INTERNATIONAL JOURNAL OF

ENVIRONMENTAL RESEARCH
AND PUBLIC HEALTH

Article
The Relationship between Economic Growth and
Air Pollution—A Regional Comparison between
China and South Korea
Min Jiang, Euijune Kim * and Youngjin Woo
Department of Agricultural Economics and Rural Development and Research Institute of Agricultural and Life
Sciences, Seoul National University, Seoul 08826, Korea; [email protected] (M.J.); [email protected] (Y.W.)
* Correspondence: [email protected]

Received: 6 March 2020; Accepted: 14 April 2020; Published: 16 April 2020

Abstract: This paper analyzes the interaction between regional economic growth and air pollution
in China and Korea. The relationship between gross regional product per capita and industrial
emission of sulfur dioxide emission is examined at the regional level using simultaneous equation
models covering 286 cities in China and 228 cities and counties in South Korea of the period
2006–2016. The results find that regional di erences existed in the relationship between air
pollution and economic growth in two countries. In both countries, an inverted U-shaped pattern
was found in metropolitan areas while a U-shaped pattern of non-metropolitan areas. Although the
emissions of pollutants in metropolitan areas of both countries have shown a downward trend in
recent years, there is still a large gap between the overall emission levels of China and South
Korea. Moreover, the level of pollutant emissions of China’s metropolitan areas is much higher
than in non-metropolitan areas, while the opposite result has occurred in Korea. In China, there
was an inverted U-shaped relationship of the eastern and northwest region, while U-shaped
relationships existed in the southwest, central and northeast regions.

Keywords: air pollution; regional economic growth; international comparison; China; South Korea

1. Introduction
The relationship between economic growth and environmental pollution has been a longstanding
global concern since the 1970s. According to the Environmental Kuznets Curve (EKC) theory first
proposed by Grossman and Krueger [1] and Panayotou [2], there is an inverted u-shaped relationship
between income level and environmental degradation. In the first stage, economic growth is associated
with environmental deterioration. At this stage, the increase in industrial activity in countries with lower
levels of economic development leads to an increase in energy-intensive production and an increase in
pollutant emissions. With the development of services and knowledge-based technology-intensive
industries, the environmental degradation trend gradually declines due to environmental awareness,
coupled with changes in production and stricter environmental regulations. The EKC literature suggests
that economic growth may affect environmental welfare through three different channels: scale effects,
composition effects and technique effects. The growth of the economic scale would result in a
proportional growth in environmental pollution, and the changes in the industrial structure would lead to
the reduction of pollution intensity [3]. Further economic growth causes technological progress through
which dirty and obsolete technologies are replaced by upgraded and cleaner technologies that improve
environmental quality [4]. Extensive previous works found that the impacts of economic growth on
environmental pollution can be generally divided into four relationships: an inverted U-shaped relationship
[1], a monotonically increasing relationship [5–10], a U-shaped relationship [4,9,11–14] and a N-shaped

Int. J. Environ. Res. Public Health 2020, 17, 2761; doi:10.3390/ijerph17082761 www.mdpi.com/journal/ijerph
Int. J. Environ. Res. Public Health 2020, 17, 2761 2 of 20

relationship [1,15–17]. However, the reverse effect of the environment on economic growth has
received relatively little attention. Another limitation of existing research is that most of the previous
research discussed the relationship between economic growth and environmental pollution from the
national level viewpoint, with little attention paid to regional perspectives.
Clarifying the relationship between environmental pollution and economic growth at the regional
level is critical for two reasons. First, the regional economic gap is a key obstacle to effective control of
pollutant emissions because different economic conditions and industrial structures are closely related to
local emission patterns and trends. The regional economic gap determines different economic growth
patterns and industrial structures. For instance, Beijing, the capital city and a first-tier city in east China,
had a relatively high level of gross regional product (GRP) per capita of US$ 19,769 in 2018. The service
industry was highly developed, accounting for 81.0% of the total GRP, and the manufacturing industry
only accounted for 18.6%. In the three typical third-tier cities of east, west and central China, namely
Shantou city, Xianyang city, and Chenzhou city, the per capita GRPs in 2018 were US$ 6308, US$ 8261
and US$ 7182, respectively, with manufacturing accounting for 50.8%, 56.9% and 46.1% of the total
GRP. In the more developed regions, manufacturing accounts for a relatively small proportion of GRP
and pollutant emissions are relatively low. However, in less developed areas, manufacturing enterprises
are still welcomed because they are needed to stimulate the local economy and create more jobs [18–
20]. Manufacturing industries would exert more significant impacts on air pollution than other industries
due to relatively intensive energy use in the production process and high pollutant emissions. Thus, the
regional economic gap and difference in industrial structures will inevitably lead to varying air pollution
across regions. Second, the intensities of China’s environmental regulations vary from region to region.
For instance, the pollutant discharge fee rate of air pollutants in Beijing was about US$ 1.73 per unit of
pollution; however, the pollutant discharge fee in the western provinces, such as Shaanxi, Ningxia,
Henan, was only about US$ 0.17. Strict environmental regulations could force the emission-intensive
industries to transfer from the eastern regions to the central or western regions of China with relatively
less stringent environmental regulations which are regarded as “pollution paradise” [21–23]. Therefore,
regional economic development levels and environmental regulations vary from region to region, resulting
in a gap between the pollution levels. Regarding the relationship between economic growth and
environmental pollution, national-level analysis provides only a general understanding of how variables
are broadly related and thus provide little guidance for policymaking [24]. Understanding different patterns
of economic growth and environmental pollution at the regional level and determinant factors of local
pollution will help to make decisions on regional pollution control policies based on regional realities.

Different from previous studies, this paper is conducted with a primary focus on the following questions.
How does the relationship between air pollution and economic growth differ by region with different stages of
development? In different countries, whether there are diverse regional patterns of the relationship between air
pollution and economic growth? Will the environmental quality and economic performance affect each other?
The purpose of this paper is to analyze the interaction between regional economic growth and air pollution in
China and South Korea (hereafter abbreviated as simply Korea). Both countries face serious air pollution
problems, and the air quality of the two countries is closely related to each other. The linkages between
economic growth and sulfur dioxide (SO 2) emissions are estimated at the city and county level using
simultaneous equation models covering 286 cities in China and 228 cities and counties in Korea during the
period 2006–2016. The focus is on SO2 because SO2 emissions are recognized as a typical transboundary air
pollution problem which has already induced various countries to cooperate through supranational institutions
[25]. Particularly, SO2 is one major pollutant that poses significant risks in many developing countries

undergoing a process of industrialization. SO2 pollution causes severe respiratory problems and significant
ecosystem degradation due to acid rain formation [26,27]. Because this paper aims to examine the regional
differences rather than the overall characteristics at the national level, the division of the region is a key point of
this paper. To explore whether there are diverse regional patterns in different countries with respect to the
relationship between
Int. J. Environ. Res. Public Health 2020, 17, 2761 3 of 20

air pollution and economic growth, this paper introduces two models of each country: metropolitan areas and
non-metropolitan areas in China and Korea. Because of both geographic and regional endowment differences,
economic development is quite uneven among different regions in China. Therefore, this paper divides China
into five economic regions: eastern China, central China, northwest China, southwest China, and northeast
China. The rest of this paper is organized as follows: Section 2 is a review of the literature review. Section 3
introduces the method and data source. Section 4 describes estimation results. Section 5 discusses the policy
implications. The last section summarizes the conclusions.

2. Background
The most widely used method for analyzing the relationship between economic growth and
environmental pollution is the EKC hypothesis. After the first empirical EKC study of Grossman and
Krueger [1], Panayotou [2] called the inverted U-shaped pattern an Environmental Kuznets Curve
after the original Kuznets curve, which describes an inverted U-shaped relationship between income
and income equality Kuznets [28]. The EKC argues that in the early stage of economic
development, the quality of the environment tends to decline until the average income reaches a
certain level in the development stage and then improves. The validity of the environmental Kuznets
curve hypothesis has been widely examined by di erent studies over the years. The conclusion is
various because of di erent choices of types of data, pollutant, country and approach (as shown in
Table 1). There were a few criticisms on the EKC hypothesis. First, the evidence in favor of an
inverted U-shaped relation is not robust and the locations of the turning points are sensitive to both
slight variations in the data and to reasonable permutations of the econometric specification [29].
Second, the empirical evidence supporting EKC crucially depends on the selected pollutant, the
sample composition and the period considered [30].
Based on the EKC hypothesis, many other studies believed that the relationship between economic
growth and environmental pollution is N-shaped [1,15–17]. Compared to the inverted U-shaped pattern,
the N-shaped pattern implies that in the early stages of economic development, environmental pressure
tends to rise with economic growth and then decline but rises again after reaching a critical level of
economic development. Leib [15] pointed out that an N-shaped pattern exists due to the external shocks,
internalization of the pollution externality, the exhaustion of abatement opportunities and decreasing
returns to scale of abatement technology. With continuous economic growth, the environmental carrying
capacity reaches its optimal level, and technological progress also reaches the saturation point and thus
further increase in production cannot mitigate the adverse impact on the environment.
One of the di erent views on the EKC hypothesis is that there is a monotonically increasing curve
that exists between pollution and growth [5–10]. Holtz-Eakin and Selden [5] examined the relationship
between economic development and carbon dioxide emissions using the global panel data of 130
counties. They suggest that global carbon dioxide emissions growth will continue to grow at an annual
rate of 1.8% in the foreseeable future and continued increase in emissions is due to the fastest growth in
output and population in low-income nations. Ang [7] provided evidence for a robust long-run relationship
between pollutant emissions and output in France from 1960 to 2000. The results showed that in the long
run, output growth can promote both CO 2 emissions and energy consumption. Using the time-series data
from 1960 to 2005, Halicioglu [8] concluded that income was the most significant variable to explain the
CO2 emissions in Turkey, followed by energy consumption and foreign trade. Chandran and Tang [ 9]
found that the inverted U-shape EKC hypothesis is not applicable to Indonesia, Malaysia, and Thailand.
The relationship between CO2 emissions and income tends to be linear in Indonesia and a normal U-
shaped curve in Malaysia and Thailand. Al-Mulali et al. [10] revealed that the EKC hypothesis does not
exist because the relationship between GRP and pollution is a monotonically increasing curve in both the
short and long run in Vietnam during the period 1981–2011.
Another point of view contrary to the EKC hypothesis provides evidence in support of a U-shaped
relationship between pollution and economic growth, indicating that pollution emission decreases with
economic growth initially and then increase. Moomaw and Unruh [11] concluded that there is a
Int. J. Environ. Res. Public Health 2020, 17, 2761 4 of 20

U-shaped and N-shaped relationship between CO2 emissions and income that exist in 16 OECD
(The Organisation for Economic Co-operation and Development) countries in the period from 1950
to 1992. They indicated that it is misleading to interpret EKC results as a process of income growth
that all countries must pass through. Kaufmann et al. [12] indicated a U-shaped relation between the
GRP per capita and the atmospheric concentration of sulfur dioxide for 23 counties between 1974
and 1989. The concentration of sulfur dioxide tended to decrease as per capita GRP rises from $
3000 to $ 12,500 but after that the concentration of SO 2 increased. Dinda et al. [4] found a U-
shaped shift between per capita income and the annual mean concentration of suspended
particulate matter and SO2 in 49 cities of 33 countries from 1979 to 1990. They pointed out that
without environmental degradation, further increases in income cannot be because of the technical
limits of industrial pollution control. In addition, Ozcan [13], Chandran and Tang [9] and Wang et al.
[14] also found more similar cases of U-shaped relationship between CO 2 emissions and economic
growth in the Middle East and Asian countries. Table 1 shows the empirical research results
published between 1995 and 2017. The EKC hypothesis was tested using di erent pollutants by
variable economic indicators, such as energy consumption, gross domestic product, trade
openness, industrial output, urbanization, population density, and foreign direct investment.

Table 1. Reviews on the Relationship between Economic Growth and Pollution.

Patterns Authors Dependent Variables Independent Variables

Holtz-Eakin and Selden [5], Gross Regional Product (GRP) per capita
Dasgupta et al. [6], Ang [7], and square, Energy Consumption, Output,
Monotonic
Halicioglu [8], Chandran Annual emissions of CO2 Foreign Direct Investment (FDI), Transport
rising curve
and Tang [9], energy consumption, Labor Force, Exports
Al-Mulali, et al. [10] and Imports

Annual emissions of GRP per capita and square, Population

Inverted Grossman and Krueger [1],
NO2, SO2, suspended density, Industry Shares in GRP,
U-shaped Panayotou [2]
particulate matter Trade Intensity

Moomaw and Unruh [11],

GRP per capita, Population growth, Spatial
Kaufmann et al. [12], Annual emissions of
intensity of economic activity, Energy
U-shaped Dinda et al. [4], Ozcan [13], CO2, SO2, suspended
consumption, FDI, Transport
Chandran and Tang [9], particulate matter
energy consumption
Wang et al. [14]

Annual emissions of GRP per capita, GRP, GRP square, GRP

Grossman and Krueger [1],
CO2, NO2, SO2, Soil cubic, Temperature, Import shares,
Aslanidis and
N-shaped quality index, Water Share of the tertiary industry,
Xepapadeas [16],
quality index, Environmental degradation index
Babu and Datta [17]
Air quality index and Population
 Although many researchers have carried out key studies on the relationship between economic
growth and air pollution, most have focused on the overall characteristics of air pollution at the national
level, using cross-country data or provincial level data of an individual country. In fact, due to regional
heterogeneity, di erent countries have diverse patterns of economic growth and air pollution at the
regional level. Using a panel data set of Malaysian states, Vincent [31] found that the EKC hypothesis
between the emission of six pollutants and income was not fulfilled. The pollution-income relationships
showed di erent trends of di erent pollutants because of di erent characteristics of the processes
generating pollution, natural resource endowments, shifting patterns of population, and the impacts of
environmental policies. List and Gallet [32] and Milimet et al. [33] used the same data set state-level
panel data from the United States from 1929 to 1994 to examine the relationship between air pollutants
SO2 and NO2 and per capita income. The results provided evidence that an inverted U-shape
characterizes the relationship between per capita emissions and per capita incomes at the state level, but
di erences in the level of the turning points across states. Park and Lee [34] explored the relationship
between economic growth and air pollution by using the annual panel data of 16 metropolitan areas in
Korea from 1990 to 2005. The results showed that GRP is negatively related to
Int. J. Environ. Res. Public Health 2020, 17, 2761 5 of 20

air pollution caused by SO2, failing to verify an inverted U-shaped nor N-shaped curves. In the case
of China, Jiang et al. [35] found an inverted U-shaped relationship between per capita income and
per capita emissions for waste gas emissions from fuel burning and wastewater, while a U-shaped
relationship for waste gas emissions from production using provincial panel data from 1985 to 2005.
They concluded that the less developed central and western regions appear to have turning points
occurring at lower income levels than the developed coastal region. Song et al. [ 36] examined the
relationship between waste gas, wastewater, and solid waste pollution and economic growth from
1985 to 2005 using provincial data in China and found that all three pollutants showed an inverted
U-shaped pattern. They also found that only a few high-income regions have reached the stage of
environmental improvement, while most provinces have more severe environmental degradation.
Wu et al. [37] analyzed the impacts of various carbon emission factors in the four classified regions,
namely, a high economy and high carbon intensity region, a high economy and low carbon intensity
region, a low economy and low carbon intensity region, and a low economy and high carbon
intensity region. They concluded that capita carbon emission increases monotonically with per
capita GRP in all regions, and the most significant factor of emission in four regions is the industrial
structure, energy intensity, population size and per capita GRP, respectively.
These regional analyses show that there are multiple patterns of pollution-growth relationships
across regions, and it is valuable to compare regions with di erent economic levels. This paper di ers from
previous studies in two ways. First, a cross-country comparative perspective is adopted to examine the
relationship between air pollution and economic growth at the regional level. Compared with single-
country research, this paper applies the multinational analysis of two Asian countries facing severe air
pollution problems, namely China and Korea, to explore whether exists a di erent regional pattern among
counties. Second, unlike most provincial-level studies, this paper uses city-level data and concern on the
relationship between air pollution and economic growth not only among regions at di erent stages of
development but also among cities at di erent scales. Previous regional studies in China generally
consider the di erences between eastern, central, and western China. Based on existing regional
research, this paper adopts a more detailed regional division method. In addition, this paper also focuses
on the regional di erence between metropolitan and non-metropolitan areas.

3. Methods

3.1. Model Specification

This paper uses the simultaneous equation model (SEM) methodology to analyze the
relationship between regional economic growth and air pollution covering 286 cities in China and
228 cities and counties in Korea during the period of 2006–2016 (Figure 1). An important criticism of
some existing empirical EKC studies is the failure to consider the feedback e ect of pollution on
economic growth. Perrings [38] proposed that the economy and its environment are jointly
determined. It is inappropriate to estimate a single equation model assuming unidirectional causality
from economy to environment. Barbier [39] found clear evidence that in the early stages of
development in many developing countries, when the environment deteriorated, rapid growth
attempts could be counterproductive and unsustainable. Pollutant emissions may limit the supply of
environmental inputs through environmental degradation, or they may reduce workday losses due to
health problems caused by pollution and reduce production. Therefore, it is more appropriate to
analyze the relationship between environmental pollution and economic growth by SEM which
simultaneously estimates the parameters of the whole system.
Int. J. Environ. Res. Public Health 2020, 17, 2761 6 of 20

Figure 1. Regional Division of Two Counties.

To handle the potential endogeneity caused by the bilateral causality between economic growth and
air pollution, this paper applies the SEM to test the relationship between per capita GRP and the annual

emission of SO2 using the Three-Stage Least Squares (3SLS) method (the Two-Stage Least Squares
(2SLS) method developed by Theil [40] is the basic approach for the SEM. 3SLS estimation introduced
by Zellner and Theil [41] is more advanced than 2SLS because 3SLS takes into consideration the
contemporaneous correlation of disturbances across the equations in the SEM. In addition, as Kennedy
[42] stressed, 3SLS is generally believed to be more consistent and asymptotically more e cient;
therefore, it is preferred to 2SLS if the disturbances of the separate equations are correlated). In this
paper, the choice of control variables is based on evidence from the existing literature. Extensive
empirical studies have attempted to analyze the causal relationships between pollutant emissions, energy
consumption and economic growth [7,43,44]. Kraft and Kraft [45] proposed that the imbalance of
industrial structure had a significant impact on the relationship between power consumption and
economic growth. Many studies have confirmed that the impact of industrial structure is significant
[43,46–48]. Therefore, energy consumption and industrial structure are also tested as the impact factors
of the relationship between growth and pollution. Before estimating the basic model, this paper tests the
cross-sectional dependence, followed by panel unit root and cointegration tests for these variables in all
models. Relying on the assumptions of cross-sectional independence may lead to ine cient and
inaccurate estimation results if the panel data are cross-sectionally dependent. Therefore, this paper uses
the CD (Cross-sectional Dependence) test developed by Pesaran [49] to the analyzed variables to
investigate whether each panel data is cross-sectionally independent. The CD test strongly rejects the
null hypothesis of no cross-sectional dependence (Table A1). To cover this issue, this paper applies the
UO (Ucar and Omay ) nonlinear unit root test [50], which allows for cross-sections dependence and
nonlinearity. Ucar and Omay [50] propose a nonlinear panel unit root test by combining the nonlinear
framework in Kapetanios et al. [51] with the panel unit root testing procedure of Im et al. [52]. The test
investigates the unit root null against the alternative hypothesis that at least one individual series follow a
nonlinear stationary process. The nonlinear cointegration tests could be fulfilled through investigating the
unit root for the residuals based on the UO method.
The results of the panel unit root tests (Table A2) show that the original series are non-
stationary sequences, and all the variables are first-order di erence stationary. The UO nonlinear
test refuses the null of no cointegration at the 1% significance level indicating that the null
hypothesis of no cointegration can be rejected and providing support for the long-term relationships
among study variables (Table A3):

ln GRPi,t = 0 + 1 ln EMPi,t + 2 ln ECi,t + 3ln ISi,t + 4 ln Emissioni,t + "1,i,t (1)

Int. J. Environ. Res. Public Health 2020, 17, 2761 7 of 20

2
ln Emissioni,t = 0+ 1 ln GRPi,t + 2 ln GRPi,t + 3 ln Pi,t + 4 ln ISi,t + 5 ln ECi,t + "2,i,t (2)

GRPi,t : Real GRP per capita of city i in period t

Emissioni,t: Per capita SO2 emissions of city i in period t
ISi,t: proportion of output value of the manufacturing industry in GRP of city i in period t
ECi,t: Electricity consumption of city i in period t
Pi,t: Population density of city i in period t
EMPi,t: The number of employed people of city i in period t
"i,t: Error term

3.2. Data Description

In the case of China, the empirical examination of the relationship between regional economic
growth and regional characteristics of air pollution is conducted at two geographic scales. This paper
divides China into five regions: eastern China, central China, northwest China, southwest China, and
northeast China (as shown in Figure 1) based on the classification by the National Bureau of Statistics of
the Republic of China [53] (the regional division method is based on the National Bureau of Statistics of
the Republic of China [53]. The 10 provinces (cities) in the east include Beijing, Tianjin, Hebei, Shanghai,
Jiangsu, Zhejiang, Fujian, Shandong, Guangdong, and Hainan; the six central provinces include Shanxi,
Anhui, Jiangxi, Henan, Hubei, and Hunan; and the 12 western provinces (districts and cities) including
Inner Mongolia, Guangxi, Chongqing, Sichuan, Guizhou, Yunnan, Tibet, Shaanxi, Gansu, Qinghai,
Ningxia and Xinjiang; the three northeastern provinces include Liaoning, Jilin and Heilongjiang. This
paper divides the western region into two regions: southwest and northwest). Next, in order to test
whether there are di erent regional patterns of pollution-economic growth relationship across countries,
this paper introduces two models of each country: metropolitan areas and non-metropolitan areas in
China and Korea. In China, the metropolitan areas consist of 19 first-tier cities and 30 second-tier cities,
and non-metropolitan areas are defined as the rest 237 third-fourth-fifth-tier cities. Following Wang et al.
[54] and Yang and Dunford [55], this paper uses the city-tier classification method which is most
commonly used for city ranking in China in recent years (This method was originally published by Yicai
Global and the National Bureau of Statistics of the Republic of China [ 56] also uses the same city
classifications to conduct National Residential Sales Price Surveys. First-tier cities cover 19 cities,
including 4 municipalities (Beijing, Shanghai, Tianjin and Chongqing) and 15 provincial capital cities (New
first-tier cities). These cities have a strong economic base, a population of more than 6 million and
abundant educational resources, advanced technology, and convenient transportation. Second-tier cities
cover 30 medium-sized prefecture cities equip with a good economic base and a population of 4–7
million, including the most eastern coastal open cities and few capital cities in the central and western
China. Third-fourth and fifth-tier cities cover the rest 237 cities, mainly composed of small-sized cities
located in central and western China equip with the relatively underdeveloped economic condition and a
population below 5 million.
In Korea, metropolitan areas cover 74 counties and cities in seven major cities, including Seoul,
Busan, Daegu, Incheon, Gwangju, Daejeon, and Ulsan. Non-metropolitan areas are composed of the
remaining 154 cities and counties (Figure 1). The Chinese data comes mainly from the City Statistical
Yearbook of China from 2006 to 2016, published by the National Bureau of Statistics of China. The
Korean data is collected from the Korean Statistical Information Service operated by Statistics Korea and
the National Institute of Environmental Research. Because the absence of per capita GRP data at the
county level for Seoul Special City, this paper calculated the GRP in every district of Seoul by multiplying
the proportion of employment in each sector by the total added value. Table 2 reports summary statistics
of the variables that are used in the estimation of region-specific models.
Int. J. Environ. Res. Public Health 2020, 17, 2761 8 of 20

Table 2. Data Analysis.

China Korea
East Central Southwest Northwest Northeast Metropolitan Non-Metro Metropolitan Non-Metro
Mean SD Mean SD Mean SD Mean SD Mean SD Mean SD Mean SD Mean SD Mean SD
GRP 6182 0.801 3244 0.642 3944 0.951 2536 0.654 4551 0.657 10,356 0.586 3,851 0.712 19,020 0.632 17,779 1.236
EMISSION 42 1.198 37 0.858 37 1.200 30 1.186 31 0.958 70 0.408 34 0.559 9 1.653 33 2.066
EMP 54 0.855 33 0.623 19 0.741 24 0.797 30 0.619 134 0.867 31 1.459 367 0.239 314 0.563
IS 36 8.466 42 9.875 37 14.470 41 9.697 40 13.088 48 7.398 49 11.939 8 1.098 13 1.300
EC 1173 1.132 491 1.246 622 2.065 364 1.218 995 1.181 2083 0.673 592 1.028 2569 0.961 4829 1.262
POP 54 0.606 42 0.593 10 1.009 28 0.696 16 0.719 39 1.312 28 1.476 77 1.029 10 0.967
Notes: GRP represents the GRP per capita (US$); EMISSION represents per capita emission (ton/10,000people); EMP represents employment (per 1000people); IS represents the share of
2
manufacturing industry in GRP (%); EC represents the energy consumption (kWh/person); POP represents population density (10,000 people/km ).SD is the standard deviation.
Int. J. Environ. Res. Public Health 2020, 17, 2761 9 of 20

3.3. Robustness Tests

It is necessary to check how robust the analysis is because the EKC hypothesis is assessed and
tested in various directions, including alternative functional forms, di erent econometric methods, the
inclusion of additional explanatory variables [57–59]. First, this paper tested the relationship between per
capita GRP and three emission indices, including per capita SO 2 emission (Total SO2 emissions
emissions/population), SO2 emission intensity (Total SO2 emissions/industrial output value) and SO 2
emissions density (Total SO2 emissions/land area). The results show that the relationship between the
three indexes and per capita GRP shows the same patterns by region, although the pollution index is
changed, the parameter symbols of ln(GRP) and its quadratic term are consistent in all panel models.
Because only the per capita SO 2 emission index is significantly correlated with the per capita GRP
variable and its quadratic term under the significant level of 10% in all models, we only reported the
results of the models using the index of per capita emission. Second, for choosing the functional form,
this paper uses T-test to check the statistical significance quadratic term and cubic terms of per capita
GRP and omit the cubic terms. Because all coe cients are insignificant in the cubic estimation, but coe
cients in the quadratic estimation are highly significant, we only consider the quadratic terms in our
model. The third step is to choose the estimation method. After examining the endogeneity of the
explanatory variables by Hausman’s test, the per capita GRP variables and its quadratic terms were
rejected as exogenous in the simultaneous equation model. Therefore, this paper uses the Three-stage
least squares method (3SLS) and the interaction terms of exogenous variables and their quadratic terms
were chosen as the instrumental variable for the per capita GRP and its quadratic term. Besides, this
paper applies the over-identification test and weak instrument test to make sure all the selected
instruments are valid. Through the three dimensions of robustness tests above, this paper confirms the
final model which tests the relationship between the per capita SO 2 emission index and per capita GRP
and its quadratic term. Moreover, the Mann-Whitney U test was used to compare the di erences in the
levels of GRP per capita and pollution emission levels across metropolitan and non-metropolitan areas in
both Chinese case and Korean case. In both counties, the results of the Mann-Whitney U test (as shown
in the Table A4) show that the null hypothesis of no statistical di erences is rejected with the value of
significance (0.001) is much smaller than (0.05). It can be concluded that there are significant di erences
in the levels of economic growth and pollution emission between metropolitan areas and non-
metropolitan areas in China and Korea.

4. Results
Table 3 presents the empirical results of the SEM. The results show that not only does regional
economic growth have a significant impact on air pollution, but also the impact of air pollution on regional
economic growth is significant (Figures A1 and A2). More importantly, the relationship between air
pollution and economic growth varies in di erent regions. From the respective of equations of GRP per
capita, the growth of GRP per capita in all models is positively correlated with per capita emission. The
economic growth was much more associated with the increase in pollutant emission of less developed
regions than developed regions. For instance, over the last decade, a 1% increase of per capita emission
could result in a 0.02% increase in the GRP per capita on average per year in east China, while 0.52%,
0.67% and 0.28% for central, northwest and northeast China. At the same time, a 1% increase in per
capita emission of non-metropolitan areas resulted in 0.44% and 0.26% of growth in per capita GRP in
China and Korea, respectively, which is 0.36% and 0.16% larger than that in the metropolitan region. It
indicates that in both countries, per capita emission has a stronger e ect on economic growth in non-
metropolitan areas than metropolitan areas. This result implies that the development of manufacturing
industries, in particular, pollution-intensive industries in non-metropolitan areas has contributed more to
economic growth than metropolitan areas.
From the respective of equations of per capita emission, the results illustrated with the Table 3
models (1)–(5) show that di erent patterns were found in China at the regional level. Only in the east and
northwest region, the sighs of coe cients of the GRP per capita and its square terms are negative
Int. J. Environ. Res. Public Health 2020, 17, 2761 10 of 20

and positive, respectively, which suggests that the regional pollution level tended to follow an inverted U-
shape pattern in the period 2006–2016. Such a pattern implies that pollutant emissions decrease with
economic growth after experienced a stage of environmental degradation in the two regions (Figure 2).
According to the China Statistical Yearbook, the proportion of output values of manufacturing to total
GRP from 2006 to 2016 decreased by 19.77% in the eastern region and 17.04% in the northwest
respectively. However, the decline in the proportion of manufacturing in total GRP only realized 8.20% in
the southwest and 7.76% in the central region. According to the National Bureau of Statistics of the
Republic of China, from 2006 to 2016, the proportion of the total output of manufacturing to total GRP
dropped from 46.58% to 36.62% in the eastern region and decreased from 39.47% to 28.94% in the
northwest region. In the southwest, the proportion decreased from 36.10% to 30.84%; the central region
decreased from 42.08% to 38.46%, and in the northeast region decreased from 44.09% to 30.77%. In the
east region and northwest region, energy-intensive industries gradually concentrated in the sectors with
lower energy intensity. Another reason for the declining trend in the developed eastern regions is strict
environmental regulations for polluting enterprises. As mentioned before, the pollutant discharge fee rate
of air pollutants in the eastern region, such as Beijing, was about ten times higher than that in the western
provinces. Stricter environmental regulations in the eastern region have led to the booming of
environmentally friendly enterprises and innovations in industries.
However, the relationship between air pollution and economic growth in the rest of China was
di erent. The results of models (2), (4) and (5) demonstrate a U-shaped curve relationship in
southwest, central and northeast region indicating that the pollutant emissions increase with
economic growth after the per capita GRP level reaches a certain level. To eliminate the economic
gap in China, the central government implemented the Western Development Strategy and Rise of
Central China Plan from 2000 to 2004, and a large amount of investment promoted the regional
output, e ectively increasing the GDP of the western and central region.
Moreover, with the adjustment of China’s economic development strategy in recent years, the
western region has received less attention than before, and environmental problems have become
more and more serious [60]. In order to pursue economic development goals and attract investment,
local governments have reduced environmental protection investment and weakened environmental
regulations. As a result, the problems of environmental pollution are increasingly severe with
economic growth. In particular, the proportion of output values of manufacturing to total GRP
declined by 28.44% in the northeast from 2006 to 2016. Even though there is a U-shaped curve, the
average GRP per capita level was still at the declining stage of the U curve indicating that the
pollution emission level had been decreased with economic growth in the period of 2006-2016 in the
northeast. Northeast is China’s largest old industrial base.
Since the implementation of the revitalization policy of the old industrial bases in the northeast
in 2003, the proportion of SO 2 high-load sectors in total industrial output declined from 23.87% in
2005 to 20.36% in 2013 due to the transformation of the regional industrial structure and the
technological progress [61]. Therefore, adjustment and optimization of the industrial structure were
found to be an e ective mean to reduce the air pollution.
The results of models (6)–(9) show that the pattern of economic growth and pollutant emissions
among metropolitan and non-metropolitan areas is reversed in both countries. From 2006 to 2016,
there was a common inverted U-shaped pattern is found in metropolitan areas of China and Korea.
Overall, it indicates that in metropolitan areas of both countries, the pollutant emission decreased
with economic growth. Conversely, in non-metropolitan areas, the U-shaped pattern implied that the
pollutant emission overall increased with economic growth. Such a di erent result confirms that the
impacts of economic growth on pollutant emission tend to be spatially heterogeneous among cities
with di erent scales in China and Korea.
The di erent patterns between metropolitan and non-metropolitan areas in both two countries could
be explained by the more stringent pollution regulation in metropolitan areas. With the rapid urbanization,
air pollution problems have been increasing concern in China. In order to better improve
Int. J. Environ. Res. Public Health 2020, 17, 2761 11 of 20

the regional air quality throughout metropolitan areas, the Chinese government has implemented a series
of national control policies to reduce the emissions of air pollutants since 2005. For instance, the Ministry
of Environmental Protection has issued the action of “Joint Prevention and Control of Air Pollution” in
2011, which aims to establish a joint prevention and control system and e ectively improve the regional
air quality. It was firstly implemented in three key regions which cover the major metropolitan areas in
China, including the Beijing-Tianjin-Hebei Region, the Yangtze River Delta and the Pearl River Delta
during the period of the 12th Five-Year Plan (2011–2015). The decrease in SO 2 from 2006 to 2016 in the
metropolitan areas reflects the success of China’s air pollution control program.

Table 3. Estimation Results of Simultaneous Equations by Three-Stages Least Squares Method.

China Korea

Non- Non-
East Central Northwest Southwest Northeast Metropolitan Metropolitan
Metropolitan Metropolitan
Model (1) Model (2) Model (3) Model (4) Model (5) Model (6) Model (8)
Model (7) Model (9)

Equation of GRP per capita

Parameter Parameter Parameter Parameter Parameter Parameter Parameter Parameter Parameter

(SE) (SE) (SE) (SE) (SE) (SE) (SE) (SE) (SE)

6.853 *** 2.886 *** 1.358 5.762 *** 5.123 *** 6.224 *** 1.170 *** 6.285 *** 5.838 ***
Intercept
(15.28) (2.63) (0.81) (3.39) (5.06) (0.73) (0.89) (28.93) (16.81)

0.233 *** 0.0552 0.062 0.118 *** 0.179 *** 0.039 *** 0.485 *** 0.685 *** 0.372 ***
EMP
(7.55) (111) (0.67) (1.22) (2.58) (0.73) (13.20) (16.66) (2.98)

0.480 *** 0.2514 *** 0.222 *** 0.209 *** 0.229 *** 0.599 *** 0.377 *** 0.122 *** 0.218 ***
EC
(33.19) (10.57) (8.5) (7.33) (6.80) (16.16) (19.99) (7.27) (6.46)

0.012 *** 0.0021 0.005 0.012 ** 0.004 0.015 0.003 *** 0.011 0.001
IS
(-3.05) ( 0.43) (0.95) (2.53) (1.47) (-3.05) (1.43) ( 1.02) ( 0.04)

0.022 *** 0.5231 *** 0.669 ** 0.184 0.2837 ** 0.083 *** 0.441 *** 0.103 *** 0.261 ***
EMISSION
(0.34) (4.05) (3.62) (0.93) (2.57) (0.28) (3.54) (7.82) (4.99)

Equation of Per capita emission

Parameter Parameter Parameter Parameter Parameter Parameter Parameter Parameter Parameter

(SE) (SE) (SE) (SE) (SE) (SE) (SE) (SE) (SE)

Intercept 446.000 *** 125.429 *** 203.22 ** 285.475 *** 150.181 * 10.719 *** 15.356 *** 25.286 *** 24.014 ***
( 3.79) (3.02) ( 2.2) (2.85) (1.94) (0.61) (0.31) (4.50) (0.54)

GRP 84.002 *** 23.915 *** 41.967 ** 57.604 *** 27.563 * 56.317 *** 0.941 *** 1.371 *** 2.336 ***
(3.84) ( 2.87) (2.27) ( 2.81) ( 1.86) (0.63) ( 0.11) ( 4.61) ( 0.65)
 3.857 *** 1.241 *** 2.051 ** 2.976 *** 1.328 * 2.447 *** 0.046 *** 0.068 *** 0.118 ***
2
GRP

( 3.80) (2.96) ( 2.23) (2.86) (1.89) ( 0.63) (0.14) (10.16) (0.77)

0.002 0.138 *** 0.171 ** 0.272 *** 0.174 *** 0.115 *** 0.023 *** 0.785 *** 0.632 ***
EC
( 0.01) (3.46) (2.2) (3.35) (2.88) ( 0.66) (0.64) (6.64) (12.29)

0.006 0.013 *** 0.009 0.0001 0.005 0.042 *** 0.060 *** 0.168 *** 0.442 ***
IS
( 0.610) (4.24) (0.7) ( 0.01) ( 0.85) (0.81) (12.46) (1.96) (5.56)

0.577 ** -0.510 *** 0.182 0.211 ** 0.594 *** 0.154 *** 0.102 *** 0.632 *** 0.026
P
(1.99) ( 5.99) ( 0.9) (1.80) (4.96) ( 0.52) ( 4.30) (10.25) (1.68)

R-square 0.7483 0.6680 0.5703 0.5842 0.5706 0.6844 0.5564 0.5139 0.7611

Number of
101 88 38 46 33 49 237 74 154
Cross sections

Time series length: 2006–2016

Turning point ($) 7882 2249 4080 2348 4725 14624 4050 20856 19682

Inverted U Inverted U Inverted U Inverted U

Pattern U shape U shape U shape U shape U shape
shape shape shape shape

Notes: The curve shapes between per capita SO2 and GDP per capita can be determined from the signs of
parameters: U-shape (If 1h0, 2i0), inverted U-shape (if 1 > 0, 2 < 0). All variables are in log form *** 1%
significance level; ** 5% significance level; * 10% significance level. Standard errors are in brackets.

In the case of Korea, the government legislated a special act named “Improvement of Air Quality in
Seoul Metropolitan Areas (SMA, areas including Seoul and Incheon metropolitan cities, Gyeonggi
province)” in December 2003 to improve the air quality in SMA. The main focus of the air quality
improvement plan for SMA (2005–2014) was to regulate the total amount of emissions in the workplace,
to supply low-emission vehicles, and to strengthen gas emission management regulations [62]. After the
implementation of the first phase of an air quality improvement plan, the annual concentration of main
pollutants has significantly decreased in Seoul and Incheon until 2013 [63]. Again, the government
adopted the second phase of the air quality improvement plan from 2015 to 2024 for SMA [64].
Int. J. Environ. Res. Public Health 2020, 17, 2761 12 of 20

Figure 2. Pollution-Growth Patterns at Five Regions in China.

There is still a large gap between the overall emission levels of China and South Korea,
although pollutant emissions in both metropolitan areas show a downward trend. Figure 3 shows
that China not only has a higher overall level of pollutant emissions than Korea, but also a larger
disparity in regional economic levels. For instance, the average GRP per capita level was found to
be US $10,356 and US $3851 for metropolitan and non-metropolitan areas in China, whereas US
$19,020 and US $17,779, respectively, in Korea (Table 2). Another significant di erence is that the
level of per capita pollutant emission in China’s metropolitan areas was still much higher than in
non-metropolitan areas, while the opposite result has occurred in Korea. The overall level of
pollutant emission in Korea’s metropolitan areas was obviously lower than in non-metropolitan
areas. This can be interpreted as that higher pollutant emission levels in non-metropolitan areas
were associated with a relatively higher proportion of manufacturing industry in GRP. The average
proportion of the manufacturing industries in non-metropolitan areas was 13% from 2006 to 2016,
which is 5% higher than that in metropolitan areas (Table 2), indicating that more manufacturing
enterprises in Korea prefer to locate in non-metropolitan areas. In China, manufacturing still
occupied an important position in metropolitan areas (48%) in the last decade. In other words,
although pollution levels in metropolitan areas of the two countries have shown a downward trend in
the past ten years, there is still a large gap between China and Korea in terms of the overall level of
pollutant emissions, especially the level of pollutant emissions of metropolitan areas.
From the respective of determinants of pollutant emission, this paper compares the elasticities of
pollutant emission with respect to energy consumption, industrial structure and population density based
on the reduced form of simultaneous equation models. After controlling the effect of economic growth, the
main determinants of air pollution are energy consumption and industrial structure in metropolitan and
non-metropolitan areas, respectively. For instance, with a 1% increase in the proportion of manufacturing
industry output in GRP, industrial SO 2 emission in non-metropolitan areas increased by 0.04% and
0.30% in China and Korea, larger than that in metropolitan areas (0.01% and 0.15%). It suggests that
industrial structure has a relative stronger effect on pollutant emission in non-metropolitan areas than
metropolitan areas. On the other hand, in the case of Korea, with a 1% increase in energy consumption,
industrial SO2 emission in non-metropolitan areas increased by 0.38% and 0.17% in metropolitan areas
and non-metropolitan areas, respectively. In the case of China, a 1% increase in energy consumption
was associated with a 0.48% increase in pollutant emission of metropolitan areas and a 0.03% increase
in non-metropolitan areas. The results find that impacts of energy consumption on pollutant emission
were larger in metropolitan areas than non-metropolitan areas in both China and Korea. In addition, the
results found that there was a negative effect of population density of the pollutant emission of most
models; however, the relationship between air pollution and population density was uncertain for the
northwest region in China and non-metropolitan areas in Korea.
 Int. J. Environ. Res. Public Health 2020, 17, 2761 13 of 20

Int. J. Environ. Res. Public Health 2020, 17, x 14 of 21

110
105
100
people)

)
95

people
(ton/10000

90

(ton/10000
85
SO2

80
Percapita

75

SO2
70

65

Percapita
55

60
0 10,000 20,000 30,000 40,000 50,000

GRP per capita($) 50

Metropolitan Non-Metropolitan 45
40
0 10,000 20,000 30,000

GRP per capita($)

Metropolitan Non-Metropolitan

Figure 3. Pollution-Growth Patterns at Di erent City Scales in Two Countries.

Figure 3. Pollution-Growth Patterns at Different City Scales in Two Countries.
5. Policy Implications
From the respective of determinants of pollutant emission, this paper compares the elasticities
According to the analysis results of this paper, the policy implications could be summarized from of pollutant emission with respect to
energy consumption, industrial structure and population
the following three respects: the national level, the regional level, and the city level. The inverted density based on the reduced form of
simultaneous equation models. After controlling the effect of U-shape patterns of the two countries suggest that pollution levels in metropolitan
areas have shown a economic growth, the main determinants of air pollution are energy consumption and industrial
downward trend in recent years, whereas there is still a large gap between the overall level of pollutant structure in metropolitan and non-metropolitan areas,
respectively. For instance, with a 1 % increase
emissions of China and Korea. This gap could be attributed to the di erence in economic development in the proportion of manufacturing industry output in GRP, industrial SO2 emission in non-

stages and the implementation of environmental policies of the two counties. Si et al. [65] proposed that metropolitan areas increased by 0.04% and 0.30% in
China and Korea, larger than that in metropolitan
many environmental inChina are communicated inthe form ofgovernment documents
areas (0.01%and0.15%)policies.Ituggests that industrial structure has arelative strongereffect onpollackingutant

clearmissionlegal inprovisionsnon-metropolitanandregulationsareasthan.Themetropolitanupdateof areasexisting.Onenvironmentaltheotherhand,lawsintheandcaseregulationsofKorea,

iswithalso relatively1%increaseslow, inandenergyenvironmentalconsumption,standardsindustrialcannotSOadapt2emissiontochangiinnongenvironmentaltropolitan- areasissues. InincreasedKorea,environmentalby0.38%andpolicy0.17%
toolsinmetropolitanaresetup inareaswayandthatnonnot-metropolitanonlycontrolsareas,therespectivelywholeprocess,.In thealso
controlscaseof beforeChina, and1%afterincreasetheevent;inenergybesides,consumptionreasonablewasenvirassonmentalciatedwithlaws0and.48%regulations,increasein andpollutanttimely

updatingemissionmechanismofmetropolitanprovideareas andguarantee0.03%forincrthease ectiveinnon-imetropoliplementationareasof.environmentalTheresultsfindpolicythat

toolsimpacts.Drawingofenergyontheconsumptionexperienceonof pSouthllutantKorea,emissionifthe wereChineselargergovernmentinmetropolitanpays moreareas attethantionnon-to

improvingmetropolitanthe enviroareasinmentalbothChinaleglandsystem,Koreait .willInaddition,havean theectiveresultsimpactfoundonthatimprovingtherewasair negativepollution,

effectof population density ofthepollutant emission of most models; however, the relationship
including formulating various market -oriented policy tools, such as environmental taxes andfees,and
to
between air pollution and population density was uncertain for the n rthwest region in China and a more detailed emissions trading market, and encourage enterprises establish
their own pollution

monitoringnon-metropolitansystems.areas in Korea.
Different from the previous regional research suggesting that there was a similar pattern in the
5. Policy Implications
relationship between air pollution and economic growth across regions in China [35–37]. This paper uses a

moreAccordingdetailreionaltothedivisionanalysismethodresultsandofthisfindspaper,that diversethepolicypatternsimplicationsexistin couldfiveregionsbesummarofChizednaat

differentfromthestagesfollowingofdevelopmentthreerespects:fromthe 2006nationalto2016level,. Thetheregionalxperiencelevel,ofandthe thedevelopmcityleventl.Themodelinvertedofthe

easternU-shaperegionpattercansprovideofthetwoinspirationcountriesforsuggestthebalancedthatpollutiongrowth andlevelspollutioninmetrofpolitanthesouthwestareashaveandshowncentral

regionsadownwardfromthetrendfollowingrecentthreeyearsaspects,whe.First,easasthereairpolluisstillonalevelslarge ingapthebetweencentralandthesouthwestveralllevelregionsf

deteriorpollutantte withemissionseconomicofChinagrowth,andtheKoreagovernment.Thisgapneedscouldto bepayattributedmoreattentiontothetodifferencechangetraditionalineconomhighc

development stages and t e implementation ofenvironmental poli ies ofthe two count .Sietal .
energy -consuming and high - polluting production methodsofthe local energy - intensive industries to

[65] proposed that many environmental policies in China are communicated in the form of reduce pollutant emissions. At the same time, it is necessary to
strengthen the improvement of production
government documents lacking clear legal provisions and regulations. The update differentiatedoexisting technology, promote the innovation and promotion of clean production
technology. Second,
environmental laws and regulations is also relatively slow, and environmental standards cannot regional environmental policies should be designed for
different stages of economic development. In the
adapt to changing environmental issues. In Korea, environmental policy tools are set up in a way that central and southwestern regions, the
main sources of air pollution are the emissions of industrial pollutants. not only controls the whole process, also controls before and after the
event; besides, reasonable The environmental regulations for polluted enterprises could be further strengthened in areas where the
environmental laws and regulations, and timely updating mechanism provide a guarantee for the
Int. J. Environ. Res. Public Health 2020, 17, 2761 14 of 20

heavy industries and highly polluting industries are densely located. The appropriate environmental
regulations could help to accelerate industrial upgrading and economic transition to a more
sustainable style. Besides, the strong pollution control policy is still needed even there is a clear
downward trend of pollutant emission in the eastern, northeastern and northwest region. Third,
increase investment in pollution control for the southwestern and central region. The U shape
pattern of these two regions means that a large amount of environmental pollution generated in the
process of pursuing rapid economic growth. Even though the local government has started to
increase investment in pollution control, but it has not played a substantial role in recent years [ 60].
In the western and central region, it is urgent to increase financial support and arrange for specific
financial resources to invest in environmental protection, especially the improvement of air quality,
and strengthen supervision and management of investment in environmental protection.
The gaps between China’s metropolitan areas and non-metropolitan areas in the terms of both
economic development level and pollutant emission level are much larger than those of Korea. To
improve regional air quality, the actual situation and major determinants of air pollution in both urban and
non-urban areas in China need to be considered. Since the major determinant of air pollutant emissions
in metropolitan areas is the energy consumption, it is crucial to develop new energy industries to reduce
pollutant emissions. It not only relies on the government’s financial subsidies, tax incentives and other
incentive policies, but also promoting the technological progress and industrial upgrading of renewable
energy through the market mechanism. In the relatively backward non-metropolitan areas, it is important
to continuously adjust the industrial structure by encouraging the development of the service industry, as
well as promote the optimization of the internal structure of the secondary industry and encourage the
development of low-pollution and low-energy-consuming industries.

6. Discussion and Conclusions

The purpose of this paper is to analyze the relationship between regional economic growth and air
pollution in China and Korea to explore whether there are diverse regional patterns of different countries. The
major findings of this paper are that regional differences existed in the relationship between air pollution and
economic growth in China and Korea. In both countries, an inverted U-shaped pattern was found in
metropolitan areas while a U-shaped pattern exists in non-metropolitan areas. Moreover, there exist different
patterns in five regions of China, which are influenced by different levels of economic development and different
regional environmental policies. There is still a large gap between the overall emission levels of China and
Korea although the emissions of pollutants in metropolitan areas of both countries have shown a downward
trend in recent years. The level of pollutant emissions of China’s metropolitan areas is much higher than in non-
metropolitan areas, while the opposite result has occurred in Korea.
This paper contributes to the literature from three respects. First, in contrast to the extensive
literature on the relationship between air pollution and economic growth, this study reveals the
importance of regional-level research. Based on the existing regional studies, this paper attempts to
adopt a more detailed regional division method and points out that the patterns of air pollution vary
significantly between regions, filling a gap in relevant research fields at the regional level. Second, this
paper focuses on the differences between cities of different sizes. Different patterns of pollution
development in metropolitan and non-metropolitan areas are compared. In addition, this paper makes a
comparison between two Asian countries, China and South Korea, which are facing serious air pollution
problems as well as large regional differences in economic development levels. One of the limitations of

this paper is that it only focuses on SO 2, leaving room for further study of other pollutants in the future.
Industrial development in different regions is closely related to local natural resources, and our analysis
may ignore the impact of production activities in some pollution-intensive industries with other pollutants
as the main pollutants on air quality. For further research, more diverse pollution phenomena need to be
paid attention to, including other air pollutants, water pollution, and solid pollution. This paper mainly
analyzes the impact of economic growth, industrial structure, energy consumption and population factors
on air pollution in the two countries. However, the environmental policies of the two countries also play
Int. J. Environ. Res. Public Health 2020, 17, 2761 15 of 20

a very important role in regional air pollution. Since it is difficult to develop a common policy variable
available for both countries, more policy factors have not been taken into account. In future
research, it is necessary to continue to pay attention to this direction of research. In addition, the
limitation of the methodology in this paper is that, due to the limitation of data, the model only
considers short-term analysis and cannot establish the long-term dynamic analysis model.

Author Contributions: Conceptualization, M.J., E.K. and Y.W.; methodology, M.J.; software, M.J.; validation,
M.J. and E.K.; formal analysis, M.J. and Y.W.; investigation, M.J. and Y.W.; resources, M.J.; data curation,
M.J.; writing—original draft preparation, M.J.; writing—review and editing, M.J. and E.K.; visualization, M.J.;
supervision, E.K. and Y.W.; project administration, E.K.; funding acquisition, E.K. All authors have read and
agreed to the published version of the manuscript.
Funding: This research was funded by the Ministry of Education of the Republic of Korea and the National
Research Foundation of Korea, Grant Number NRF-2017S1A3A2066771.
Conflicts of Interest: The authors declare no conflict of interest.

Appendix A
Table A1. Results of Cross-sectional Dependence Test.

Panel A Panel B
Pesaran CD 39.773 *** 23.724 ***
p-value 0.000 0.000
Number of observations 286 228
Notes: Panel A includes 285 cities of China, 2006–2016; Panel A includes 285 cities of China, 2006–2016. ***
1% significance level.

Table A2. Results of Panel Unit Root Test.

Panel A Panel B
Level First di erences Level First di erences
EMISSION 2.930 5.003 *** 1.114 4.558 ***
GRP 2.488 9.015 *** 1.649 8.570 ***
GRP2 2.090 4.002 *** 7.173 24.447 ***
EMP 1.269 1.705 *** 5.192 5.260 ***
EC 0.930 2.347 *** 1.028 2.792 ***
IS 3.924 8.580 *** 4.641 9.025 ***
POP 2.314 3.259 *** 2.737 16.704 ***
Notes: Panel A includes 286 cities of China, 2006–2016; Panel B includes 228 cities and counties of Korea, 2006–2016.
*** 1% significance level.

Table A3. Results of Cointegration test.

Statistic p-Value
Panel A.
Equation of GRP per capita 0.063 *** 0.000
Equation of per capita Emission 0.236 *** 0.000
Panel B.
Equation of GRP per capita 1.115 *** 0.000
Equation of per capita Emission 0.664 *** 0.000

Notes: Panel A includes 286 cities of China, 2006–2016; Panel B includes 228 cities and counties of Korea, 2006–2016.
*** 1% significance level.
 Statistic p-value
Panel A.
Equation of GRP per capita 0.063*** 0.000
Equation of per capita Emission 0.236*** 0.000
Panel B.
Int. J. Environ. Res. Public Health 2020, 17, 2761 16 of 20
Equation of GRP per capita 1.115*** 0.000
Equation of per capita Emission 0.664*** 0.000
Notes: Panel A includes 286 cities of China, 2006–2016; Panel B includes 228 cities and counties of
Table A4. Results of Mann-Whitney U-test.
Korea, 2006–2016. ***1% significance level.

U-Value Z-Statistics p-Value

Table A4. Results of Mann-Whitney U-test.
Korea

U -value Z-Statistics p-value

GRP 193,144 6.4637 *** <0.00001
Korea
Emission 175,605 8. *** <0.00001
6696
GRP 193,144 − .4637*** <0.00001

Emission 175,605 −8.6696*** <0.00001

China

China
GRP 220,681 15.0719 *** <0.00001
GRP 220,681 15.0719*** <0.00001
Emission 972,885 16.6324 *** <0.00001
Emission 972,885 16.6324*** <0.00001
Notes:Notes:AvalueAvalueofpof < 0p.<050.was05wasconsideredredtotobebesignificantficant..****1%1%significancesignifica level.

Model 1 Model 2 Model 3

7.4 14 14.8

14.6

14.4

7.3 14.2

13

14

7.2 13.8

13.6

12

7.1 13.4

13.2

Int. J. Environ. Res. Public Health 2020, 17 , x 13 18 of 21

11 12.8

7 10

7 10 7 10

Figure A1. Cont.

Model 4 Model 5 Model 6
 9 9 10.4

8.9 8.9

10.2

8.8 8.8

10

8.7 8.7

8.6 8.6

9.8

8.5 8.5

9.6

8.4 8.4

8.3 8.3 9.4

7 10 13 7 10 13 7 10

Model 7 Model 8 Model 9

12.2 8.25 10.4

12 8.2

10.2

11.8

8.15

11.6

8.1 10

11.4

8.05

11.2 9.8

11

9.6

10.8 7.95

10.6 7.9 9.4

7 10 13 7 10 13 16 7 10 13 16

Figure A1. Growth-Pollution Patterns. Notes: The X-axis is per capita SO2 emission (ton), Y-axis is
Figure A1. Growth-Pollution Patterns.
GDP per capita ($).
Notes: The X-axis is per capita SO2 emission (ton), Y-axis is GDP per capita ($).

Model 1 Model 2 Model 3

90 70 80

80

70

60

70
60
60
50
50
40
50

40

40

30

30

30

20

20 20

10

10 TP=7882 10 TP=4080

TP=2249

0 0 0
 10.8 7.95
10.6 7.9 9.4
7 10 13 7 10 13 16 7 10 13 16

Figure A1. Growth-Pollution Patterns.

Int. J. Environ. Res. Public Health 2020, 17, 2761 17 of 20
Notes: The X-axis is per capita SO2 emission (ton), Y-axis is GDP per capita ($).

Model 1 Model 2
0
20,
0
90 70 00
200 60 0 10,000 0
Figure A2.
80 0 04000 00 Cont.

60 Mo
del Mod
70 7 el 8

5
50 5 20

60
20

5
40
4
50

20
40
5
30 3

30 20

5
20 2

20 19

5
10 1
10 TP=7882
19
TP=2249
5
0
0 0
19
0 10,000 20,000 0 2000 4000
4
9
Model 4 Model 5 19
TP=405
TP=20,
0 856

70 100
4
8 19

90

4
60 7 18
10 20,
80 ,0 00 20,0 40,00 60,0
0 00 0 0 00 0 00

50 70

60

40

50

30 40

30

20

20

Int. J. Environ. Res. Public Health 2020, 17 ,x

10
TP=2348 TP=4725
10
 Model 3
80
80

60
70

40
60

50
20 19 of 21
TP=14,624
40

0
30
0 10,000 20,000 30,000

20

Model 9
10
TP=4080
32

31

31
0
30
0 3000 6000 9000
30

29
Model 6
29
28 TP=19,682
120
0 10,000 20,000 30,000 40,000

100

Figure A2. Pollution-Growth PatternsFigure. A2Notes:.PollutionTheX--GrowthaxisisGRPPatternspercapita. ($), Y-

axis is per capita SO2 emission (ton).TP means the level of GRP per capita at the turning points of each region.
Notes: The X-axis is GRP per capita ($), Y-axis is per capita SO 2 emission (ton).TP means the level of GRP per
Referencescapitaatthe turning points of each region.
1. Grossman, G.M.; Krueger, A.B. Environmental Impacts of a North American Free Trade Agreement; NBER Working
Reference
Paper, No. 3914; National Bureau of Economic Research: Cambridge, MA, USA, 1991.
2. 1. Panayotou,Grossman,TG.Empirical.M.;KruegTestsr,Aand.B. PolicyEnvironmentalAnalysis ImpactsofEnvironmentalfNorthDegradationAmericanFreeatDiTradeerentAgreementStagesof; EconomicNBER

WorkingPaper No.3914; National Bureau of EconomicResearch: Cambridge, MA,USA, 1991 .

Development ;Working Paper, Technology and Employment Programme; International Labor O ce: Geneva,
2. Switzerland, Panayotou, T. Empirical Tests and Policy Analysis of Environmental Degradation at Different Stages of Economic
1993.
Development; Working Paper, Technology˙ and Employment Programme; International Labor Office:

3. Akbostancı, E.; Turut-Asik, S.; Tunç, G.I. The relationship between income and environment in Turkey:
Geneva, Switzerland, 1993.

Is there an environmental Kuznets curve. Energy Policy 2009, 37, 861–867. [CrossRef]
3. Akbostancı, E.; Turut-Asik, S.; Tunç, G.İ. The relationship between income and environment in Turkey: Is

4. Dinda, S.; Coondoo, D.; Pal, M. Air Quality and Economic Growth: An Empirical Study. Ecol. Econ. 2000, 34,
there an environmental Kuznets curve. Energy Policy 2009, 37, 861–867.

409–423. [CrossRef]
4. Dinda, S.; Coondoo, D.; Pal, M. Air Quality and Economic Growth: An Empirical Study. Ecol. Econ. 2000,
5. Holtz-Eakin, D.; Selden, T.M. Stoking the Fires? CO2 Emissions and Economic Growth. J. Public Econ. 1995,
34, 409–423.
57, 85–101. [CrossRef]
5. Holtz-Eakin, D.; Selden, T.M. Stoking the Fires? CO 2 Emissions and Economic Growth. J. Public Econ. 1995, 57,
85–101.
6. Dasgupta, S.; Laplante, B.; Wang, H.; Wheeler, D. Confronting the Environmental Kuznets Curve. J. Econ.
Perspect. 2002, 16, 147–168.
7. Ang, J.B. CO2 Emissions, Energy Consumption, and Output in France. Energy Policy 2007, 35, 4772–4778.
8. Halicioglu, F. An Econometric Study of CO2 Emissions, Energy Consumption, Income and Foreign Trade in
Turkey. Energy Policy 2009, 37, 1156–1164.
 Int. J. Environ. Res. Public Health 2020, 17, 2761 18 of 20

6. Dasgupta, S.; Laplante, B.; Wang, H.; Wheeler, D. Confronting the Environmental Kuznets Curve. J. Econ.
Perspect. 2002, 16, 147–168. [CrossRef]
7. Ang, J.B. CO2 Emissions, Energy Consumption, and Output in France. Energy Policy 2007, 35, 4772–
4778. [CrossRef]
8. Halicioglu, F. An Econometric Study of CO2 Emissions, Energy Consumption, Income and Foreign Trade
in Turkey. Energy Policy 2009, 37, 1156–1164. [CrossRef]
9. Chandran, V.G.R.; Tang, C.F. The Impacts of Transport Energy Consumption, Foreign Direct Investment and Income
on CO2 Emissions in ASEAN-5 Economies. Renew. Sustain. Energy Rev. 2013, 24, 445–453. [CrossRef]
10. Al-Mulali, U.; Saboori, B.; Ozturk, I. Investigating the Environmental Kuznets Curve Hypothesis in Vietnam.
Energy Policy 2015, 76, 123–131. [CrossRef]
11. Moomaw, W.R.; Unruh, G.C. Are Environmental Kuznets Curves Misleading Us? The Case of CO2
Emissions. Environ. Dev. Econ. 1997, 2, 451–463. [CrossRef]
12. Kaufmann, R.K.; Davidsdottir, B.; Garnham, S.; Pauly, P. The Determinants of Atmospheric SO 2 Concentrations:
Reconsidering the Environmental Kuznets Curve. Ecol. Econ. 1998, 25, 209–220. [CrossRef]
13. Ozcan, B. The Nexus Between Carbon Emissions, Energy Consumption and Economic Growth in Middle
East Countries: A Panel Data Analysis. Energy Policy 2013, 62, 1138–1147. [CrossRef]
14. Wang, S.S.; Zhou, D.Q.; Zhou, P.; Wang, Q.W. CO2 Emissions, Energy Consumption and Economic
Growth in China: A Panel Data Analysis. Energy Policy 2011, 39, 4870–4875. [CrossRef]
15. Lieb, C.M. The Environmental Kuznets Curve: A Survey of the Empirical Evidence and of Possible Causes; Discussion
Paper Series No. 391; University of Heidelberg, Department of Economics: Heidelberg, Germany, 2003;
Available online: www.uni-heidelberg.de/md/awi/forschung/dp391.pdf (accessed on 8 November 2019).
16. Aslanidis, N.; Xepapadeas, A. Smooth ‘Inverted-V-Shaped’ & Smooth ‘N-Shaped’ Pollution-Income Paths;
Working Papers 0405; University of Crete, Department of Economics: Rethymno, Greece, 2004.
17. Babu, S.S.; Datta, S.K. The Relevance of Environmental Kuznets Curve (EKC) In a Framework of Broad-
Based Environmental Degradation and Modified Measure of Growth—A Pooled Data Analysis. Int. J. Sustain.
Dev. World Ecol. 2013, 20, 309–316. [CrossRef]
18. Dong, L.; Dong, H.; Fujita, T.; Geng, Y.; Fujii, M. Cost-E ectiveness Analysis of China’s Sulfur Dioxide
Control Strategy at the Regional Level: Regional Disparity, Inequity and Future Challenges. J. Clean. Product.
2015, 90, 345–359. [CrossRef]
19. Shi, H.; Zhang, L. China’s Environmental Governance of Rapid Industrialization. Environ. Polit. 2006, 15,
271–292. [CrossRef]
20. Van Rooij, B.; Lo, C.W.H. Fragile Convergence: Understanding Variation in the Enforcement of China’s
Industrial Pollution Law. Law Policy 2010, 32, 14–37. [CrossRef]
21. Zhao, X.; Zhao, Y.; Zeng, S.; Zhang, S. Corporate Behavior and Competitiveness: Impact of
Environmental Regulation on Chinese Firms. J. Clean. Product. 2015, 86, 311–322. [CrossRef]
22. Lian, T.; Ma, T.; Cao, J.; Wu, Y. The E ects of Environmental Regulation on the Industrial Location of
China’s Manufacturing. Nat. Hazards 2016, 80, 1381–1403. [CrossRef]
23. Yin, J.H.; Zheng, M.Z.; Chen, J. The E ects of Environmental Regulation and Technical Progress on CO2
Kuznets Curve: An Evidence from China. Energy Policy 2015, 77, 97–108. [CrossRef]
24. Fodha, M.; Zaghdoud, O. Economic Growth and Pollutant Emissions in Tunisia: An Empirical Analysis of
the Environmental Kuznets Curve. Energy Policy 2010, 38, 1150–1156. [CrossRef]
25. Bruyn, S.M. Explaining the Environmental Kuznets Curve: Structural Change and International
Agreements in Reducing Sulphur Emission. Environ. Dev. Econ. 1997, 2, 485–503. [CrossRef]
26. Guttikunda, S.K.; Carmichael, G.R.; Calori, G.; Eck, C.; Woo, J.H. The contribution of megacities to
regional sulfur pollution in Asia. Atmos. Environ. 2003, 37, 11–22. [CrossRef]
27. Sun, W.Y.; Ding, S.L.; Zeng, S.S.; Su, S.J.; Jiang, W.J. Simultaneous absorption of NO x and SO2 from
flue gas with pyrolusite slurry combined with gas-phase oxidation of NO using ozone. J. Hazard. Mater. 2011,
192, 124–130. [CrossRef]
28. Kuznets, S. Economic Growth and Income Inequality. Am. Econ. Rev. 1955, 45, 1–28.
29. Harbaugh, W.T.; Levinson, A.; Wilson, D.M. Reexamining the Empirical Evidence for an Environmental
Kuznets Curve. Rev. Econ. Stat. 2002, 84, 541–551. [CrossRef]
30. Magnani, E. The Environmental Kuznets Curve: Development Path or Policy Result? Environ. Modell.
Software 2001, 16, 157–165. [CrossRef]
 Int. J. Environ. Res. Public Health 2020, 17, 2761 19 of 20

31. Vincent, J.R. Testing for Environmental Kuznets Curves within a Developing Country. Environ. Dev. Econ.
1997, 2, 417–431. [CrossRef]
32. List, J.A.; Gallet, C.A. The Environmental Kuznets Curve: Does One Size Fit All? Ecol. Econ. 1999, 31,
409–423. [CrossRef]
33. Millimet, D.L.; List, J.A.; Stengos, T. The Environmental Kuznets Curve: Real Progress or Misspecified
Models? Rev. Econ. Stat. 2003, 85, 1038–1047. [CrossRef]
34. Park, S.; Lee, Y.M. Regional Model of EKC for Air Pollution: Evidence from the Republic of Korea. Energy
Policy 2011, 39, 5840–5849. [CrossRef]
35. Jiang, Y.; Lin, T.; Zhuang, J. Environmental Kuznets Curves in the People’s Republic of China: Turning Points and
Regional Di erences; ADB Economics Working Paper Series; Asian Development Bank: Mandaluyong
City, Philippines, 2008.
36. Song, T.; Zheng, T.; Tong, L. An Empirical Test of the Environmental Kuznets Curve in China: A Panel
Cointegration Approach. China Econ. Rev. 2008, 19, 381–392. [CrossRef]
37. Wu, Y.; Shen, L.Y.; Zhang, Y.; Shuai, C.Y.; Yan, H.; Lou, Y.G.; Ye, G. A New Panel for Analyzing the Impact
Factors on Carbon Emission: A Regional Perspective in China. Ecol. Indic. 2019, 97, 260–268. [CrossRef]
38. Perrings, C.A. Economy and Environment: A Theoretical Essay on the Interdependence of Economic and Environmental
Systems; Cambridge University Press: Cambridge, UK, 1987.
39. Barbier, E.B. Valuing environmental functions: Tropical wetlands. Land Econ. 1994, 70, 155–173. [CrossRef]
40. Theil, H. Estimation and Simultaneous Correlation in Complete Equation Systems. In Henri Theil’s
Contributions to Economics and Econometrics; Springer: Dordrecht, The Netherlands, 1992; pp. 65–107.
41. Zellner, A.; Theil, H. Three-Stage Least Squares: Simultaneous Estimation of Simultaneous Equations.
In Henri Theil’s Contributions to Economics and Econometrics; Springer: Dordrecht, The Netherlands,
1992; pp. 147–178.
42. Kennedy, P. A Guide to Econometrics; MIT Press: Cambridge, MA, USA, 2003.
43. Hamilton, C.; Turton, H. Determinants of Emissions Growth in OECD Countries. Energy Policy 2002, 30,
63–71. [CrossRef]
44. Soytas, U.; Sari, R. Energy Consumption, Economic Growth, and Carbon Emissions: Challenges Faced by
an EU Candidate Member. Ecol. Econ. 2009, 68, 1667–1675. [CrossRef]
45. Kraft, J.; Kraft, A. On the Relationship between Energy and GNP. J. Energy Dev. 1978, 3, 401–403.
46. Viguier, L. Emissions of SOx, NOx, and CO2 in Transition Economies: Emission Inventories and Divisia
Index Analysis. Energy J. 1999, 20, 59–87. [CrossRef]
47. Selden, T.M.; Forrest, A.S.; Lockhart, J.E. Analyzing Reductions in US Air Pollution Emission: 1970 to
1990. Land Econ. 1999, 75, 1–21. [CrossRef]
48. Zhang, Z. Decoupling China’s Carbon Emissions Increase from Economic Growth: An Economic Analysis
and Policy Implications. World Dev. 2000, 28, 739–752. [CrossRef]
49. Pesaran, M.H. General Diagnostic Tests for Cross Section Dependence in Panels; Center for Economic
Studies and Ifo Institute (CESifo): Munich, Germany, 2004.
50. Ucar, N.; Omay, T. Testing for unit root in nonlinear heterogeneous panels. Econ. Lett. 2009, 104, 5–8.
[CrossRef]
51. Kapetanios, G.; Shin, Y.; Snell, A. Testing for a unit root in the nonlinear STAR framework. J. Econom.
2003, 112, 359–379. [CrossRef]
52. Im, K.S.; Pesaran, M.H.; Shin, Y. Testing for unit roots in heterogeneous panels. J. Econom. 2003, 115,
53–74. [CrossRef]
53. National Bureau of Statistics of the Republic of China. Statistical System and Classification Criteria;
National Bureau of Statistics of the Republic of China: Beijing, China, 2018. (In Chinese). Available online: http:
//www.stats.gov.cn/tjzs/cjwtjd/201308/t20130829_74318.html (accessed on 11 January 2019).
54. Wang, Q.; Zhao, Z.; Shen, N.; Liu, T. Have Chinese Cities Achieved the Win-Win between Environmental
Protection and Economic Development? From the Perspective of Environmental E ciency. Ecol. Econ. 2015, 51,
151–158. [CrossRef]
55. Yang, Z.; Dunford, M. City shrinkage in China: Scalar Processes of Urban and Hukou population Losses.
Reg. Stud. 2018, 52, 1111–1121. [CrossRef]
 Int. J. Environ. Res. Public Health 2020, 17, 2761 20 of 20

56. National Bureau of Statistics of the Republic of China. National City CPI and City Classification. 2018. (In
Chinese). Available online: http://www.stats.gov.cn/tjfw/tjzx/tjzxbd/201811/t20181121_1634999 (accessed on 11
January 2019).
57. Merlevede, B.; Verbeke, T.; De Clert, M. The EKC for SO2: Does firm size matter? Ecol. Econ. 2006, 59,
451–461. [CrossRef]
58. Galeotti, M.; Lanza, A.; Pauli, F. Reassessing the Environmental Kuznets Curve for CO2 Emissions: A
Robustness Exercise. Ecol. Econ. 2006, 57, 152–163. [CrossRef]
59. Galeotti, M.; Manera, M.; Lanza, A. On the Robustness of Robustness Checks of the Environmental
Kuznets Curve Hypothesis. Environ. Res. Econ. 2009, 42, 551–574. [CrossRef]
60. Wang, F.C.; Guo, Q.Y. The Influence of Economic Growth to Environmental Pollution and Regional
Heterogeneity. J. Shanxi Financ. Econ. Univ. 2014, 36, 14–27.
61. Li, H.; Haynes, K.E. Economic Structure and Regional Disparity in China: Beyond the Kuznets Transition.
Int. Reg. Sci. Rev. 2011, 34, 157–190. [CrossRef]
62. Korea Ministry of Environment. Basic Plans for Air Quality Improvement Plan for Seoul Metropolitan Areas;
Korea Ministry of Environment: Sejong City, Korea, 2005.
63. Han, C.W.; Lim, Y.H.; Yorifuji, T.; Hong, Y.C. Air Quality Management Policy and Reduced Mortality Rates in
Seoul Metropolitan Area: A Quasi-Experimental Study. Environ. Int. 2018, 121, 600–609. [CrossRef]
64. Korea Ministry of Environment. Basic Plans for Second Phase of Air Quality Improvement Plan for Seoul
Metropolitan Areas (2015–2024); Korea Ministry of Environment: Sejong City, Korea, 2013.
65. Si, L.P.; Jin, Y.J.; Wu, Z.Q. Comparison of Environmental Policy Tools Between China and South Korea. J.
Shandong Univ. Adm. 2017, 1, 20–26. (In Chinese)

' 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open
access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license
(http://creativecommons.org/licenses/by/4.0/).