1"|"P age"
%
A Thesis Paper
On the URBAN AIR POLLUTION:
AN ANALYSIS OF THE
ENVIRONMENTAL KUZNETS CURVE
IN ASIAN CITIES
Submitted by:
GATDULA, Valerie B.
TOLENTINO, Richmay Anne C.

To:
Dr. Rosalina Tan

problem as they are indicative of the continuous growth in economic activity (Anderson &
Brooks, 1996).
Sustainable development has significant implications on the extent of economic
activity in the future. Anderson and Brooks (1996) elaborate saying, “scientific basis
supporting the relationship between business activity, resource depletion and the environment
has grown stronger in recent years.”After all, economic activity is limited and defined by the
state of the environment in which businesses operate, get raw materials from, etc.
The call for sustainable development has been even more urgent for Asian countries
where majority of economic growth is happening and where two of the most populous
countries in the world China and India are located. Anderson and Brooks (1996) note the
implications of having the two most populous cities of the world in Asia the exponential
increase in pollution levels given the magnitude of economic activity in the area, as well as
the alarming damage it may cause to human beings given the high population level in the
region.
In spite of the magnitude of importance of studying and determining the mechanisms
between income and the environment in Asia, there have been limited studies on the subject
matter. As discussed during an interview with Ms. May Ajero of Clean-Air initiative for
Asian Countries (CAI-ASIA), there is no quantitative study yet which analyzes the empirical
relationship between income and air pollution levels (2010).
3"|"P age"
%
B. Objective of the Research
In line with the importance of establishing or disproving the income-environment
relationship in Asia, this paper will conduct a regression analysis of three air pollutants (PM-
10, SO2, and NO2) on per capita income through the Environmental Kuznets Curve (EKC)
equation. The regression will be made for a panel data of seven Asian countries observed for
a period of eight years. The contribution of this paper is the creation of a scientific
relationship between income and pollution backed by empirical data. This is not only of
academic importance; rather, it brings significant policy implications. After all, research
studies are one of the bases of policies made. For instance, observations of the EKC in certain


5"|"P age"
%
D. Methodology
Air pollution measurements for seven Asian countries (China, Hong Kong, India,
Japan, South Korea, Singapore and Thailand) over eight years (1998-2005) were obtained
through CAI-ASIA. There were three air pollutants observed: PM-10, SO2 and NO2.
National income and population levels used to compute for income per capita were obtained
from the World Bank database. Population density, industrialization level, R&D expenditure,
Gross capital formation and road sector energy consumption data were obtained from the
World Bank database. The pollutants were regressed on the of income per capita (its square
and cubic forms), 3-year lag GDP per capita (its square and cubic), population density,
industrialization level, R&D expenditure, gross capital formation and road sector energy
consumption levels of the seven countries for eight years. The regression equation used was
the Grossman and Krueger EKC equation. Panel regression was conducted while holding for
fixed effects to control for time-constant factors that affect Y.
However, as the cubic coefficients are observed to be insignificant, they are dropped
altogether and analysis focuses on the squared form of the equation.

A. Theoretical
According to Bruvoll and Medin (2002), the Environmental Kuznets Curve (EKC)
was postulated due to the increasing concern on the relationship between economic growth
and the environment (i.e., increase air, water and land pollution, etc).
The EKC describes the relationship between the concentrations of pollution to a
country’s income per capita; as a country starts to develop, air pollution level rise. However,
after a certain income per capita, pollution levels begin to decrease as the country is able to
invest in more efficient technologies and production methods.

Figure 1: Environmental Kuznets Curve having an inverted-U shape. Shows the relationship of air
pollution relative to the level of development of a country. (Peters & Murray, 2006)
The EKC is associated with the development stages of a country. During the
agricultural stage, a country has low levels of income per capita at the same time it also has
low levels of pollution. As it approaches the industrial stage, there is an increase in the
production of goods and as such increase in air pollution. This is mainly brought about by
factory outputs and the use of excessive fossil fuel to run the machines for production. An
improvement in air quality begins to follow as a country stars to invest in technology. This is
clearly depicted by the diagram below. As one can see, the quality of air pollution depends on
the level of income per capita. Furthermore, based on the theory it follows an inverted-U
shape.
8"|"P age"
%
Stage 1
Air pollution concentration
Stage 2
Start of
industrial
development
Initiation of
emissions

refers to the technology as a percent of GDP. A higher technology composition effect
improves the state of the environment as there are more efficient means of manufacturing and
producing goods. A higher technology composition is assumed to imply more sophisticated
end efficient technology that is beneficial for the environment. Last, is the technique effect.
Technique pertains to the research and development (R&D) of a country. Countries with
better techniques experience improving environment conditions as R&D enables the country
to discover means and ways of doing things that are more efficient. That is, technique leads
to the substitution of crude production processes to more efficient and cleaner ones. The first
9"|"P age"
%
effect demonstrates the negative effect of development on the environment, happening during
the early stages. On the other hand, the last two shows how the environment would improve
as brought about by more economic progress.
Furthermore, Borghesi (1999) suggested that market signals or the ‘existence of an
endogenous self-regulatory market mechanism for the use of natural resources’ may also
explain the shape of the EKC. According to him, during the early stages of development there
is heavy exploitation of natural resources leading to a reduction of natural capital. However,
at a certain time, there comes an increase in the price of natural resources. This leads to a
reduction in its exploitation. Furthermore, there is an ‘accelerated shift towards less resource-
intensive technologies’.
In addition, (Yandle et al., 2004) offers another reason as to why the EKC is shaped,
as it is. According to his reasoning, environment quality is a luxury good at higher levels of
income. This indicates that ‘the income elasticity of demand for environmental resources
varies with the level of income’. As a country is at its early stages of development, the
income elasticity for such is less than one. However, after a certain threshold the income
elasticity becomes greater than one. That is, the change in demand for high quality
environment becomes larger than the change in income. The increasing demand for good
quality environment results in an improvement in the environment.
Regarding the limitations of the EKC, Stern (2004) offers a comprehensive study
regarding of its theoretical confines. First, there is ‘no feedback from environmental damage

%
B. Empirical
In an empirical analyses of the EKC, two topics are of main interest: first, the
calculation of the threshold where environmental quality improves with rising per capita
income and second, whether a given indicator of environmental degradation displays an
inverted- U relationship in association with rising levels of per capita income.
In terms of the calculated threshold, studies done by Grossman and Krueger (1991),
Shafik and Bandopadhyay (1992) and Selden and Song (1994) would be used as basis of
comparison due to their extensive research and well documented study.
Grossman and Krueger (1991) analyzed the EKC relationship in the context of the
North American Free Trade Agreement (NAFTA) and used the EKC-based hypothesis to
argue that a NAFTA-based trade expansion would protect the environment. They used sulfur
dioxide and dark matter (smoke) suspended in the air in order to estimate the environmental
conditions. Their results showed that turning point came when per capita GDP was in the
range of $4,000 to $5,000 measured in 1985 U.S. dollars, which is approximately $6,700 to
$8,450 in 2003 dollars. Unlike the relationship found for sulfur dioxide and smoke, no
turning point was found for suspended particulates. In this case, the relationship between
pollution and GDP was monotonically increasing. As GDP per capita rose, so did this form of
pollution. (Yandle et al., 2004). Furthermore, Grossman and Krueger’s study looked into the
effect of other factors such as population density and the type of land on pollution levels. As
these factors are not correlated to income level, they are not necessary to make the equation
unbiased. However, Grossman and Krueger noted that the addition of these variables “reduce
residual variance and make the coefficients more precise.” Lastly, the Grossman and Krueger
included the “cubic of average GDP per capita in the preceding three years to proxy for the
effect of permanent income, and because past income is likely to be a relevant determinant of
current environmental standards” (Grossman and Kruegar, 1995). That is, income three years
12"|"P age"
%
before the time period analyzed has an effect on the time period’s pollution levels as
machinery, equipment and activities employed at the current time is a product of income in

and Krueger (1991), Shafik and Bandyopadhyay (1992) and Selden and Song (1994)
presented evidence that some pollutants have historically followed an inverted U-curve with
respect to income. Although these and other empirical studies point to a correlation between
income and pollution, the causal relation is not observed for all sets of data. That is, there
seems to be a highly specific and controlled environment under which the EKC condition can
be observed. As such, some research like those done by Harbaugh, Levinson and Wilson
(2002), Carson (2009), etc. do not agree with the EKC model due to the limitations of the
theory and the assumptions incorporated in it.
In terms of the shape of the EKC, debates and further studies have shown other
variations from the inverted-U shape originally proposed: cubic function and L-shaped curves.
Torras and Boyce (1998) suggested that instead of a quadratic function, the EKC
actually follows a cubic one. This allows for the possibility that a downturn in pollution (at
the peak of the inverted U) can be followed by a later upturn, that is, a reversal of the
tendency for pollution levels to decline with further increases in per capita income. These
findings imply that beyond some point, high-income levels, rather than being conducive to
further improvement in air and water quality, can have the opposite effect. One possibility is
that the scale effect overshadows the composition and technology effects.
14"|"P age"
%

Figure 3: Environmental Kutznets Cruve: Cubic Function (Torras and Boyce 1998)
Furthermore, a study done by Lucinda Peters and Frank Murray (2006) revealed an L-
shaped EKC as compared to the traditional inverted-U when applied to the Asian context. Air
quality and Gross National Income (GNI) per capita data were collected to create simple


16"|"P age"
%
C. Contribution of Paper to the Study of Economics
This research paper, given the recent data of the different air pollutants (PM-10, SO
2

and NO
2
) in several cities in Asia, will utilize the Environmental Kutznets Curve in order to
determine a two-fold goal: first, if the EKC model exists in the Asian context and second, the
GNI per capita that each pollutant would start to decline if ever it does exist. As compared to
the study published by Lucinda Peters and Frank Murray in 2006, this paper is grounded on
scientific data obtained from the Clean Air Initiative- Asia. That is, it will be able to establish
an econometric relationship between income and environment levels for Asian countries.
Furthermore, it would elaborate on possible explanations based on the existence or non-
existence of the EKC. It will more specifically define the relationship between income and
pollution, as well as the impact of Research and Development (R&D) Expenditure, Road
Energy consumption, Capital Formation, and Population Density on pollution levels. The
highly specific relationship that would be obtained could greatly help in the formulation of
timely and essential policies to improve the state of the environment.

the environment. For this study, 3 of the most widely monitored pollutants are used: PM10,
SO2 and NO2. Taking the logarithm of the pollutant results in a slightly different
interpretation of results the coefficient would be indicative of the effect of the change in the
independent variable on the change in the pollutant. A positive coefficient means that an
increase in the rate of change of per capita GDP results in a similar increase in the rate of
change of pollutant levels. This is different from the interpretation for level variables where a
positive coefficient implies that an increase in the independent variable results in an increase
in the dependent variable. The squaring of the GDP per capita will allow for the
determination of an inverse-U shape as it will reveal the decreasing effect of high levels of
per capita GDP on pollution levels. After all, at high income per capita levels (GDP per
capita) the negative sign of the coefficient of GDP per capita squared will have a decreasing
effect on pollutant levels resulting in the downward turn of the U-shaped curve.
18"|"P age"
%
For this study, the Grossman and Krueger variation of the EKC equation was used
to incorporate more factors that could affect income. A more in depth discussion will be
conducted in the methodology section.


in the inverse-U shaped curve. The Cubic part is there only to provide for a more accurate
measure of the relationship. If indeed there is a U-shaped curve, β
3
will have the same
negative sign as β
2,
implying it will continue to decrease pollution levels. Or alternatively, an
insignificant β
3
also shows that the square is a sufficient indicator of the income-pollution
relationship.
20"|"P age"
%
In addition to the population density factor looked into by Grossman and Krueger,
this paper added 3 other variables: Industry value (% of GDP from industrial sector), Road
Sector Energy Consumption, and Gross Capital Formation. R&D Expenditure pertains to the
amount of money invested in R&D in relation to the country’s total GDP. The World Bank
defines Road Sector Energy consumption as the “total energy used in the road sector
including petroleum products, natural gas, electricity, and combustible renewable waste.”
This is indicative of the level of vehicle activity which inevitable affects pollution levels. Last,
this paper also looks into gross capital formation or the “outlays on additions to the fixed
assets of the economy plus net changes in the level of inventories.” (World Bank) The level
of capital formation is indicative of industrial and economic activity including “land
improvements (fences, ditches, drains, and so on); plant, machinery, and equipment
purchases; and the construction of roads, railways, and the like, including schools, offices,
hospitals, private residential dwellings, and commercial and industrial buildings. Inventories
are stocks of goods held by firms to meet temporary or unexpected fluctuations in production
or sales.” That is, higher capital formation is indicative of higher industrial and economic
construction resulting in pollution a few years hence.


it
+ c G
it
^2+ d G
it
^3+ e G
it
+ f G
it
^2+ g G
it
^3 + X
it
+ u
SO
2
= a + b G
it
+ c G
it
^2+ + d G
it
+ e G
it
^2+ X
it
+ u
NO
2
= a + b G

Git^3 and Xit. Grossman and Kruegar used Xit as a representation for population density,
land use, and distance from desert areas. For this study, Xit stands for population density,
industrial level, Road Sector Energy Consumption and Gross Capita Formation.
Environmental condition is depicted by the level of air pollutant while development is
captured by the income per capita. The square measures nonlinearities in the time path of
pollution while the cube allows for flexibility in determining the relationship between income
and pollution. An EKC relationship will result in an insignificant coefficient for the cubic
factors, as well as a positive sign for income per capita and lagged income per capita, and a
negative sign for their respective squares. Thus, supporting the inversed U-shape of the
theory.
The potential source for the fragility of the results can be brought about by multi-
collinearity. As might be expected, there is a high degree of correlation between the per
capita income, its square, cube and lagged versions. 22"|"P age"
%
B. Data For the Yit or the pollutant level, Particulate Matter (PM-10), Sulfur Dioxide (SO
2
)
and Nitrogen Dioxide (NO
2
) levels would be utilized. The data was gathered from Clean Air
Initiative- Asia (CAI-Asia, 2010).
According to literature, among air pollutants, these three are the more documented
ones as these are some of the ones earlier discovered leading to the development of capacity
to measure such compounds. Also, it is important to consider the three pollutants due to their

of adverse effects on the respiratory system (US EPA, 2010).
The Gross Domestic Product was calculated at purchaser's prices is the sum of gross
value added by all resident producers in the economy plus any product taxes and minus any
subsidies not included in the value of the products. It is calculated without making deductions
for depreciation of fabricated assets or for depletion and degradation of natural resources.
Data are in current U.S. dollars. Dollar figures for GDP are converted from domestic
currencies using single year official exchange rates. For a few countries where the official
exchange rate does not reflect the rate effectively applied to actual foreign exchange
transactions, an alternative conversion factor is used. The data was sourced from World Bank
national accounts data, and OECD National Accounts data files. Using this data, the lagged
variables were easy to determine (World Bank, 2011).
Industry value, Road Sector Energy consumption and Gross Capital Formation data
were all obtained from the World bank database.
The seven countries observed for PM-10 and SO
2
are: China, Hong Kong, India,
Japan, Korea, Singapore, and Thailand. For NO
2
, the five countries observed are: Hong Kong,
India, Japan, Singapore and Thailand. While the NO2 regression involved: Hong Kong,
Japan and Singapore. For all regressions, data for years 1998-2005 were used. The timeframe
of the paper’s analysis is only for eight years due to the constraints in pollutant level data.
24"|"P age"
%
Furthermore, the capital of the country was chosen to represent that nation’s status of air
quality as this city contains the most data.
Regression for panel data was conducted, controlling for time differences through the
fixed effects model. That is, the data points were taken as non-random occurrences, and the
model controlled for factors in such a way that characteristics of the data do not change over
time.

-8.14665e-08
Population Density
-0.00123877
-0.00843698
-0.0102944
Road Sector
Energy
Consumption
-1.62134**
-1.72679**
0.354297
Gross Capital
Formation (-9)
-1.93209**
-0.495216
0.0049***
R-squared
0.975041
0.909464
0.951947
Adjusted R-
squared
0.958402
0.849107
0.906297

Given that β
3
exponents were not significant, these were no longer taken into
consideration. The summarized results of regression with cubic parameters can be seen in the