ENERGY CONSUMPTION, CO2
EMISSION AND ECONOMIC
GROWTH IN NIGERIA
Mohammed I. SHUAIBU &Mutiu A. OYINLOLA
Department of Economics
University of Ibadan
Ibadan, Nigeria
Preamble
• Proponents of global warming have argued that global
emissions of carbon dioxide (CO2) remain a major
source of global warming.
• Indeed, CO2 emission has increased by 3% in 2011,
reaching an all-time high of 34billion tonnes (NEAA,
2012).
• The alarming threat and attendant consequence of
climate change on economies make this nexus
contentious.
• Energy Use: Crucial to economic prosperity.
• YET adequate attention to optimal energy use to
minimise pollutant emission
Preamble
• Nigeria: 6% average growth (last decade).
• In spite of this remarkable development, the
supply of electricity which remains the main
source of energy in Nigeria is fitful.
• This has led to a shift to other alternative
sources of power which require burning of
fossil fuels and has consequently increased
toxic emissions.
• Worrisome trend!
Preamble
• The energy consumption index in Nigeria increased
from 2.8% in 2010 relative to the increase and dip of
4.9% and 1.9% observed accordingly in 2008 and 2009.
• In absolute terms, energy consumed stood at
19.1million tonnes of coal equivalent (TCE) in 2010,
from 20.4million and 18.3million TCE in 2008 and 2009,
respectively (CBN, 2009, 2010).
• Consequent upon above, emission of green house
gases stood at a staggering 49.6 and 41.2million tonnes
in 2008 and 2009.
• In sum, the average change of pollutants emitted
between 1990 and 2009 was 41.3%.
Motivation
•
•
•
•
•
•
•
This rather appalling scenarios, inter alia, stimulate this study.
Similar studies: Binh, 2011, for Vietnam; Omotor, 2008; Omisakin, 2008; Odularu
and Okonkwo, 2009; Dantama et al. 2012; Olusanya, 2012; for Nigeria).
Focus: the effect of energy and/or electricity consumption on economic growth,
neglecting the consequence of persistent energy depletion in form of CO2
emission on the economy.
However, the study by Chebbi (2009) for Tunisia, Menyah and Wolde-Rufael (2010)
for South Africa, Tiwari (2011a, 2011b) for India, Akpan and Akpan (2012) for
Nigeria, are notable exceptions.
Methodology: Johansen and Juselius or Engle and Granger cointegration and/or
Granger causality approach especially for studies carried out in Nigeria.
We depart from these approaches and propose to use the Gregory-Hansen
cointegration test and the Toda-Yomamato causality test.
Why?: These studies ignore the possible effects of structural breaks which may
have an effect on energy consumption.
Motivation Cont’d
• The Gregory and Hansen (1996) tests for structural break
cointegration allows for cointegrating vectors to change at
an unknown time period.
• This is in view of the fact that, in general, failure to account
for breaks can produce misleading tests leading to incorrect
inference.
• The Granger non-causality test using the Toda-Yomamoto
(T-Y) procedure which is applicable regardless of whether a
series is I(0), I(1) or I(2), not-cointegrated or cointegrated of
any arbitrary order.
• This implies that it avoids the potential bias associated with
unit root and cointegration tests (see Rambaldi and Doran,
1996) which most studies have used.
Objectives and Expected Outcomes
• To develop and test an econometric model to identify
the main economic fundamentals surrounding the
interplay between economic growth, CO2 emission and
energy consumption in Nigeria.
• To examine the long- and short-run effect of CO2 and
energy consumption on economic growth.
• Apriori, we expect a positive relationship between
energy consumed and growth while a negative
relationship is expected between CO2 emission and
growth. For labour and capital, the other control
variables, we expect a positive association with growth.
Stylized Facts
• Climate change is a global phenomenon and its impact in form of
unusual and extreme weather conditions is evident in Nigeria.
• This has primarily been induced by depletion and consumption of
energy resources which in turn result in CO2 emissions.
• Nigeria ranks 38 amongst the global community of nations in terms
of CO2 emissions.
• The Nigerian economy has over the years been heavily dependent
on oil as it account for over 95% of governments’ export earnings
and 40% of revenue inflows. The consumption of refined petroleum
products, biomass and electricity has over the years dominated
Nigeria’s energy consumption mix.
• While significant and stable growth rates have been recorded in the
last decade, emission of green house gases has been on the rise
given the expansive energy consumption that accompanied growth.
Stylized Facts Cont’d
•
•
•
•
•
•
•
Emissions in metric tonnes per million US$ of GDP in Nigeria as at 2004 stood at
762 at purchasing power parity in 2000USD while the emissions per capita (tons
per person) was 0.73 (World Bank, 2007).
This was probably due to a faster increase in the use of gas (especially) and oil
than of coal. Nigeria and Brazil are examples where this effect was dominant in the
total change of emissions (ibid.).
In addition, CO2 emissions have on the average grown by 41.3% between 1990
and 2009 albeit the downward trend observed between 1992 and 1995.
Specifically, emissions from the use of coal/peat fell by almost 90% indicating a
drastic reduction in the use of coal.
During the same period, the use of oil increased and its emission rose by almost
30% while the percentage change observed for natural gas emission stood at 82%.
International marine bunkers also had a significant share in emissions as the
period average percentage change was a staggering 237.3%. I
nternational aviation bunkers emission was slightly less with about 109.3%.
power
Energy use (kg of CO2
emissions CO2 emissions CO2 emissions
cons oil equivalent per from gaseous fuel from liquid fuel from solid fuel
capita)
cons (kt)
cons(kt)
cons (kt)
GDP growth
Year
Electric
(kWh)
1970-1974
1926250000.00
624.21
489.91
7440.34
631.46
11.83
1975-1979
3266142857.14
654.64
2561.14
10127.73
663.20
3.91
1980-1984
5344800000.00
711.99
6118.76
28487.46
255.96
-3.85
1985-1989
7853000000.00
716.79
6723.81
33289.03
235.42
5.72
1990-1994
9297000000.00
735.83
9395.59
41192.88
229.55
3.63
1995-1999
9335200000.00
723.03
10892.46
27739.39
184.82
2.50
2000-2004
12443600000.00
737.17
14659.20
33262.62
49.14
6.19
2005-2009
18390800000.00
727.19
21935.99
29523.02
29.34
6.21
2010
2011
21624000000.00
na
na
na
na
na
na
na
na
na
7.82
6.67
Literature Review
• Theoretically, the links between economic growth,
environment and climate change are complex, multidimensional and dynamic such that the implications on the
economy are numerous: changes in productivity, resource
endowments, and production and consumption patterns.
• The traditional and new growth models have been widely
used to examine the implications of environment on
growth. In such models, energy consumption is considered
as inputs to the growth process of an economy (Tsani,
2010).
• The Ramsey-Cass-Koopmans model of growth has also
formed the basis of much related work on climate change
economics Fankhauser and Tol (2005) and (Millner and
Dietz (2011)
Literature Review
• Another strand of literature, the ecological economic
theory, states that energy consumption is a limiting factor
to economic growth, especially in modern economies
(Binh, 2011).
• However, Stern (2000) had argued that technological
progress and other physical inputs could not possibly
substitute the vital role of energy in production processes.
• In more recent times renewed interest in examining this
nexus have focused on the Environmental Kuznets Curve
(EKC) or what is also termed as the Carbon Kuznets Curve
(CKC) hypothesis (See Aslandis, 2009 and Galleotti et al.,
2009), Tiwari (2011a and 2011b) and Akpan and Akpan
(2012).
Literature Review
• Broad application of different methodologies!
• CGE (Static: Bosello et al., 2007 and Dynamic: Eboli et al., 2010).
• Econometric Analysis (Tiwari (2011a) granger causality and vector
error correction, Tiwari (2011b) VAR framework and variance
decomposition and impulse response function. Binh (2011)
threshold cointegration , vector error correction models and
Granger causality tests, Dantama et al, (2012) ARDL approach to
cointegration analysis and error correction model, Akpan and Akpan
(2012) Multivariate Vector Error Correction (VECM), Chebbi (2009)
Johansen cointegration, error-correction modeling, impulse
response functions and Granger causality; Menyah and WoldeRufael (2010) bound test approach to cointegration, granger
causality test and variance decomposition analysis.
Literature Review Cont’d
• Empirical results have been mixed!
• More studies lend support to the positive relationship between energy
consumption and growth (See Odularu and Okonkwo, 2009, Chebbi, 2009)
• Menyah and Wolde-Rufael (2010) found a long-run relationship among the
variables with a positive and statistically significant relationship between
pollutant emissions and economic growth.
• Tiwari (2011a) found that CO2 granger causes GDP while energy
consumption does not granger cause GDP. Also, GDP does not granger
cause CO2 but energy consumption granger causes CO2 emissions.
• Tiwari (211b) found that CO2 emissions have positive impact on energy
use and capital but negative impact on population and GDP.
• Binh’s (2011) indicated that there is a strong uni-directional causality
running from economic growth to energy consumption, but not vice versa.
• Dantama et al, (2012) found that a long-run relationship between
economic growth and energy consumption exists.
Theoretical Framework and
Methodology
• This paper adopts and adapts the dynamic endogenous
growth model used by Menyah and Wolde-Rufael
(2010) as the basis for our empirical specification.
• However, Romer (2006) had argued that since
environmental considerations are absent in such
models, many now believe, following Malthus’s (1798)
classic argument that these considerations are critical
for long-run economic growth.
• It is against this backdrop that we extend the model to
include CO2 emission as an additional dependent
variable.
The Model
LNGDPt  0  1LNPOPt  2 LNINVt  3 LNCOEt  4 LNECON  et (1)
• Where LNGDP represents log of gross domestic
product, PLNOP represents log of population
growth, LNINV denotes capital (gross fixed capital
formation is used as a proxy), LNCOE stands for
CO2 emissions, LNECON represents the log of
aggregate energy consumption while is the error
term assumed to be white noise.
• Apriori, we expect >0, >0, <0 and >0.
Estimation Procedures
• Unit root test developed by Zivot and Andrew (1992).
p
Yt    2 DU t ( b )   T  Yt 1  i Yt i  et
(2)
pi 1
Yt     DTt ( b )   T  Yt 1  i Yt i  et
i 1
(3)
p
Yt     DU t ( b )   T   DTt ( b )  Yt 1  i Yt i  et
(4)
i 1
• Where DUt and DTt are dummy variables for a mean
shift and a trend shift respectively; DUt(τb) = 1 if t >
and 0 otherwise, and DTt( ) = t- if t > τb and 0
otherwise. In other words, DUt is a sustained dummy
variable that captures a shift in the intercept, and DTt
represents a shift in the trend occurring at time.
Estimation Procedure
• Gregory and Hansen (1996) tests for cointegration where
the structural break is test-determined and the
cointegrating vectors are allowed to change at an unknown
time period. T
y1t  1  2t   y2t  et
(5)
y1t  1  2t   t   T y2t  et
y1t  1  2t   T y2t   T y2tt  et
(6)
(7)
• The standard methods to test the null hypothesis of no
cointegration are residual-based and are obtained when
equations (5, 6 and 7) are estimated using OLS and the unit
root tests are applied to the regression errors (Gregory and
Hansen, 1996).
Estimation Procedure
• Toda-Yomamoto (T-Y) Granger non-causality technique
ln GDPt 
ln GDPt 1 
ln GDPt 2 
ln GDPt 3 
ln GDPt 4   ln GDPt 

ln POP 
ln POP 
ln POP 
ln POP 
ln POP  
t
t 1 
t 2 
t 3 
t  4   ln POPt 






ln INVt   A0  A1 ln INVt 1   A2 ln INVt 2   A3 ln INVt 3   A4 ln INVt 4    ln INV 

t









 
ln COEt 
ln COEt 1 
ln COEt 2 
ln COEt 3 
ln COEt 4   ln COEt 

ln ECON 
ln ECON 
ln ECON 
ln ECON 
ln ECON  

t
t 2 
t 3 
t  4  






 ln ECONt 
• In Eq. (8), A1…A4 are four 5×5 matrices of coefficients
with A0 being the 5×1 identity matrix, εs are the
disturbance terms with zero mean and constant
variance.
• From Eq. (8) we can test the hypothesis of granger
causality amongst the variables with the following
hypothesis: H  a  a  a  a  0 and an opposite of noncausality with the following hypothesis: H  a  a  a  a  0
0
1
ij
2
ij
3
ij
4
ij
0
1
ji
2
ji
3
ji
4
ji
Empirical Analysis and Findings
UNIT ROOT TESTs
• First, we conclude that the structural breaks in
the series are not sturdy enough to generate any
divergence with the results of conventional unit
root tests.
• ADF test carried out showed that all the variables
are non-stationary at levels but became
stationary after their first difference was taken. It
is pertinent to note that the ADF test treats
regime shifts as exogenous.
Zivot-Andrews Unit Root Test Result
Z-A (1992)
Model A
Variable
Model B
Model C
t
Breakpoint
Lag
t
Breakpoint
Lag
t
Breakpoint
Lag
LNCOE
-5.15**
2000
0
-4.21**
1997
0
-5.02**
2000
0
LNGDP
-4.22
1992
1
-2.39
1982
1
-3.72***
1992
1
LNINV
-5.08**
1983
3
-4.85*
1986
3
-5.82*
1983
3
LNPOP
-9.14*
1993
3
-6.60*
2005
3
-6.05*
2005
3
-3.29***
2001
4
-3.47
2007
2
-4.05*
2001
2
LNECON
Empirical Analysis and Finding Cont’d
COINTEGRATION TEST
• We find evidence of a significant long-run
relationship amongst the variables considered as
the ADF and PP test statistic exceed the critical
values at the 1%, 5% and 10% level.
• The Engle and Granger cointegration test
conducted validate the G-H cointegration test
results as it shows the significance of the ADF
statistic of the residuals of the estimated model.
Gregory-Hansen Cointegration Tests
Level Shift Model
Level Shift with Trend Model
ADF Procedure
-5.30
2
1991
Phillips Procedure
-38.48
Regime Shift Model
t-stat
Lag
Break
-4.14
0
1992
-6.45
1
1992
Za-stat
-25.26
Zabreak
Zt-stat
1992
-4.19
1992
-5.85
1992
-7.00
Ztbreak
1992
1992
1992
-45.33
Empirical Analysis and Finding Cont’d
 CAUSALITY TEST
• We find that higher growth rates are associated with pollutant
emissions and this is significant at 1% significance level.
• As expected, increased energy use intensifies emission of CO2 at the
5% level.
• We present reasonable evidence showing that within a neoclassical growth model, energy consumption and emission of
carbons do not lead to economic growth in Nigeria.
• What makes our finding differ with other previous similar studies
may be the fact that they relied on the environmental Kuznets
curve theory to examine the nexus while we considered the
relationship within the new growth model.
• In addition, we considered additional variables which are expected
to drive Nigeria’s long run growth.
Toda-Yomamoto Causality Test Result
Excluded
LNECON
LNGDP
LNINV
LNPOP
All
Excluded
LNCOE
LNGDP
LNINV
LNPOP
All
Excluded
LNCOE
LNECON
LNINV
LNPOP
All
Excluded
LNCOE
LNECON
LNGDP
LNPOP
All
Excluded
LNCOE
LNECON
LNGDP
LNINV
All
Dependent variable: LNCOE
Chi-sq
9.762006
11.1201
1.941714
2.460189
39.61857
Dependent variable: LNECON
Chi-sq
1.747484
11.4951
12.05518
7.350929
16.79229
Dependent variable: LNGDP
Chi-sq
0.252004
0.366448
1.128372
1.047775
7.162158
Dependent variable: LNINV
Chi-sq
0.323052
15.98737
12.77942
11.05991
33.08963
Dependent variable: LNPOP
Chi-sq
17.95404
3.628073
13.18016
17.20822
90.12281
df
3
3
3
3
12
Prob.
0.0207
0.0111
0.5846
0.4825
0.0001
df
3
3
3
3
12
Prob.
0.6264
0.0093
0.0072
0.0615
0.1576
df
3
3
3
3
12
Prob.
0.9688
0.9471
0.7702
0.7897
0.8467
df
3
3
3
3
12
Prob.
0.9556
0.0011
0.0051
0.0114
0.0009
df
3
3
3
3
12
Prob.
0.0004
0.3045
0.0043
0.0006
0.0000
Empirical Analysis and Finding Cont’d
• The model is dynamically stable as shown by
the Inverse root of AR Characteristic
Polynomial
Inverse Roots of AR Characteristic Polynomial
1.5
1.0
0.5
0.0
-0.5
-1.0
-1.5
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
Policy Implications and Concluding
Remarks
• This study examined the relationship between energy consumption,
CO2 emissions and economic growth in Nigeria.
• We relied on the new growth model as the theoretical
underpinnings of the study and thus basis of our empirical model
specification.
• The study found the existence of a long-run relationship between
the variables considered.
• However, we did not find a causal relationship running from energy
consumption and CO2 emissions to growth.
• This finding departs from other empirical studies for Nigeria and the
reason may be attributed to the fact that most previous studies for
Nigeria have relied on the Environmental Kuznets Curve hypothesis
which does not account for traditional growth determinants.
Policy Implication and Concluding
Remarks
• We also find a significant causal relationship running from
growth expansion to CO2 emissions and energy
consumptions.
• In other words, the economy expands, an increment is
observed in per capita energy consumption and thus
emissions of pollutants also increase.
• The policy implication of our finding buttresses the need
for government to green its energy policies and diversify
the countries energy sources.
• This may be done by considering renewable energy sources
such as biomass, wind and solar amongst others such that
CO2 emissions are reduced to the barest minima and the
long-run balanced growth path can be sustained.