions. Vehicle HC emissions were divided into three emission zones, which corresponded to low emissions, medium emissions, and high emissions. Overall, vehicle pollution emission efficiency and per capita GDP have a significant inhibitory effect on the three main air pollutants from vehicles (NOx, HC, CO). Both passenger and freight turnover have significant roles in promoting the three air pollutants from vehicles (NOx, HC, CO). Road density and road carrying capacity have a significant role in promoting vehicle HC and CO emissions. Increasing truck proportion inhibits vehicle CO emissions and promotes vehicle NOx emissions. The urbanization rate has a positive effect on vehicle HC and CO emissions. Moreover, there is obvious heterogeneity in different emission zones of the three air pollutants from vehicles (NOx, HC, CO).

1. Introduction

The negative effects of air pollution on human health have been widely reported in the literature [1,2,3,4,5,6,7]. Air pollution causes about 3.7 million premature deaths every year in the world, and the air pollutant emission from ground transportation is an important factor affecting urban environmental quality [8,9]. Vehicle emissions mainly refer to harmful oxides produced by the incomplete combustion of fuel; such emissions can harm human health and plant growth to varying degrees. The composition of vehicle exhaust is complex; it often contains hundreds of different substances, including, carbon monoxide (CO), hydrocarbons (HC), nitrogen oxides (NOx), sulfur dioxide (SO₂), particulate matter (PM), and aldehydes. Emissions from diesel engines are mainly CO, HC, and NOx, while those from gasoline are mainly HC, NOx, and PM. NOx can irritate the nose, eyes, lungs, and throat, increasing the risk of viral infection, and can also adversely affect plant growth. NOx and HC are both important sources of photochemical smog, which directly harms animals, plants, and the human body. CO can cause dizziness, headaches, and rapid heartbeat; when the CO concentration is high, it can damage the central nervous system and cause suffocation.

Many studies have clear evidence to prove the harm of air pollution to human health [10,11,12,13,14,15,16,17]. According to the China Vehicle Environmental Management Annual Report, published by the Ministry of Environmental Protection of the People’s Republic of China, total vehicle emissions of CO, HC, NOx, and PM during the period 2011–2016 show decreasing trends; however, the annual reduction rate is low, with an average annual decrease of just 0.6%, while absolute emission values remain severe [18]. In 2016 alone, vehicles emitted 34.391 million tons of CO, 4.22 million tons of HC, and 5.778 million tons of NOx. Therefore, it is necessary to investigate the driving factors of air pollutants from vehicles in China.

Previous researches have conducted various in-depth studies and analyses on the driving factors of air pollutants from vehicles [19,20,21,22,23,24]. With the increasing severity of air pollution caused by vehicle emissions worldwide so far, many scholars have devoted themselves to the research of vehicle emissions providing a scientific basis for the treatment of vehicle emission. Liu et al. [25] found that the economy is the basic driving force for vehicle emissions, and that the NOx emissions of vehicles rises with economic development. With extensive urbanization, China’s vehicle ownership is increasing rapidly, and vehicle exhaust pollution has become the main source of urban air pollution. Regarding research focused on vehicle technology level, Huo et al. [26] found that the level of vehicle technology significantly influences vehicle emission factors. In terms of vehicle structure and vehicle emissions. For the operational level of vehicles, Montag [27] developed and analyzed a simple technical and behavioral mechanism model for determining the volumetric emissions produced by automobiles. Sugihara et al. [28] found that the number of passenger cars and road length have a positive effect on air pollutants from motor vehicles. Limanond et al. [29] reported that GDP plays an important role in the amount of energy consumed by the transportation sector. Liang et al. [30] reported the relationship between energy structure (efficiency) and the transportation industry. Alshehry et al. [31] found that economic growth has an important impact on CO₂ emissions from the transportation industry in Saudi Arabia. Cheng et al. [32] and Fan and Lei [33] studied the impact of urban traffic management policies, population, energy intensity and other factors on air pollution caused by the transportation industry in Kaohsiung and Beijing.

In brief, previous studies have explored the social and economic factors of vehicle emissions from multiple perspectives but failed to investigate the impact of transportation influencing factors on air pollutants from vehicles. Previous studies have mainly focused on economic and social factors; data from transportation has been lacking. In fact, transportation influencing factors have a great influence on the treatment of air pollutants from vehicles in China. China’s total road mileage and highway density continue to grow. With the rapid development of highway construction and of the economy, vehicle numbers and both passenger and road freight turnover have shown clear upward trends. Although China’s car ownership rate is lower than that of developed countries, issues associated with fuel quality, car quality, and road conditions mean that vehicle exhaust emissions are much higher. The transportation department is closely related to the emissions of vehicles. It is more effective to reduce the air pollutants from vehicles by analyzing the transportation influencing factors. Meanwhile, the existing literature is mainly focused on specific areas; few studies have included the whole of China. In this paper, the research scope is in 31 provinces in China. Furthermore, when studying air pollutants from vehicles in China, previous researches have overlooked comparative analysis of three main vehicle emission species (NOx, HC, CO). In terms of research methods, few previous literatures have systematically discussed the impacts of transportation influencing factors on air pollutants from vehicles with different emission zones in China.

This paper hopes to investigate, in a comprehensive way, how transportation influencing factors affect three air pollutants from vehicles, and the following problems are discussed. What are the spatial and temporal distributions of the three air pollutants from vehicles (NOx, HC, CO)? How does transportation influencing factors affect three air pollutants from vehicles in China? Given the heterogeneity of different emission zones, the impact of transportation influencing factors on three air pollutants from vehicles with different emission zones in China needs further consideration. The in-depth study of the above issues is helpful to investigate the role of transportation influencing factors in improving air quality and has profound implications for policymaking of transportation departments.

In response to the above issues, this paper utilizes balanced panel data from 31 provinces in China from 2006 to 2016 to investigate the relationship between transportation influencing factors and three kinds of air pollutants from vehicles (NOx, HC, CO), and attempts to make progress are as follows: (1) Using the emission factor method, this paper calculates three kinds of air pollutants from vehicles in China (NOx, HC, CO). (2) using the fixed effect method, the impact of transportation influencing factors on the three kinds of air pollutants from vehicles (NOx, HC, CO) in China is investigated. (3) The panel threshold model is utilized to investigate the effects of traffic factors on transportation influencing factors on the three kinds of air pollutants from vehicles (NOx, HC, CO) in China in different emission zones.

The innovations for this paper in the following three aspects: (1) The spatial distribution characteristics of three air pollutants from vehicles from 2006 to 2016 are analyzed by ArcGIS 10.2. (2) This paper investigates the relationship between transportation influencing factors and three air pollutants from vehicles (NOx, HC, CO) by using the panel data of 31 provinces in China. (3) Given the heterogeneity of different emission zones, the impact of transportation on three air pollutants from vehicles with different emission zones in China is analyzed. The existing research focuses on the impact of socio-economic factors on single air pollutants from vehicles in China, while this research focuses on the relationship between transportation influencing factors and three air pollutants from vehicles (NOx, HC, CO) in different emission zones.

The remaining sections of this study are organized as follows: Section 2 is methodology and data. Section 3 and Section 4 are results and discussion, respectively. The conclusions and policy implications are presented in the last section.

2. Methodology and Data

2.1. Panel Threshold Regression Model

To study the influence of transportation sector factors on the atmospheric pollutants of Chinese vehicles under different pollution levels, it is necessary to first classify the three main pollutants (NOx, HC, CO) emitted by vehicles. Previous studies have often ignored the heterogeneity of samples [34]; however, the panel threshold regression method uses strict statistical inference methods for critical parameter estimation and hypothesis testing [35,36,37], which can accurately reveal the relationship between dependent and independent variables in different groups. The threshold regression method has previously been used to estimate the thresholds of the main drivers of carbon dioxide emissions [38]; this suggests that the method is also suitable for determining the thresholds of the three major pollutants emitted by vehicles (NOx, HC, CO) and the influence of transportation sector factors on these pollutants under different pollution levels.

The basic form of the panel threshold regression model is as follows:

$$y_{it}=u_i+{\beta }_1{x}_{it}I\left(q_{it}\le \gamma \right)+{\beta }_2{x}_{it}I\left(q_{it}>\gamma \right)+{\epsilon }_{it}$$

(1)

where in region i and year t, and $q_{it}$ and $\gamma$ denote the threshold variable and the unknown threshold, respectively. The ${\epsilon }_{it}~\left(0,{\sigma }^2\right)$ is the random disturbance error term, and I($·$) is the indicator function. The regression slope of $x_{it}$ is ${\beta }_1$ or ${\beta }_2$ depending on whether $q_{it}$ is smaller or larger than $\gamma$; ${\mu }_i$ is the individual interception indicating that the model is an individual fixed effects model.

Based on the basic form, this study established a relationship model between transportation sector factors and the three major pollutants (NOx, HC, and CO) emitted by motor vehicles, as shown in Equation (2):

$$\{\begin{array}{l}ln\left(y_{it}\right)={\mu }_{it}+{\theta }_{1}ln(Z_{it})+{\epsilon }_{it},y_{it} \le \gamma \\ ln(y_{it})={\mu }_{it}+{\theta }_{2}ln(Z_{it})+{\epsilon }_{it},y_{it}>\gamma \end{array}$$

(2)

This is an individual fixed effect model where $y_{it}$ is the three major pollutants (NOx, HC, and CO) emitted by motor vehicles in region i and year t, ${\mu }_{it}$ is an individual fixed effect, and ${\epsilon }_{it}$ is a random interference term. The $Z_{it}$ is a set of explanatory variables, and $\theta$ is the coefficient of the explanatory variables. In this study, the three main pollutants (NOx, HC, and CO) emitted by motor vehicles are taken as threshold variables, and $\gamma$ is the threshold to be estimated. When $y_{it} \le \gamma$, the coefficient of $Z_{it}$ is ${\theta }_1$; when $y_{it}>\gamma$, the coefficient of $Z_{it}$ is ${\theta }_2$. In addition, the analysis framework of this paper is shown in Figure 1.

2.2. Variables Selection

2.2.1. Dependent Variable

Pollutant emissions of vehicles (NOx, HC, and CO): This study used the emission factor method to calculate the pollutant emissions of vehicles, according to the Technical Guidelines for the Preparation of Air Pollutant Emission Inventory for Road Vehicles (Trial). The road vehicles emissions (E) mainly include exhaust emissions (E₁) and HC evaporative emissions (E₂). The vehicle HC emissions are calculated based on Formula (3) and the vehicle NOx (or CO) emissions are calculated based on Formula (4). The formula for road vehicle emissions is as follows:

$$\mathrm{E}=E_1+E_2$$

(3)

The formula for calculating the exhaust emissions of road vehicles is as follows:

$$E_1=E_m=\mathsf{\Sigma }(P_{m,i}\times E{F}_{m,i}\times VK{T}_i)\times {10}^{-6}$$

(4)

Among them, $E_1$ represents NOx (or CO, HC) exhaust emissions (tons), m represents the 31 provinces in China, $E_m$ represents the NOx (or CO, HC) exhaust emissions in the m area (tons), and P_m,i represents the quantity of the i-type vehicle in the m area (car). $E{F}_{m,i}$ is the pollutant concentration per unit mileage of the i-type vehicle in the m-zone (g/km) and $VK{T}_i$ is the average annual mileage of the i-type vehicle (km/car).

E₂ is the annual HC evaporative emissions during driving and parking are calculated according to the following formula:

$$E_2=\left(E{F}_1\times \frac{VKT}{V}+E{F}_2\times 365\right)\times \mathrm{P}\times {10}^{-6}$$

(5)

where $E_2$represents the HC evaporation emission during the annual driving and parking period (tons)$,E{F}_1$ is the evaporative emission coefficient during the driving process of the vehicle (g/h), VKT is the average annual mileage of the vehicle (km/car), V is the average running speed of vehicles (km/h$),E{F}_2$ is the comprehensive emission coefficient during the parking period (g/d) and P is the average daily traffic flow of gasoline-fueled motor vehicles on the road (car).

2.2.2. Transportation Variables

Road carrying capacity (RCC): The road carrying capacity is measured by the number of vehicles per kilometer of road, which reflects local road levels and affects vehicle emissions. According to international standards, when the number of vehicles per kilometer in a region reaches 270 vehicles/km, vehicle possession is considered to have reached saturation. If the number of vehicles continues to increase, it will cause problems such as slower speeds and traffic accidents. Therefore, in this study, the ratio of the number of civil vehicles to road length was used to measure the road carrying capacity in a region.

Road density (RD): The road density is the length of the road construction per unit area. This indicator also reflects the local road level. When the road density increases at a lower level than the growth rate of vehicles, it will lead to traffic congestion and increasing vehicle emissions [34]. Therefore, in this study, the ratio of road length to the area was used to measure the road density in a region.

Traffic pollutant emission efficiency (NEE, HEE, and CEE): Transportation emission efficiency can reflect the impact of vehicle technological progress on vehicle emissions. There are two main forms of technology progress: the improvement of vehicle emission reduction technology and high-efficiency vehicle technology to reduce vehicle emissions [26]. In this study, the ratio of passenger and freight transport turnover to the emissions of the three main vehicle pollutants (NOx, HC, CO) was used to measure the traffic NOx emission efficiency (NEE), traffic HC emission efficiency (HEE), and traffic CO emission efficiency (CEE) in a region.

Proportion of trucks (TRUCK): The contribution of different vehicle types to different vehicle emissions is obvious. The main air pollutants in this study are NOx of vehicles, and the main contributor to NOx emissions is trucks [25]. Therefore, this study selected truck proportion to represents the impact of vehicle structure. The ratio of the number of trucks to the number of civilian vehicles was used to measure the proportion of trucks in a region.

Road passenger turnover (RPT): Highway passenger traffic volume represents the road passenger traffic multiplied by the average transport distance. This indicator not only expresses the cause of transportation more comprehensively, but also reflects the distance travelled by the vehicle. This indicator is used to characterize the impact of vehicle mileage on vehicle pollution.

Road freight turnover (RFT): The road freight turnover is the mass of goods carried by the vehicle in a certain period of time multiplied by the corresponding transport distance. This index can represent the impact of vehicle mileage on vehicle pollution.

2.2.3. Control Variables

Urbanization rate (UR): In the process of urbanization, the change from ‘farmer’ to ‘resident’ has a profound impact on production and consumption behavior. With urbanization, urban motorization has increased, and the scale of vehicle ownership has been continuously expanding [39]. At the same time, urban road construction brought about by urbanization has an impact on the road system [40]. In this study, the ratio of urban population to the total population was used to measure the urbanization rate in a region [41].

Per capita GDP (PGDP): The per capita GDP reflects the local economic level. On the one hand, with the improvement of economic level year by year, demand for car purchase increases, which may lead to the increasing vehicle pollutant emissions [25]. On the other hand, economic growth may change the car market’s consumption habits. For example, the promotion of new energy vehicles will continue to reduce the emissions of vehicle units [42] and alleviate vehicle pollution emissions. In this study, the ratio of urban population to total population was used to measure the urbanization rate in a region. Therefore, we used the ratio of domestic (provincial) GDP to domestic (provincial) resident population to measure the per capita GDP of a region, so as to measure the impact of economic development on vehicle emissions.

2.3. Data Sources

This study considered data on vehicle NOx, HC, and CO emissions in 31 provinces and cities of China from 2006 to 2016, while Hong Kong, Macao, and Taiwan were excluded from this paper. The data used to calculate vehicle atmospheric emission mainly come from the China Statistical Yearbook (2007–2017) and the statistical yearbooks of the relevant provinces (2007–2017), including data on the total number of vehicles in each province. Data such as mileage and average speed of vehicles come from specific literature and from the Technical Guidelines for the Preparation of Air Pollutant Emission Inventory for Road Vehicles (Trial). Data on vehicle Sulphur content and fuel standards were derived from specific policy documents. In addition, the various indicators affecting vehicle emissions, including passenger numbers, freight turnover, truck ownership, road length, and regional area, were derived from the National Bureau of Statistics (2007–2017).

Table 1. Descriptive statistics of dependent and explanatory variables.

Definition	Variables	Mean	Std. Dev.	Min	Max	Unit
NOx emission	NOx	140,288	90,781	21,858	427,946	tons
HC emission	HC	70,784	53,932	1256	352,387	tons
CO emission	NO	140,288	90,781	21,858	427,946	tons
Road carrying capacity	RCC	132,192	71,656	10,392	324,138	car/km
Road density	RD	0.884	0.565	0.0276	2.742	km/km2
NOx emission efficiency	NEE	423.9	232.5	29.35	1763	km
HC emission efficiency	HEE	350.5	410.8	9.634	3928	km
CO emission efficiency	CEE	33.45	30.85	0.728	195.7	km
The proportion of trucks	TRUCK	0.252	0.276	0.0403	2.186	%
Road passenger turnover	RPT	6397	25,422	18.60	196,871	man-kilometer
Road freight turnover	RFT	1414	1645	25.40	7959	ton-kilometer
Urbanization rate	UR	51.78	15.07	0.222	89.60	%
Per capita GDP	PGDP	3.790	2.237	0.615	11.81	yuan

lists descriptive statistics of all variables.

3. Estimation Results

This section may be divided by subheadings. It should provide a concise and precise description of the experimental results, their interpretation, as well as the experimental conclusions that can be drawn.

3.1. Exploratory Temporal Analysis of Vehicle Air Pollutants

Figure 2 compared the changes of NOx, CO, and HC emissions from vehicles in China from 2006 to 2016. The total vehicle NOx emissions increased from 3.26 million tons in 2006 to 4.29 million tons in 2016. The vehicle NOx emission trend is divided into two time periods. From 2006 to 2012, annual vehicle NOx emissions showed an upward trend, reaching a peak in 2012. After 2012, total emissions showed a downward trend. Except for a significant increase in total emissions during 2009 and 2010, the total emission value varied little between year. Compared with vehicle HC, the NOx emission reduction pressure remains huge [43]. Therefore, more attention is needed to reduce vehicle NOx emissions [26]. The total amount of vehicle HC emissions decreased with time and the annual change is large. For example, in 2008, 2014, and 2016, total emissions increased significantly compared with the previous year. The trend of total vehicle CO emissions is the same as that of vehicle HC emissions (decreasing over time). The annual variation of vehicle CO emissions is small. The year-on-year growth rate of vehicle NOx (or CO, HC) emissions shows the growth rate of vehicle NOx (or CO, HC) emissions between survey year t − 1 and t, for each t = 2006–2016. In other words, the year-on-year growth rate of vehicle NOx (or CO, HC) emissions in year t is the value of vehicle NOx emissions in year t minus vehicle NOx emissions in year t − 1 divided by vehicle NOx emissions in year t − 1, for each t = 2006–2016.

3.2. Spatial Distribution of Vehicle Air Pollutants

The spatial distribution of vehicle CO, NOx and HC emissions is complex. From Figure 3, the vehicle CO emissions across most provinces fell from 2006 to 2016, as did the number of high emission regions. vehicle CO emissions is mainly distributed in Shandong, Hebei, Henan, Yunnan, Guangdong, Jiangsu, and Zhejiang.

For Figure 3, regions with high vehicle NOx emissions experienced a process of growth and then decline from 2006–2016. From 2006 to 2010, vehicle NOx emissions rose in most provinces, with high-emission areas increasing over time. Conversely, the vehicle NOx emissions across most provinces fell from 2010 to 2016, as did the number of high emission regions. This is consistent with the overall emission trend. Vehicle NOx emissions are mainly concentrated in Shandong and Hebei.

For Figure 3, the number of high vehicle HC emissions areas fell from 2006 to 2016, but the spatial distribution remained relatively stable with emissions concentrated in Shandong, Hebei, Henan, Yunnan, Guangdong, Jiangsu, and Zhejiang.

3.3. Regression Results for Vehicle Air Pollutants

For

Table 2. Fixed effects of vehicle NOx, CO and HC emissions.

Variable	NOx Emission	CO Emission	HC Emission
RPT	0.2412 ***	0.310 ***	0.290 ***
RFT	0.5989 ***	0.685 ***	0.647 ***
RCC	−0.0070	0.154 ***	0.161 ***
RD	−0.0262	0.166 **	0.189 **
TRUCK	0.5725 ***	−0.0858 *	−0.0352
PGDP	−0.0673 *	−0.280 ***	−0.276 ***
UR	0.0020	0.380 ***	0.420 ***
NEE	−0.8418 ***
CEE		−0.981 ***
HEE			−0.944 ***
Obs	341	341	341
Number of provinces	31	31	31

Note: * p < 0.1, ** p < 0.05, *** p <0.01.

, traffic pollutant emission efficiency (NEE, CEE, and HEE) and per capita GDP (PGDP) have a significant inhibitory effect on the three main pollutants (NOx, HC, CO). Road density (RD) and road carrying capacity (RCC) have a significant role in promoting vehicle HC and CO emissions, but no significant impact on vehicle NOx emissions. Increasing truck proportion (TRUCK) significantly inhibits vehicle CO emissions and plays a significant role in promoting the vehicle NOx emission; however, it has no impact on HC emission of vehicles. In terms of vehicle operation level, both road passenger turnover (RPT) and road freight turnover (RFT) have significant roles in promoting the three pollutants (NOx, HC, CO) and are the main causes of vehicle air pollution. The urbanization rate (UR) has a significant effect on HC and CO emissions, but no significant effect on vehicle NOx emissions. Observations (Obs) are 341.

3.4. Threshold Effect Estimation Results

In order to examine whether the impact of transportation influencing factors on air pollutants from vehicles at different emission zones, the existence of three main vehicle pollutants (NOx, HC, CO) threshold was tested firstly.

Table 3. Estimation results of threshold value.

Variable	Threshold Value	F Value	Estimation Value (ton)
NOx emission	Triple threshold	45.776 ***	95,037; 159,395; 294,241
HC emission	Double threshold	33.459 ***	19,505; 54,757
CO emission	Triple threshold	32.888 ***	205,787; 783,075; 1,275,881

Note: *** p < 0.01.

shows the threshold value results. Based on the F-statistic and p-value (p < 0.01), the estimated results of vehicle NOx and CO emissions reject the null hypothesis of no three-threshold effects at 1% significance level, indicating that three thresholds exist. Hence, it can be inferred that vehicle NOx and CO emissions have a three-threshold effect. The three thresholds for vehicle NOx emissions are 95,037, 159,395 and 29,4241. Vehicle NOx emissions can be divided into four zones: NOx ≤ 95,037 (low emission zones), 95,037 < NOx ≤ 1593,95 (medium emission zones), 159,395 < NOx ≤ 294,241 (high emission zones), and NOx > 294,241 (extra-high emission zones). The three thresholds for vehicle CO emissions are 205,787, 783,075 and 1,275,881. Vehicle CO emissions can be divided into four zones: CO $\le$ 205,787 (low emission zones), 205,787 < CO $\le$ 783,075 (medium emission zones), 783,075 < CO $\le$ 1,275,881 (high emission zones), and CO > 1,275,881 (extra-high emission zones). Based on the F-statistic and p-value (p < 0.01), the estimated results of vehicle HC emissions reject the null hypothesis of no double threshold effects at 1% significance level, indicating that two thresholds exist. Hence, it can be inferred that vehicle HC emissions have a double threshold effect. The two thresholds for vehicle HC emissions are 19,505 and 54,757. Vehicle HC emissions can be divided into three zones: HC ≤ 19,505 (low emission zones), 19,505 < HC ≤ 54,757 (medium emission zones), and HC > 54,757 (high emission zones).

4. Discussion

4.1. Threshold Effects of Vehicle NOx Emissions

Table 4. Threshold effects of vehicle NOx emissions.

Variables	Model 1	Model 2	Model 3	Model 4
	NOx ≤ 95,037	95,037 < NOx ≤ 159,395	159,395 < NOx ≤ 294,241	NOx > 294,241
RPT	0.335 ***	0.230 ***	0.346 ***	0.306 ***
RFT	0.581 ***	0.588 ***	0.514 ***	0.801 ***
NEE	−0.877 ***	−0.820 ***	−0.794 ***	−1.092 ***
RCC	−0.100 ***	0.0447 *	0.0815	0.306
RD	0.111	−0.0515	−0.116	0.679
TRUCK	0.613 ***	0.306	0.806 ***	−0.286
PGDP	−0.0529	−0.0492	−0.0621	−0.293
UR	−0.0211	−0.0198 **	−0.281	3.072 ***
Obs	110	110	88	33
Number of provinces	10	10	8	3

Note: * p < 0.1, ** p < 0.05, *** p < 0.01.

shows the estimation results NOx emissions threshold effects. For every 1% increase in the urbanization rate, vehicle NOx emissions in extra-high-emission zones increase by 3.072%. The increase in urban population has led to an increase in urban roads. When this increase in roads does not match the development of the urbanization process, traffic congestion occurs, travel time costs increases, and vehicle pollutant emissions increase [44]. On this basis, road systems should be improved to reduce the impact of urbanization on extra high vehicle NOx emission zones.

The proportion of trucks is the most important promoting factor for vehicle NOx emissions in high and low emission areas. Excessive emissions exceed standards and have led to increasing vehicle NOx emissions from trucks in China. It is necessary to increase controls on vehicles that exceed standards, continue to promote the elimination system of yellow-label vehicles, and reduce the impact of trucks on vehicle NOx emissions. In medium and low emission zones, the road carrying capacity also has a significant impact on NOx emissions. In low emission zones, with increasing road carrying capacity, vehicle NOx emissions fall. In medium emission areas, with increasing road carrying capacity, vehicle NOx emissions rise. Road passenger turnover and road freight turnover are the main reasons for increasing vehicle NOx emissions for all four pollution zones. Traffic NOx emission efficiency is the main reasons for the reduction of vehicle NOx emissions in the four pollution zones.

4.2. Threshold Effects of Vehicle CO Emissions

Table 5. Threshold effects of vehicle CO emissions.

Variables	Model 1	Model 2	Model 3	Model 4
	$\mathrm{CO}\le$ 205,787	$205,787<\mathrm{CO}\le$ 783,075	$783,075<\mathrm{CO}\le$ 1,275,881	CO > 1,275,881
RPT	1.118 **	0.466 ***	0.537 ***	0.412 ***
RFT	0.159	0.416 ***	0.429 ***	0.371 ***
CEE	−1.226 **	−0.522 ***	−0.638 ***	−0.464 ***
RCC	1.195	−0.0504	0.315 ***	0.117
RD	0.155	−0.347 ***	−0.290 ***	1.354
TRUCK	−2.815	−0.987 **	0.166	−4.223 *
PGDP	−0.389	−0.529 ***	−0.525 ***	−0.649
UR	1.257	2.457 ***	0.309	3.025 **
Obs	99	77	132	33
Number of provinces	9	7	12	3

Note: * p < 0.1, ** p < 0.05, *** p < 0.01.

shows the threshold effects of vehicle CO emissions. Overall, the road passenger turnover and traffic CO emission efficiency results are similar to those from the overall fixed regression. For all four vehicle CO emission intervals, road passenger turnover has a significant role in promoting vehicle CO emissions, while traffic pollution emission efficiency has a significant role in inhibiting vehicle CO emissions. Therefore, the Chinese government should pay attention to the traffic CO emission efficiency and road passenger turnover in various pollution intervals. Except for low vehicle CO emission zones, road freight turnover plays an important role in increasing the vehicle CO emissions.

From Table 5, the road carrying capacity is the main reason for increasing vehicle CO emissions in high emission zones. Road density and GDP per capita are the main inhibiting factors for vehicle CO emissions in medium and high emission zones. The proportion of trucks is the main inhibiting factor for vehicle CO emissions in medium and extra-high emission zones. Urbanization rate is the main reason for increasing vehicle CO emissions in medium and extra-high emission zones. Therefore, road passenger turnover is the main reasons for increasing vehicle CO emissions for all four pollution zones and traffic NOx emission efficiency is the main reasons for the reduction of vehicle CO emissions in the four pollution zones.

4.3. Threshold Effects of Vehicle HC Emissions

Table 6. Threshold effects of vehicle HC emissions.

Variables	Model 1	Model 2	Model 3
	$\mathrm{HC}\le$ 19,505	$19,505<\mathrm{HC}\le$ 54,757	HC > 54,757
RPT	0.725 ***	0.635 ***	0.673 ***
RFT	0.165	0.585 ***	0.321 ***
HEE	−0.873 ***	−0.747 ***	−0.834 ***
RCC	1.119 **	0.425 ***	−0.0682 **
RD	0.977 ***	−0.385 ***	−0.0317 **
TRUCK	4.918 *	0.446	0.376
PGDP	−1.599	−0.731 ***	0.206 ***
UR	−1.769 **	0.781 **	−0.0229
Obs	44	99	198
Number of provinces	4	9	18

Note: * p < 0.1, ** p < 0.05, *** p < 0.01.

shows the threshold effects of vehicle HC emissions. In high vehicle HC emission zones, all indicators have passed the significance test except for the proportion of trucks and urbanization. Road passenger turnover, road freight turnover, and per capita GDP all increase vehicle HC emissions in high emission zones, while traffic HC emission efficiency, road carrying capacity, and road density reduce vehicle HC emissions in high emission zones.

In medium emission zones, only the proportion of trucks did not pass the significance test. Road passenger turnover, road freight turnover, road carrying capacity, and urbanization rate all increase HC emissions in medium emission zones; traffic pollutant discharge efficiency, road density and GDP per capita inhibit HC emissions in medium emission zones.

In low vehicle HC emission zones, all indicators have passed the significance test except for road freight turnover and GDP per capita. Road passenger turnover, road carrying capacity, the proportion of trucks, and road density all increase HC emissions in medium emission zones; traffic pollutant discharge efficiency and urbanization rate inhibit HC emissions in low emission zones. Therefore, road passenger turnover is the main reasons for increasing vehicle HC emissions in the three pollution zones and traffic HC emission efficiency is the main reasons for the reduction of vehicle HC emissions in the three pollution zones.

5. Conclusions and Policy Implications

5.1. Conclusions

Based on a literature review of vehicle air pollution and vehicle transportation systems, this study calculated the emissions of three atmospheric pollutants in 31 provinces and cities of China from 2006 to 2016; we used the emission factor method according to technical guidelines for the preparation of road vehicle air pollutants list (Trial). We found that vehicle NOx emission trends in China can be divided into two periods: 2006 to 2012 (upward trend) and 2012–2016 (downward trend). The CO and HC emissions from vehicles in China decreased during the sample period. In addition, the spatial distribution of CO, HC, and NOx emissions was generally scattered.

The fixed effect model was established to analyze the impact of different traffic system indicators on air pollution emissions. Overall, vehicle pollution emission efficiency and per capita GDP have a significant inhibitory effect on the three main air pollutants from vehicles (NOx, HC, CO). Both passenger and freight turnover have significant roles in promoting the three air pollutants from vehicles (NOx, HC, CO). Road density and road carrying capacity have a significant role in promoting vehicle HC and CO emissions. Increasing truck proportion inhibits vehicle CO emissions and promotes vehicle NOx emissions. The urbanization rate has a positive effect on vehicle HC and CO emissions. Moreover, there is obvious heterogeneity in different emission zones of the three air pollutants from vehicles (NOx, HC, CO).

For the prevention and control of the severe air pollution in China, China’s State Council [45] released the “Atmospheric Pollution Prevention and Control Action Plan” on 12 September 2013. In this action plan, the control of vehicle emission was a significant point. For example, limit the vehicles possession, optimize the transportation structure and management, promote the upgrading of related industries and products, and improve vehicle emission efficiency. Those main control policies mentioned above were closely connected with the traffic influencing factors in this paper, which were also coincident with the quantitative evaluation results in this work. So, this work also can be useful for Chinese government to improve relevant policy and control strategy to intervene and eliminate the occurrence of motor vehicle pollutants in China.

5.2. Policy Implications

The above conclusions have significant policy implications.

First, given that the emission efficiency of traffic pollutants is a key factor for all three major vehicle pollutants, it is necessary for the Chinese government to improve the technical level of vehicle emissions reduction. The improvement of vehicle emission reduction technologies is mainly achieved by improving the quality of oil products. At present, the main problems in China are insufficient supervision of oil quality and a lack of strict penalties. In the implementation of strict fuel standards, it is necessary to match vehicle exhaust emission standards, improve the refining level through technical support, and strengthen the enforcement of fuel standards.

Second, the Chinese government should strengthen the supervision of trucks and strictly control the pollutant emissions of trucks. Trucks have a far-reaching impact on China’s vehicle NOx emissions at the national level, and the impact on HC and CO is also obvious. Although China has implemented regular truck inspections, it lacks a road supervision mechanism. Emissions exceeding the standards have led to increasing NOx emissions from trucks in China. At present, the vehicle scrapping system lacks mandatory control measures and there are low standards for scrap subsidies. Therefore, it is necessary for the Chinese government to formulate regulatory measures and continue to promote the yellow-label vehicle retirement system.

Third, in the process of promoting urbanization, the Chinese government needs to make reasonable road system planning. The empirical results have shown that the urbanization rate plays an important role in promoting HC and CO emissions. Urbanization in China has resulted in unreasonable urban spatial layouts and a lack of suitable transportation systems. The Chinese government should change the concept of urban development and abandon car-oriented urban planning, construction, and management concepts in order to build a sustainable urban development model.

Finally, the Chinese government should vigorously develop rail and water transport infrastructure to reduce the volume of road transport. The empirical results have shown that road passenger turnover and road freight turnover have significant promotional effects on the NOx, HC, and CO emitted by vehicles. National railways have sustained losses in recent years [46]. In terms of railway transportation, the Chinese government should reduce the cost and improve efficiency. In terms of water transport, China’s water transport market access management model still adheres to the approval system mode; however, this pattern requires further improvement.

It also should be noticed that only one significant key factor category (transportation influencing factors) was considered in this work and identified the heterogeneity of different emission zones. Based on the previous research mentioned above, transportation influencing factors were undoubtedly the key factors that lead to severe air pollution in many provinces in relevant countries. However, there are still some other important factors including meteorological condition can be taken into account [47]. Therefore, it would be meaningful to comprehensively analyze the influential factors of motor vehicle pollutants from different perspectives (such as meteorological conditions), which may be carried out in our future work.