THE JOURNAL OF ENERGY

AND DEVELOPMENT

Fakhri Issaoui, Hassen Toumi, and Wassim Touili,

“The Effects of Carbon Dioxide Emissions

on Economic Growth, Urbanization,
and Welfare: Evidence from
Middle Eastern and North African Countries,”
Volume 41, Number 2

Copyright 2016
THE EFFECTS OF CARBON DIOXIDE EMISSIONS
ON ECONOMIC GROWTH, URBANIZATION, AND
WELFARE: APPLICATION TO COUNTRIES IN THE
MIDDLE EAST AND NORTH AFRICA

Fakhri Issaoui, Hassen Toumi, and Wassim Touili*

Introduction

S ince the 1970s, economic and environmental issues increasingly seem in-
separable making it difficult today to talk about the environment without in-
corporating the economy and vice versa. Indeed, the importance of addressing

*Fakhri Issaoui, Associate Professor at the Ecole Supérieure de Sciences Economiques et

Commerciales, University of Tunis (Tunisia), received his Ph.D. in economics from the
University of Tunis El Manar. His research interests include public and labor economics,
economic development, economic justice, macroeconomics, and growth challenges in
transitioning economics. The author has published more than 30 articles in journals such as
Panoeconomicus, The Journal of Energy and Development, Journal of Interdisciplinary Mathe-
matics, International Journal of Computational Economics and Econometrics, IUP Journal of
Applied Economics, IUP Journal of Marketing Management, Global Business and Management
Research, African Sociological Review, Global Journal of Management and Business Research,
Global Journal of Human Social Research, Global Journal of Research in Engineering, The
Economic Research Guardian, and The Global Journal of Strategies and Governance, as a sampling.
Hassen Toumi, Assistant Professor at the University of Economics and Management of Sfax
(Tunisia), received his Ph.D. in economics from the University of Sfax. His research interests include
public and labor economics, economic development, energy economics, banking economics, and
econometrics. The author’s works have been published in numerous journals such as The Journal
of Energy and Development, Environmental Science and Pollution Research, Journal of the
Knowledge Economy, and The Economic Research Guardian.
Wassim Touili is a doctoral candidate in economic science at the Ecole Supérieure de
Commerce, University of Manouba (Tunisia). His research focuses on economic develop-
ment. The author has published articles in The Journal of Energy and Development and The
Economic Research Guardian.

The Journal of Energy and Development, Vol. 41, Nos. 1 and 2

Copyright
2016 by the International Research Center for Energy and Economic Development
(ICEED). All rights reserved.
223
224 THE JOURNAL OF ENERGY AND DEVELOPMENT

environmental issues (global warming, pollution, deforestation, overexploitation

of natural resources, etc.) continues to attract the interest of researchers and ac-
ademicians. The fundamental reason underlying such attention lies in the fact that
carbon dioxide (CO2) emissions are growing globally and commanding a two-
tiered approach—both economic and environmental—in the evaluation and as-
sessment of what policy options are needed and viable.
The economic approach considers CO2 emissions as the logical consequence
of industrial activities, which—although they are polluting—are a by-product
of value-added processes and robust economic growth. The environmental
approach often differs from the economic perspective in asserting that sus-
tainable development cannot be based on polluting industries as long as the
realized growth is offset by a loss of social welfare. Some academicians, like
N. Stern, warn that global warming due to the proliferation of greenhouse gas
(GHG) emissions—with CO2 being the primary culprit—is the main threat to
humanity.1
An alternative view in the field of economics has emerged in recent years
calling into question the traditional thesis that finds a positive relationship between
CO2 emissions and growth. For example, W. Nordhaus and J. Boyer suggest that
a warming of two degrees Celsius could lead to a five-percent decrease in the
average annual per-capita consumption in Africa and Asia.2
As long as the growth models based on polluting energy are generalized, the
countries of the Middle East and North Africa (MENA) are facing the same basket
of problems and trade-offs (growth versus the environment) as those experienced
in developed nations. Indeed, CO2 emissions in MENA countries are becoming
increasingly salient since the Rio Summit in 1992 and potentially threaten the
well-being of the populace. This reality is confirmed by the World Bank, which
estimated that the cost of pollution makes up 7 percent of the gross domestic
product (GDP) in MENA.3 Figures 1A through 1J display the trend of increasing
CO2 emission levels in selected countries of the MENA region (Algeria, Bahrain,
Egypt, Jordan, Kingdom of Saudi Arabia, Morocco, Qatar, Tunisia, United Arab
Emirates, and Yemen) over the 1990-2010 period.
At this stage of the analysis, it is important to note that the relationship between
growth and CO2 emissions has been well researched with most studies focusing on
the trade-off between economic growth and emissions. However, studies that have
examined the impact of CO2 emissions on welfare are relatively sparse and more
recent.
Therefore, to cope for this deficiency, in our article we will attempt to in-
corporate within a single model the economic and environmental approaches.
Indeed, we try to explain the relationship that is established between CO2 emis-
sions and a vector of variables involving economic growth, energy consumption
(first approach), urbanization, and life expectancy (second approach).
 EFFECTS OF EMISSIONS IN MENA COUNTRIES 225

Figure 1A
CARBON DIOXIDE EMISSION LEVELS FOR ALGERIA, 1990–2010
(in kilotons-kt)

Source: World Bank, World Bank Indicators (Washington, D.C.: World Bank, 2016).

Figure 1B
CARBON DIOXIDE EMISSION LEVELS FOR BAHRAIN, 1990–2010
(in kilotons-kt)

Source: World Bank, World Bank Indicators (Washington, D.C.: World Bank, 2016).
226 THE JOURNAL OF ENERGY AND DEVELOPMENT

Figure 1C
CARBON DIOXIDE EMISSION LEVELS FOR EGYPT, 1990–2010
(in kilotons-kt)

Source: World Bank, World Bank Indicators (Washington, D.C.: World Bank, 2016).

Figure 1D
CARBON DIOXIDE EMISSION LEVELS FOR JORDAN, 1990–2010
(in kilotons-kt)

Source: World Bank, World Bank Indicators (Washington, D.C.: World Bank, 2016).
 EFFECTS OF EMISSIONS IN MENA COUNTRIES 227

Figure 1E
CARBON DIOXIDE EMISSION LEVELS FOR THE KINGDOM OF SAUDI ARABIA, 1990–2010
(in kilotons-kt)

Source: World Bank, World Bank Indicators (Washington, D.C.: World Bank, 2016.)

Figure 1F
CARBON DIOXIDE EMISSION LEVELS FOR MOROCCO, 1990–2010
(in kilotons-kt)

Source: World Bank, World Bank Indicators (Washington, D.C.: World Bank, 2016).
228 THE JOURNAL OF ENERGY AND DEVELOPMENT

Figure 1G
CARBON DIOXIDE EMISSION LEVELS FOR QATAR, 1990–2010
(in kilotons-kt)

Source: World Bank, World Bank Indicators (Washington, D.C.: World Bank, 2016).

Figure 1H
CARBON DIOXIDE EMISSION LEVELS FOR TUNISIA, 1990–2010
(in kilotons-kt)

Source: World Bank, World Bank Indicators (Washington, D.C.: World Bank, 2016).
 EFFECTS OF EMISSIONS IN MENA COUNTRIES 229

Figure 1I
CARBON DIOXIDE EMISSION LEVELS FOR THE UNITED ARAB EMIRATES, 1990–2010
(in kilotons-kt)

Source: World Bank, World Bank Indicators (Washington, D.C.: World Bank, 2016).

Figure 1J
CARBON DIOXIDE EMISSION LEVELS FOR YEMEN, 1990–2010
(in kilotons-kt)

Source: World Bank, World Bank Indicators (Washington, D.C.: World Bank, 2016).
230 THE JOURNAL OF ENERGY AND DEVELOPMENT

This paper is divided as follows: we begin with a review of the literature on the
effects of CO2 emissions on growth and welfare and then present our methodo-
logical approach. The subsequent sections will present the econometric estimation
approach and an analysis of the results. The last and final section provides our
conclusions.

Literature Review
Literature Review of the CO2 Effect on Growth: The review of the literature
can be divided into four research categories: (1) testing for the existence of the
environmental Kuznets curve; (2) testing the causal relationship between energy
consumption and economic growth; (3) establishing a dynamic relationship among
carbon emissions, energy consumption, and economic growth; and (4) focusing on
the dynamic relationship among carbon emissions, energy consumption, economic
growth, trade liberalization, and urbanization.
Regarding the first category, which addresses the Kuznets curve, we can refer
to the seminal work of G. Grossman and A. Krueger,4 who concluded that there
was an absence of evidence showing the existence of a causality relation between
gradual environmental degradation and country growth. A contrary conclusion
was offered by N. Shafik, whose results implied that CO2 emissions were in-
creasing in parallel with economic growth.5
D. Stern et al.’s research showed that CO2 emissions began to decrease
when the economy reached a well-defined income threshold.6 The results of
E. Akbostanci et al. do not comply with the principles of the environmental
Kuznets curve hypothesis, which asserts that environmental degradation tends
to get worse as modern economic growth occurs until an average threshold
income is reached.7 I. Martinez-Zarzo and A. Bengochea-Morancho showed
that CO2 emissions and income level are negatively related in low-income
countries, but they are positively related in high-income countries.8 We find
that the results of the research on the environmental Kuznets curve are di-
vergent and often contradictory. Moreover, in the majority of cases, researchers
have failed to confirm the existence of a typical Kuznets curve, which should be
an inverted U-shape.
Regarding the second research category, there have been numerous stud-
ies analyzing the causality between growth and energy consumption. Indeed,
according to the works of A. Akarca and T. Long, E. Yu and B. Hwang, E. Yu and
J. Choi, U. Erol and E. Yu, D. Stern, and B. Cheng, no causal relationship was
found between energy consumption and GDP.9 Nevertheless, the works of J. Asafu
Adjaye, H. Yang, U. Soytas and R. Sari, R. Morimoto and C. Hope, G. Altinay and
E. Karagol, and P. Narayan and R. Smyth supported the existence of a causal
relationship between the two variables—energy consumption and growth.10 Other
research has found a unidirectional causal relationship as shown by Y. Glasure and
 EFFECTS OF EMISSIONS IN MENA COUNTRIES 231

A. Lee, B. Cheng, J. Chang and J. Wong, U. Soytas and R. Sari, and P. Narayan
and R. Smyth.11 Still other academics, including A. Masih and R. Masih, J. Asafu
Adjaye, Y. Glasure, and W. Oh and K. Lee, found energy consumption and growth
to have a bidirectional relationship.12
With regard to the research on the relationship between carbon emissions,
energy consumption, and economic growth, we can refer to the work of U. Soytas
et al., which demonstrates the absence of a causal relationship between growth and
carbon emissions, on the one hand, and income and energy consumption, on the
other.13 Contrary to this, a causal relationship among these variables was suggested
in the research of E. Neumayer and M. Cole, M. Shahbaz et al., F. Halicioglu, and
G. Akpan and U. Akpan.14 The publications of U. Soytas et al., H. E. Chebbi and
Y. Boujelbene, J. Ang, U. Soytas and R. Sari, and X. Zhang and X. Cheng provided
varied results.15
Concerning the final category of our literature review, which focuses on
the dynamic relationship between carbon emissions, energy consumption, and
economic growth, we can refer to the study of S. Hossain who has shown
the absence of a long-term causality relation between these variables.16
However, in the short term, the author confirms the existence of a set of uni-
directional causality relations among economic growth, trade openness, and CO2
emissions.

Literature Review of the CO2 Effect on Well-Being: As mentioned pre-

viously, the majority of works have focused on the trade-offs between economic
growth and CO2 emissions. Although the studies that have examined the impact
of CO2 emissions on well-being are relatively few, we can refer to those of
N. Georgescu-Roegen and D. Meadows et al., which assert that economic activity
necessarily leads to the accumulation of CO2 emissions and, consequently, causes
environmental degradation and decreased social welfare.17
E. Lambin studied the interactions between welfare and environmental deg-
radation and concluded that environmental problems can threaten social well-
being.18 C. Declercq et al. found that life expectancy would increase by up to 22
months if major European cities could reduce air pollution.19 Additionally,
Y. Chen et al. have shown the existence of a negative correlation between lon-
gevity and environmental degradation.20 According to the authors, since 1950
and due to a growing pollution, the life expectancy in northern China has de-
creased by five-and-a-half years.
The United Nations Environmental Programme report on the future of the
global environment asserted that air pollution adversely affects the well-being in
almost all regions of the world.21 The World Health Organization estimates that
over 1 billion people in Asian countries are exposed to air pollutants.22 A. Ali and
K. Ahmad have studied the impact of CO2 emissions on life expectancy for Oman
and concluded that emissions have a positive and insignificant relationship with
232 THE JOURNAL OF ENERGY AND DEVELOPMENT

life expectancy in the long run but, in the short run, it has a negative and significant
relationship with life expectancy.23
A. Assadzadeh et al. examined the impact of CO2 emissions on life expectancy
in eight oil-exporting countries from 2000 to 2010. The short-run elasticities
revealed that per-capita GDP and CO2 emissions have negative and insignificant
effects on life expectancy.24
M. Mehrara and M. R. Masoumi have studied the relationship between life
expectancy, CO2 emissions, and GDP growth rate for 108 developing countries for
the time span of 1995–2012.25 They concluded that the environmental Kuznets
curve hypothesis has not been confirmed in developing countries and, if they in-
clude CO2 emissions as an endogenous variable in their model, their effect on health
will increase, and, if the GDP increases, it will result in longer life expectancies.
In summation, the deterioration of the environment is a negative externality for
economic growth and decreases the benefits of health improvements. Moreover,
a labor force with poor health conditions does not increase productivity; therefore,
economic growth will decline.

Methodology and Data

To study the effects of economic growth on CO2 emissions and well-being, we

have selected a sample of 10 MENA countries (Algeria, Bahrain, Egypt, United
Arab Emirates, Jordan, Saudi Arabia, Morocco, Qatar, Tunisia, and Yemen)
covering the period 1990 to 2010.
This research is based on the following equation, in which the explanatory
variables were selected from varied literature sources:
CO2 = f ðGDPC, EC, URB, LEXPÞ
The variables used in our study are:
CO2: Carbon dioxide emissions, measured by per-capita metric tons, are those
stemming from the burning of fossil fuels and the manufacture of cement. They
include carbon dioxide emissions produced during consumption of solid, liquid,
and gas fuels and gas flaring.
GDPC: The GDP per capita (measured in U.S. dollars) evaluates the impact of
the level of development on the environment. Theoretically, the assumptions of
the environmental Kuznets curve (EKC) hypothesis postulate that environmental
degradation will accelerate in developing countries until a certain level of income
is reached and then the trend will be reversed, resulting in an improvement in the
environmental situation. Given the poor economic performance associated with
low technological development of the countries in our study, we assume that any
unit increase in GDP per capita is associated with an increase in total carbon
dioxide emissions.
 EFFECTS OF EMISSIONS IN MENA COUNTRIES 233

EC: Energy consumption, measured in kilograms of oil equivalent per capita,

refers to the use of coal, oil, rock oil, and natural gas as energy sources. At the global
level, energy consumption is the second source of greenhouse gas emissions.
URB: This variable (the percentage of urban population relative to the overall
population) is also an important determinant of environmental quality. Indeed, it is
assumed that the increase in population and, in particular, those living in urban
areas induces rising food needs, which results in over-exploitation and depletion of
natural resources and a resulting CO2 emissions increase.26
LEXP: Life expectancy at birth indicates the number of years a newborn infant
would live if prevailing patterns of mortality at the time of its birth were to stay the
same throughout its life. This variable is critical in the determination of envi-
ronmental quality. The United Nations Environmental Programme report on the
future of the global environment stated that “environmental degradation un-
dermines development and threatens all aspects of human well-being.” 27 The
report continued by linking environmental degradation to human health problems,
including certain types of cancers, vector-borne diseases, zoonosis, nutritional
deficiencies, and respiratory illnesses.28
The database that we will use covers the period 1990 to 2010 and includes
10 countries belonging to the MENA region—Algeria, Bahrain, Egypt, Jordan,
Kingdom of Saudi Arabia, Morocco, Qatar, Tunisia, United Arab Emirates, and
Yemen. To find the relationship between CO2 emissions (CO2), energy con-
sumption (EC), GDP per capita (GDPC), urbanization (URB), and life expectancy
(LEXP), the following model is proposed in equation (1):

CO2it = A0 ECita1i GDPCita2i URBa3i a4i

it LEXPit ð1Þ

The logarithmic transformation of equation (1) is given in equation (2):

lnðCO2it Þ = a0 + a1i lnðECit Þ + a2i lnðGDPCit Þ + a3i lnðURBit Þ

+ a4i lnðLEXPit Þ + eit ð2Þ

where a0 = A0; the index (i) and (t) represent the country and time period, re-
spectively; and a1, a2, a3, and a4 represent the long-term elasticities.
Table 1 provides the overall descriptive statistics of all variables included in the
model for the ten MENA countries in our study.

Estimation Method

In this section we will present the results of our estimation of equation (1) using
different methods such as the fully modified ordinary least squares (FMOLS) and
234 THE JOURNAL OF ENERGY AND DEVELOPMENT

Table 1
a
INDIVIDUAL AND GROUP PANEL DESCRIPTIVE STATISTICS

CO2 LEXP GDPC EC URB

Algeria Mean 3.03528 68.83138 2367.96 905.04 59.89095

Median 3.00672 68.93020 1743.35 872.67 59.91900
Std. Dev. 0.23522 1.21195 1054.01 100.76 4.82388
Probability 0.75920 0.49920 0.13668 0.27211 0.51901
Bahrain Mean 23.92757 74.50603 13801.28 8593.23 88.39048
Median 21.83855 74.58924 12846.45 8691.60 88.38600
Std. Dev. 3.65638 1.18564 4649.08 422.50 0.07444
Probability 0.27006 0.50052 0.30901 0.33743 0.00000
Egypt Mean 1.96024 68.05932 1346.08 712.30 42.97171
Median 1.90753 68.59124 1249.49 654.37 42.98100
Std. Dev. 0.44346 1.81375 562.28 149.88 0.21076
Probability 0.46222 0.39807 0.08054 0.25810 0.57363
Jordan Mean 3.36807 71.76074 2143.29 1076.56 79.29895
Median 3.28586 71.77976 1763.00 1018.52 79.80800
Std. Dev. 0.26590 1.07411 920.99 111.51 2.55053
Probability 0.97759 0.54313 0.04437 0.21267 0.24583
Morocco Mean 1.28272 67.89000 1638.97 378.03 53.30267
Median 1.18095 68.13859 1362.53 359.15 53.33500
Std. Dev. 0.22547 1.57583 624.24 69.51 2.61313
Probability 0.38932 0.53608 0.17766 0.41579 0.73181
Qatar Mean 53.70485 76.77919 35449.18 18220.45 96.13129
Median 58.34635 76.82015 28666.61 18319.97 96.31100
Std. Dev. 10.68711 0.85622 22747.73 2567.26 1.72273
Probability 0.08365 0.56999 0.24524 0.61853 0.59554
Saudi Arabia Mean 14.90136 72.43208 10689.97 5232.88 79.74862
Median 15.18525 72.61476 8656.17 5029.88 79.84800
Std. Dev. 2.02371 1.82582 4079.43 811.27 1.55906
Probability 0.40314 0.51425 0.10450 0.73130 0.63001
Tunisia Mean 2.05928 72.12904 2641.42 756.47 63.00338
Median 2.08321 72.50000 2336.08 763.94 63.43200
Std. Dev. 0.24916 1.36336 898.26 108.88 2.42197
Probability 0.47719 0.71254 0.35744 0.42284 0.41754
United Arab Mean 27.19258 74.32040 33589.40 10887.17 80.62481
Emirates Median 28.78999 74.40568 32984.74 11305.89 80.23600
Std. Dev. 5.44599 1.52080 6780.04 1521.60 1.95700
Probability 0.57141 0.51552 0.29603 0.15001 0.36236
Yemen Mean 0.90167 60.35113 669.29 277.09 26.31710
Median 0.90689 60.45785 546.54 270.90 26.26700
Std. Dev. 0.11429 1.45305 359.11 50.76 3.30276
Probability 0.60636 0.53582 0.15691 0.43902 0.55537

(continued )
 EFFECTS OF EMISSIONS IN MENA COUNTRIES 235

Table 1 (continued)
a
INDIVIDUAL AND GROUP PANEL DESCRIPTIVE STATISTICS

CO2 LEXP GDPC EC URB

Panel Mean 13.23336 70.70593 10433.68 4703.92 66.96800

Median 3.23091 71.40626 2796.77 998.20 70.40850
Std. Dev. 16.92878 4.65187 14866.52 5871.39 20.91096
Probability 0.00000 0.00000 0.00000 0.00000 0.00136

a
CO2 = carbon dioxide emissions measured by per-capita metric tons; LEXP = life expectancy at
birth; GDPC = the gross domestic product per capita measured in U.S. dollars; EC = energy
consumption measured in kilograms of oil equivalent per capita; URB = the percentage of urban
population relative to the overall population; and Std. Dev. = standard deviation.

dynamic ordinary least squares (DOLS) approaches. These methods allow us to

exceed the limits of other techniques used in the context of panel data modeling
such as fixed or random effects models, which do not take into account the
presence of unit roots in the series and can lead to biased estimates and statistical
tests that do not follow the standard Student distribution.

Unit Root Testing: There are a variety of unit root tests used in panel data
modeling. As an example, we can refer to the works of J. Breitung, K. Hadri,
I. Choi, A. Levin et al. (this panel root test is often referred to as the LLC method
representing all its authors), K. Im et al. (the panel root test is often referred to as
the IPS test representing all of its authors), and J. Carrion-i-Silvestre et al.29 Next,
we consider the following autoregressive specification:
yit = ri yit – 1 + di xit + eit ð3Þ

with i = 1, ..., N for each country in the model panel; t = 1, ..., T denotes the time
period; xit represents the exogenous variables in the model; ri represents the
autoregression coefficients; and eit represents the stationary error terms. If ri > 1,
yit is considered as having a stationary trend, whereas if ri = 1, then yit will have
a unit root. J. Breitung and A. Levin et al. assume a homogeneous autoregressive
unit root under the alternative hypothesis, while K. Im et al. assume the existence
of a heterogeneous autoregressive unit root under the alternative hypothesis.30
According to G. Maddala and S. Wu and I. Choi, the unit root tests can be de-
termined using the nonparametric Fisher statistics.31 K. Hadri and J. Carrion-i-
Silvestre et al. assume that panel unit root tests examine the null hypothesis of
stationarity of the panel data.32
In the case of dynamic models of panel data, the recognition of the hetero-
geneity of parameters is important in order to avoid potential biases that may
236 THE JOURNAL OF ENERGY AND DEVELOPMENT

arise due to poor specification of the model. K. Im et al. assume that the unit root
test in dynamic models of panel data is used for heterogeneous autoregressive
coefficients.33 Such heterogeneity may occur due to the heterogeneity of dif-
ferent economic conditions and stages of economic development of each
country.
Thus, K. Im et al. suggest the average of the augmented Dickey-Fuller (ADF) test,
allowing serial correlations of error terms to the different orders, represented as:

Xpi
eit = j=1
rij eit – j + uit ;

which we substitute into equation (3) and obtain equation (4):

Xpi
yit = ri yit – 1 + u e + di xit + uit
j = 1 ij it – j
ð4Þ

where ri is the number of lags in the ADF regression. The null hypothesis is that
each series in the panel data contains a unit root (H0: ri = 1 "i). The alternative
hypothesis is that at least one of each of the series in the panel data is stationary
(H0 : ri < 1 "i). The statistical T-bar specified by K. Im et al. is the average of
individual statistics ADF as shown in the following:34
XN
tNTðri Þ = 1/ N t ðr Þ
i = 1 iT i

The alternative statistic “t-bar” allows for the testing of the null hypothesis of
the existence of unit root for all individuals, with tiT (ri) representing the estimated
ADF, N the number of individuals, and T as the number of observations. K. Im
et al. propose the use of the following standardized statistic:35

Zi = ðN Þ1=2 ðtNT – EðtNT ÞÞ=ðvarðtNT ÞÞ1=2

where E(tNT) represents the arithmetic mean and var (tNT) denotes the variances of
the ADF individual statistics.
Table 2 displays the results of the various unit root tests we performed under
four cases utilizing six different tests: LLC (referring to the work of Levin, Lin,
Chu),36 Breiting,37 IPS (referring to the work of K. Im, M. Pesaran, and Y. Shin),38
ADF-Fisher, PP-Fisher, and Hadri39 on the static relationships of LNCO2, LNEC,
LNGDPC, LNURB, and LNLEXP. The results show that the majority of the unit
root tests are lower than the critical value at the 1-percent level; thus, the null
hypothesis is accepted.
 EFFECTS OF EMISSIONS IN MENA COUNTRIES 237

Table 2
a
PANL UNIT ROOT TEST RESULTS

Augmented Fisher-
Levin, Lin, Im, Pesaran, Dickey-Fuller Phillips-
and Chu (LLC) Breiting and Shin (IPS) (ADF)-Fisher Perron Test Hadri
Test [Prob.] Test [Prob.] Test [Prob.] Test [Prob.] [Prob.] Test [Prob.]

Table 2 shows the statistics of the panel unit root tests. The values in brackets correspond to the
p values. LN = natural logs; CO2 = carbon dioxide emissions measured by per-capita metric tons;
LEXP = life expectancy at birth; GDPC = the gross domestic product per capita measured in U.S.
dollars; EC = energy consumption measured in kilograms of oil equivalent per capita; URB = the
percentage of urban population relative to the overall population; and Prob. = probabilities.
Case 1: Model with constant terms (level form)
LNCO2
-2.1685 -2.0801 -1.9001 33.5720 32.6261 5.4851
[0.0151]** [0.0188]** [0.0287]** [0.0292]** [0.0371]** [0.0000]***
LNEC
0.7055 1.3904 3.3516 6.3828 6.2921 8.2477
[0.7598] [0.9178] [0.9996] [0.9983] [0.9984] [0.0000]***
LNGDP
2.4667 0.4421 5.1147 2.8488 1.7619 8.4485
[0.9932] [0.6708] [1.0000] [1.0000] [1.0000] [0.0000]***
LNURB
-3.7341 0.0324 -2.0126 87.9771 197.0970 9.9938
[0.0001]*** [0.5129] [0.0221]** [0.0000]*** [0.0000]*** [0.0000]***
LNLEXP
-10.7103 4.5749 -3.2057 191.1610 611.4120 9.9264
[0.0000]*** [1.0000]*** [0.0007]*** [0.0000]*** [0.0000]*** [0.0000]***
Augmented Fisher-
Levin, Lin, and Im, Pesaran, Dickey-Fuller Phillips-
Chu (LLC) Test Breiting and Shin (IPS) (ADF)-Fisher Perron Test Hadri
[Prob.] Test [Prob.] Test [Prob.] Test [Prob.] [Prob.] Test [Prob.]
Case 2: Model with individual effects, individual linear trends
LNCO2
-2.3111 - -2.8525 38.8876 45.1752 3.9255
[0.0104]** - [0.0022]** [0.0069]** [0.0010]** [0.0000]***
LNEC
-0.9428 - 0.1784 20.2201 19.9228 6.4584
[0.1729] - [0.5708] [0.4442] [0.4628] [0.0000]***
LNGDP
-2.8928 - -2.7807 39.7857 16.7556 6.2915
[0.0019]** - [0.0027]** [0.0053]** [0.6688] [0.0000]***
LNURB
3.0951 - -4.5838 66.1302 327.9840 6.8418
[0.9990] - [0.0000]*** [0.0000]*** [0.0000]*** [0.0000]***
LNLEXP
1.6734 - -0.2699 59.3903 326.7760 6.9048
[0.9529] - [0.3936] [0.0000]*** [0.0000]*** [0.0000]***

(continued )
238 THE JOURNAL OF ENERGY AND DEVELOPMENT

Table 2 (continued)
a
PANL UNIT ROOT TEST RESULTS

Augmented Fisher-
Levin, Lin, and Im, Pesaran, Dickey-Fuller Phillips-
Chu (LLC) Test Breiting and Shin(IPS) (ADF)-Fisher Perron Test Hadri
[Prob.] Test [Prob.] Test [Prob.] Test [Prob.] [Prob.] Test [Prob.]
Case 3: Model with only constant term (first difference)
LNCO2
-13.8323 -6.4288 -14.8899 188.3490 433.1680 2.5287
[0.0000]*** [0.0000]*** [0.0000]*** [0.0000]*** [0.0000]*** [0.0057]**
LNEC
-12.3592 -3.4999 -11.2735 140.5540 149.0630 2.8348
[0.0000]*** [0.0002]*** [0.0000]*** [0.0000]*** [0.0000]*** [0.0023]**
LNGDP
-10.6037 -4.3815 -9.3092 111.1760 109.9110 2.6352
[0.0000]*** [0.0000]*** [0.0000]*** [0.0000]*** [0.0000]*** [0.0042]**
LNURB
-0.0252 3.1210 -4.4076 62.7401 145.8320 5.7800
[0.4900] [0.9991] [0.0000]*** [0.0000]*** [0.0000]*** [0.0000]***
LNLEXP
-9.2050 0.6484 -6.8267 173.4750 268.6810 5.9880
[0.0000]*** [0.7416]*** [0.0000]*** [0.0000]*** [0.0000]*** [0.0000]***
Augmented Fisher-
Levin, Lin, and Im, Pesaran, Dickey-Fuller Phillips-
Chu (LLC) Test Breiting and Shin(IPS) (ADF)-Fisher Perron Test Hadri
[Prob.] Test [Prob.] Test [Prob.] Test [Prob.] [Prob.] Test [Prob.]
Case 4: Model with only constant term (first difference) with trends
LNCO2
-12.5762 - -13.0353 134.7350 150.1050 5.9128
[0.0000]*** - [0.0000]*** [0.0000]*** [0.0000]*** [0.0000]***
LNEC
0.1729 - -10.4481 108.7650 134.1920 3.5202
[0.0000]*** - [0.0000]*** [0.0000]*** [0.0000]*** [0.0002]***
LNGDP
-10.5190 - -8.3636 92.8672 91.7768 4.9748
[0.0000]*** - [0.0000]*** [0.0000]*** [0.0000]*** [0.0000]***
LNURB
-18.0141 - -9.2828 50.4115 51.9622 5.8452
[0.0000]*** - [0.0000]*** [0.0002]*** [0.0001]*** [0.0000]***
LNLEXP
-4.5410 - -10.0883 96.5730 67.4709 12.1012
[0.0000]*** - [0.0000]*** [0.0000]*** [0.0000]*** [0.0000]***

a
*** = significance at the 1-percent level; ** = significance at the 5-percent level; and * =
significance at the 10-percent level.
 EFFECTS OF EMISSIONS IN MENA COUNTRIES 239

We deduce that the residuals of the static relationship among LNCO2, LNEC,
LNGDP, LNURB, and LNLEXP have a unit root; therefore, they are non-stationary.
The series of different variables are non-stationary in level but are stationary in the
first difference. Therefore, the series are cointegrated I(1). Thus, we can deduce
the existence of a long-run relationship between CO2 emissions in the MENA
countries and LNEC, LNGDPC, LNURB, and LNLEXP. It is possible to estimate
the error-correction model (ECM) because the error-correction term is stationary
at level. Table 3 provides the unit root test results for the ECM.
Cointegration: To study the existence of a cointegration relationship, we have
referred to the work of P. Pedroni,40 where the null hypothesis is to test for the
absence of cointegration based on the unit roots tests applied to estimated re-
siduals. P. Pedroni suggests two types of residual-based tests for the test of the null
of no cointegration in heterogeneous panels. The first category of tests is referred
to as the within-dimension panel tests and consist of four tests based on pooling the
residuals of the regression along the within-dimension of the panel. These are the
panel v-statistic, the panel rho-statistic (also denoted as the panel r-statistic using
the Greek letter), the panel PP-statistic, and the panel ADF-statistic. The second
category of tests is referred to as the between-dimension panel tests and consists
of three tests based on pooling the residuals of the regression along the between-
dimension of the panel. The group tests include the group rho-statistic (also denoted
as the group r-statistic), group PP-statistic, and group ADF-statistic. Each of the
seven statistics follows a standard normal distribution for sufficiently large N and T:
pffiffiffiffi
zNT – m N
pffiffiffi ! N ð0; 1Þ
d

with ZNT one of the seven statistics; m and d are the values tabulated by Pedroni’s
method.
The Pedroni cointegration test results, as displayed in table 4, show that all panel
statistics (v, rho, ADF, and PP) and group statistics (rho, ADF, and PP) are less than

Table 3
a
UNIT ROOT TEST RESULTS OF THE ERROR-CORRECTION MODEL (ECM)

Statistic Probability

Levin, Lin, and Chu t* -3.4548 0.0003***

Im, Pesaran, and Shin W-stat -4.4602 0.0000***
ADF-Choi Z-stat -4.3082 0.0000***
PP-Choi Z-stat -4.3882 0.0000***

a
*** = coefficients are significant at the 1-percent level; ** = coefficients are significant at the
5-percent level; and * = coefficients are significant at the 10-percent level.
240 THE JOURNAL OF ENERGY AND DEVELOPMENT

Table 4
a
PEDRONI COINTEGRATION TEST RESULTS

Within Dimension
Statistic Prob. Statistic Prob.

Panel v-Statistic -0.079959 0.5319 -0.429636 0.6663

Panel rho-Statistic 1.394150 0.9184 0.487075 0.6869
Panel PP-Statistic -1.475531 0.0700** -4.091995 0.0000***
Panel ADF-Statistic -2.719920 0.0033** -4.049483 0.0000***
Between Dimension
Statistic Prob.
Group rho-Statistic 1.308420 0.9046
Group PP-Statistic -4.917524 0.0000***
Group ADF-Statistic -3.010200 0.0013**

a
*** = significance at the 1-percent level and ** = significance at the 5-percent level. The total
number of observations is 200. ADF = augmented Dickey-Fuller and PP = Phillips-Perron.

the critical value of the normal distribution for a threshold of 5 percent. Thus, we
can conclude that all of these tests confirm the existence of a long-term coin-
tegration relationship between CO2 emissions and LNEC, LNGDPC, LNURB,
and LNLEXP.
Cointegration Relationships: To estimate systems of cointegrated variables
on panel data and to identify tests on cointegration vectors, it is essential to apply
an effective estimation methodology.
Indeed, there are several estimation techniques one could use such as the fully
modified ordinary least squares (FMOLS) method employed by P. Pedroni, the dy-
namic ordinary least squares (DOLS) method, and the generalized method of moments
(GMM). P. Pedroni,41 P. Phillips and H. Moon,42 and C. Kao and M.-H. Chiang43 have
shown that, in the case of panel data, the FMOLS and DOLS techniques lead to
asymptotically-distributed estimators according to a standard normal distribution.
However, C. Kao and M.-H. Chiang44 assume that the OLS estimation in finite
samples leads to biased estimators relative to the FMOLS method. Moreover, they
highlight the superiority of the DOLS method in comparison to the FMOLS
method. The authors consider FMOLS as the most effective technique in esti-
mating cointegration relationships using panel data, the theorem of the repre-
sentation of Engel and Granger establishes equivalence between the existence of
a long-term relationship and the error-correction model of CO2 emissions (see
table 5 for the cointegration test results using the Kao and Chiang methodology).
Thus, we examine the direction of causality between variables in a group setting.
The two-step method of Engle and Granger45 shows that if two non-stationary
 EFFECTS OF EMISSIONS IN MENA COUNTRIES 241

Table 5
a
COINTEGRATION TEST OF KAO AND CHIANG

t-Statistic Prob.

ADF -5.7995 0.0000***

Residual variance 0.0134
HAC variance 0.0116

a
*** = the critical value at the 1-percent significance level for the panel; the ADF statistics is
–5.799502; and the total number of observations is 200. ADF = augmented Dickey-Fuller and HAC
variance = heteroscedasticity and autocorrelation consistent variance estimation.

variables are cointegrated, a vector autoregression (VAR) in first differences will be

poorly specified.
Therefore, as we found a long-term equilibrium relationship between CO2
emissions (CO2) and LNEC, LNGDPC, LNURB, and LNLEXP, we need to specify
a model with a dynamic representation of the error-correction term. This means
that the traditional VAR model is increased with an error-correction term delay of
a period (ECTT–1), which is obtained from the model based on cointegration OLS.
The Granger causality test is based on the following regression:
X X
DlnCO2it = A1i + p
b 10ip DlnC02 it – p + b DlnECit – p
p 11ip
X X
+ b DlnGDPCit – p +
p 12ip
b DlnURBit – p
p 13ip
X
+ b DlnLEXPit – p + u1i ECTT – 1
p 14ip
ð5Þ

where D represents the first difference of the variable and p the number of lags.
The importance of the first differentiated variables is that they indicate the short-
term direction of Granger causality, while the t-statistics on a delayed period in
terms of error correction represent the long-term Granger causality. In the CO2
consumption of equation (5), short-run causality from energy consumption, GDP,
urbanization, and the life expectancy are tested, respectively, based on H0: b11ip =
0"ip, H0: b12ip = 0"ip, H0: b13ip = 0"ip, and H0: b14ip = 0"ip. The null hy-
pothesis of no long-run causality for each element of equation (5) is tested by
examining the significance of the p-value for the coefficient on the respective
error-correction term represented by ECT.

Results and Interpretation

Short-Run Effects: The error-correction model (ECM) results are statistically

significant at the 1-percent level. This implies, relatively speaking, a rapid speed
242 THE JOURNAL OF ENERGY AND DEVELOPMENT

of adjustment to the long-run equilibrium. The values of the ECM are negative and
statistically significant. Indeed, the negative values of the ECM indicate the speed
of convergence from the short to the long run. The coefficients of the ECM show
that short-run deviations are corrected by 37.2 percent for the group of countries
constituting the sample.
When looking at our panel causality test results in table 6, we note that when
we take the totality of the panel countries in our sample, CO2 emissions are
explained in the short term by only two variables: energy consumption and eco-
nomic growth per capita. Indeed, the coefficient of the energy consumption
(0.666) is positive and significant at the 1-percent level. This seems logical be-
cause of the high correlation between energy consumption and CO2 emissions.
Indeed, any increase in the energy consumption by one unit generates an increase
in CO2 emissions by 0.66 units. Also, the per-capita economic growth appears to
have, in the short term, a positive and significant effect at the 10-percent level. In
other words, the relative increase in GDP per capita of a unit causes a relative
increase in CO2 emissions of 0.123 units.
However, at the individual level, we notice that the results are relatively het-
erogeneous. With regard to the effect of energy consumption on CO2 emissions,
the results show that the said effect is established for only three countries—Jordan
(coefficient of 0.94 with a significance at the 1-percent level), Saudi Arabia
(coefficient of 1.15 with a significance at the 5-percent level), and Morocco (coef-
ficient of 1.018 with a significant at the 1-percent level). A priori, in these three
nations, it turns out that energy consumption is the main cause of CO2 emissions.
However, for the remaining countries of Algeria, Bahrain, Egypt, the United Arab
Emirates, Qatar, Tunisia, and Yemen the estimated coefficients were not signif-
icant. This allows us to say, without much risk, that there is not a causality re-
lationship between energy consumption and CO2 emissions.
As for the effect of life expectancy on CO2 emissions, we find that it is gen-
erally not significant (both at the group and individual level) except in the case of
Jordan. Indeed, in this country, the life expectancy has a negative and significant
effect at the 10-percent level on CO2 emissions. This means that the increase in life
expectancy cannot be achieved solely at the expense of lowering CO2 emissions.
Regarding the effect of GDP per capita, we find that it is positive and signif-
icant at the 10-percent level for the group of MENA countries. At the individual
level, said effect is positive and significant at the 10-percent level only for the
cases of Qatar and Yemen. Concerning the urbanization effect, we see that it is not
significant for the group as a whole, while it is positive and significant at the
5-percent level for Jordan and Qatar.
Long-Run Effects: Following P. Pedroni, the fully modified OLS (FMOLS)
and the dynamic OLS (DOLS) technique for heterogeneous cointegrated panels
are estimated.46 Table 7 displays the FMOLS long-run estimation results and table
8 displays the DOLS long-run estimation results.
 Table 6
a
PANEL CAUSALITY TEST RESULTS FOR THE MIDDLE EAST AND NORTH AFRICA REGION (INDIVIDUAL AND GROUP), 1990–2010

Short-Run Elasticities (ln CO2 is the dependent variable)

DLNEC DLNLEXP DLNGDPC DLNURB ECM(-1)
Country Coeff. Prob. Coeff. Prob. Coeff. Prob. Coeff. Prob. Coeff. Prob.
***
Algeria 0.0624 0.8880 107.090 0.2149 -0.0531 0.6527 -49.787 0.1228 -0.8694 0.0012
***
Bahrain 0.5534 0.2768 59.278 0.3018 0.2348 0.3013 -2.8124 0.9477 -0.9392 0.0006
***
Egypt 0.1501 0.5328 11.910 0.2529 0.1852 0.1001 14.2200 0.2101 -1.2052 0.0002
*** * ** ***
Jordan 0.9411 0.0000 -98.484 0.0643 -0.1614 0.1876 6.5992 0.0461 -1.2870 0.0000
*** **
Morocco 1.0182 0.0004 4.942 0.4953 0.0665 0.4826 -2.1080 0.5830 -0.6942 0.0544
* ** **
Qatar 0.2710 0.4294 -36.756 0.8724 0.3769 0.0562 125.9700 0.0115 -0.6088 0.0132
** *
Saudi Arabia 1.1583 0.0153 -17.997 0.6438 0.3360 0.1423 12.3700 0.5918 -0.3759 0.0997
***
Tunisia -0.1094 0.6303 0.948 0.4097 -0.0165 0.8899 3.0031 0.1353 -0.8755 0.0009
**
United Arab Emirates 0.2000 0.8589 24.071 0.8984 0.1490 0.7911 1.7092 0.9475 -0.6681 0.0532
* ***
Yemen 0.3393 0.4542 18.596 0.8683 0.4004 0.0991 1.5090 0.9662 -1.1873 0.0026
*** * ***
Panel 0.6660 0.0000 2.284 0.3470 0.1230 0.0736 1.2477 0.2661 -0.3720 0.0000

a
The Probs. are the standard errors with *** = significance at the 1-percent level; ** = significance at the 5-percent level; and * significance at the
10-percent level. D ( ) is the first difference operator. Prob. = probability; ECM = error-correction model; Coeff. = coefficients; LN = natural logs;
EFFECTS OF EMISSIONS IN MENA COUNTRIES

LEXP = life expectancy at birth; GDPC = the gross domestic product per capita measured in U.S. dollars; EC = energy consumption measured in
kilograms of oil equivalent per capita; and URB = the percentage of urban population relative to the overall population.
243
 244

Table 7
FULLY MODIFIED ORDINARY LEAST SQUARES (FMOLS) LONG-RUN ESTIMATION RESULTS FOR THE MIDDLE EAST AND NORTH
a
AFRICA REGION (INDIVIDUAL AND GROUP), 1990–2010

Long-Run Elasticities (ln CO2 is the dependent variable)

LNEC LNLEXP LNGDPC LNURB
Country Coeff. Prob. Coeff. Prob. Coeff. Prob. Coeff. Prob.

Algeria 0.3412 0.5577 -0.3059 0.7394 0.0218 0.8980 -0.0215 0.9711

***
Bahrain 1.4155 0.0018 -0.1484 0.9689 -0.1921 0.3043 -1.6054 0.6179
*** *** ***
Egypt 0.6623 0.0000 2.6962 0.0001 0.0579 0.1208 -4.1161 0.0000
*** *** ** ***
Jordan 0.9024 0.0000 -3.0930 0.0000 -0.0854 0.0145 2.0094 0.0015
*** ***
Morocco 1.1047 0.0001 -2.1296 0.0077 -0.0710 0.2527 0.8043 0.4271
*** *** *** ***
Qatar 0.4701 0.0004 -45.7510 0.0000 -0.4384 0.0000 44.3440 0.0000
** **
Saudi Arabia 0.5375 0.3194 -16.7620 0.0452 0.2399 0.1568 15.4520 0.0409
***
Tunisia 0.8123 0.0013 -0.7687 0.2170 0.0991 0.1305 -0.5181 0.5333
*** * *
United Arab Emirates 0.7884 0.0043 -11.2470 0.0695 0.1607 0.4986 9.7355 0.0903
** ***
Yemen 0.5218 0.0254 -0.6236 0.0042 0.1053 0.1277 -0.3535 0.1339
***
Panel 0.8199 0.0000 1.4247 0.2335 -0.0892 0.1022 -0.3864 0.2686

a
The Probs. are the standard errors with *** = significance at the 1-percent level; ** = significance at the 5-percent level; and * significance at
the 10-percent level. LN = natural logs; CO2 = carbon dioxide emissions measured by per-capita metric tons; LEXP = life expectancy at birth; GDPC = the
THE JOURNAL OF ENERGY AND DEVELOPMENT

gross domestic product per capita measured in U.S. dollars; EC = energy consumption measured in kilograms of oil equivalent per capita; URB = the
percentage of urban population relative to the overall population; Coeff. = coefficients, and Prob. = probabilities.
 Table 8
DYNAMIC ORDINARY LEAST SQUARES (DOLS) LONG-RUN ESTIMATION RESULTS FOR THE MIDDLE EAST AND NORTH AFRICA
a
REGION (INDIVIDUAL AND GROUP), 1990–2010

Long-Run Elasticities (ln CO2 is the dependent variable)

LNEC LNLEXP LNGDPC LNURB
Country Coeff. Prob. Coeff. Prob. Coeff. Prob. Coeff. Prob.

Algeria 0.0784 0.8992 0.2868 0.7360 0.0925 0.5593 -0.3304 0.5135

**
Bahrain 1.4792 0.0322 -0.0572 0.9926 -0.1914 0.5317 -1.8236 0.7254
*** *** ***
Egypt 0.6506 0.0000 2.9155 0.0007 0.0527 0.3619 -4.3323 0.0000
*** *** * ***
Jordan 0.8814 0.0000 -3.0007 0.0005 -0.0845 0.0964 1.9501 0.0068
*** **
Morocco 1.0902 0.0047 -2.4922 0.0210 -0.0863 0.3680 1.2393 0.3954
** ** **
Qatar 0.4803 0.3151 -41.2150 0.0309 -0.4072 0.0286 39.9410 0.0353
* *
Saudi Arabia 0.7620 0.2217 -21.3350 0.0661 0.2912 0.2051 19.3790 0.0666
***
Tunisia 0.7441 0.0094 -0.9321 0.1611 0.1000 0.2423 -0.2426 0.7824
*
United Arab Emirates 0.7245 0.0625 -9.3172 0.2441 0.1170 0.7693 8.0820 0.2766
**
Yemen 0.4998 0.1639 -0.6319 0.0279 0.0973 0.3214 -0.2897 0.4131
*** *** ***
Panel 2.7870 0.0000 -2.4882 0.0000 -0.7474 0.0000 -0.3072 0.2434

a
The Probs. are the standard errors with *** = significance at the 1-percent level; ** = significance at the 5-percent level; and * significance at
the 10-percent level. LN = natural logs; CO2 = carbon dioxide emissions measured by per-capita metric tons; LEXP = life expectancy at birth; GDPC = the
EFFECTS OF EMISSIONS IN MENA COUNTRIES

gross domestic product per capita measured in U.S. dollars; EC = energy consumption measured in kilograms of oil equivalent per capita; URB = the
percentage of urban population relative to the overall population; Coeff. = coefficients, and Prob. = probabilities.
245
246 THE JOURNAL OF ENERGY AND DEVELOPMENT

At the whole group level, we notice that the coefficient associated with the
consumption of energy is positive and significant at the 1-percent level, while the
coefficients associated with life expectancy and per-capita income are negative
and significant at the 1-percent level (table 8). The positive and significant effect
of energy consumption on CO2 emissions is explained, in the long run, by the fact
that the countries constituting the MENA group will maintain an energy strategy
with the consumption of higher polluting energy products. Indeed, the estimation
shows that a relative increase of 1 percent in energy consumption will lead to an
increase of 2.787 percent of CO2 emission units. This allows us to assume that the
energy basket of goods that are currently being consumed by these countries will
have, in the long term, a multiplier effect on pollution.
The effect of life expectancy on CO2 emissions is negative and significant at
the 1-percent level. The relative increase in life expectancy by a unit decreases the
CO2 emission units by 2.4882 percent. This result seems logical and is consistent
with theoretical expectations, which converge on the hypothesis that an increase in
life expectancy requires fewer greenhouse gas emissions, including CO2. The
latter is often accused of being either the primary cause or the stimulus of several
respiratory diseases and other illnesses.
Additionally, the effect of per-capita income on CO2 emissions is negative and
significant at the 1-percent level. This implies that when per-capita income in-
creases by 1 percent, CO2 emission units will decrease by 0.7474 percent. This
allows us to say that, in the long term, the MENA group should attempt to generate
economic growth from economic activity sectors that are less polluting and
cleaner. This finding seems to be logical since the European Union—the primary
economic partner for almost the majority of MENA countries—has become in-
creasingly rigorous with regard to environmental regulations and imposes envi-
ronmental standards on its commercial partners with respect to meeting CO2
emission reduction goals.
At the individual level, we find that the positive relationship between energy
consumption and CO2 emissions is strongly verified in the majority of countries in
our sample (Bahrain, Egypt, Jordan, Tunisia, Morocco, and the United Arab
Emirates). In the longer term, this can be explained by the governments continuing
their current strategies that maintain the existing energy consumption patterns or,
alternatively stated, the absence of a willingness to substitute the current con-
sumption of energy goods with those that are less polluting.
Regarding the effect of life expectancy on CO2 emissions, the results are
negative and significant, respectively, at the levels of 1 percent (Jordan), 5 percent
(Morocco, Qatar, and Yemen), and 10 percent (Saudi Arabia), and positive and
significant at the 1-percent level only in the case of Egypt. As already mentioned,
as the majority of MENA countries work toward developing new strategies and
policies to improve living conditions and protect the environment, this will ne-
cessitate efforts to reduce CO2 emissions.
 EFFECTS OF EMISSIONS IN MENA COUNTRIES 247

However, the positive and significant effect exerted by life expectancy on

emissions of CO2 in the Egyptian case, despite the fact that it does not conform to
our theoretical expectations, can be explained by the fact that, over the long run,
the improvements in living standards should be achieved through consumption
and the use of more polluting energy products. In other words, Egypt will not
change its long-term strategy for energy use and improving living conditions;
therefore, in the long run, it remains dependent on its current energy policy. Thus,
increasing life expectancy by one unit requires increased CO2 emissions by 2.6962
units. This means that a portion of said emissions are the result of ameliorative
activities on living conditions (health, education, infrastructure, etc.).
As for the effect of per-capita income on CO2 emissions, we notice that it is
negative and significant for the cases of Jordan and Qatar and not significant for
the remainder of the sample. Despite this result being inconsistent with theoretical
expectations, we can say that it can be explained by the fact that the future eco-
nomic strategy of Jordan and Qatar are not based upon current energy structures
but seek to invest more in economic sectors that are less polluting and more
aligned with international environmental standards.
The effect of urbanization seems to be negative and significant at the 1-percent
significance level in the case of Egypt and positive and significant at the 1-percent
level in the case of Jordan, 5-percent level for Qatar, and 10-percent level for
Saudi Arabia. A priori, when the urban population increases, it requires and needs
to consume more goods and social services that are significant generators of CO2
emissions.

Conclusion

The main objective of this article is to explain the relationship between CO2
emissions and a set of economic and socio-economic variables within a select
number of MENA countries. The empirical results have covered two time hori-
zons: the short and long term.
In the short term, our results found that for all countries in our sample, CO2
emissions can be explained by energy consumption and economic growth per
capita, which exert positive and significant effects. Indeed, it seems that in the
short term there is a causal relationship between energy consumption (which
generates more CO2 emissions) and economic growth. So this leads us to note that
this group of countries is continuing, in the short run, to use the same energy
trajectory for end-use consumption by households and the intermediate stages of
consumption. Therefore, this energy consumption vector increases per-capita in-
come and CO2 emissions.
However, in the longer term, there will be a structural change in the economic
course of all selected MENA countries in our sample. Thus, we noticed that CO2
248 THE JOURNAL OF ENERGY AND DEVELOPMENT

emissions are still positively influenced by energy consumption. This means that
this group of nations will retain, in the long run, the same current energy trajectory
(in terms of consumption). The effect of life expectancy is negative and significant
on CO2 emissions, which means that improving living conditions in the long run
cannot continue without the lowering of CO2 emissions. In addition, the per-capita
income effect is negative and significant, which means that the long-term economic
strategy of these countries is based on more activity in non-polluting sectors. In
other words, the growth-generating economic potential is located in non-polluting
sectors and not in sectors that are generators of greenhouse gases.
In terms of recommendations and suggestions, we can say, based upon the esti-
mation results, if in the short term the strategy of MENA countries is based on
a trajectory of consumption and production of polluting energy products, there will be
in the longer term a strategic change within this group as they look toward promoting
lower polluting activities and sectors. This is no longer a choice but a constraint
(imposed by the European Union requiring exporting countries to comply with en-
vironmental standards for CO2 and other emissions). So long as life expectancy is
negatively correlated to CO2 emissions, then it would be logical that these countries
would invest in improving living conditions and reducing CO2 emissions.

NOTES
1
N. Stern, The Stern Review: The Economics of Climate Change (Cambridge, United Kingdom:
Cambridge University Press, 2007).
2
W. Nordhaus and J. Boyer, Warming the World: Economic Models of Global Warming
(Cambridge, Massachusetts: MIT Press, 2000).
3
The World Bank, The World Bank Annual Report 2010 (Washington, D.C.: The World Bank, 2010).
4
G. Grossman and A. Krueger, “Economic Growth and the Environment,” Quarterly Journal of
Economics, vol. 110, no. 2 (1992), pp. 353–77.
5
N. Shafik “Economic Development and Environmental Quality: An Econometric Analysis,”
Oxford Economic Papers, vol. 46, no. 5 (1994), pp. 757–73.
6
D. Stern, M. Common, and E. Barbier, “Economic Growth and Environmental Degradation:
The Environmental Kuznets Curve and Sustainable Development,” World Development, vol. 24,
no. 7 (1996), pp. 1151–160.
7
E. Akbostanci, S. Turut-Asik, and G. I. Tunc, “The Relationship between Income and Envi-
ronment in Turkey: Is There an Environmental Kuznets Curve?” paper delivered at the International
Conference on Economics, Turkish Economic Association, Ankara, Turkey, September 11–13, 2006.
8
I. Martinez-Zarzo and A. Bengochea-Morancho, “Pooled Mean Group Estimation of an En-
vironmental Kuznets Curve for CO2,” Economics Letters, vol. 82, no. 1 (2004), pp. 121–26.
9
A. T. Akarca and T. V. Long, “On the Relationship between Energy and GNP: A Reexami-
nation,” The Journal of Energy and Development, vol. 5, no. 2 (spring 1980), pp. 326–31; E. S. H. Yu
 EFFECTS OF EMISSIONS IN MENA COUNTRIES 249

and B. K. Hwang, “The Relationship between Energy and GNP: Further Results,” Energy Eco-
nomics, vol. 6, no. 4 (1984), pp. 186–90; E. S. H. Yu and J. Y. Choi, “The Causal Relationship
Between Energy and GNP: An International Comparison,” The Journal of Energy and Development,
vol. 10, no. 2 (spring 1985), pp. 249–72; U. Erol and E. S. H. Yu, “On the Causal Relationship
between Energy and Income for Industrialized Countries,” The Journal of Energy and Development,
vol. 13, no. 1 (autumn 1988), pp. 113–22; U. Erol and E. S. H. Yu, “Time Series Analysis of the
Causal Relationships between US Energy and Employment,” Resources and Energy, vol. 9, no. 1
(1987), pp. 75–89; D. I. Stern, “Energy Use and Economic Growth in the USA: A Multivariate
Approach,” Energy Economics, vol. 15, no. 2 (1993), pp. 137–50; and B. S. Cheng, “An Investigation
of Cointegration and Causality between Energy Consumption and Economic Growth,” The Journal of
Energy and Development, vol. 21, no. 1 (autumn 1995), pp. 73–84.
10
J. Asafu-Adjaye, “The Relationship between Energy Consumption, Energy Prices and Eco-
nomic Growth: Time Series Evidence from Asian Developing Countries,” Energy Economics, vol.
22, no. 6 (2000), pp. 615–25; H. Y. Yang, “A Note of the Causal Relationship between Energy and
GDP in Taiwan,” Energy Economics, vol. 22, no. 3 (2000), pp. 309–17; U. Soytas and R. Sari,
“Energy Consumption and GDP: Causality Relationship in G-7 Countries and Emerging Markets,”
Energy Economics, vol. 25, no. 1 (2003), pp. 33–7; R. Morimoto and C. Hope, “The Impact of
Electricity Supply on Economic Growth in Sri Lanka,” Energy Economics, vol. 26, no. 1 (2004),
pp. 77–85; G. Altinay and E. Karagol, “Electricity Consumption and Economic Growth: Evidence
From Turkey,” Energy Economics, vol. 27, no. 6 (2005), pp. 849–56; and P. K. Narayan and R.
Smyth, “Energy Consumption and Real GDP in G7 Countries: New Evidence from Panel Coin-
tegration with Structural Breaks,” Energy Economics, vol. 30, no. 5 (2008), pp. 2331–341.
11
Y. U. Glasure and A. R. Lee, “Cointegration, Error Correction and the Relationship between
GDP and Energy: the Case of South Korea and Singapore,” Resource and Energy Economics, vol.
20, no. 1 (1998), pp. 17–25; B. S. Cheng, “An Investigation of Cointegration and Causality between
Energy Consumption and Economic Growth;” B. S. Cheng, “Causality between Energy Con-
sumption and Economic Growth in India: An Application of Cointegration and Error-Correction
Modeling,” Indian Economic Review, vol. 34, no. 1 (1999), pp. 39–49; J. Chang and J. F. Wong,
“Poverty, Energy and Economic Growth in Singapore,” Working Paper no. 37, Department of
Economics, National University of Singapore, 2001; U. Soytas and R. Sari, “Energy Consumption
and GDP: Causality Relationship in G-7 Countries and Emerging Markets;” and P. K. Narayan and
R. Smyth, op. cit.
12
A. M. Masih and R. Masih, “Energy Consumption, Real Income and Temporal Causality:
Results from Multi-Country Study Based on Cointegration and Error-Correction Modeling Tech-
niques,” Energy Economics, vol. 18, no. 3 (1996), pp. 165–83; J. Asafu-Adjaye, op. cit.; Y. U.
Glasure, “Energy and National Income in Korea: Further Evidence on the Role of Omitted Vari-
ables,” Energy Economics, vol. 24, no. 4 (2002), pp. 355–65; and W. Oh and K. Lee, “Causal
Relationship between Energy Consumption and GDP Revisited: The Case of Korea 1970–1999,”
Energy Economics, vol. 26, no. 1 (2004), pp. 51–9.
13
U. Soytas, R. Sari, and B. T. Ewing, “Energy Consumption, Income and Carbon Emissions in
the United States,” Ecological Economics, vol. 62, no. 3 (2007), pp. 482–89.
14
E. Neumayer and M. A. Cole, “Examining the Impact of Demographic Factors on Air Pol-
lution,” Pollution and the Environment, vol. 26, no. 1 (2004), pp. 5–21; M. Shahbaz, C. F. Tang,
and M. Shabaz Shabbir, “Electricity Consumption and Economic Growth Nexus in Portugal Using
Cointegration and Causality Approaches,” Energy Policy, vol. 39, no. 6 (2011), pp. 3529–536;
250 THE JOURNAL OF ENERGY AND DEVELOPMENT

F. Halicioglu, “An Econometric Study of CO2 Emissions, Energy Consumption, Income and
Foreign Trade in Turkey,” Energy Policy, vol. 37, no. 3. (2009), pp. 1156–164; and G. E. Akpan
and U. F. Akpan, “Electricity Consumption, Carbon Emissions and Economic Growth in Nigeria,”
International Journal of Energy Economics and Policy, vol. 2, no. 4 (2012), pp. 292–306.
15
U. Soytas et al., op. cit.; H. E. Chebbi and Y. Boujelbene, “CO2 Emissions, Energy Con-
sumption and Economic Growth in Tunisia,” paper presented at the 2008 International Congress,
European Associate of Agricultural Economists, Ghent, Belgium, August 26–29, 2008; J. B. Ang,
“Economic Development, Pollutant Emissions and Energy Consumption in Malaysia,” Journal of
Policy Modeling, vol. 30, no. 2 (2008), pp. 271–78; U. Soytas and R. Sari, “Energy Consumption,
Economic Growth, and Carbon Emissions: Challenges Faced by an EU Candidate Member,”
Ecological Economics, vol. 68, no. 6 (2009), pp. 1667–675; and X.-P. Zhang and X.-M. Cheng,
“Energy Consumption, Carbon Emissions, and Economic Growth in China,” Ecological Eco-
nomics, vol. 68, no.10 (2009), pp. 2706–712.
16
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. 6991–999.
17
N. Georgescu-Roegen, The Entropy Law and the Economic Process (Cambridge, Massa-
chusetts: Harvard University Press, 1971), and D. H. Meadows, D. L. Meadows, J. Randers, and
W. W. Behrens III, The Limits to Growth (New York: Universe Books, 1972).
18
E. Lambin, Une écologie du Bonheur (Paris: Editions Le Pommier, 2009), p. 267.
19
C. Declercq, M. Pascal, O. Chanel, M. Corso, A. Ung, L. Pascal, M. Blanchard, S. Larrieu, and
S. Medina, Impact sanitaire de la pollution atmosphérique dans neuf villes françaises. Résultats
du projet Aphekom (Saint-Maurice: Institut de Veille Sanitaire, 2011).
20
Y. Chen, A. Ebenstein, M. Greenstone, and H. Li, “Evidence on the Impact of Sustained
Exposure to Air Pollution on Life Expectancy from China’s Huai River Policy,” Proceedings of the
National Academy of Sciences, vol. 110, no. 6 (2013), pp. 12936–41.
21
United Nations Environment Programme (UNEP), Global Environmental Outlook: GEO-4
Environment for Development (Nairobi, Kenya: UNEP, 2007).
22
Programme des Nations Unies pour l’environnement (PNUE), L’avenir de l’Environnement
Mondial (Nairobi, Kenya: PNUE, 1999), available at http://www.unep.org/geo/GEO2000/pdfs/ov-fr.pdf.
23
A. Ali and K. Ahmad, “The Impact of Socio-Economic Factors on Life Expectancy in Sul-
tanate of Oman: An Empirical Analysis,” Middle-East Journal of Scientific Research, vol. 22, no. 2
(2014), pp. 218–24.
24
A. Assadzadeh, F. Bastan, and A. Shahverdi, “The Impact of Environmental Quality and
Pollution on Health Expenditures: A Case Study of Petroleum Exporting Countries,” in the Pro-
th
ceedings of the 29 International Business Research Conference, Sydney, Australia, November
24–25, 2014.
25
M. Mehrara and M. R. Masoumi, “The Relationship between Income, Health, and the Envi-
ronment,” Indian Journal of Fundamental and Applied Sciences, vol. 4, no. S4 (2014), pp. 737–44.
 EFFECTS OF EMISSIONS IN MENA COUNTRIES 251
26
T. Azomahou, F. Laisney, and P. Nguyen-Van, “Economic Development and CO2 Emissions:
A Nonparametric Panel Approach,” Journal of Public Economics, vol. 90 (2006), pp. 1347–363.
27
Programme des Nations Unies pour l’environnement (PNUE), op. cit., p. 38.
28
Ibid.
29
J. Breitung, “The Local Power of Some Unit Root Tests for Panel Data,” in Advances in
Econometrics, Volume 15: Nonstationary Panels, Panel Cointegration, and Dynamic Panels, ed.
B. H. Baltagi (Amsterdam: JAY Press, 2000), pp. 161–78; K. Hadri, “Testing for Stationarity in
Heterogeneous Panel Data,” Econometric Journal, vol. 3, no. 2 (2000), pp. 148–61; I. Choi, “Unit
Root Tests for Panel Data,” Journal of International Money and Finance, vol. 20, no. 2 (2001), pp.
249–72; A. Levin, C.-F. Lin, and C.-S. Chu, “Unit Root Tests in Panel Data: Asymptotic and Finite-
Sample Properties,” Journal of Econometrics, vol. 108, no. 1 (2002), pp. 1–24; K. Im, M. Pesaran,
and Y. Shin, “Testing for Unit Roots in Heterogeneous Panels,” Journal of Econometrics, vol. 115,
no. 1 (2003), pp. 53–74; and J. L. Carrion-i-Silvestre, T. del Barrio-Castro, and E. Lopez-Bazo,
“Breaking the Panels: An Application to GDP Per Capita,” Econometrics Journal, vol. 8, no. 2
(2005), pp. 159–75.
30
J. Breitung, op. cit.; A. Levin et al., op. cit.; and K. Im et al., op. cit.
31
G. S. Maddala and S. Wu, “A Comparative Study of Unit Root Tests with Panel Data and
a New Simple Test,” Oxford Bulletin of Economics and Statistics, no. 61 (1999), pp. 631–52, and I.
Choi, op. cit.
32
K. Hadri, op. cit., and J. L. Carrion-i-Silvestre et al., op. cit.
33
K. Im et al., op. cit.
34
Ibid.
35
Ibid.
36
A. Levin et al., op. cit.
37
J. Breitung, op. cit.
38
K. Im et al., op. cit.
39
K. Hadri, op. cit.
40
P. Pedroni, “Critical Values for Cointegration Tests in Heterogeneous Panels with Multiple Re-
gressors,” Oxford Bulletin of Economics and Statistics, vol. 61, no. 0 special edition (1999), pp. 653–70,
and “Panel Cointegration: Asymptotic and Finite Sample Properties of Pooled Time Series Tests with an
Application to the PPP Hypothesis,” Econometric Theory, vol. 20, no. 3 (2004), pp. 597–625.
41
P. Pedroni, “Fully Modified OLS for Heterogeneous Cointegrated Panels and the Case of
Purchasing Power Parity,” Indiana University Working Papers in Economics 96-020, 1996.
42
P. Phillips and H. R. Moon, “Non-Stationary Panel Data Analysis: An Overview of Some
Recent Developments,” Econometric Reviews, vol. 19, no. 3 (2000), pp. 263–86.
252 THE JOURNAL OF ENERGY AND DEVELOPMENT
43
C. Kao and M.-H. Chiang, “On the Estimation and Inference of a Cointegrated Regression in
Panel Data,” in Nonstationary Panels, Panel Cointegration, and Dynamic Panels, Advances in
Econometrics, Vol. 15, ed. B. Baltagi (Amsterdam: JAY Press, 1997), pp. 179–222.
44
Ibid.
45
R. F. Engle and C. W. J. Granger, “Co-Integration and Error Correction: Representation,
Estimation, and Testing,” Econometrica, vol. 55, no. 2 (1987), pp. 251–76.
46
P. Pedroni, “Fully Modified OLS for the Heterogeneous Cointegrated Panels,” Advances in
Econometrics, vol. 15 (2000), pp. 93–130.