Escolar Documentos
Profissional Documentos
Cultura Documentos
AND DEVELOPMENT
Copyright 2014
*Mohamed Arouri is a Researcher at EDHEC Business School, France. His research areas are
energy economics and risk management. The authors publications have appeared in Journal of
Banking & Finance, Journal of Macroeconomics, Energy Economics, Journal of International
Money and Finance, Ecological Economics, Manchester School, Journal of Energy and Development, Macroeconomic Dynamics, and Applied Economics, as a sampling.
Muhammad Shahbaz is an Assistant Professor in the Department of Management Science,
COMSATS Institute of Information Technology, Lahore, Pakistan. His research areas are development and energy economics. The authors publications have appeared in Energy Policy,
Energy Modelling, Energy Economics, Renewable and Sustainable Energy Reviews, Energy: The
International Journal, The Renewable Energy-Journal, and Defense and Peace Economics.
Rattapon Onchang is an Assistant Professor at the Department of Environmental Science,
Faculty of Science, Silpakorn University, Thailand. His research areas are pollution, climate change,
and environmental and natural resource management. The authors publications have appeared in
Environmental and Natural Resource Journal, Journal of Environmental Sciences, and The Journal
of Energy and Development, as a sampling.
Faridul Islam is a Professor and Chair of the Department of Economics at Morgan State
University. His research areas are development and energy economics. The author has published in
Economics Letters, Empirical Economics, Journal of Economic Education, Journal of Asian
Economics, Economic Change and Restructuring, The International Trade Journal, International
Journal of Social Economics, and Indian Economic Review.
Frederic Teulon is the Executive Head of Research at IPAG Business School (Paris, France). His
research areas are international economics, energy economics, and financial markets. The authors
publications have appeared in Energy Policy, Economic Modelling, International Economics,
Review of Accounting and Finance, Journal of Applied Business Research, and The Journal of
International Financial Markets, Institutions & Money.
The Journal of Energy and Development, Vol. 39, Nos. 1 and 2
Copyright 2014 by the International Research Center for Energy and Economic Development
(ICEED). All rights reserved.
149
150
151
from 16.2 kilo-tons of oil equivalent (ktoe)/billion Thai baht in 2005 to 12.1 ktoe/
billion Thai baht by the end of 2030that is a 25 percent decline. This will result
in a reduction of overall CO2 emissions by 976 million tons.6 The target is to
reduce the use of fossil fuels and CO2 emissions.
Urbanization plays a critical role in energy consumption and, thus, CO2 emissions.7 Over the past decade, population growth in Thailand has slowed. The total
Thai population is projected to increase from 63.4 million in 2008 to 71 million by
2028. The share of urban population relative to the total population also has increased as a result of economic growth, especially in the industrial areas, which
have become the centers of economic growth.8 This has contributed further to the
acceleration of urban population growth and the expansion of cities. It is forecasted
that approximately one-half of the Thai population will move to urban areas during
the next 50 years.9 Per-capita income in Thailand is still relatively low. The Thai
government likely will have to exert significant efforts to boost economic growth in
order to meet the needs of an exploding urban population. Moreover, a higher urban
population has implications for energy use and, similarly, on the environment.
S. Hossain investigated the dynamic causal relationships among carbon dioxide
emissions, energy consumption, economic growth, trade openness, and urbanization
for a panel of nine newly industrialized countriesnamely, Brazil, China, India,
Malaysia, Mexico, Philippines, South Africa, Thailand, and Turkey.10 The results
found that over time higher energy consumption in the panel of newly industrialized
countries gives rise to more CO2 emissions; subsequently, our environment will be
polluted more. The findings further suggest that variables such as trade openness
and urbanization have significant and negative impacts on carbon emissions. P. K.
Narayan and S. Narayan tested the environmental Kuznets curve (EKC) hypothesis
for 43 developing countries using both time-series and panel data estimation
techniques.11 They find that only data for the Middle Eastern and South Asian panels
show emissions are lower in the long run compared to the short run, implying that
CO2 emissions have fallen with a rise in income. A. Kukla-Gryz analyzed the
impact of economic growth and international trade on the level of air pollution by
using the estimation of the structural equation model with two factors describing the
structure of economic activity and air pollution intensity.12 The results suggest that
in developing countries, both international trade and per-capita income lead to
changes in the structure of economic activity andas a consequenceto an increase in air pollution. In addition, results indicate that the impact of economic
growth on air pollution intensity in the developing countries occurs through the
change of the structure of economic activity.
Although there are a number of studies that examined the existence of EKC for
Thailand, e.g., S. Hossain, P. K. Narayan and S. Narayan, and A. Kukla-Gryz, all
of the research incorporated Thailand only among a set of comparative countries.
This shows that applications devoted to Thailand and use single country-specific
time series of multiple causalitieseconomic growth, energy consumption,
152
urbanization, international trade, and CO2 emissionsseem not to have been carried
out. Therefore, this study contributes to the existing energy literature in four ways:
(i) it is the first comprehensive effort to explore the relationship among economic
growth, energy consumption, and CO2 emissions by incorporating trade openness
and urbanization in a CO2 emissions function for the case of Thailand; (ii) we
apply the Zivot-Andrews (ZA) structural break unit root test to assess the order of
integration of the variables; (iii) a dummy is used to accommodate structural
breaks while applying the autoregressive distributed lag (ARDL) bounds methodology testing the long-run relationship between the variables; and (iv) the vector
error-correction model (VECM) is used to detect the direction of causality between the variables. Our results found cointegration between the variables and the
EKC exists for Thailand, which is confirmed further by unidirectional causality
running from economic growth to CO2 emissions.
In the next section we will review the relevant literature, which will be followed by
an explanation of the modeling and estimation strategy. We then present the interpretations of our results and end by offering our conclusions and policy implication.
Review of Literature
As an outgrowth of the literature on the links among energy consumption,
economic growth, and environmental degradation, the environmental Kuznets
curve has gained significant popularity. The EKC relationship posits that as
economic growth rises (measured by per-capita income), at the initial stage energy
pollutants increase; but, after a certain threshold income has been reached, pollutants start falling. This produces an inverted U-curve. Researchers have examined the validity of EKC both for individuals as well as for groups of nations. The
conceptual framework owes much to the work of G. Grossman and A. Krueger.13
Since then, a burgeoning literature has emerged.14 Despite this, no clear consensus
has occurred. Using cross-national panel data, T. Selden and D. Song examined
the relationship between economic growth and environmental degradation.15 They
considered four air pollutants: particulate matter, sulfur dioxide, nitrogen oxides,
and carbon monoxide. They discovered an inverted-U relationship between emissions and per-capita GDP.
M. Cole et al. examined the relationship between per-capita income and a wide
range of environmental indicators using cross-country panel data to test for the
existence of the EKC.16 The results suggest that EKC is found for local air pollutants even as indicators with a more global or indirect impact either increase
monotonically with income or else have predicted turning points at high per-capita
income levels. These findings offer valuable insight for crafting sustainable environmental policy. The inverted-U relationship between economic growth and
CO2 emissions estimated from panel data need not hold for specific individual
153
countries over time.17 They found that the time patterns for air pollutants and CO2
emissions correlate positively with economic growth and that emissions reduction
requires structural and technological changes in the economy. They point out that
sustainable growth implies the rate of economic growth, which does not further
aggravate pollution. Concern about climate change has prompted several analysts
to examine the relationship between economic activity and energy consumption
and/or carbon emissions, e.g., F. Halicioglu, and N. Apergis and J. Payne.18 D.
Holtz-Eakin and T. Selden were the first to explore the relationship between
economic growth and CO2 emissions.19 Estimates derived from global panel data
(130 countries) suggest a diminishing marginal propensity to emit CO2 as GDP per
capita rises. The relationship exhibited a monotonic, not inverted-U, type.
Various studies investigated the relationship among economic growth, energy
consumption, and CO2 emissions and confirmed the existence of the environmental
Kuznets curve. For example, H. Pao and C. Tsai for Brazil, Russia, India, and
China; I. Ozturk and A. Acaravci for Turkey; A. Acaravci and I. Ozturk for Europe;
M. Nasir and F. Rehman and M. Shahbaz et al. in Pakistan; H. Iwata et al. in 28
developed and developing countries; B. Saboori et al. in Indonesia; B. Saboori et al.
in Malaysia; M. Shahbaz et al. for Romania; and A. Tiwari et al. for India, validated
the presence of the EKC.20
Many of the earlier studies of EKC applied bivariate models in the case of
Thailand and, thus, potentially suffer from the problem of omitted variables bias.
D. Stern pointed out that the EKC results have a very flimsy statistical foundation.21
In particular, he argues that a new generation of decomposition and efficient frontier
models might help disentangle the true relations between the series, which may even
lead to the demise of the EKC. Contemporary approaches that have succeeded in
avoiding the pitfalls include cointegration and Granger causality within a multivariate framework. With the advent of alternative statistical models like the autoregressive distributed lag procedure (ARDL), the research has been extended to
several nations. The models have been implemented for both individual and groups
of developing countries.
154
Ct f Yt ; Yt2 ; Et ; Tt ; Ut :
We use a non-linear functional form (in variables) to specify the relation. All series are
transformed into logarithms. The estimable equation, thus, takes the following form:
lnCt b1 bY lnYt bY 2 lnYt2 bE lnEt bT lnTt bU lnUt mt
where, C refers to carbon emissions per capita; E is energy consumption per capita;
Y and (Y2) refer to real GDP per capita and its square, respectively; T is trade
openness per capita; U is urban population as a share of the total population; and
m is a random error term, obeying the standard assumptions. The EKC hypothesis
requires that bY > 0 and bY2 < 0. We expect energy use to increase pollutants, bE > 0.
The sign of the coefficient of T can go either way. If bT < 0, pollution is reduced due
to strict environmental laws and helped by the importation of capital and technology
that is environmentally favorable. G. Grossman and A. Krueger and F. Halicioglu
argue that if bT > 0, then polluting industries in developing economies might be
shifting production that generates more emissions, which would be consistent with
the safe-haven hypothesis.24 Urbanization leads to higher demands for energy and,
thus, environmental degradation. We expect bU > 0.
We use the ARDL bounds test approach to examine cointegration among the
series developed by M. Pesaran et al.25 The ARDL model applies regardless of the
order of integration.26 Most macroeconomic variables are I(1). A. Haug argues that
the ARDL approach for cointegration is preferable because it is better at small sample
properties compared to other methods.27 Also, the unrestricted error-correction model
(UECM) with appropriate lags captures the data-generating process within the
general-to-specific framework.28 An appropriate modification of the orders of the
ARDL model is sufficient to simultaneously correct for residual serial correlation
and endogeneity problems.29 The following UECM is used to examine the long- and
short-run relationships among the series.
Xp
Xq
lnCt a0 aDUM DUM
b
DlnC
d DlnYt
t1
i
i1
i0 i
Xr
X
X
u
v
s DlnYt 2
e DlnEt
v DlnTt
io i
i0 i
i0 i
Xs
In the first part of equation (3), b, d, e, s, and v capture the short-run parameters, while lC, lY, lY2, lE, lT, and lU capture the long-run relations among
the series and DUM is the dummy variable.30 The hypothesis of no cointegration,
H0: lC = lY = lY2 = lE = lT = lU = 0, is tested against the alternate of cointegration, i.e., Ha: lC 6 lY 6 lY2 6 lE 6 lT 6 lU = 0. The cointegration decision is
155
based on the critical bounds tabulated by M. Pesaran et al. against the computed
F-statistic.31 The upper critical bound (UCB) assumes that the variables are
integrated at I(1) and the lower critical bound (LCB) assumes they are I(0). If UCB
is less than the F-statistic, the decision favors a long-run relationship among the
variables. If the F-statistic is less than LCB, there is no cointegration. The decision
about cointegration will be inconclusive if the F-statistic lies between UCB and
LCB.32
After establishing cointegration exists among the series, we turn to test
the direction of causality between the pairs of economic growth, energy consumption, trade openness, urbanization, and carbon emissions. The vector
error-correction method is suitable for causal relations between the variables if
the series are found to be stationary at I(1).33 All the series are endogenous in
the system of VECM, which shows that the response variable is explained both
by its own lags and the lags of independent variables as well as the errorcorrection term and residual term. The VECM for the five variables case can be
written as follows:
Xl
Xm
a
DlnC
a22 DlnYtj
DlnCt a01
11
ti
Xn i1
Xo j0
Xp
2
a
DlnY
a
DlnE
a DlnTts
33
44
tr
tk
r0
s0 55
Xk0
q
a DlnUtu h1 ECTt1 m1 i
u0 66
Xl
Xm
DlnYt b01
b
DlnY
b22 DlnCtj
ti
11
Xn i1
Xo j0
Xp
2
b
DlnY
b
DlnE
b DlnTts
tr
33
44
tk
r0
s0 55
Xk0
q
b DlnUtu h2 ECTt1 m2 i
u0 66
Xl
Xm
2
f
DlnY
f22 DlnCtj
DlnYt2 f01
11
t1
Xn i1
Xo j0
f33 DlnYtk
f DlnEtr
Xk0
Xqr0 44
p
f DlnTts
f DlnUtu h3 ECTt1 m3 i
s0 55
u0 66
Xl
Xm
DlnEt u01
u
DlnC
u22 DlnYtj
ti
11
Xn i1
Xo j0
2
u DlnYtk
u44 DlnEtr
k0 33
Xp
Xqr0
u DlnTts
u DlnUtu h4 ECTt1 m4 i
s0 55
u0 66
156
Xl
Xm
DlnTRt d01
d
DlnTR
d DlnCtj
11
ti
i1
j0 22
Xn
X0
2
d DlnYtk
d DlnYtk
k0 33
r0 44
Xp
Xp
d
DlnE
Xl
Xm
DlnUt q01
q
DlnU
q22 DlnCtj
11
ti
Xn i1
Xo j0
2
q DlnYtk
q DlnYtk
k0 33
r0 44
Xp
Xq
q
DlnE
where uit are error terms and assumed to be identically, independently, and normally distributed (N;(iid)). The statistical significance of the lagged error term,
i.e., ECTt1, further validates a long-run relationship between the series. The estimated ECTt1 also shows the speed of convergence from the short-run toward the
long-run equilibrium path. The VECM helps to test the causal relation once series
are found to be cointegrated. In such cases, causality must be present at least from
one direction. Furthermore, the VECM can distinguish between short- and longrun causal relationships and detect causality in the long run, short run, or both.34
A statistically significant t-statistic on an estimated lagged error term, i.e.,
ECTt1, with a negative sign confirms the existence of a long-run causal relation.
Short-run causality is indicated by the joint c2 test on the estimated first differenced lagged independent variables. For example, a significant a22,i 6 0"i implies
that economic growth Granger causes CO2 emissions and causality runs from the
latter to the former. A similar inference can be drawn about other causality hypotheses. Finally, we use the Wald or F-tests to test the joint significance of estimated lagged terms of independent variables and the error-correction term,
which further confirms the existence of short- and long-run causality, known to
measure strong Granger causality.35 In addition to sensitivity analysis, parameter
stability and goodness of fit for the ARDL model are checked by the cumulative
sum of recursive residuals (CUSUM) and the cumulative sum of squares of recursive residuals (CUSUMSQ).
Results and Empirical Discussions
While not necessary, the first step is to find out the order of integration of the
variables prior to applying the ARDL bounds testing for the long-run relationships
between the series. This is needed to insure that none of the series is I(2). In
the literature, various methods are available to test the stationarity properties of
the variables, such as the augmented Dickey-Fuller (ADF) by D. Dickey and
W. Fuller, the Dickey-Fuller Generalized Least Squares (DF-GLS) method by
157
G. Elliott et al., and the Ng-Perron by S. Ng and P. Perron.36 The results of these
tests imply that the variables are integrated at I(1).37 C. Baum points out that these
aforementioned tests may be biased in the presence of structural breaks.38 To
overcome this issue, we use the Zivot-Andrews structural break unit root test.39
The Zivot-Andrews test allows information about one structural break in the series
to have one structural break point. The results are robust with the findings of ADF,
DF-GLS, and Ng-Perron. The results reported in table 1 suggest that the variables
are first difference stationary with intercept and trend, i.e., I(1). This sets the stage
for applying the ARDL bounds testing approach to cointegration for a long-run
relationship among the variables for Thailand over the period 19712010.
The ARDL bounds test statistic is sensitive to the selection of lag length.
H. Lutkepohl documents that dynamic link among the series can be captured if the
proper lag is used.40 The selected lag length of 2, using the Akaike information
criterion (AIC), is reported in table 2.
The next step is to examine a long-run relationship among the variables. The
results of the ARDL bounds testing approach to cointegration are reported in
table 3. Our findings show that the calculated F-statistics, i.e., 11.6357, 28.0861,
and 7.4809, exceed the upper critical bounds at the 1-percent and 10-percent
levels of significance when CO2 emissions, energy consumption, and trade openness are used as predicted variables. Our sample consists of 40 observations
(19712010); thus, the critical values from M. Pesaran et al. are inappropriate.41
As such, we chose to use the lower and upper critical bounds generated by
P. Narayan.42 We find three cointegration vectors and, thus, a long-run relationship
among economic growth, energy consumption, trade openness, urbanization, and
CO2 emissions for Thailand over the study period of 19712010.
Table 1
ZIVOT-ANDREWS STRUCTURAL BREAK UNIT ROOT TEST RESULTS
st
At 1 Difference
At Level
Variable
T-Statistic
Time Break
T-Statistic
Time Break
ln Ct
ln Yt
2
ln Yt
ln Et
ln Ut
ln Tt
2.917(1)
3.726 (1)
3.683 (1)
3.148 (1)
3.381 (0)
3.937 (1)
1990
1988
1988
1989
1988
1976
5.253(0)**
4.876(0)***
4.913 (0)***
6.587 (0)*
7.150 (0)*
7.575 (0)*
1987
1997
1997
1984
1987
1981
158
LogL
LR
FPE
AIC
SC
HQ
236.0789
561.4300
602.0704
NA
530.8359
53.4742*
2.22e-13
5.54e-20
5.02e-20*
12.1094
27.3384
27.5826*
11.8508
25.5284*
24.2212
12.0174
26.6944*
26.3867
* indicates lag order selected by the criterion; LR = sequential modified LR test statistic (each
test at the 5-percent significance level); FPE = the final prediction error; AIC = the Akaike
information criterion; SC = the Schwarz information criterion; HQ the Hannan-Quinn information
criterion; and NA = not available.
In addition to the ARDL approach, we also implemented the Johansen cointegration approach to test the robustness of the long-run relationships. The results
in table 4 show two cointegration vectors, as confirmed by trace test statistics and
maximum Eigen values at the 1-percent and 5-pecent level of significance. This
validates a long-run relationship, which is robust in the long-run results.
We now discuss the marginal impacts of economic growth, energy consumption, trade openness, and urbanization on CO2 emissions in the long run as well as
in the short run. The long-run results reported in table 5 indicate that both linear
and non-linear terms of real GDP per capita confirm an inverted-U relationship
between economic growth and CO2 emissions. The results suggest that a 1-percent
Table 3
AUTOREGRESSIVE DISTRIBUTED LAG PROCEDURE (ARDL)
a
COINTEGRATION ANALYSIS
Variable
F-statistics
Break year
Critical values #
Lower bounds
Upper bounds
2
Adjusted R
F-statistic
a
ln Ct
ln Yt
ln Yt
ln Et
ln Tt
ln Ut
11.6375*
1990
1% level
10.150
11.130
0.9703
19.8202*
1.5483
1988
5% level
7.135
7.980
0.9994
26.0811*
1.5345
1988
10% level
5.950
6.680
0.9994
24.7990*
28.0861*
1989
7.4809***
1976
4.2474
1988
0.9702
19.6792*
0.6410
3.5003**
0.9367
21.7236*
159
Trace Statistic
161.0758*
81.7760*
44.7516
21.1809
9.0893
0.0072
79.2998*
37.0243**
23.5707
12.0915
9.0821
0.0072
rise in real GDP per capita will increase CO2 emissions per capita by 7.2369
percent while a negative sign of the estimate, i.e., 0.3057 of the squared term,
establishes the inverted relationship between CO2 emissions and real GDP per
capita. This evidence confirms that CO2 emissions increase at the initial stage of
economic growth and decline after a threshold point, that is, EKC exists. In other
words, GDP influences the efficient consumption of energy, resulting in a reduction of CO2 emissions. This implies that energy demand-side-management
policies should be launched by applying energy-efficient technology by public
organizations to reduce energy consumption. The use of energy-efficient machines
and appliances helps to accomplish this goal, something that potentially can be
achieved at higher levels of income. These findings are consistent with those of
T. Song et al., F. Halicioglu, M. Fodha and O. Zaghdoud, H. Lean and R. Smyth,
M Shahbaz et al. (2012 and 2013), and A. Tiwari et al.43 The impact of energy
consumption on energy pollutants is positive and statistically significant at the
1-percent level. We find that a 0.6424-percent increase in CO2 emissions is linked
with a 1-percent increase in energy consumption. Energy consumption is a major
contributor to environmental degradation. CO2 emissions have increased from
power generation and industrial sectors that are using more natural gas and lignite/
coal. In 2011, total lignite/coal consumption was 19 million tons, an increase of
5.5 percent from 2010 (based on the heating value); natural gas consumption in
2011 rose by 2.6 percent compared to that in 2010.44 Although air pollution
emissions control technologies for these fuels are in place, the desired reduction
in CO2 emission levels remains unsatisfactory. The dilemma for Thailand is to
balance the needs of economic growth with energy consumption and greenhouse
gas emissions in the long run. It is forecasted that natural gas demand in 2011
(excluding the use of feedstock in gas separation plants) will increase by 6.1
percent compared with that in 2010.
160
Dependent Variable: ln Ct
Long-Run Results
Variable
Coefficient
Constant
40.3455
7.2369
ln Yt
2
ln Yt
0.3057
ln Et
0.6424
ln Tt
0.1699
ln Ut
2.1404
Short-Run Results
Variable
Coefficient
Constant
0.0201
ln Yt
7.0400
2
ln Yt
0.3005
ln Et
0.4182
ln Tt
0.0962
ln Ut
2.3064
0.6136
ECMt1
2
R
0.7880
2
Adjusted R
0.7456
F-statistic
18.5899*
Diagnostic test
F-statistic
2
x NORMAL
0.0357
2
x SERIAL
1.2763
2
x ARCH
0.7092
2
x WHITE
1.1725
2
x REMSAY
0.7594
Standard Error
7.5585
1.5350
0.0772
0.1634
0.0972
0.4031
t-Statistic
5.3377*
4.7144*
3.9580*
3.9292*
1.7477***
5.3090*
Standard Error
0.0135
3.3750
0.1601
0.1638
0.0728
1.1179
0.1473
t-Statistic
1.4921
2.0859**
1.8763***
2.5525*
1.3219
2.0630*
4.1656*
Prob. value
0.9822
0.2948
0.7046
0.3470
0.4537
161
162
The straight lines represent critical bounds at the 5-percent significance level.
extenuating factors may have influenced other break points during the period such
as urbanization in 1988, energy consumption in 1989, CO2 emissions in 1990, and
a significant reduction in the incidence of poverty. The number of people living in
poverty decreased substantially from nearly 18 million in 1988 to a little less than 7
million in 1996.49 The average annual rate of change of the urban population between 1985 and 1990 was 2.44 percent.50 The demand for energy in Thailand grew
along with an expanding economy. During the period of rapid economic growth,
energy consumption increased by 8 to 10 percent annually and added to CO2
emission from about 165 teragrams (Tg) in 1990 to 202 Tg in 1994.51
The CUSUM and CUSUMsq graphs are not always dependable, so Y. Leow
suggests that we should apply the Chow forecast test to check for structural breaks
in a series.52 Therefore, we applied the test. The results are reported in table 6 and
indicate no evidence of a structural break in the Thai economy over the period of
19841990. Thus, the long- and short-run parameters are stable and appear reliable for policy-making purposes in Thailand.
VECM Granger Causality Analysis: Given the existence of cointegration among
163
Figure 2
PLOT OF CUMULATIVE SUM OF SQUARES (CUSUMsq) OF RECURSIVE RESIDUALS
The straight lines represent critical bounds at the 5-percent significance level.
Knowledge about the direction of causality between the series can help policy
makers in crafting an integrated and sustainable environmental policy. C. Granger
suggested that if the series are first difference stationary and cointegrated to employ
the VECM.53 The results are reported in table 7.
Table 7 reveals that in the long term, there is bidirectional causality between
CO2 emissions and energy consumption, energy consumption, and trade openness,
and trade openness and CO2 emissions. We find unidirectional causality from
economic growth and urbanization to CO2 emissions. The statistical significance
of ECTt1 implies the speed of adjustment in the trade openness equation is
0.7664 as compared to CO2 emissions and energy consumption equations, which
are 0.6574 and 0.6573, respectively.
Table 6
CHOW FORECAST TEST
Chow Forecast Test
F-statistic
Likelihood ratio
Value
P-Value
0.9193
71.8695
0.6169
0.0000
164
ln Ct
ln Yt
4.7830**
[0.0179]
ln Yt
ln Tt
2.7059*** 3.7464**
[0.0871]
[0.0384]
0.8874
[0.4248]
0.8958
[0.4258]
5.4765** -0.6574*
[0.010]
[-3.8751]
3.0029**
[0.0222]
0.8161
[0.4540]
0.2812
[0.7573]
3.8441**
[0.0356]
1.0611
[0.3617]
0.3668
[0.6967]
3.6399**
[0.0416]
4.1918** 3.0008**
[0.0275] [0.0229]
ln Yt
Long
Run
ln Et
ln Ct
Dependent variable
ln Yt
ln Ut
ECTt1
ln Et
1.0491
[0.3657]
0.1160
[0.8909]
0.1568
[0.8557]
0.4417
[0.6480]
0.1054
[0.9004]
-0.6573*
[-3.0071]
ln Tt
0.1286
[0.8799]
1.0231
[0.3746]
1.1403
[0.3364]
0.0479
[0.9533]
0.0132
[0.9869]
-0.7664*
[-2.9441]
ln Ut
0.1047
[0.9010]
0.1154
[0.8944]
0.1086
[0.8975]
0.2746
[0.7622]
3.0177***
[0.0678]
In the short run, the feedback hypothesis is found between economic growth
and CO2 emissions. Urbanization Granger causes CO2 emissions and economic
growth. The unidirectional causality is found from trade openness to urbanization. Overall, the results indicate that unidirectional causality runs from
economic growth (ln Yt and ln Yt2) to CO2 emissions (ln Ct) in the long and the
short run. This lends support to the existence of the environmental Kuznets
curve. These findings are consistent with D. Maddison and K. Rehdanz for
North America; X. Zhang and X.-M. Cheng and A. Jalil and S. Mahmud for
China; S. Ghosh for India; P. Narayan and S. Narayan for developing economies; M. Nasir and F. Rehman and M. Shahbaz et al. for Pakistan; and Tiwari
et al. for India.54
Conclusions and Future Research
This paper examines the existence of EKC for Thailand over the period of
19712010 by implementing the ARDL bounds test approach in the presence of
165
166
NOTES
1
2
Government of Thailand, Ministry of Information and Communication Technology, National
Statistical Office (NSO), Gross National Product, Gross Domestic Product and National Income at
Current Market Prices by Economic Activities: 2000 2009 (Bangkok: NSO, 2011), available at
http://service.nso.go.th/nso/nsopublish/BaseStat/basestat.html.
3
Government of Thailand, Office of Natural Resources and Environmental Policy and Planning
(ONEP), Thailands Second National Communication under the United Nations Framework Convention
on Climate Change (Bangkok: Ministry of Natural Resources and Environment, 2010), p. 102.
4
Government of Thailand, Ministry of Energy, Energy Policy and Planning Office (EPPO),
Draft of 20 Years (2011-2030) Energy Conservation Plan (Bangkok: Ministry of Energy, 2011), p. 94.
7
I. Martnez-Zarzoso and A. Maruotti, The Impact of Urbanization on CO2 Emissions: Evidence from Developing Countries, Ecological Economics, vol. 70, no. 7 (2011), pp. 1344353.
8
Government of Thailand, Office of Natural Resources and Environmental Policy and Planning
(ONEP), Thailands Second National Communication under the United Nations Framework
Convention on Climate Change.
9
Government of Thailand, Ministry of Interior, Department of Public Works and Town and
Country Planning (DPT), National Landscape Planning Project: Policy for Area Development in
Thailand (Bangkok: Ministry of Interior, 2006).
10
S. Hossain, Panel Estimation for CO2 Emissions, Energy Consumption, Economic Growth,
Trade Openness and Urbanization of Newly Industrialized Countries, Energy Policy, vol. 39, no.
11 (2011), pp. 6991999.
11
P. K. Narayan and S. Narayan, Carbon Dioxide Emissions and Economic Growth: Panel Data
Evidence from Developing Countries, Energy Policy, vol. 38, no. 1 (2010), pp. 66166.
12
A. Kukla-Gryz, Economic Growth, International Trade and Air Pollution: A Decomposition
Analysis, Ecological Economics, vol. 68, no. 5 (2009), pp. 1329339.
13
S. S. Wang, D. Q. Zhou, P. Zhou, and Q. W. Wang, CO2 Emissions, Energy Consumption and
Economic Growth in China: A Panel Data Analysis, Energy Policy, vol. 39, no. 9 (2011), pp. 4870875.
167
15
T. M. Selden and D. Song, Environmental Quality and Development: Is There a Kuznets
Curve for Air Pollution Emissions? Journal of Environmental Economics and Management, vol.
27, no. 2 (1994), pp. 14762.
16
H. T. Pao and C. M. Tsai, CO2 Emissions, Energy Consumption and Economic Growth in
BRIC Countries, Energy Policy, vol. 38, no. 2 (2010), pp. 7850860; I. Ozturk and A. Acaravci,
CO2 Emissions, Energy Consumption and Economic Growth in Turkey, Renewable and Sustainable Energy Reviews, vol. 14, no. 9 (2010), pp. 3220225; A. Acaravci and I. Ozturk, On the
Relationship between Energy Consumption, CO2 Emissions and Economic Growth in Europe,
Energy, vol. 35, no. 12 (2010), pp. 5412420; M. Nasir and F. U. Rehman, Environmental Kuznets
Curve for Carbon Emissions in Pakistan: An Empirical Investigation, Energy Policy, vol. 39, no. 3
(2011), pp. 1857864; M. Shahbaz, H. H. Lean, and M. S. Shabbir, Environmental Kuznets Curve
Hypothesis in Pakistan: Cointegration and Granger Causality, Renewable and Sustainable Energy
Reviews, vol. 16, no. 5 (2012), 2947953; H. Iwata, K. Okada, and S. Samreth, A Note on the
Environmental Kuznets Curve for CO2: A Pooled Mean Group Approach, Applied Energy, vol. 88,
no. 5 (2011), pp.1986996; B. Saboori, J. B. Sulaiman, and S. Mohd, An Empirical Analysis of the
Environmental Kuznets Curve for CO2 Emissions in Indonesia: The Role of Energy Consumption
and Foreign Trade, International Journal of Economics and Finance, vol. 4, no. 2 (2011), pp.
24351; B. Saboori, J. B. Sulaiman, and S. Mohd, Economic Growth and CO2 Emissions in
Malaysia: A Cointegration Analysis of the Environmental Kuznets Curve, Energy Policy, vol. 51
(December 2012), pp. 18491; M. Shahbaz, M. Mutascu, and P. Azim, Environmental Kuznets
Curve in Romania and the Role of Energy Consumption, Renewable and Sustainable Energy
Reviews, vol. 18, issue c (2013), pp. 16573; and A. K. Tiwari, M. Shahbaz, and Q. M. A. Hye,
The Environmental Kuznets Curve and the Role of Coal Consumption in India: Cointegration and
Causality Analysis in an Open Economy, Renewable and Sustainable Energy Reviews, vol. 18,
issue c (2013), pp. 51927.
21
D. I. Stern, The Rise and Fall of the Environmental Kuznets Curve, World Development,
vol. 32, no. 8 (2004), pp. 1419439.
22
23
World Bank, World Development Indicators (Washington, D.C.: World Bank, 2011).
168
United States, Ecological Economics, vol. 62, no. 3 (2007), pp. 48289; A. Jalil and A. Mahmud,
Environmental Kuznets Curve for CO2 Emissions: A Cointegration Analysis for China, Energy
Policy, vol. 37, no. 12 (2009), pp. 5167172; F. Halicioglu, op. cit.; and M. Shahbaz, H. H. Lean, and
M. S. Shabbir, Environmental Kuznets Curve Hypothesis in Pakistan: Cointegration and Granger
Causality.
24
25
M. H. Pesaran, Y. Shin, and R. J. Smith, Bound Testing Approaches to the Analysis of Level
Relationship, Journal of Applied Econometrics, vol. 16, no. 3 (2001), pp. 289326.
26
M. H. Pesaran and B. Pesaran, Microfit 4.0 (Windows Version) (New York: Oxford University
Press, 1997).
27
A. Haug, Temporal Aggregation and the Power of Cointegration Tests: A Monte Carlo
Study, Oxford Bulletin of Economics and Statistics, vol. 64, no. 4 (2002), pp. 399412.
28
J. Laurenceson and J. C. H. Chai, Financial Reform and Economic Development in China
(Cheltenham, United Kingdom: Edward Elgar, 2003).
29
The dummy is based on the Zivot-Andrews structural break unit root test.
31
32
In such a situation, we may have to rely on the lagged error correction term (ECMt) for a longrun relationship.
33
D. Dickey and W. Fuller, Distribution of the Estimators for Autoregressive Time Series with
A Unit Root, Journal of the American Statistical Association, vol. 74, no. 366 (June 1979), pp.
42731; G. Elliott, T. J. Rothenberg, and J. H. Stock, Efficient Tests for an Autoregressive Unit
Root, Econometrica, vol. 64, no. 4 (1996), pp. 81336; and S. Ng and P. Perron, Lag Length
Selection and the Construction of Unit Root Tests with Good Size and Power, Econometrica, vol.
69, no. 6 (2001), pp. 1519554.
37
The results of the ADF, DF-GLS, and Ng-Perron are available from the authors upon request.
169
38
C. F. Baum, A Review of Stata 8.1 and Its Time Series Capabilities, International Journal of
Forecasting, vol. 20, no. 1 (2004), pp. 15161.
39
E. Zivot and D. Andrews, Further Evidence of the Great Crash, the Oil Price Shock and
Unit Root Hypothesis, Journal of Business and Economic Statistics, vol. 10, no. 3 (1992), pp.
25170.
40
T. Song, T. Zheng, and L. Tong, An Empirical Test of the Environmental Kuznets Curve in
China: A Panel Cointegration Approach, China Economic Review, vol. 19, no. 3 (2008), pp.
38192; F. Halicioglu, op. cit.; M. Fodha and O. Zaghdoud, Economic Growth and Pollutant
Emissions in Tunisia: An Empirical Analysis of the Environmental Kuznets Curve, Energy Policy,
vol. 38, no. 2 (2010), pp. 1150156; H. H. Lean and R. Smyth, CO2 Emissions, Electricity
Consumption and Output in ASEAN, Applied Energy, vol. 87, no. 6 (2010), pp. 1858864;
M. Shahbaz, H. H. Lean, and M. S. Shabbir, Environmental Kuznets Curve Hypothesis in Pakistan:
Cointegration and Granger Causality, Renewable and Sustainable Energy Reviews, vol. 16, no. 5
(2012), pp. 2947953; M. Shahbaz, M. Mutascu, and P. Azim, Environmental Kuznets Curve in
Romania and the Role of Energy Consumption, Renewable and Sustainable Energy Reviews, vol.
18, issue c (2013), pp. 16573; and A. K. Tiwari et al., op. cit.
44
Government of Thailand, Ministry of Energy, Energy Policy and Planning Office (EPPO),
Energy Statistics 2012 (Bangkok: Ministry of Energy, 2012), available at http://www.eppo.go.th/
info/index-statistics.html.
45
S. Khalil and Z. Inam, Is Trade Food for Environment? A Unit Root Cointegration Analysis,
The Pakistan Development Review, vol. 45, no. 4 (2006), pp. 1187196, and F. Halicioglu, op. cit.
46
M. Bahmani-Oskooee and A. Nasir, ARDL Approach to Test the Productivity Bias Hypothesis, Review of Development Economics, vol. 8, no. 3 (2004), pp. 48388.
48
International Monetary Fund (IMF), World Economic Outlook Database (Washington, D.C.:
IMF, 2011), available at: http://www.imf.org/external/data.htm.
170
49
Government of Thailand, Office of Natural Resources and Environmental Policy and Planning
(ONEP), Thailands Initial National Communication under the United Nations Framework Convention
on Climate Change (Bangkok: Ministry of Science, Technology and Environment, 2000), p. 100.
50
United Nations (U.N.), World Urbanization Prospects, The 2009 Revision (New York: U.N.,
2011), available at http://esa.un.org/unpd/wup/index.htm.
51
Government of Thailand, Office of Natural Resources and Environmental Policy and Planning
(ONEP), Thailands Initial National Communication under the United Nations Framework Convention on Climate Change.
52
G. M. Ghani, Does Trade Liberalization Affect Energy Consumption? Energy Policy, vol. 43,
no. 4 (2012), pp. 28590; and M. A. Cole and R. J. R. Elliott, Determining the Trade-Environment
Composition Effect: The Role of Capital, Labor and Environmental Regulations, Journal of
Environmental Economics and Management, vol. 46, no. 3 (2003), pp. 36383.
56