Impact of New and Old Driving Force Conversion on Air Quality: Empirical Analysis Based on RDD

Abstract

This paper evaluates the impact of the conversion of new and old driving forces of economic growth on air quality. Based on the policy shock of the establishment of the Shandong pilot zone, this paper took the monthly air quality index (AQI) as a measurement indicator and adopted the Regression Discontinuity Design (RDD) method. It was found that the conversion of new and old driving forces significantly improved air quality, and this effect was still robust for different bandwidths. The examination of individual pollutants revealed that the improvement in air quality (AQI) came mainly from the reduction in the levels of five pollutants in the air: PM_2.5, PM₁₀, SO₂, CO, and NO₂. Moreover, through the convergence analysis, it was found that the air pollution governance effect of the conversion of new and old driving forces had significant $\beta$ convergence characteristics. The convergence characteristics were still robust even considering control variables and spatial factors, thus indicating that the conversion of new and old driving forces had a synergistic governance effect on air pollution control.

1. Introduction

Since the implementation of the reform and opening-up policy, China’s economic development has marked great achievements. However, such extensive economic development has also led to serious environmental problems, particularly air pollution [1,2]. According to authoritative statistics, in 2020 a total of 135 Chinese cities, out of 337, accounting for 40.1% of the total, failed to meet the air quality standard. Furthermore, the concentration of ${\mathrm{P}\mathrm{M}}_{2.5}$ in one-third of Chinese cities did not achieve the national secondary standard, with frequent episodes of heavy pollution weather events at a regional level.

The negative externalities caused by environmental pollution demonstrate the market failure in the governance of private sectors’ pollution. Therefore, it has become a common choice for governments to provide this public good (i.e., a high-quality environment) to residents. In the Chinese political and economic context, the environmental protection policies formulated by the central government are mainly implemented by local governments, as is the use of environmental protection funds. However, given the long-term emphasis given in China to the prioritization of economic construction, local governments normally choose a strategy of extensive development to achieve short-term economic growth at the expense of environmental quality. In recent years, China’s economy has slowed down and shifted; the original extensive growth momentum has been declining, and the factor-driven development model is currently unsustainable. In terms of industrial structure, those industries characterized by high consumption, high pollution, high emissions, and low value-added have been gradually replaced by those in surrounding developing countries, such as Vietnam and Cambodia, characterized by lower labor costs. As such, China’s economy began to shift from an extensive growth model, characterized by excessive capital accumulation and environmental plundering, to an innovation- and human capital-based model, such that conversion is now the driving force of economic growth.

In January 2018, the State Council of China approved the “Overall plan for the construction of Shandong conversion of new and old driving force comprehensive pilot zone”, which is the first regional development strategy for the construction of a comprehensive pilot zone including the conversion of new and old driving forces in China. The conversion of new and old driving forces is a concept of development economics; it refers to the conversion of driving forces in the process of economic growth of a country. In industry, it mainly refers to the introduction of new technology, materials, and energy to replace old ones. This pilot zone aims to achieve coordinated economic, social, and environmental development; accordingly, phased goals have been set, that is, by 2020 emerging industries (new information technology, new energy, and new materials) will be developed by dissolving excess production capacity and eliminating backward production capacity (mainly related to steel, coal, electrolytic aluminum, thermal power, building materials, and other industries). Meanwhile, the plan also clearly points out that the pilot area will force enterprises to make a green transformation with higher pollution emission standards, thus further improving the regional linkage mechanism and the collaborative governance policy of air pollution prevention and control, realizing the initial conversion of new and old driving forces of economic growth, and promoting high-quality economic development. In this context, this study aimed to evaluate the environmental effect of the conversion of new and old driving forces from the perspective of air pollution governance.

High-consumption, high-pollution crude economic growth will bring an unbearable load to the environment and resources. Changing the economic growth momentum and promoting economic structural transformation is the fundamental way to achieve an environment-friendly and resource-saving economy. Since the establishment of the pilot zone for the conversion of new and old driving forces, whether it has produced good environmental benefits has not yet been effectively proven. In important areas and key links, the policy effectiveness achieved in the pilot zone will have significant policy reference significance for China and the world’s economic growth power transformation. In view of this, this paper uses air quality as the entry point and adopts a Regression Discontinuity Design (RDD) to empirically test whether the establishment of the pilot zone has produced good environmental benefits and provide new empirical evidence for the next policy direction setting.

2. Policy Background and Literature Review

2.1. Policy Background

The experience of economic development at the global level showed that if a country can achieve a perfect transition between the traditional and the emerging driving forces of economic growth, then it can achieve long-term economic development; otherwise, it will fall into the “middle-income trap”. With the launch of a new round of global scientific and technological revolution and industrial reform, China, as the largest developing country, began to optimize its economic structure and transform its growth driving force, changing from a high-speed to a high-quality economic development model. In 2017, the general office of the State Council of China issued the “Opinions on innovating management, optimizing services, cultivating and expanding new driving forces of economic development, and accelerating the continuous conversion of new and old driving forces of economic development”. These Opinions pointed out that accelerating the cultivation and development of new driving forces, and transforming and upgrading the traditional driving forces of economic development are two important ways to promote economic structural transformation and real economic upgrading. The government encourages qualified places to try first, radiating a new vitality from traditional driving forces and creating regional brands of new industries and new formats.

The Shandong Province is a strategic node of the opening-up from south to north, as well as of the gradient of development from east to west in China, and occupies an important position in the regional development pattern. Its economic structure is highly similar to that of the whole country; its industrial structure is generally based on heavy industries, and traditional industries still occupy a leading position. In addition, the total amount of emerging industries is lower and the role of new driving forces in supporting economic growth has not been brought into full play; hence, the tasks of resolving excess capacity and eliminating backward capacity remain arduous. To address these issues, in January 2018 the State Council of China approved the “Overall plan for the construction of Shandong’s comprehensive pilot zone for the conversion of new and old driving forces”, taking Shandong Province as a pilot area for the conversion of new and old driving forces of national economic growth. The key task of this initiative is to convert the driving forces of economic growth, with the policy goal to eliminate the excess and backward capacity in industries such as iron and steel, coal, electrolytic aluminum, thermal power, and building materials before 2020, encourage some heavily-polluting and highly resource-depleting industries to withdraw from the market and focus on cultivating and expanding new technologies, new industries, new business formats, and new modes. Initially, a scientific and effective pattern is planned to be established to transform and upgrade traditional industries, to obtain replicable and popularized pilot experience in the conversion of the driving forces of economic growth for the whole country.

As a set of methodologies summarized in China’s incremental reform, this policy pilot plays a unique advantage in promoting China’s economic and social development, as well as political and institutional innovation [3]. Its objectives are to encourage qualified places to try first for national policy and innovate boldly; then, the experience of pilot policy implementation will be summarized and finally promoted from one region to the whole country. Therefore, it is particularly important to evaluate the phased policy effect in the pilot area.

2.2. Literature Review

For a long time, scholars have been studying how to weigh the impact of economic growth on the environment. The “Environmental Kuznets Curve (EKC)” hypothesis is an important theory to describe this relationship, creatively put forward by Grossman and Krueger in the early 1990s. These authors argued that there is an inverted “U-shaped” correlation between pollution emissions and per capita GDP. Subsequently, several scholars carried out a large number of studies around the Environmental Kuznets Curve. Selden and Song found that four types of air pollutants (i.e., ${\mathrm{S}\mathrm{O}}_2$, ${\mathrm{C}\mathrm{O}}_2$, ${\mathrm{P}\mathrm{M}}_{2.5}$, and ${\mathrm{N}\mathrm{O}}_2$) had an inverted “U-shaped” correlation with per capita GDP, thus verifying the existence of the EKC for gaseous pollutants [4]. By calculating the air pollution index of 122 countries, Buehn further supported the hypothesis of the inverted “U-shaped” correlation between economic growth and environmental pollution [5]. However, some scholars argued that the EKC hypothesis is too restrictive. By including spatial factors and investigating 290 cities in China, He found heterogeneity in the EKC curve across different pollutants and regions [6]. Further studies found a significant correlation between economic growth and air pollution, although not necessarily an inverted “U-shaped” correlation. Along with the economic and social development, a “Re-link Effect” was found between economic growth and environmental pollution, and other curve relationships may exist such as “U-shaped”, “N-shaped”, or inverted “N-shaped” curves [7].

In recent years, the world has entered a deep adjustment period. New technologies, new industries, new formats, and new models are at the critical point of generating an industrialization breakthrough, and a new round of industrial revolution is rapidly developing. In this context, the conversion of new and old driving forces has become the fundamental strategy for China to enter a new era of economic development. Yang defined the new and old driving forces of economic growth from the perspective of alternating processes and analyzed the dynamic evolution of the conversion of new and old driving forces from the two-dimensional perspective of technological efficiency and technological progress [8]. The existing literature shows that since the opening-up and reform, China has experienced two conversions of the driving forces of economic growth. The first conversion began with China’s accession to the World Trade Organization (WTO)in 2001, whereby China changed from a domestically oriented to an export-oriented economy. The second conversion started with China’s “four trillion” investment after the global financial crisis in 2008, by expanding domestic demand to promote steady economic growth. Each conversion of the driving forces of economic growth has obvious historical characteristics [9]. Li deeply explored the conversion of new and old driving forces of Chinese economic growth by building a comprehensive index and evaluating the present historical stage and future trends for each province [10]. Pei summarized a series of important assertions by General Secretary Xi Jinping, pointing out that the current round for the conversion of driving forces is focused on improving the quality and efficiency of economic growth, reducing the supply of industries with low added value, high emissions, high pollution, and high consumption, and advocating resource conservation and environmentally-friendly development, that is, green development [11]. Furthermore, by assessing the characteristics of spatial-temporal evolution between green development and the conversion of driving forces, Gai found that the level of coupling and coordination between the two is increasing, although with spatial heterogeneity [12].

The existing literature generally investigated the relationship between economic growth and environmental pollution, and research on the policy-shock effect brought by the conversion of the driving forces of economic growth is insufficient, especially on environmental pollution reduction. Most studies remained at a theoretical level of analysis and lacked empirical research. The purpose of this study was to deeply explore the governance effect of the conversion of new and old driving forces in economic growth from the perspective of air pollution. The contribution of this study lies in the fact that it complements existing literature on the empirical analysis of this field, introducing the RDD into the policy evaluation of the conversion of new and old driving forces, thereby further expanding the research perspectives of this field.

3. Research Design

3.1. Econometric Equations

The Regression Discontinuity Design (RDD) was used in this study to evaluate the environmental governance effect of the conversion of the driving forces of economic growth. The RDD has been widely used in air pollution literature, including taking time as a breakpoint to investigate whether there is a sudden change before and after a shock event [13,14,15,16]. The basic idea of Regression Discontinuity Design (RDD) is that there is a continuous variable, and under the influence of a certain exogenous shock policy, the continuous variable gets different treatment probabilities on both sides of the critical point, which also fits the purpose of this paper, and this idea is the reason for adopting the RDD in this paper. Specifically, the dependent variable of this paper, the air quality index, as a continuous variable, has a sudden change before and after the establishment of China’s New Old Power Conversion Pilot Zone, and other factors affecting air quality are continuously changing before and after, then we have reasons to believe that the sudden change in air quality is caused by the establishment of the pilot zone. This idea of quasi-natural experiments can greatly avoid the endogeneity problem caused by reverse causality in traditional econometric models, and can effectively analyze the treatment effects caused by policy interventions or unexpected external events. Following the method employed by Lee and Lemieux [17], the estimation equations of the Local Average Treatment Effect (LATE) of regression discontinuity, that is, the average experimental effect at the breakpoint, were expressed as follows:

$$AQ{I}_{cd}=b_0+b_{1}experimen{t}_{cd}+b_{2}f(x)+b_{3}experimen{t}_{cd}f(x)+l{X}_{cd}+e_{cd}$$

(1)

$$x=d-d_0$$

(2)

$$d_0-h\le x\le d_0+h$$

(3)

where c represents the city; d represents the month; $d_0$ indicates the month of the establishment of the pilot zone (i.e., January 2018); $h$ represents the bandwidth; ${AQI}_{cd}$ is the air quality index of city $c$ in month $d$; and ${experiment}_{cd}$ is the dummy variable (treatment variable) representing the establishment of the pilot zone. 1 means after the establishment month $d_0$ of the pilot zone in city $c$ and 0 means before the establishment month $d_0$ of the pilot zone in the city $c$. $x$ is the runner variable, which was used to represent the number of months from the establishment of the pilot zone; and $f\left(x\right)$ is a polynomial function with $x$ (runner variable) as the independent variable. $X_{cd}$ indicates the control variables, including the weather control variable and the time control variable. The former includes monthly average temperature and monthly average precipitation, which were used to control the impact of changes in weather factors on air quality, while the latter includes legal holidays, and was used to control the impact of residents’ working time arrangement on air quality. ${\epsilon }_{cd}$ is the random disturbance term. ${\beta }_1$ in Equation (1) is the main concern of this study, as it captures the difference in air quality before and after the establishment of the pilot zone.

3.2. Sample Data

This study covered the period from 1 January 2016 to 31 December 2019. The 16 pilot cities undergoing the conversion of the new and old driving forces of economic growth were identified according to the “Overall plan for the construction of Shandong’s comprehensive pilot zone for the conversion of new and old driving forces”, approved by the State Council. The source of air quality data was the AQI published by the data center of the Ministry of Environmental Protection of China. The AQI is a comprehensive index reflecting the overall pollution degree, and including six pollutants, namely ${\mathrm{P}\mathrm{M}}_{2.5}$, ${\mathrm{P}\mathrm{M}}_{10}$, ${\mathrm{S}\mathrm{O}}_2$, $\mathrm{C}\mathrm{O}$, ${\mathrm{N}\mathrm{O}}_2$, and ${\mathrm{O}}_3$. Increasing values of the AQI indicate a worsening of air quality, while decreasing values indicate improvements in air condition. The original temperature and precipitation (mm) data come from the National Meteorological Science Data Sharing Service Platform—China Surface Climate Data Daily Dataset (V3.0), which were processed for monthly average treatment using the Barnes method and the Inverse Distance Weighted (IDW)method, respectively. In addition, the control variables of legal holidays were set according to the holiday arrangements issued by the General Office of the State Council in China over the years. The descriptive statistics of all variables are presented in

Table 1. Descriptive statistics of the variables.

Variable	Obs.	Mean	Std. Dev.	Min.	Max.
AQI	768	98.779	25.644	39	192
x	768	−0.5	13.862	−24	23
experiment	768	0.5	0.500	0	1
temperature	768	13.873	9.750	−3.551	28.250
precipitation	768	62.298	60.575	2.319	313.153
holiday	768	0.5	0.500	0	1

.

4. Empirical Results

4.1. Benchmark Regression Results

Table 2. Benchmark Regression Results.

	Local Polynomial Regression
	Benchmark Regression	Adding the Control Variable
${\beta }_1$	−32.985 **	−50.362 ***
	(14.868)	(14.383)
Control Variables	No	Yes
Bandwidth	5.248	5.248
Sample Size	768	768

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

shows the estimation results of the RDD. In this paper, the commonly used minimization mean square error (MSE) method was used to select the optimal bandwidth, a time window of 5 months (5.248) before and after was determined, and the optimal lag order of the RDD polynomial estimation was determined to be 2nd order according to the AIC and BIC information criteria (see

Table 3. Polynomial Regression Order Selection.

	Second Order	Third Order	Fourth Order
AIC	446.295	584.213	586.353
BIC	451.909	592.849	594.988

for details). The regression results in Table 2 show that the establishment of the test area has a significant negative effect on the AQI, i.e., it significantly reduces the air pollution level and improves the air quality condition, and the results remain robust after adding control variables such as weather factors and holiday schedules, which implies that the establishment of the test area is likely to have improved the air quality to some extent.

While conducting regression estimation, this paper draws scatter plots based on the values before and after the breakpoints, and the trend of the non-parametric fit plots in Figure 1 shows that the second-order polynomial fit has a significant breakpoint at the time point of the test area establishment, and the fit plots further verify that it is appropriate to use breakpoint regression to estimate the local treatment effect of the test area establishment on air pollution control in this paper.

4.2. Robustness Tests

4.2.1. Bandwidth Sensitivity Analysis

Bandwidths may affect the robustness of RDD estimation results [18]. Therefore, this study established bandwidths of 6, 8, 10, and 12 months before and after the establishment of the pilot zone for robustness analysis. The estimation results are shown in

Table 4. Bandwidth Sensitivity Analysis.

	Selection of Results for Different Bandwidths
${\beta }_1$	−61.399 **	−56.456 ***	−70.997 ***	−60.141 ***
	(24.029)	(17.099)	(13.588)	(11.465)
Control Variables	Yes	Yes	Yes	Yes
Bandwidth	6	8	10	12
Sample Size	768	768	768	768

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

. The regression results indicate that the establishment of the pilot zone had a significant negative impact on the AQI across all bandwidths, apparently notably reducing the degree of air pollution and improving air quality. This also shows that the regression results of this study are also robust across different bandwidths.

4.2.2. Validity Test: Sample Non-Random Test

The random distribution of samples is the key component of the RDD. Only if the sample is randomly distributed in the field of cut-point, the average treatment effect (ATE) at the critical value estimated by the model is the unbiased estimator of policy implementation for the differences between the treatment group and the control group.

With regard to the execution variable, this study used the sample non-random test method proposed by Bugni to verify whether the execution variable had an accurate manipulation effect near the critical value, by constructing the statistics of the $g$ order [19]. Compared to other methods, this method can be more asymptotically effective under weaker conditions. When the execution variable meets the continuous distribution, it satisfies the local randomization assumption of RDD. The original hypothesis of this test was that the density function of the executive variable is a continuous function at the critical value. In this study, the executive variable is $x$ (i.e., the number of months before the establishment of the pilot zone); the test results are shown in

Table 5. Sample Non-Random Test.

RDD Non-Randomized Approximate Sign Test
Running variable:	X
Cutoff c = 0	Left of c	Right of c	Number of obs. = 768
Number of obs.	384	384
Eff. number of obs.	74	86
Eff. neighborhood	−5	5
p-value	0.385

. The results show that the p-value of the regression discontinuity non-random test of these samples at the critical value was 0.160. This implies that the null hypothesis cannot be rejected; it also indicates that the density function of the executive variable $x$ is a continuous function at the critical value, and there is no human intervention.

In addition, the RDD continuity design assumes that other characteristic variables or control variables are smooth at the critical value. Therefore, based on the available data, this study carried out a continuity test on covariates with a certain degree of explanation force (i.e., precipitation and temperature), that is, covariates were used to replace the respective dependent variables in Equation (1); the results are shown in

Table 6. Covariate Jump Test.

X	Coef.	Std. Err.	z	P > z
precipitation	0.448	8.566	0.052	0.958
temperature	−0.448	0.810	−0.553	0.580

. The change of these variables at the discontinuity point was not significant and was without a certain upward or downward trend. For these variables that should not be affected by the establishment of the pilot zone, the policy-shock effect had no significant results on these variables; this further indicates that the RDD adopted in this study is appropriate.

4.3. Further Analysis

4.3.1. Analysis of Single Pollutants

From the results in

Table 7. Regression Results of Itemized Pollutants.

	PM2.5	PM10	SO2	CO	NO2	O3
Experiment	−32.67 *	−70.215 ***	−14.578 *	−0.756 ***	−20.516 ***	64.463 ***
	12.658	19.570	6.511	0.251	6.860	3.578
Bandwidth	5.248	5.248	5.248	5.248	5.248	5.248
Sample Size	768	768	768	768	768	768

Note: *** p < 0.01, * p < 0.10.

, it can be seen that the improvement of air quality (AQI) mainly comes from the reduction of the five pollutants ${\mathrm{P}\mathrm{M}}_{2.5}$, ${\mathrm{P}\mathrm{M}}_{10}$, ${\mathrm{S}\mathrm{O}}_2$, $\mathrm{C}\mathrm{O}$, and ${\mathrm{N}\mathrm{O}}_2$ in the air, compared to the increase of ${\mathrm{O}}_3$ pollution in the air instead of reduction. Specifically, in near-earth cities, ${\mathrm{O}}_3$ is mainly produced by the chemical reaction between nitrogen oxides and oxygen from automobile exhaust and industrial emissions, indicating that there are still problems of substandard or even illegal emissions in industrial production, and the air pollution caused by automobile exhaust has become more serious with the growth of automobile use, which also indicates that the popularity of new energy electric vehicles in China still needs to be improved (as of 2020 It also shows that the popularity of new energy electric vehicles in China still needs to be improved (less than 2% of the total vehicle ownership in 2020).

4.3.2. Test of the Synergistic Governance Effect

The results of the Regression Discontinuity showed that the conversion of new and old driving forces had a significant effect on air pollution governance. However, considering the spatial fluidity and spillover of air pollution [20], joint prevention and control must be carried out among different regions to improve the air quality of all regions. Therefore, the air pollution governance effect observed from the analysis of this study may be characterized by a synergistic governance effect. From the perspective of the phenomenon, the existence of such a synergistic governance effect indicates that the regions with a high level of pollution governance drive those with a low level. At the same time, it also shows that despite different air pollution governance levels, air quality improved in all regions, and the governance level between them tended to narrow.

Following the economic convergence theory in the neoclassical growth theory, this study selected the $\beta$ convergence model to test the existence of a synergistic governance effect of air pollution, which would be proved by the observation of a convergence characteristic. As the pilot zone was established in January 2018, this study selected samples after this time point (i.e., from January 2018 to December 2019) to test whether the air pollution governance effect caused by the conversion of new and old driving forces had a synergistic governance effect. The AQI was used to indicate the degree of air pollution; higher values indicate greater air pollution. Based on previous analysis, the reciprocal of the AQI, i.e., IAQI, was calculated and recorded as a proxy variable of the treatment effect. As IAQI is equal to 1/AQI, greater air pollution during the sample period would indicate a worse treatment effect and a lower IAQI value. In contrast, lower air pollution would indicate a greater treatment effect and a higher IAQI value.

$\beta$ convergence includes absolute convergence and conditional convergence. The absolute $\beta$ convergence model was constructed as follows:

$$\ln \left(\frac{IAQ{I}_{i,t+1}}{IAQ{I}_{it}}\right)=\alpha +\beta \ln \left(IAQ{I}_{\mathrm{it}}\right)+{\nu }_t+{\mu }_i+{\epsilon }_{it}$$

(4)

On the left side of Equation (4), the logarithmic difference was used to calculate the growth rate of IAQI, while on the right side, $v_t$ indicates the time-fixed effect, $u_i$ indicates the urban fixed effect, and ${\epsilon }_{it}$ is a random error term.

In conditional $\beta$ convergence, several control variables are added to absolute $\beta$ convergence. Based on data availability, this study added monthly average temperature and monthly average precipitation as control variables and constructed the following $\beta$ convergence model equation:

$$\ln \left(\frac{IAQ{I}_{i,t+1}}{IAQ{I}_{it}}\right)=\alpha +\beta \ln \left(IAQ{I}_{\mathrm{it}}\right)+{\gamma }_1\ln (temperatur{e}_{it})+{\gamma }_2\ln (precipitatio{n}_{it}){\upsilon }_t+{\nu }_t+{\mu }_i+{\epsilon }_{it}$$

(5)

For both absolute $\beta$ convergence and conditional $\beta$ convergence, if $\beta$< 0 and passes the significance level test, there will be a convergence trend in the effectiveness of air governance in these regions, otherwise there will be a divergence trend.

Based on Equation (4), this study used the panel fixed effect to test the $\beta$ convergence mechanism of the air pollution degree in the pilot zone. The results of the absolute convergence mechanism test, presented in

Table 8. Results of the $\beta$ Convergence Test.

	Absolute Convergence	Conditional Convergence	Spatiotemporal Convergence
β	−0.7159 *** (0. 0406)	−0.7131 *** (0. 0403)	−0.5474 ** (0.1898)
Constant Term	3.2705 *** (0. 2041)	3.1187 *** (0.2179)
Control Variables	NO	YES	YES
Annual Fixed Effect	YES	YES	YES
Urban Fixed Effect	YES	YES	YES
R2	0.6144	0.6552	0.2003

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

, show that the absolute convergence coefficient $\beta$ in the pilot zone was negative and passed the 1% statistical significance test. This in turn indicates that there was $\beta$ absolute convergence of the air pollution governance effect.

Furthermore, taking regional weather factors into account, based on Equation (5), this study employed the two-way fixed effect to test the $\beta$ conditional convergence. The results are shown in Table 8. It is found that the convergence coefficient $\beta$ passes at least a 1% significance level test and the result is negative, which indicates that under the condition of considering other influencing factors such as control variables, the effect of air pollution governance still has significant convergence characteristics. The result shows that over time, the air quality has been improved and the governance level of air pollution in different regions tends to be narrow.

The abovementioned $\beta$ conditional convergence test was based on the premise that the research samples are independent of each other. However, as air pollution has spatial fluidity, the air pollution governance effect among the cities investigated may be geographically dependent. Therefore, this study further constructed the $\beta$ conditional convergence test equation under the Spatial Panel Durbin Model (SDM). The corresponding equations are as follows:

$$\ln (\frac{IAQ{I}_{i,t+1}}{IAQ{I}_{it}})=\alpha +\beta \ln IAQ{I}_{it}+{\beta }_1{W}_{ij}\ln IAQ{I}_{it}+\delta X_{it}+\eta W_{ij}{X}_{it}+{\mu }_i+{\upsilon }_t+{\epsilon }_{it}$$

(6)

$${\epsilon }_{it}=\lambda W_{ij}{\epsilon }_{it}+{\eta }_{it}$$

(7)

The geographic distance weight matrix was adopted in this study; the spatial weight matrix $W_{ij}$ was expressed as follows:

$$W_{ij}=\{\begin{array}{ll}0& i=j\\ 1/d_{ij}& i\ne j\end{array}$$

(8)

In the geographical distance weight matrix of Equation (8), $d_{ij}$ represents the geographical distance between cities $i$ and $j$, which was measured by the distance between urban geographical centers. Accordingly, the maximum likelihood method proposed by Lee and Yu [21] was used to test the $\beta$ conditional convergence of air pollution; the results are shown in Table 7. It can be seen that the value of conditional convergence coefficient $\beta$ under spatial econometrics was significantly negative; this means that after adding spatial factors, the convergence conclusion was still robust, that is, it still showed the $\beta$ convergence characteristics.

5. Discussion

Economic growth may generate serious environmental pollution problems, especially in relation to air pollution. Changing the current model of economic growth and promoting the conversion of new and old driving forces are fundamental to improving environmental quality. This is consistent with the research conclusions of Hu and Zhou [22], Qin and Hu [23], Zhao [24], and Xiao [25]. These authors believe that the extensive model of economic growth is no longer applicable to the current development stage of China and that it is essential to change the driving force of economic growth and explore a green and sustainable growth pattern. Since 2018, China has begun to change the driving forces of its economic growth by setting up a comprehensive pilot zone for the conversion of new and old driving forces. There is no consensus in the existing literature about whether the establishment of this pilot zone has brought significant environmental improvement effects. Hence, this study used the RDD to evaluate this policy effect, demonstrating that the establishment of the pilot zone for the conversion of new and old driving forces has significantly improved air quality.

Different from the traditional EKC study, from the fitting trend of the scatter plot in Figure 1, there are obvious breakpoints at the time point of the establishment of the pilot zone (January 2018), i.e., there may be a breakpoint correlation between the transformation of economic growth dynamics and environmental benefits, combined with the regression results, the establishment of the pilot zone has a significant negative impact on the air quality index, and the conclusions are more robust under different bandwidth choices, which also complements the relevant studies on the relationship between economic growth and environmental benefits.

In the analysis of individual pollutants, it was found that the improvement of air quality (AQI) in the pilot zone mainly came from the reduction of the content of five pollutants in the air, namely ${\mathrm{P}\mathrm{M}}_{2.5}$, ${\mathrm{P}\mathrm{M}}_{10}$, ${\mathrm{S}\mathrm{O}}_2$, $\mathrm{C}\mathrm{O}$, and ${\mathrm{N}\mathrm{O}}_2$. After the establishment of the pilot zone, the local government promoted the solution of excess capacity in steel, coal, electrolytic aluminum, thermal power, and building materials as a key task, and guided various heavy-polluting industries to exit the market in an orderly manner in the short term. The regression results show that such measures play a significant role in reducing the emissions of PM_2.5, PM₁₀, SO₂, CO, and NO₂. Further tests reveal the spatial synergy of the air pollution control effects brought about by the shift in economic growth dynamics. Specifically, this paper uses convergence tests and finds that the air pollution control effect has an absolute convergence feature, which is still robust when weather and spatial factors are taken into account. It indicates that the regions with higher air pollution control levels drive the regions with lower air pollution control levels, and the spatial synergistic control effect exists as the control level between them tends to narrow while maintaining the improvement of air quality.

6. Conclusions and Policy Recommendations

This study focused on the policy-shock effect of the establishment of the comprehensive pilot zone for the conversion of new and old driving forces and used the RDD to evaluate its effect on air quality. The results showed that the conversion of new and old driving forces significantly reduced the degree of air pollution and improved air quality. This improvement effect mainly came from the reduction of ${\mathrm{S}\mathrm{O}}_2$ and $\mathrm{C}\mathrm{O}$ emissions. Furthermore, through the convergence analysis, it was found that the air pollution governance effect had significant $\beta$ convergence characteristics. In addition, when considering the control variables and spatial factors, the convergence characteristics were still robust, which indicates that the conversion of new and old driving forces had a synergistic governance effect on air pollution control.

The conclusions of this study have the following policy implications. Firstly, under the background of a new round of global scientific and technological revolution and industrial reform, characterized by a new trend of multi-field and interdisciplinary breakthroughs, China’s economy has changed from high-speed growth to high-quality development. China is currently in a key moment of transformation of its development pattern, optimization of the economic structure, and change in its growth drivers. The empirical results of this research showed that the conversion of new and old driving forces can significantly improve air governance. This means that the conversion of new and old driving forces in economic growth is not only a replacement process of old driving forces for new driving forces but also a process of promotion of green economic development. Moreover, the conversion of new and old driving forces will help China develop into a resource-saving and environmentally-friendly society. Therefore, local governments must seize the opportunity to promote the growth and development of new technologies, new industries, new formats, and new models, injecting new driving forces into economic development and promoting the conversion of the new and old driving forces to a higher development level. Secondly, this paper finds that the air pollution control effect is mainly reflected in the five pollutants PM_2.5, PM₁₀, SO₂, CO, and NO₂, while the content of O₃ in the air is still found to be rising. Therefore, the governance process of air pollution cannot be generalized, and local governments should not only focus on the treatment of different single pollutants and carry out joint prevention and control, but also deeply analyze the causes of different single pollutants and periodically evaluate the effectiveness of their governance. Thirdly, in the process of promoting regional air pollution governance, local governments should focus on joint prevention and control across regions, and pay more attention to the convergence or divergence characteristics of regional governance effect development. In cases of convergence characteristics in these regions, local governments should pay attention to the gap in the governance effects among regions and, subsequently, analyze the comparative advantages of these different regions and accurately implement policies. In parallel, they should design specific measures in line with local development conditions, focusing on improving the governance level of high-pollution regions, and optimizing the regional linkage mechanism between air pollution prevention and governance.