Article 6 Nonrenewable Energy, Renewable Energy, Carbon Dioxide
Article 6 Nonrenewable Energy, Renewable Energy, Carbon Dioxide
Article 6 Nonrenewable Energy, Renewable Energy, Carbon Dioxide
art ic l e i nf o a b s t r a c t
Article history: This paper mainly investigates the relationship among energy consumption, carbon emissions and
Received 9 January 2015 economic growth in China from 1952 to 2012. First, we implement unit root and cointegration analysis to
Received in revised form test the stationary. Furthermore, we analyze the mutual influence among energy consumption, carbon
26 May 2015
emissions and economic growth through Granger causality analysis. Next, we also conduct static and
Accepted 29 July 2015
dynamic regression analysis on the determinants of carbon emissions and economic growth. Last, we
predict the future influence of different energy consumption on carbon emissions and economic growth.
Keywords: We find that coal has dominant impact on economic growth and carbon emissions. GDP (Gross Domestic
Nonrenewable and renewable energy Product) has bi-directional relationship with CO2 (carbon dioxide) emission, coal, gas, and electricity
CO2 emission
consumption. It is imperative to change energy consumption structure in China. We had better decrease
Economic growth
coal consumption rate. It is significant to develop hydro and nuclear power in China.
& 2015 Elsevier Ltd. All rights reserved.
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 680
2. Literature review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 681
3. Data and methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 682
4. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 684
4.1. Unit root. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 684
4.2. Cointegration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 685
4.3. Granger causality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 685
4.4. Static and dynamic regression analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 686
4.5. Impulse response function analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 686
5. Conclusion and policy implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 687
Acknowledgment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 688
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 688
1. Introduction
http://dx.doi.org/10.1016/j.rser.2015.07.176
1364-0321/& 2015 Elsevier Ltd. All rights reserved.
X. Long et al. / Renewable and Sustainable Energy Reviews 52 (2015) 680–688 681
Fig. 1. GDP, CO2 emission and Energy consumption in China. (a) The unit of GDP is Fig. 3. The installed capacity of nuclear power consumption and its rates in China.
100 million RMB; (b) the Unit of CO2 is 10,000 t; (c) the Units of coal, oil, gas and (a) Nuclear power is indicated by the left value with the unit of 10,000 t coal
electricity are 10,000 t coal equivalent. (d) GDP has been deflated based on 1990. equivalent, and its rates are indicated by right value with the unit of percentage.
(e) the data are from China Compendium of Statistics 1949–2008 [3], and China (b) The data are from China Compendium of Statistics 1949–2008 [3], and China
Statistical Yearbook [4]. Statistical Yearbook [4].
hypotheses which have been adduced from the studies conducted short-run elasticity of energy consumption and energy R&D to
on the nexus between energy consumption and economic growth. variations in oil prices and income for 20 OECD (Organization for
First is unidirectional causality hypothesis, which incorporates the Economic Co-operation and Development) member countries
conservative and economic growth hypotheses [8,9]. The former from 1980 to 2010. Satti et al. [30] further investigated coal
indicates that economic growth has significant positive impact consumption and economic growth in Pakistan using error
upon energy consumption, while the latter means that energy correction model (ECM) and Granger causality test from 1974
consumption rather exerts positive influence upon economic to 2010. They found bidirectional relationship between them.
growth. The second form of relationship is called bi-directional Inglesi-Lotz [31] explored the influence of renewable energy
causality hypothesis. This is also known as the feedback hypothesis consumption upon economic growth of four OECD countries from
[10]. The third relational form is called the neutral causality 1990 to 2010. The results indicated that renewable energy con-
hypothesis, which suggests that there is virtually no significant sumption has positive impact on economy growth through unit
relationship between economic growth and energy consumption root, cointegration of panel data, and fixed-effect regression
[8]. There are some variations of the hypotheses testing due to analysis. Omri et al. [32] examined the different impacts of nuclear
model specification, econometric methods, country development energy, renewable energy on economic growth in 17 developed
levels (Sebri [11]). and developing countries through dynamic simultaneous-
Some of the previous studies on this subject matter, explored equation panel data analysis. It is important to establish the
Granger-causality time series approach, unit root, and cointegra- coordinated management mechanism of integrated planning and
tion tests to determine the relationship between economic growth transmission of renewable energy in China (Zeng et al. [33]).
and energy consumption using various frameworks. Kraft and Shahbaz et al. [34] studied the impact of renewable energy
Kraft [12] for instance, in earlier times used Granger-causality to consumption on economic growth in Pakistan through the quarter
examine the relationship between GNP (Gross National Product) data from 1972Q1 to 2011Q4. They found renewable energy
growth and energy consumption in the United States. Zhang and consumption has mutual influence upon economic growth.
Cheng [13] used a multivariate model for economic growth, energy With the increase of carbon dioxide emission, “3-E” (energy,
use, carbon emission, capital and urban population in China from environment and economy) has become a hot topic. Many pre-
1960 to 2007. Chiou-Wei et al. [14] applied linear and nonlinear vious studies just paid more attention on “2-E” (such as Pao and
Granger causality tests to six Asian industrializing countries and Fu, [24]; Shahbaz et al. [34]). Also, while most previous studies
the United States. According to Lin et al. [15], renewable energy mainly focused on renewable energy or non-renewable energy,
consumption in China has mutual Granger causality relationship this paper investigates different roles of renewable and non-
with economic growth. Ma et al. [16] surveyed the renewable renewable energy consumption on economic growth and carbon
energy economy of China from the aspects of reviewing the emissions. What is more, we also analyze hydro and nuclear
situation of renewable energy development, assessing the coun- power in China. Most literatures adopt static regression analysis,
try's potential for renewable energy. Sun and Peng [17] analyzed but this paper combines static and dynamic regressions to explore
the relationship between energy consumption and economic the impacts of energy consumption on economic growth and
growth in the Yangtze River Delta of China from 1990 to 2008. carbon emissions.
Zhang [18] examined the nexus between energy consumption and The potential innovations of this paper are as follows: first and
economic growth through co-integration analysis in the BRICS foremost, this paper mainly investigates the relationship among
countries (Brazil, Russia, India, China, South Africa). energy consumption, carbon emissions and economic growth in
Zhang and Ren [19] investigated the mutual relationship China from 1952 to 2012. More importantly, we also explore the
between energy consumption and economic growth in Shandong different impacts of renewable and non-renewable energy con-
Province of China from 1980 to 2008. Ying et al. [20] analyzed the sumption on carbon emissions and economic growth in China. For
energy consumption and economic growth nexus in China using renewable energy, we choose three different forms of indices:
co-integration analysis and state-space model. Lin et al. [21] hydro power, nuclear power and their summation. Next, we
investigated the relationship between energy consumption and implement unit root and cointegration analysis to test the sta-
economic growth in South Africa using a nonparametric testing tionary, because our data is from 1952 to 2012 and it tends to walk
approach. Sadorsky [22] found that real per capita income and per randomly with drift. Furthermore, we analyze the mutual influ-
capita renewable energy consumption of emerging economies are ence among energy consumption, carbon emissions and economic
positively and significantly related through panel co-integrated growth through Granger causality analysis. We also conduct static
estimates in the study of renewable energy consumption and and dynamic regression analysis on the determinants of carbon
income in emerging economies. emissions and economic growth. Last, we predict the future
Chang et al. [23] reviewed China's energy production, con- influence of different energy consumptions on carbon emissions
sumption and the prospects for renewable energy use. Further- and economic growth. Coal has dominant impact on economic
more, Pao and Fu [24] employed cointegration test to investigate growth and carbon emissions. Through Granger causality analysis,
renewable energy, non-renewable energy and economic growth in we find that GDP has bi-directional relationship with CO2 emis-
Brazil. Hu et al. [25] studied the relationship amongst China's sion, coal, gas, and electricity consumption. It is imperative to
economic growth, energy production and consumption vis-à-vis adjust energy consumption structure in China. We had better
their connection with air emission from 2000 to 2011 based on a decrease the coal consumption rate, and increase the renewable
system indicator. Yasar et al. [26] examined how biogas production energy consumption rate such as hydro, solar, and ocean power
from spent wash of sugar industrial waste has been very helpful in among others.
reducing CO2 emission in Jhang Punjab district of Pakistan, but
decried the labor intensive nature of the work. The relationship
between energy consumption and economic growth might not be
fixed, due to policy changes in times series data, structural breaks, 3. Data and methodology
technology and economic changes [27].
Ozturk and Bilgili [28] investigated the nexus between We compute real GDP by deflating nominal GDP according to
economic growth and biomass consumption for 51 nations in the base year of 1990. We use employment numbers as labor.
Sub-Saharan Africa. Wong et al. [29] estimated long-run and Capital stock is estimated by perpetual inventory method.
X. Long et al. / Renewable and Sustainable Energy Reviews 52 (2015) 680–688 683
Table 1
CO2 emission coefficient estimates for coal, oil and gas.
Energy NCV CEF(IPCC 2006) COF Conversion factors from physical unit to coal equivalent CO2 emission coefficient
Coal Soft coal 20,908 kj/kg 25.8 kg/1,000,000 kg 0.99 0.7143 kg coal equivalent/kg 2.763 kg/kg coal equivalent
Anthracite coal 20,908 kj/kg 26.8 kg/1,000,000 kg 0.99 0.7143 kg coal equivalent/kg 2.145 kg/kg coal equivalent
Weighted mean 20,908 kj/kg 26 kg/1,000,000 kg 0.99 0.7143 kg coal equivalent/kg 1.642 kg/kg coal equivalent
Oil 41,816 kj/kg 20 kg/1,000,000 kg 1 1.4286 kg coal equivalent/kg 2.145 kg/kg coal equivalent
Gas 38,931 kj/ m3 15.3 kg/1,000,000 kg 1 1.33 kg coal equivalent/m3 1.642 kg/kg coal equivalent
Table 2
Descriptive analysis.
Table 3
Correlation analysis.
GDP 1.0000
CO2 0.9791 1.0000
Labor 0.7953 0.8828 1.0000
Capital 0.9722 0.9176 0.6541 1.0000
coal 0.9765 0.9997 0.8837 0.9133 1.0000
Oil 0.9747 0.9935 0.8934 0.9078 0.9905 1.0000
Gas 0.9667 0.9363 0.6969 0.9868 0.9330 0.9227 1.0000
Electricity 0.9956 0.9811 0.8045 0.9718 0.9785 0.9750 0.9746
Capital stock is estimated by perpetual inventory method, Following Chen [39], CO2 emissions are estimated by Eq. (3),
K t ¼ I t þ ð1 δÞK t 1 ð1Þ X
3 X
3
CO2 ¼ CO2 ¼ Ei NCVi CEFi COFi ð44=12Þ ð3Þ
In Eq. (1), K, I and δ represent capital stock, investment, and i¼1 i¼1
depreciation rate respectively. Following Zhang et al. [35], we treat
depreciation rate as 9.6%. Following Wang et al. [36], we use total where CO2 indicates carbon dioxide emission of coal, oil and
social fixed asset investment as investment variable. Following natural gas. Electricity used is renewable, which is generated by
Chow and Lin [37] and Kim [38], the initial capital stock is wind, water and nuclear. Thus, we exclude such electricity to
estimated as Eq. (2): compute carbon emission.
Energy consumption just incorporates coal, oil and gas. NCV is
It
K0 ¼ ð2Þ net calorific value provided by China Energy Statistics Yearbook
ðdt þ g t Þ
[40] in 2007. CEF is carbon emission factor provided by Inter-
where d and g are “depreciation rate” and “growth rate”, respectively. governmental Panel on Climate Change (IPCC) [41]. Following
We estimate coal, oil, gas and electricity (generated by wind, Chen [39], coal consumption mainly includes soft coal (about
water and nuclear), hydro power and nuclear power through 75–80%) and anthracite coal (about 20%) in China. IPCC [41] does
products between total energy consumption and their percentages not provide CEF of coal directly. We compute CEF of coal through
respectively. weighted mean of soft coal (80%) and anthracite coal (20%) [41].
When we estimated carbon dioxide emissions, we used coal, oil COF is carbon oxidization factor, and (44/12) reveals the molecular
and gas consumption. Here, electricity is generated by wind, hydro weights ratio between carbon dioxide and carbon (see Table 1).
and nuclear, and we treat it as the proxy variable of renewable Correlation analysis is used to explore the mutual relationship
energy. In other words, we only considered the CO2 emissions for between two different variables. It is computed as Eq. (4).
fuel combustion, rather than heat, cold consumption and electri-
CovðX; YÞ
city consumption. Our data is from China Compendium of Statis- CorrðX; YÞ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð4Þ
tics 1949–2008, and China Statistical Yearbook VarðXÞVarðYÞ
684 X. Long et al. / Renewable and Sustainable Energy Reviews 52 (2015) 680–688
According to the average value, coal consumption is much more lnGDP 1.446 5.366 3.49 1.546 5.18 3.49
than oil, electricity, gas, hydro power and nuclear power con- lnlabor 0.013 7.071 3.49 0.133 7.071 3.49
sumption in China. This means that China's economy depends lncapital 0.066 2.968 3.49 0.453 3.161 3.49
heavily on coal consumption. lnCO2 3 4.938 3.49 3.213 4.838 3.49
lncoal 3.031 4.943 3.49 3.286 4.849 3.49
According to the correlation analysis in Table 3 below, the
lnoil 2.337 5.113 3.49 2.269 4.965 3.49
correlation coefficient of CO2 emission is 0.9791, which indicates that lngas 2.459 4.496 3.49 2.431 4.359 3.49
some GDP of China come from environmental pollution. When we lnelectricity 3.368 8.234 3.49 3.374 8.246 3.49
compare the four energy consumptions, we find that electricity
consumption has the closest correlation with GDP, while Coal, oil Notes: (a) Both ADF and PP tests contain time trend. (b) ADF test employs t-
statistics. (c) PP test uses adjusting t-statistics.
and gas consumptions have fallen behind electricity consumption.
Augmented Dickey–Fuller (ADF) [42] and Phillips and Perron
(PP) [43] tests are often employed to test whether time series data α2i a 0 (for some i¼1,…,k); a causal relationship of X3 Y, when
is stationary or not. Consider the following three different models β1j ¼0 (for all j¼1,…,k) and α2i ¼0 ( for all i¼ 1,…k).
by ADF [42];
X
p 1
4. Discussion
Model1 : Δyt ¼ ηyt 1 þ δ βi yt i þ εt t ¼ 1; 2; …; T ð5Þ
i¼1
We also implemented unit root, cointegration, Granger causality,
X
p 1 static and dynamic regressions and impulse response function to
Model2 : Δyt ¼ ηyt 1 þ α þ δ βi yt i þ εt t ¼ 1; 2; …; T ð6Þ analyze the relationship among nonrenewable and renewable energy
i¼1 consumption and economic growth in China from 1952 to 2012.
pX
1
Model3 : Δyt ¼ ηyt 1 þ α þ λt þ δ βi yt i þ εt t ¼ 1; 2; …; T ð7Þ 4.1. Unit root
i¼1
Our data extends from 1952 to 2012, which covers 61 years.
In Eqs. (5)–(7), Δyt indicates the differentiation of y, yt 1 is one Because different years have unique disturbances, the long-period
order lagged variable, yt i is i order lagged variable, and ε is time series data tends to experience random drift. In other words,
random disturbance. Model 1 contains no time trend and constant, the data is non-stationary time series. Unit root tends to lead to
Model 2 includes constant, and Model 3 has both constant and the following demerits. The estimator of auto-regression is not
time trend. Null hypothesis is H0: δ ¼ 0, and alternative hypothesis approximate to normal distribution. Even the central limit theo-
is H1: δ o 0. A null hypothesis reveals there is at least one unit root. rem is not suitable to non-stationary time series. Because
Alternative hypothesis indicates there is no unit root. t-statistics value is not approximate to normal distribution, t-test
Compared with ADF [42], PP [43] just contains one order lagged and hypothesis test will be inefficient. Unit root tends to cause
variable of y. Consider the following three different models, spurious regression, in which the regression is not reliable any
Model1 : Δyt ¼ ηyt 1 þ εt t ¼ 1; 2; …; T ð8Þ more. Thus, it is necessary to implement unit root test from
different energy consumption, carbon emissions and economic
Model2 : Δyt ¼ ηyt 1 þ α þ εt t ¼ 1; 2; …; T ð9Þ growth from 1952 to 2012. In this paper, we use Augmented
Dickey–Fuller and Phillips and Perron methods to investigate
Model3 : Δyt ¼ ηyt 1 þ α þ λt þ εt t ¼ 1; 2; …; T ð10Þ whether energy consumption, carbon emissions and economic
growth in China are stationary or not from 1952 to 2012.
Model 1 has no constant and time trend. Model 2 contains In order to test for the presence of stationary, we implemented
constant. Model 3 includes both constant and time trend. Null unit root analysis upon the logarithm of GDP, labor, capital stock,
hypothesis is H0: δ ¼ 0, and alternative hypothesis is H1: δ o 0. CO2 emission, coal, oil, gas and electricity consumptions. The unit
Null hypothesis means that time series is unstable with at least root test contains ADF and PP, both of which employ level and first
one unit root, whereas alternative hypothesis indicates that time difference analysis (see Table 4). We chose critical value level to be
series is stationary without unit root. at 5%. All t-values of level analysis in ADF are more than critical
Following Granger et al. [44] [45], the causal relationship of value at 5% significance, which accepts null hypothesis and
two variables X and Y is indicated as follows. indicates the existence of unit root. However, all t-values of first
Xk Xk difference analysis in ADF are less than critical value at 5%
Yt ¼ α Y þ j ¼ 1 β1j X t j þ ε1t
i ¼ 1 1i t i
ð11Þ
significance level, which rejects null hypothesis and indicates the
absence of unit root.
Xk Xk
Xt ¼ i¼1
α2i Y t i þ j¼1
β2j X t j þ ε2t ð12Þ All adjusting t-values of level analysis in PP are over critical
value at 5% significance, which accepts null hypothesis and reveals
where Yt i and Xt j indicate the lagged variables of Y and X the existence of unit root. However, all adjusting t-values of first
respectively. There are four different relationships between X and difference analysis in PP are less than critical value at 5% sig-
Y. A causal relationship of “X ) Y”, when β1j a0 (for some j¼ 1,…, nificance, which rejects null hypothesis and indicates no unit root
k) and α2i ¼0 ( for all i ¼1,…k); a causal relationship of “Y ) X”, existence. In all, level values of the eight variables are non-
when β1j ¼ 0 (for all j¼1,…,k) and α2i a 0 (for some i¼1,…,k); a stationary at 5% significance, but their first differences are sta-
causal relationship of X3 Y, when β1j a 0 (for some j¼1,…,k) or tionary. Hence, GDP, labor, capital stock, CO2 emission, coal, oil, gas
X. Long et al. / Renewable and Sustainable Energy Reviews 52 (2015) 680–688 685
Table 5
Cointegration analysis.
Trace test
Max-Eigen test
Number of cointegrations Eigenvalue Max-Eigen statistic 5% Critical value Probability
Nonen 0.708568 71.51104 52.36261 0.0002
At most 1n 0.628494 57.43097 46.23142 0.0022
At most 2n 0.540075 45.04808 40.07757 0.0127
At most 3 0.399215 29.55205 33.87687 0.1507
At most 4 0.332153 23.41438 27.58434 0.1565
At most 5 0.255210 17.08984 21.13162 0.1678
At most 6 0.180602 11.55272 14.26460 0.1286
At most 7 0.018062 1.057185 3.841466 0.3039
n
Notes: denotes rejection of the hypothesis at the 0.05 significance level.
Table 7 According to regression analysis from 1952 to 2012, we can see that
Static and dynamic regression analysis. coal consumption has the largest impact upon economic growth and
carbon emissions in static and dynamic regressions among all energy
Time period: 1952–2012
consumption. It indicates that coal is the dominant energy consump-
Static regression Dynamic regression tion in China. On one hand, coal has promoted the rapid economic
growth through thermal power plants, iron and steel plants and
Variables lnGDP lnCO2 Variables ΔlnGDP Δln CO2 others; on the other hand, coal consumption also has serious
Constant 2.247 nn
1.917nnn
Constant 0.032 0.007nnn
environmental pollution.
(1.049) (0.198) (0.02) (0.002) Oil consumption has positive impacts on economic growth and
lnlabor 0.249n 0.087nnn Δlnlabor 0.364 0.018 carbon emissions in China. It seems that oil consumption has
(0.136) (0.026) (0.281) (0.033) promoted the development of chemical and transport, which also
lncapital 0.392nnn 0.014nnn Δlncapital 0.261nn 0.041nnn
causes the emissions of great environmental pollutants. Gas consump-
(0.025) (0.005) (0.121) (0.014)
lncoal 0.312nnn 0.914 nnn
Δlncoal 0.408nnn 0.896nnn tion has negative impact on economic growth from 1952 to 2012.
(0.054) (0.010) (0.083) (0.01) Maybe China's importation of more gas from other countries has
lnoil 0.157nnn 0.135 nnn
Δlnoil 0.125 0.120nnn increased financial budget of the country. Compared with coal and oil,
(0.056) (0.012) (0.091) (0.011) gas consumption has negative impact on carbon emission. Renewable
lngas 0.232nnn 0.014nn Δlngas 0.166nnn 0.002
(0.033) (0.006) (0.032) (0.004)
energy consumption has positive influence on economic growth and a
lnelectricity 0.249nnn 0.017 Δlnelectricity 0.086 0.001 negative impact on carbon emissions even though they are insignif-
(0.079) (0.015) (0.055) (0.007) icant. It is important to increase the consumption rate of renewable
Obs. 61 61 Obs. 60 60 energy and decrease the consumption rate of coal.
2
R 0.998 0.999 R2 0.67 0.998
Based on the regression analysis from 1994 to 2012, coal
Time period: 1994–2012 consumption also has the largest influence on economic growth
Static regression Dynamic regression and carbon emissions, which is consistent with the previous
Variables lngdp lnCO2 Variables ΔlnGDP Δln CO2
Constant 15.65nn 3.034nnn Constant 0.033 0.002
result. Oil consumption has positive effects upon economic growth
(6.546) (0.326) (0.032) (0.002) and carbon emissions. Gas consumption has positive impact upon
lnlabor 1.764nn 0.135nnn Δlnlabor 0.61 0.160n economic growth and carbon emissions, which is inconsistent
(0.615) (0.031) (1.391) (0.076) with the previous result. China's gas storage is not more than coal;
lncapital 0.342nnn 0.004 Δlncapital 0.215 0.003
hence, the gas consumption was not so much that its carbon
(0.049) (0.002) (0.122) (0.007)
lncoal 0.206nnn 0.793nnn Δlncoal 0.107 0.799nnn emissions could be serious. However, after 1994, gas consumption
(0.057) (0.003) (0.074) (0.005) has increased greatly, which also polluted the natural environment
lnoil 0.116 0.180nnn Δlnoil 0.005 0.175nnn more. The impacts of hydro power and nuclear power are not
(0.105) (0.005) (0.088) (0.005) significant for economic growth and carbon emissions. Compared
lngas 0.172nnn 0.0247nnn Δlngas 0.12 0.019nnn
(0.039) (0.002) (0.078) (0.004)
with coal consumption, the rates of hydro and nuclear power
lnhydro 0.027 0.002 Δlnhydro 0.015 0.001 consumption are still lower. It is necessary to decrease the coal
(0.034) (0.002) (0.027) (0.002) consumption rate, and increase the rates of hydro and nuclear
lnnuclear 0.021 0.001 Δlnnuclear 0.016 0.001 power consumption.
(0.015) (0.001) (0.016) (0.001)
Obs. 19 19 Obs. 18 18
R2 0.999 0.999 R2 0.539 0.999 4.5. Impulse response function analysis
Notes: (a) Standard errors are in parentheses. (b) Δ represents difference. (c)nnn, nn, n The response of carbon dioxide emissions will decline towards
indicate the significance levels of 1%, 5% and 10%, respectively. economic growth in the following 20 years (see Fig. 4). The response of
coal consumption will also decrease for economic growth, which
that oil storage is not rather abundant in China, which depends more reveals that China should reduce the rate of coal consumption to abate
on oil importation from oil-producing countries. Gas consumption has carbon emissions. The gas and hydropower consumption will have
two-way Granger causality relationship with economic growth. Eco- relatively higher impact on economic growth. It is significant to shift
nomic growth has bi-directional Granger causality links with electri- energy consumption rate, to increase the consumption rate of gas and
city. Economic growth has unidirectional causality with hydro power hydropower. Oil consumption also has positive influence on economic
consumption, which seems to be related to low rate of hydro power growth. Because oil shortage is obvious in China, we had better
use. It is significant to increase the hydro power consumption rate. increase oil import from other countries, more importantly, China
Nuclear power consumption has one-way causality with economic should increase oil exploration offshore.
growth. We had better use nuclear power safely, and increase the R&D Nuclear power also has positive impact on economic growth in
expenditure of nuclear power. Renewable energy consumption can China, even though its impact is still not stronger. Because fossil
decrease carbon emissions and other environmental pollutants, which energy has severe environmental pollution, it is important to
can also facilitate the sustainable economic development of China. expand the consumption rate of nuclear power. Coal consumption
and gas consumption have rather higher impact upon carbon
4.4. Static and dynamic regression analysis emissions in the further twenty years. It is necessary to control
and decrease fossil energy consumption, and rather increase
We also conduct static and dynamic regression analysis on the renewable energy consumption rates. The response of economic
determinants of economic growth and carbon emissions (see Table 7). growth has declined for carbon emission in the future. Compared
The purpose of dynamic regression analysis is to explore the changing with coal and gas consumption, the impact of oil consumption is
effects of nonrenewable and renewable energy consumption on not so strong for carbon emissions. The response of hydropower is
economic growth and carbon emissions. Because the data of hydro not so big for carbon emissions, compared with other fossil energy.
and nuclear power started from 1994, thus we have two different time Nuclear power has weak influence upon carbon emissions.
series data: one is from 1952 to 2012, which uses the summation data The State Council of China put forward “Energy Development
of hydro and nuclear power; the other is from 1994 to 2012, which Strategy Action Initiative (2014–2015)”. By this strategy, China is
employs the data of hydro and nuclear power consumptions. determined to reduce coal consumption rate of total energy
X. Long et al. / Renewable and Sustainable Energy Reviews 52 (2015) 680–688 687
.08 .08
.05
.04 .04
.00
.00 .00
-.05
-.04 -.04
.06
.08
.2
.04
.04
.02 .0
.00
.00
-.2
-.04
-.02
.2
.05 .04
.1
.00 .0 .00
-.1
-.05 -.04
-.2
consumption within 62% by 2020; increase natural gas consumption guarantee the installed capacity of solar power electricity reaching
rate of total energy consumption over 10% by 2020; increase the 0.1 billion KW. We had better develop geothermal energy, biomass
proportion of non-fossil energy of primary energy consumption to energy and marine energy positively [47]
about 15% by 2020. It is significant to develop hydroelectricity power,
with the aim to increase the installed capacity of wind electricity to
0.35 billion KW by 2020. China also aims to generate 0.2 billion KW 5. Conclusion and policy implications
wind electricity power by 2020, and invest wind power base and
related outflow project in Jiuquan, west of inner Mogolia, east of Inner This paper mainly investigates the relationship among renew-
Mogolia, North of Hebei, Jilin, Heilongjiang, Shandong, Hami and able energy, carbon emissions and economic growth in China from
Jiangsu. It is also important to develop solar electricity power, to 1952 to 2012. More importantly, we also compare the different
688 X. Long et al. / Renewable and Sustainable Energy Reviews 52 (2015) 680–688
impacts of renewable and non-renewable energy consumption on [14] Chiou-Wei SZ, Chen CF, Zhu Z. Economic growth and energy consumption
carbon emissions and economic growth in China. revisited-Evidence from linear and nonlinear Granger causality. Energy Econ
2008;30:3063–76.
Through unit root, we find energy consumption, carbon emis- [15] Lin B, Mohamed M. Renewable energy consumption-economic growth nexus
sions and economic growth are not stationary. They have long-run for China. Renew Sustain Energy Rev 2014;40:111–7.
equilibrium relationship through cointegration analysis. Through [16] Ma H, Oxley L, Gibson J, Li W. A survey of China's renewable energy economy.
Renew Sustain Energy Rev 2010;14:438–45.
Granger causality analysis, we find that GDP has bi-directional [17] Sun J, Peng F. Empirical analysis on energy consumption and economic growth
relationship with CO2 emission, coal, gas, and electricity consump- in Yangtze River Delta-an estimation based on co-integration. Energy Proc
tion. Through static and dynamic regression, we find coal has 2011;5:520–4.
[18] Zhang YJ. Interpreting the dynamic nexus between energy consumption and
dominant impact upon economic growth and carbon emissions;
economic growth: empirical evidence from Russia. Energy Policy 2011;39:2265–72.
oil consumption has positive impact on carbon emissions. Through [19] Zhang Z, Ren X. Causal relationship between energy consumption and
impulse response function, we find the impacts of hydro and economic growth. Energy Proc 2011;5:2065–71.
nuclear power have positive impact on economic growth. [20] Ying M, Bin W, Long X. An empirical analysis for the dynamic relationship
between China's energy consumption and economic growth-Based on co-
We attempt to put forward some policy implications according integration analysis and state-space model. Energy Proc 2011;5:2252–6.
to the empirical analysis. Through static and dynamic regression [21] Lin B, Presley Jr. KW. Energy consumption and economic growth in South
analysis, we can see the impacts of hydro and nuclear power is not Africa reexamined: a nonparametric testing approach. Renew Sustain Energy
Rev 2014;40:840–50.
as strong as coal. It is important to change energy consumption [22] Sadorsky P. Renewable energy consumption and income in emerging econo-
structure. We should decrease the coal consumption rate to reduce mies. Energy Policy 2009;37:4021–8.
carbon emissions, and enhance the consumption rate of natural [23] Chang J, Leung DYC, Wu CZ, Yuan ZH. A review on the energy production,
consumption, and prospect of renewable energy in China. Renew Sustain
gas. China should explore more value-added products of non- Energy Rev 2003;7:453–68.
renewable energy. Nuclear power has unidirectional causality [24] Pao HT, Fu HC. Renewable energy, non-renewable energy and economic
upon economic growth. It means nuclear power is significant for growth in Brazil. Renew Sustain Energy Rev 2013;25:381–92.
[25] Hu H, Zhang XH, Lin LL. The interactions between China's economic growth,
long time economic development. It is important to develop
energy production and consumption and the related air emissions during
nuclear power safely. Economic growth has unidirectional caus- 2000–2011. Ecol Indic 2014;46:38–51.
ality on hydro power. The development of hydro power can reduce [26] Yasar A, Ali A, Tabinda AB, Tahir A. Waste to energy analysis of shakarganj
sugar mills; biogas production from the spent wash for electricity generation.
carbon emissions. With environmental deteriorating, it is impera-
Renew Sustain Energy Rev 2015;43:126–32.
tive to explore and utilize more renewable energy such as wind, [27] Ghosh S, Kanjilal K. Long-term equilibrium relationship between urbanization,
hydro, biofuel, solar and other, to promote the sustainable devel- energy consumption and economic activity: empirical evidence from India.
opment of economy in China. Energy 2014;66:324–31.
[28] Ozturk I, Bilgili F. Economic growth and biomass consumption nexus: dynamic
panel analysis for Sub-Sahara African countries. Appl Energy 2014;137:110–6.
[29] Wong SL, Chia WM, Chang Y. Energy consumption and energy R&D in OECD:
Acknowledgment perspectives from oil prices and economic growth. Energy policy
2013;62:1581–90.
[30] Saati SL, Hassan MS, Mahmood H, Shahbaz M. Coal consumption: an alternate
We appreciate the financial support provided by Grant from energy resource to fuel economic growth in Pakistan. Econ Model
the China Postdoctoral Science Foundation (No. 2014M551504), 2014;36:282–7.
the Joint Research Grant of NSFC-NRF Scientific Cooperation [31] Inglesi-Lotz R. The impact of renewable energy consumption to economic
growth: a panel data application. Energy Econ 2015. http://dx.doi.org/10.1016/
Program (No. 71411170250), National Research Foundation of
j.eneco.2015.01.003.
Korea Grant funded by the South Korean Government (NRF- [32] Omri A, Ben Mabrouk N, Sassi-Tmar A. Modeling the causal linkages between
2014K2A2A2000766), National Research Foundation of Korea nuclear energy, renewable energy and economic growth in developed and
developing countries. Renew Sustain Energy Rev 2015;42:1012–22.
Grant funded by the South Korean Government (NRF-
[33] Zeng M, Duan J, Wang L, Zhang Y, Xue S. Orderly grid connection of renewable
2014S1A2A2027622), National Natural Science Foundation of energy generation in China: management mode, existing problems and
China (Nos. 71471076, 71473107), and Natural Science Founda- solutions. Renew Sustain Energy Rev 2015;41:14–28.
tion of Jiangsu Higher Education Institutions of China (No. [34] Shahbaz M, Loganathan N, Zeshan M, Zaman K. Does renewable energy
consumption add in economic growth? An application of auto-regressive
15KJD610002). distributed lag model in Pakistan Renew Sustain Energy Rev 2015;44:576–85.
[35] Zhang J, Wu GY, Zhang JP. The estimation of China's provincial capital stock:
References 1952–2000. (in Chinese). Econ Res J 2004;10:35–44.
[36] Wang X, Fan G. The sustainability of China's economic growth. (in Chinese).
Beijing: Economic Science Press; 2000.
[1] Environmental Performance Index. 〈http://www.epi.yale.edu/〉; 2013. [37] Chow G, Lin A. Accounting for economic growth in Taiwan and Mainland
[2] Li KQ 〈http://news.sina.com.cn/c/2015-03-05/105331571230.shtml?qq-pf-to=pcqq. China: a comparative analysis. J Comp Econ 2002;30:507–30.
c2c〉; 2015. [38] Kim SW. Total factor productivity and regional economic development
[3] China compendium of statistics 1949–2008. China Statistics Press; 2010. disparity in China, in: regional economic development disparity and coordi-
[4] China statistical yearbook. China Statistics Press; 2014. nate development in China. (in Chinese). Shanghai People's Press; 2007. p.
[5] Zeng M, Li C, Zhou L. Progress and prospective on the policy system of 67–104.
renewable energy in China. Renew Sustain Energy Rev 2013;20:36–44. [39] Chen SY. The study of energy-saving and emission-abatement, structural
[6] China Wind Energy Association (CWEA). Statistics of China's wind power adjusting and their impact on the transformation of industrial development
installed capacity. Beijing: CWEA; 2004–2011. model in China. Peking University Press; 2011.
[7] China Electricity Council (CEC). Research report on China's new energy [40] China energy statistics yearbook, China Statistics Press; 2007.
generation development; 2012. [41] IPCC. IPCC guidelines for national greenhouse gas inventories. www.ipcc-ggip.
[8] Ozturk I. A literature survey on energy-growth nexus. Energy Policy iges. or.jp/public/2006gl/pdf/2_Volume 2/V2 _2_Ch2_Stationary_Combustion.
2010;38:340–9. pdf; 2006.
[9] Omri A. An international literature survey on energy-economic growth nexus: [42] Dickey DA, Fuller WA. Distribution of the estimators for autoregressive time
evidence from country-specific studies. Renew Sustain Energy Rev series with a unit root. J Am Stat Assoc 1979;74(366):427–31.
2014;38:951–9. [43] Phillips PCB, Perron P. Testing for a unit root in time series regression.
[10] Tigcu CT, Ozturk I, Aslan A. Renewable and non-renewable energy consump- Biometrika 1988;75(2):335–46.
tion and economic growth relationship revisited: evidence from G7 countries. [44] Granger CWJ. Testing for causality: a personal viewpoint. J Econ Dyn Control
Energy Econ 2012;34:1942–50. 1980;2:329–52.
[11] Sebri M. Use renewables to be cleaner: meta-analysis of the renewable energy [45] Granger CWJ. Some recent developments in the concept of causality. J Econ
consumption–economic growth nexus. Renew Sustain Energy Rev 2015;42:657–65. 1988;39:199–211.
[12] Kraft J, Kraft A. On the relationship between energy and GNP. J Energy Dev [46] Johansen S. Statistical analysis of cointegration vectors. J Econ Dyn Control
1978;3:401–3. 1988;12:231–54.
[13] Zhang XP, Cheng XM. Energy consumption, carbon emission and economic [47] Energy Development Strategy Action Initiative (2014–2015). 〈www.gov.cn/
growth in China. Ecol Econ 2009;68:2706–12. zhengce/content/2014-11/19/content_9222.htm〉; 2014.