Long Run Relationship among GDP, GDP per capita and ...
Tahir Mahmood (UEF) Trend and Growth Dependence of Energy Intensity in European Economies 1980 - 2006 Abstract The paper explores the European energy intensity trends both for individual countries and for the panel of rich and poor countries. A new approach based on de-trended energy intensity is introduced into growth analysis. The dependence of de-trended energy intensity on GDP growth and its main contributing sectors (i.e. industry, services and agriculture) is studied in details. The role of population growth is also noticed when the analysis is cast in GDP per capita format. The results are compared to the experience of developing countries where the population growth has the major impact on energy consumption. The impact of main economic indicators analyzed (i.e. GDP growth, growth in sector outputs, GDP per capita growth, and population growth) on de-trended energy intensity are large and energy saving for European countries compared equally large but energy using impacts in poor countries. Therefore, we found that the cost of converting energy into GDP is low in the European countries and high in poor countries. Keywords: Energy intensity, economic growth, sector outputs, panel data.
I. Introduction Energy is the main input for the economic growth. Sustainable development in economic growth underlines the importance of sustained growth of energy productivity. This is manifested by the declining energy intensity for industrial countries. Energy intensity is a measure of the energy efficiency of a nation’s economy. It is calculated as units of energy per unit of GDP. Higher energy intensity indicates a high price or cost of converting energy into GDP. Low energy intensity indicates a low price or cost of converting energy into GDP. In addition, national energy intensities change over time. Countries with higher GDP tend to have energy intensities that improve, helping to insulate them from some of the erosive effects of declining energy supplies. Countries at the bottom of the GDP scale tend to require more and more energy to produce the GDP. For the support of this argument a study by Huang (2008) found that in poor countries, 1% increase in economic growth requires more than 1% increase in energy use. On other hand rich Europen economies require less and less energy input in relative terms. For the 1
support of this argument a study by EEA (2008) found that over the period 1990-2002, European GDP grows at an annual average rate of 2.2 % and total energy consumption at annual average rate of 0.5%. As a result, total energy intensity in the EU fall as an average rate of 1.7% (see EEA 2008). In general the picture resembles closely to EKC phenomena. EKC is an “inverted U” relationship between the level of economic development and the pollution level. In poor countries, as economies improve, pollution gradually increases and as the industrial potential expands. Sooner or later, the pollution problem becomes a major concern which requires immediate actions. In general, as the income increases beyond some threshold, there is a tendency towards producing low pollution products. (See Haung et al. 2008, Grossman and Krueger 1995, Dinda 2004 and Dinda et al. 2000).
With current policies, European energy demand will grow yearly at 1.2 percent to 2020. Europe represents 17 percent of global energy consumption, less than the 22 percent of the United States, the world’s largest energy consumer, but more than China, which is at 14 percent. Yet energy intensity (the energy used to generate real GDP in PP-corrected US-dollars) varies significantly across the Europe. Northwestern Europe’s economies run at a relatively low level of energy intensity of 7,200 BTUs per GDP. The energy intensity of Northeastern Europe is almost twice as high, slightly above the global average of 12,600 BTUs per GDP. Meanwhile, Southern Europe falls between the two with an energy intensity of 8,300 BTUs per GDP, similar to that of the United States. Note that if Europe captures the full potential of increase in energy productivity, the continent could abate energy demand equivalent to 8 million barrels of oil per day, or double the final electricity consumption of all EU-25 countries today. Solely using the available techneologies offers positive returns (see EEA 2008).
Decrease in energy intensity in developed countries is a largely accepted hypothesis in energy literature. However the relationship between energy intensity and economic growth is not so clear. Typically it is assumed that for the developed countries a negative relationship between energy intensity and economic growth is valid. Thus the “energy2
saving” high GDP-level effect is extended to be valid also for economic growth. Main argument is the fact that service economy is less energy demanding than preceding economic epochs, and energy saving economy is only possible with high technology (Stern 2003). However if the technology and energy composition effects (i.e. declining trend of energy/GDP ratio) are removed, the growth effects may still be energy saving since non-trend effects are still affected by technology spill-over effects.
Interestingly, these hypotheses have not been formally tested. Today’s economic literature has focused on the issue of convergence of energy intensities. Some papers examined the issue by comparing energy intensities of developed and developing countries (Alcantra and Duro, 2004), and other conducted a convergence analysis of transitory countries (Markandya et al. 2006, Ezcurra 2007, Le Pen and Sevi 2007).
In this paper we explore the European energy intensity trends both for individual countries and for the panel of countries. The growth dependence of energy intensity and its main sector (i.e. industry, services and agriculture) outputs are studied in details. The role of population growth is also noticed when the analysis is cast in GDP per capita format. The results are compared to experience of some developing countries where the population growth has the major impact on energy consumption.
We found negative trend in energy intensity for European countries and positive trend for the poor countries. Therefore, we focus on the de-trended intensity response to economic growth and its main sectors. We de-trended the energy intensity series and found the negative impact of economic growth on de-trended energy intensity with European countries. However, we still found positive relationship between de-trend energy intensity and economic growth for the panel of poor countries. These findings support the hypothesis that energy intensity falls as a country’s economic growth takes place after a certain level of economic development. The high income group (i.e. rich countries) has low energy intensities because energy consumption in these countries grows more slowly than GDP and they have undertaken energy conservation polices. Hence, the role oil price shocks on the falling energy intestines of European economies are questioned. 3
Where as, in poor countries growth requires intensive energy use, and the business cycles and the price shocks affect energy use positively. Hence the cost of converting energy into GDP is high in poor countries. We also found that the services sector is negatively related with energy intensities in European countries but positive relationship was found in poor countries. A positive relationship was also found between population growth and de-trended energy intensity for both groups of countries.
The study is organized as follows. Section 2 presents data and variables. Section 3 reviews the theory of economic growth and energy intensity in Europe. Section 4 presents model and estimation procedures, and 5 gives the results and 6 present the conclusion.
2. Data and Variables Energy Intensity: The data on energy intensity is from the US Energy Information Administration (EIA Independent Statistic and Analysis). Data for this variable is calculated by dividing the data on total primary energy consumption in quadrillion British thermal by the gross domestic product using purchasing power parities in billions of 2000 U.S. dollars for each available country and year from Global Insight. Real GDP Per Capita: The data on Real GDP per Capita, Real GDP and Population is from Center for International Comparison at the University of Pennsylvania (Penn Tables). The definition of variables is following: Real Gross Domestic Product Per Capita and components for 1996 are obtained from an aggregation using price parties and domestic currency expenditures for consumption, investment and government of August 2001 vintage. Real Gross Domestic Product is calculated by multiplying the data on Real GDP Per Capita for each country by Population. Population: Population variable is from World Bank Development Indicators 2001, and United Nation Development Center sources. Industry Output: Industry corresponds to ISIC divisions 10-45 and includes manufacturing (ISIC divisions 15-37). It comprises value added in mining, manufacturing (also reported as a separate subgroup), construction, electricity, water, and gas. Value added is the net output of a sector after adding up all outputs and subtracting intermediate inputs. It is calculated without making deductions for depreciation of fabricated assets or 4
depletion and degradation of natural resources. The origin of value added is determined by the International Standard Industrial Classification (ISIC), revision 3. Data are in current U.S. dollars. Source: World Bank national accounts data, and OECD National Accounts data files. Service Output: Services correspond to ISIC divisions 50-99. They include value added in wholesale and retail trade (including hotels and restaurants), transport, and government, financial, professional, and personal services such as education, health care, and real estate services. Also included are imputed bank service charges, import duties, and any statistical discrepancies noted by national compilers as well as discrepancies arising from rescaling. Value added is the net output of a sector after adding up all outputs and subtracting intermediate inputs. It is calculated without making deductions for depreciation of fabricated assets or depletion and degradation of natural resources. The industrial origin of value added is determined by the International Standard Industrial Classification (ISIC), revision 3. Data are in current U.S. dollars. Source: World Bank national accounts data, and OECD National Accounts data files. Agriculture Output: Agriculture corresponds to ISIC divisions 1-5 and includes forestry, hunting, and fishing, as well as cultivation of crops and livestock production. Value added is the net output of a sector after adding up all outputs and subtracting intermediate inputs. It is calculated without making deductions for depreciation of fabricated assets or depletion and degradation of natural resources. The origin of value added is determined by the International Standard Industrial Classification (ISIC), revision 3. Data are in current U.S. dollars. Source: World Bank national accounts data, and OECD National Accounts data files. Time Period: 1980-2006, 27 years. Developed Countries: Austria, Belgium, Cyprus, Denmark, Finland, France, Greece, Iceland, Ireland, Italy, Luxembourg, Malta, Netherlands, Norway, Poland, Portugal, Spain, Sweden, and United Kingdom. Poor Developing Countries: We selected 8 SAARC (i.e. South Asian Association for Regional Cooperation) countries plus their two important bordering countries (Indonesia and Malaysia). The countries are following, Afghanistan, Bangladesh, Bhutan, India, Indonesia, Maldives, Malaysia, Nepal, Pakistan, and Srilanka. 5
3. Economic Growth and Energy Intensity in European Economies Energy intensity can be reduced by improving efficiency in the use of energy and by improving economic activities. Basically the economies with high GDP level manage to do this better than poor countries. This is due the (relative) less energy consuming service economy compared to agricultural or industry economy and efficiency of energy production. The level of economy’s technology, its energy base and conversion are key elements in the evolution of energy/GDP –ratio. Figure 1: Time cross plot of GDP –levels and Energy/GDP -ratio. BEL
8,400
11,000
8,000
FIN 14,000
PPP
GBR
8,500
13,000 8,000
0 00 ,0 0
0
PPP
FRA
6,400
5,000 40,000,000
6,200 120,000,000
200,000,000
0
500,000,000
PPP
20 ,0
0
0
00 ,0 0
PPP
00 ,0 0
6,800
6,600
15 ,0
300,000,000
8,000
7,000
6,000
5, 00
100,000,000
7,200
80 ,0 12 00, 0, 00 0 0 16 00, 0, 00 0 0 20 00, 0, 00 00 0 24 0, 0, 00 0 0 28 00, 0, 00 0 0 32 00, 0, 00 0 0 36 00, 0, 00 00 0 0, 00 0
0
9,000
7,000 6,800
10 ,0
6,000
7,600
0, 00 0
6,500
10,000
ESP 7,200
8,000
EI
12,000
DNK 9,000
EI
7,000
8,800
EI
EI
7,500
CYP
13,000
EI
AUT 8,000
1,500,000,000
PPP
GRC
IRL
10,000
7,000
9,000
9,000
6,500
8,000
8,000
6,000
7,000 EI
7,500
EI
11,000
EI
EI
EI
12,000
7,000
5,500
6,000
6,000
5,000
5,000
5,000 400,000,000
4,500
10,000 7,000 9,000 8,000 40,000,000
120,000,000
6,500 500,000,000
200,000,000
PPP
1,500,000,000
1,200,000,000
PPP
ISL
ITA
18,000
2,000,000,000
4,000 0
PPP
0 50,000,000
NLD
10,000
12,000
16,000 6,400
11,000
9,000
16,000
150,000,000 PPP
MLT
18,000
17,000
300,000,000
PPP
LUX
6,800
100,000,000
14,000 8,000
EI
12,000
EI
6,000
EI
EI
EI
10,000 15,000
9,000 14,000
10,000 5,600
7,000
13,000
8,000
8,000
0 00 ,0 0
0
7,000 0
4,000,000 8,000,000 12,000,000 PPP
40 ,0
00 ,0 0
0 00 ,0 0
POL
30 ,0
NOR
2,000,000,000
20 ,0
PPP
0
1,200,000,000
PPP
6,000
0
5,000,000 10,000,000
6,000 00 ,0 0
0
5,200 400,000,000
10 ,0
12,000
0
200,000,000
600,000,000
PPP
PPP
15,000
PRT
16,000
14,000
14,000
13,000
SWE
6,500
14,000
6,000
13,000
5,500
12,000 EI
12,000
EI
EI
EI
12,000 5,000
11,000
10,000 11,000 8,000
10,000 9,000
6,000 0
100,000,000 PPP
300,000,000
4,500
10,000
4,000
9,000
3,500 0
200,000,000
600,000,000
8,000 0
100,000,000
PPP
PPP
300,000,000
0
100,000,000
300,000,000
PPP
Average energy intensity is decreasing approximately for all 20 European countries except Greece, Iceland, and Portugal (see Fig 1). In Iceland this is caused by an increase 6
of energy intensive industries, like the aluminum industry. On other hand in Greece and Portugal, this is most likely caused by decline in economic growth (see Table 1). In general economic growth seems to have an important role in lowering energy intensity. Between 1990 and 2006, economic growth in Europe required increasing energy inputs but the growth rate was less than the GDP growth rate. Total energy consumption increased until 2004 and stabilized all the way through 2006. Over the period 1990-2006, average GDP grew and average energy intensity in the EU decrease (see Table 1).
Table 1: Economic Growth and Energy Intensity in Europe 1980 - 2006 Economic Growth
Energy Intensity
Country
1981- 90
1991-2000
2000-2006
1981-90
1991-2000
2000-2006
Austria
0.064
0.039
0.044
8.856
8.795
8.761
Belgium
0.062
0.037
0.045
9.367
9.274
9.160
Cyprus
0.101
0.059
0.059
8.894
8.963
8.893
Denmark
0.064
0.044
0.050
8.854
8.790
8.616
Finland
0.070
0.034
0.049
9.334
9.337
9.165
France
0.064
0.036
0.043
8.913
8.901
8.823
Greece
0.046
0.042
0.075
8.629
8.780
8.70
Iceland
0.064
0.045
0.073
9.479
9.605
9.730
Ireland
0.074
0.083
0.073
8.926
8.766
8.492
Italy
0.067
0.032
0.036
8.726
8.663
8.648
Luxembourg
0.087
0.066
0.077
9.387
9.022
8.860
Malta
0.082
0.069
0.041
9.079
9.040
8.883
Netherlands
0.061
0.048
0.040
9.231
9.108
9.011
Norway
0.052
0.063
0.068
9.528
9.355
9.214
Poland
0.041
0.055
0.063
-
9.430
9.029
Portugal
0.075
0.050
0.034
8.520
8.634
8.680
Spain
0.072
0.044
0.065
8.792
8.758
8.782
Sweden
0.062
0.034
0.050
9.430
9.340
9.109
United Kingdom
0.067
0.043
0.053
8.980
8.837
8.631
High average energy intensity found in era 1981-90 may still belong to the era of adaption to the high oil prices of the 1970’s. Average energy intensity and average economic growth decreased approximately for all 20 European countries in period 19902000. The period 2000-2006, should be characterized as well-managed era. Here, we observe a negative relationship between economic growth and energy intensity.
7
The negative relationship between energy intensities and economic growth of the panel of European countries indicates that the cost of converting energy into GDP is decreasing in the European nations (see Fig 2). The cost of converting energy into GDP is increasing in poor countries because the relationship between energy intensities and economic growth in the poor countries is positive (see, Appendix Fig. A3, and see also Hannesson 2002).
Figure 2: Time cross plot of Energy/GDP -ratio and dlnGDP 18,000 16,000 14,000
EI
12,000 10,000 8,000 6,000 4,000 2,000 -.10
-.05
.00
.05
.10
.15
.20
GDP
4. Model and Estimation Procedure Many results confirm that the following energy-output relationship or linear regression is valid for industrial countries: ln Et ln Yt t Where 0 1.
By adding on both side ln Yt gives the energy intensity form ln Et ln Yt ln Yt ln Yt t
8
ln( Et / Yt ) ( 1) ln Yt t .
Next we take a difference on both side of the last equation resulting in ln( Et / Yt ) ln( Et 1 / Yt 1 ) ( 1) ln Yt t .
Now if assume that past energy intensity can be approximated with an elementary trend function like ( Et 1 / Yt 1 ) exp(c1 c 2Trt 1 ) .
Thus we estimate model ln( Et 1 / Yt 1 ) c1 c 2Trt 1 t 1
to data and use the fit
ˆ 2t to ln( Et 1 / Yt 1) F cˆ 1 c Tr 1 de-trend the current energy
intensity. That is
ln( Et / Yt ) (cˆ1 cˆ2Trt ) ln( Et / Yt ) DT ( 1) ln Yt t . We expect the sign of parameter d ( 1) to be negative in following regression model
1)
ln( Et / Yt ) DT d ln Yt t .
The parameter d measures the short run dependency of energy-intensity on economic growth but it gives also an indirect estimate of long run energy output relationship, i.e.
d 1. Thus the small negative value d implies low energy intensity and large negative value of d implies high energy intensity.
Next we propose three steps of analysis in order to analyze the impact of economic growth on de-trended energy intensity.
9
Step. 1. Analyze the trends in energy intensity series Take the natural log of energy intensity (E/Y) series, and draw the graphs for each country in the panel. Figure 3. Trends in Energy Intensity in European Countries LNEI AUT
BEL
9.00
CYP
9.5
8.95
DNK
9.08
9.1
9.04
9.0
9.00
8.9
8.96
8.8
8.92
8.7
8.88
8.6
ESP 8.88
9.4
8.84
8.90 8.85
9.3
8.80
8.80
9.2
8.76
8.75 8.70 1980
1985
1990
1995
2000
2005
9.1 1980
1985
1990
FIN
1995
2000
2005
8.84 1980
1985
FRA
9.5
1990
1995
2000
2005
8.5 1980
1985
GBR
9.05
1995
2000
2005
8.72 1980
1985
GER
9.2
1995
2000
2005
2000
2005
2000
2005
2000
2005
8.9
8.95
9.0
1990
GRC
9.00
9.00 9.4
1990
8.8
8.95 8.90 9.3
8.90
8.8
8.7 8.85
8.85 9.2
8.6
9.1 1980
1985
1990
1995
2000
2005
8.75 1980
1985
1990
IRL
1995
2000
2005
8.4 1980
1985
1990
ISL
9.2
1995
2000
2005
8.75 1980
1985
1990
ITA
9.8
1995
2000
2005
8.5 1980
1985
1990
LUX
8.80
9.7
9.0
8.6
8.80
8.80
1995
MLT
9.8
9.2
9.6
9.1
8.75 9.4
9.6
9.0
8.8
8.70
9.2
9.5
8.9 9.0
8.6
8.65
9.4
8.4 1980
1985
1990
1995
2000
2005
9.3 1980
8.8
8.8
1985
NLD
1990
1995
2000
2005
8.60 1980
1985
1990
NOR
1995
2000
2005
8.6 1980
1985
1990
POL
1995
2000
2005
8.7 1980
1985
PRT
9.4
9.6
9.8
8.8
9.3
9.5
9.6
8.7
9.2
9.4
9.4
9.1
9.3
9.2
1990
1995
SWE 9.6
9.4 8.6 8.5
9.2
8.4 9.0
9.2
8.9 1980
1985
1990
1995
2000
2005
9.1 1980
9.0
9.0
1985
1990
1995
2000
2005
8.8 1980
8.3
1985
1990
1995
2000
2005
8.2 1980
1985
1990
1995
2000
2005
8.8 1980
1985
1990
1995
Fig, 3. shows clear downward trends in energy intensity for most of series. The data for some countries is very noisy with large swings over the years. Some do not show clear downward trends (Greece, Iceland, and Portugal). In all other countries energy intensity falls. The existence of a trend in energy intensity has significant implications for feasible target energy consumption (see Le Pen and Sevi 2007).
10
Note that from macroeconomic point of view the declining energy share started in the late 1980’s can be caused by oil shocks (see Blanchard and Gal 2008, Killian 2008a,b and 2009). Thus, energy intensity decline is due to these business cycle effects. However we argue that declining trend is due to the technology and better energy conversion since if the business cycle view is valid there should be a positive correlation between de-trended energy intensity and GDP growth rate. In general oil price shocks may have long run energy intensity effects but typically business cycle theories imply short run non-trend effects to be pro-cyclical. Our hypothesis is contrary to this as we argue that GDP growth effects are still energy saving, i.e. they have negative effects on the de-trended energy intensity.
Step. 2. De-trending energy intensity ln( Et / Yt ) DT series for each sample country. 2006
E Suppose a sample it of T observations on energy intensity series for each cross Yit t 1980
section country i = 1, 2, …,N.
Consider the model where ln( Eit / Yit ) is a sum of deterministic and stochastic components:
2)
3)
ln( Eit / Yit ) ci1 ci 2Trt it
it i it 1 it .
c2 is the slope of the deterministic trend measure, t is an AR(1) random variable whose with first-order autocorrelation coefficient , t is IID random variable. An important issue is whether t is stationary or not. If the series is I(0) this means that stochastic shocks have only a transitory effect on intensity series, and then t series can be used for further empirical analysis. But when | | 1 , t is a random walk type series,
11
i.e. I (1) series, and the regression model with GDP growth rate is unbalanced since ln Yt is expected to be stationary in regression model 1).
Table 2 Results from the Estimation of Model 1 (HAC t-values in parenthesis) _______________________________________________ c1 c2 Country Austria 8.92 [488.88]*** -0.006[-6.24]** Belgium 9.43[322.0]*** -0.011[-13.31]** -0.0004[-0.28] Cyprus 8.93[334.59]*** Denmark 8.96[303.87]*** -0.013[-7.44]** Finland 9.92[310.38]*** -0.009[-5.509]** France 8.96[624.91]*** -0.005[-6.64]** Greece 8.59[288.09]*** 0.007[3.83]* Iceland 9.40[417.55]*** 0.013[8.032]** 9.07[320.32] *** -0.022[-11.25]*** Ireland Italy
8.74[614.27]***
-0.004[-5.07]**
Luxembourg
9.54[290.41]***
-0.032[-10.20]***
Malta
9.12[222.82]***
-0.007[-3.35]
Netherlands
9.31[613.96]***
Norway
9.60[353.26]***
Poland
10.15[184.27]***
Portugal
8.45[180.61]***
Spain
8.79[464.07]***
-0.001[-1.03]
Sweden
9.54[331.75]***
-0.016[-7.58]**
United Kingdom
9.10[715.10]***
-0.0192[-22.74]***
Panel Regression
9.12[933.98]***
-0.008[-13.78]***
* -0.012[-11.65]*** -0.015[-11.23]*** -0.047[-15.45]*** 0.011[4.05]**
(1) *** significant at 1 % level of significance (2) ** significant at 5 % level of significance (3) * significant at 10% level of significance
Now, we eliminated the estimated trend functions from the energy intensity series, i.e. we obtained following series
ln( Eit / Yit ) cˆi1 cˆ2Trt 1 ˆit ln( Eit / Yit ) DT
Graphs of de-trended energy intensity ln( Eit / Yit ) DT series are shown in Fig. 4.
12
Figure 4. De-trended Energy Intensities in European Countries DEN AUT
BEL
.08
.08
.04
.04
.00
CYP
DNK
.15
.15
.10
.10
.05
.05
.00
.00
-.05
-.05
ESP .08
.04
.00
-.04
.00
-.04
-.08 1980
1985
1990
1995
2000
2005
-.08 1980
1985
1990
FIN
1995
2000
2005
-.10 1980
1985
FRA
.20
1990
1995
2000
2005
-.10 1980
-.04
1985
GBR
.08
.08
.04
.04
1990
1995
2000
2005
-.08 1980
1985
1990
GRC
1995
2000
2005
2000
2005
2000
2005
IRL
.12
.10
.08
.15
.05
.04
.10
.00 .05
.00
.00
.00 -.04
.00 -.04
-.08
-.04
-.05
-.05
-.12
-.10 1980
1985
1990
1995
2000
2005
-.08 1980
1985
1990
ISL
1995
2000
2005
-.08 1980
1985
1990
ITA
.10
2000
2005
-.16 1980
1985
1990
LUX
.06
1995
2000
2005
-.10 1980
1985
1990
MLT
.15
1995
NLD
.2
.10
.04
.05
1995
.08
.1
.04
.05 .02
.0
.00
.00
.00
.00
-.1 -.05
-.05
-.02
-.10 1980
1985
1990
1995
2000
2005
-.04 1980
1985
1990
NOR
1995
2000
2005
-.15 1980
-.04
-.2
-.10
1985
1990
POL
1995
2000
2005
-.3 1980
1985
PRT
.10
.10
.1
.05
.05
.0
.00
.00
-.1
1990
1995
2000
2005
2000
2005
-.08 1980
1985
1990
1995
SWE .10 .05 .00 -.05
-.05
-.05
-.2
-.10 1980
-.10 1980
-.3 1980
1985
1990
1995
2000
2005
1985
1990
1995
2000
2005
-.10
1985
1990
1995
2000
2005
-.15 1980
1985
1990
1995
Step. 3. Unit Root Tests on De-trended Energy Intensity and Economic Growth In order to find unbiased relationship between energy intensity and economic growth, we apply following tests. a. Apply unit root test on de-trend energy intensity ln( Et / Yt ) DT series. b. Apply unit root test on real GDP growth (ln Yt ) series. The analysis of non-stationary panels is similar to analysis of non-stationary time series of the 1980s. However non-stationary panels include some unique issue such as crosssectional heterogeneity and correlation. We use LLC test (see Levin, Lin, and Chu 2002), Fisher-ADF and Fisher –PP test (see Maddala & Wu 1999). We test unit roots in panel of 13
series, i.e. energy intensity ln( Et / Yt ) DT series and real GDP growth (ln Yt ) series for 19 cross sections with 27 time periods. Table 3. Summary of five tests Test
Null
Alternative
LLC Fisher-ADF
Unit Root Unit Root
Possible Deterministic Components* None. F. T None. F. T
Autocorrelation Correction Method
No Unit Root Lags Some cross- sections Lags with out Unit Root Fisher PP-test Unit Root Some cross- sections None. F. T Kernel with out Unit Root *) None: no deterministic components. F: fixed cross section effects. T: individual trend effects
Table 4. Panel unit root test results
Test LLC Fisher- ADF PP- test
ln( Et / Yt ) DT Test value p-value -2.86 0.002** 54.30 0.078* 53.12 0.080*
(ln Yt )
Test value -6.01 121.64 141.36
p-value 0.000*** 0.000*** 0.000***
Note: (1) Automatic selection of lags based on minimum AIC: 0-3. (2) ***, **, and * denote rejection of null hypothesis. At 1%. 5% and 10% level of Significance, respectively (3) Deterministic components. PP, Breitung, and Fisher-ADF tests: fix cross section effects and Individual trends. IPS: fixed cross section effects.
Table 4 present the panel unit root test results of series. Table indicates that de-trended energy intensity series ln( Et / Yt ) DT and real GDP growth (lnGDP) series are stationary with all three tests. Therefore, both series are integrated of order zero, i.e. I (0) -series.
14
5. The Relationship between Economic Growth and De-trended Energy Intensity 5.1. Rich countries We use following M1 and M2 panel regression models in order to find the impact of economic growth on de-trended energy intensity
M1)
ln( E / Y )it DT i d ln Yit it
M2)
ln( E / Y )it DT i ln( E / Y ) DT it 1 d ln Yit 'it
The OLS estimation of error component panel data model with lagged dependent variable in the set of regressors produces biased coefficient estimates. The basic problem of using OLS is that the lagged dependent variable is correlated with the error term as the dependent variable ln( E / Y ) DT it is a function of 'it and it immediately follows that ln( E / Y ) DT it 1 is also a function of 'it . The fixed effect (FEM) estimators are also biased
and inconsistent unless the number of time periods is large (for details, see e.g. Baltagi 2002, pp. 129-131).
In order to cope with these problems estimators based on General Method of Moments (GMM) are employed which are consistent for with fixed T. We exploit the GMM-DIFF procedure of Arellano and Bond (1991), which suggests first to difference the model and then to use lags of the dependent and explanatory variables as instruments for the lagged dependent variable. We expect the sign of parameter d ( 1) to be negative in regression model for European countries. We use panel model for 19 European countries. The results in Table 5 show negative value of d which implies that relationship between de-trend energy intensity and real GDP growth in European countries is negative. Hence, hypothesis that the energy intensity falls as a country’s economic growth increase at certain level of development is meaning full. Also the decreasing energy intensity is not necessary a
15
result of business cycles and energy prices shocks since this hypothesis demands that coefficient d gets a positive value.
Table 5 Panel Results (N = 19, T =26, 1980-2006) Dependent variable ln( Et / Yt ) DT
M1 (LSDV)
M2 (LSDV)
M2 ( 7 ) (GMM)
0.54[5.05]
-0.13[-1.80]
(0.002)***
(0.07)*
dSR
-0.062[-1.76]
-0.17[-2.15]
-0.24[-1.79]
(0.09)*
(0.012)**
(0.08)*
SR
0.92
d LR d SR / (1 )
-0.36
LR
0.64
R2
0.30
0.50
No of observation DW-statistic Hansen test (p-Val) ( 4)
483 2.54**
483 2.15**
-0.21 0.79
400 0.90
AR1 (p-Val) (5)
0.06*
AR2 (p-Val) ( 6)
0.85
(1) *** significant at 1 % level of significance (2) ** significant at 5 % level of significance (3) * significant at 10% level of significance (4) Hansen test for over identified restrictions: H 0 :instruments do not correlate withResiduals (5) Arellano – Bond test of first-order autocorrelation, H : There is no first order-autocorrelation. 0
(6) Arellano– Bond test of second-order autocorrelation ,H :There is no second order-autocorrelation. 0
(7) Instrument: Dependent variable lagged 2 periods. Explanatory variables in current period.
Many studies such as Hannesson (2002) found that relationship between growth in energy use and GDP became weaker after the first oil shock especially in the rich countries. The relationship between economic growth and energy consumption growth will also become weaker, in the sense that any given growth rate will require less growth in energy use. Hence, in European economies falling energy intensities implies energy 16
use tend to grow more slowly than GDP. For the support of this argument an other study by Huang et al (2008) applied the dynamic panel data model in order to find causality relationship between energy consumption growth and economic growth, They found also that in rich countries income change leads energy consumption change and the overall effect of economic growth on energy consumption is negative.
A conclusion drawn in a study by the World Economic Council that the energy intensity falls as a country’s economic growth increases, and that this does not, however, occur until a country has reached a certain level of development (see WEC 1993, p.49) is only partly supported by our analysis. However it is not straightforward to conclude that the energy intensity is always higher in poor countries than in rich ones. Much depends also on how GDP is measured (see Hannesson 2002). Hence we try find out also the impact of economic growth on de-trended energy intensity for poor countries that have low GDP per capita with high population. Next we conduct the same three steps of analysis which we did for the European countries above.
5.2. Poor Countries First we analyze energy intensity trend in the sample of poor countries, i.e. we estimate Eq. 1). The relationships of energy intensity with GDP, GDP per capita and population growth in poor country is found Figures in Appendix (see Fig. A1, Fig.A2, and Table. A1). The Table 6 shows positive and significant c2 for all poor countries except Afghanistan, India, and Sri Lanka. Thus, we found positive trend in energy intensity
ln( Et / Yt ) series for majority of sample countries.
In order to get the relationship between energy intensity and economic growth in poor countries, we regressed de-trended energy intensity on the growth rate of real GDP, in panel context. Here, we used M1 and M2 (GMM) regression (Table 7).
17
Table.6
Results from the Estimation of Equation (1) (HAC t-values in parenthesis)
Poor Countries
c1
Afghanistan
c2
8.34[19.06]**
-0.04[-1.87]*
Bangladesh
6.47[255.88]***
0.02[14.50]***
Bhutan
7.35[12.23]***
India
9.09[250.09]***
0.01[2.97]** -0.001[-0.96]
Indonesia
8.71[379.27]***
0.01[4.03]***
Maldives
7.12 [7.80]***
0.03[0.041]***
Malaysia
6.83[34.22]***
0.08[6.92]***
Nepal
6.29[207.55]***
0.047[5.48]***
Pakistan
8.88[747.61]***
Sri Lanka
7.75[140.75]***
0.001[2.50]* -0.002[-1.00]
Panel Regression
7.80[135.25]***
0.023[6.14]***
(1) *** significant at 1% level of significance (2) ** significant at 5% level of significance (3) * significant at 10% level of significance
The results in Table 7 show the positive value of d. Note also that the relationship between energy intensity and real GDP in poor countries is above one ( d 1) . Hence, the increasing energy intensity is a result of economic growth that does not induce energy saving. The increasing energy intensities in poor countries may also be the result of business cycles and oil price shocks, since this hypothesis demands that coefficient of d gets a positive values. Thus the energy intensities of GDP increase in poor countries. The energy consumption has, in most case, grown more rapidly than GDP. For example in India energy consumption grown more rapidly than GDP (see Hannesson 2002). Huang et al. (2008) found that in poor countries, 1% increase in economic growth requires more than 1% increase in energy use. The cost of converting energy into GDP is high in poor countries.
18
Table 7 Panel Results (N = 10, T = 26, 1980-2006) Dependent ln( Et / Yt ) DT
M2 ( 7 ) (GMM)
variable M1 (LSDV)
0.77[43.40]
dSR
SR
(1) (2) (3) (4)
(0.000)***
0.10[1.30](0.13)
0.36[13.34] (0.000)***
1.10
d LR d SR / (1 )
1.56
LR
2.56
R2
0.15
No of observation DW-statistic Hansen test (p-Val) ( 4)
260 1.80
250 0.99
AR1 (p-Val) (5)
0.49
AR2 (p-Val) ( 6)
0.48
*** significant at 1 % level of significance, ** significant at 5 % level of significance * significant at 10% level of significance Hansen test for over identified restrictions: H 0 :instruments do not correlate with residuals
(5) Arellano – Bond test of first-order autocorrelation, H : There is no first order-autocorrelation. 0
(6) Arellano– Bond test of second-order autocorrelation ,H :There is no second order autocorrelation. 0
(7) Instrument: Dependent variable lagged 2 periods. Explanatory variables in current period.
5.3. Energy Intensity and the Main Sectors Contributing to Economic Growth Typically we can divide an economy into three main sectors that is agricultural, industrial, and services sectors1. Theoretically there exist some views how these sectors are related with economic growth. The relation of industrial sector with economic growth 1
We have divided the economy in to these three main sectors since the separate data is available on these different sectors.
19
has it roots in Kaldor views about manufacturing sector. Kaldor (1966) argued that an industrial sector is the “engine of growth”. It is widely believed that an expansion of the service sector relative to the rest of the economy leads to a reduction in the long run rate of growth of output per capita (see for example Baumol et at. 1985, Bjork 1999, Wolff 1985b, Wilber 2002, among others). Considered following relationship between GDP and its main sector outputs:
Y (t ) Ae t AG (t ) I (t ) c S (t ) | ln( ) ln Y (t ) A ' t ln AG (t ) c ln I (t ) ln S (t )
| dt
dY (t ) / dt dAG (t ) / dt dI (t ) / dt dS (t ) / dt c . Y (t ) AG (t ) I (t ) S (t )
4)
ln Yit i ln Agricultureit c ln Industryit ln Servicesit it
In order to find the impact of sectors on de-trended energy intensity we recall that
(M1)
ln( E / Y ) it
DT
d ln Yit it
By plugging the Eq.4) into model M1), we get the relationship between de-trended energy
intensity and economy sectors. We use following fixed effects and dynamic panel data models.
Fixed Effect Model: (M3) ln( Eit / Yut ) DT i 1 ln Agricultureit 2 ln Industryit 3 ln Servicesit it
20
Dynamic Panel Data Model:
ln( E / Y )it DT i ln( E / Y ) DT it 1 1 ln Agricultureit (M4)
2 ln Industryit 3 ln Servicesit it
In the earlier phases of development there is a shift away from agriculture towards heavy industry, while in the later stages of development there is a shift from the more resource intensive extractive and heavy industrial sectors towards services. It is often argued that this will result in an increase in energy used per unit of output in the early stages of economic development and a reduction in energy used per unit output in the later stages of economic development (Panayotou, 1993).
However, service sector still need large energy and resource inputs (Stern 2003). The energy consumption of the service sector comprises the energy used in buildings of the public and private service sector. This sector is also often referred to as tertiary sector. The share of the sector in the final energy consumption has increased slightly in the EU25 (13 % in 2004 vs. 12% in 1990). The energy consumption of this sector is often calculated as a residual as the balance between the total final energy consumption and the energy consumption of industry and agriculture. The economic growth is faster for services than for industry sector or the whole economy: 3.2 %/year on average from 1990 to 2004 as compared to 1.4 % for industry and 2 % for the GDP. As a consequence, the service sector has increased its contribution to the GDP: in 2004, 64 % of the GDP was generated by services in the EU-25 up from 54% in 1990 (59% in 1996). ( see IEEA 2009).
21
Table. 8 Sector Output Growth and De-trended Energy Intensity European
ln( Et / Yt ) DT
M3 (OLS)
1
2 3
[-1.68]*
(7)
M3 (OLS)
(GMM)
M4
(7)
(GMM)
- 0.55
- 0.25
[-2.90]**
[-1.60]*
-0.005
-0.002
-0.0001
[-1.80]*
[-1.10]
[1.45]*
-0.003
-0.006
-0.10
-0.01
[-1.05]
[1.30]
[1.40]
-0.12
-0.10
0.19
0.029
[-1.85]*
[-2.10]**
[1.70]*
[6.34]**
0.40
No of observation
420
DW-statistic
1.60
AR2 (p-Val) ( 6)
M4
[-1.40]
R2
Hansen test (pVal) ( 4) AR1 (p-Val) (5)
(1) (2) (3) (4)
-0.004
Poor
0.30 400
255
200
2.55* 0.34
0.90
0.05*
0.69
0.63
0.61
*** significant at 1 % level of significance ** significant at 5 % level of significance * significant at 10% level of significance Hansen test for over identified restrictions: H 0 :instruments do not correlate with residuals
(5) Arellano – Bond test of first-order autocorrelation, H : There is no first order-autocorrelation. 0
(6) Arellano– Bond test of second-order autocorrelation ,H :There is no second order-autocorrelation. 0
(7) Instrument: Dependent variable lagged 2 periods. Explanatory variables in current period.
The results in Table 8 show for European panel that the relationship between services sector and de-trended energy intensity is negative and significant ( 3 ) . The services sector has a relatively low level of final energy intensity albeit the sector is fast growing fastest and the share of services sector is highest in European countries. Drivers impacting on services final energy intensity include: improvements in energy efficiency, use of information and communication technology in offices, the average office or floor 22
space per unit of added value, climatic conditions, and insulation. These factors have much lower energy scale than the industrial processing has. Therefore, the cost of converting energy in to GDP lowest in services sector because services sector has increased its efficiency in Europe. On other hand for the panel of poor economies, the relationship between services sector and de-trend energy intensity is positive. The positive value for the services sector is highest. This is may be due to inefficiencies in services sector.
Final sector energy intensities are influenced by structural changes in the economy, i.e. shifts in the GDP structure among economic or industrial branches. For instance, an increasing share of services in the GDP, all other things being equal, results in a decrease of the final energy intensity because it requires much less energy to create one unit of GDP in the services sector than in the manufacturing industry For the same reason, a falling contribution of energy-intensive branches to the industry value added also results in a decrease of the final energy intensity (see IEEA 2009)
5.2. Energy Intensity, GDP per capita and Population Growth Two main indicators can be considered to characterize the level of activity in the service sector: GDP per capita, which is increased by almost 30 % from 1996 to 2004 and employment by 15 %. The labour productivity has been rising steadily since 1996 in the EU-25. Therefore, next we study the role of population growth and GDP per capita.
Fixed Effect Model: (M5)
ln( E / Y )it DT i 1 ln POPit 2 ln GDPcit it
Dynamic Panel Data Model:
(M6)
ln( E / Y )it DT i ln( E / Y ) DT it 1 1 ln POPit 2 ln GDPcit it
23
European countries have high and rising income levels. They have also stable or declining populations and constantly improving energy intensities. The result of all this is the high GDP per capita level. Their (active) population and energy intensity move in same directions that help insulate them from the worst effects of energy declines. The results in Table 9 support this argument. We found negative coefficient on real GDP per capita growth for European panel ( 2 ) but the pure population growth effect ( 1 ) is positive. Hannesson (2008) describes also that there is some indication that growth in energy use declines with GDP per capita, for any give growth rate of GDP.
Table. 9 GDP per capita growth, Population growth, and De-trended Energy Intensity European
ln( Et / Yt ) DT
Poor
M5 (OLS)
M6 ( 7 ) (GMM)
M5 (OLS)
M6 ( 7 ) (GMM)
0.93 [1.70]* -0.22 [-2.95]**
-0.003 [-2.02]** 0.0001 [2.05]** -0.19 [-25.68]***
1.99 [1.30] 0.10 [1.10]
0.12 [23.80]*** 0.99 [7.54]*** 0.24 [1.34]
483
464
255
200
1
2 R2 No of observation DW-statistic Hansen test (pVal) ( 4) M1(p-Val) (5) M2(p-Val) ( 6) (1) (2) (3) (4)
1.20
1.15 0.34
0.82
0.99
0.31
0.35
0.18
*** significant at 1 % level of significance ** significant at 5 % level of significance * significant at 10% level of significance Hansen test for over identified restrictions: H 0 :instruments do not correlate with residuals
(5) Arellano – Bond test of first-order autocorrelation, H : There is no first order-autocorrelation. 0
(6) Arellano– Bond test of second-order autocorrelation ,H :There is no second order-autocorrelation. 0
(7) Instrument: Dependent variable lagged 2 periods. Explanatory variables in current period.
On other hand poor countries have few energy options, their population growth is high, and energy intensities are high. Therefore, their economies tend to show worsening 24
energy intensities over time. We obtained positive relationship between energy intensity and population growth for poor countries, whereas the relationship between GDP per capita growth and de-trended energy intensities are positive for poor countries.
6. Conclusion We found negative energy intensity trend for the European countries and positive energy intensity trend for a sample of poor countries. Therefore, we de-trended the energy intensity series and analyzed the impact of economic growth, sector outputs, population, and GDP per capita on de-trended energy intensities. We found negative relationship between economic growth and de-trended energy intensity for the European countries, and positive relationship for the poor countries. The rich countries have undertaken energy conservation polices. In poor countries growth requires intensive energy use. Energy intensity correlations in poor countries can be also a result of business cycles and oil prices shocks. The relationship between sector outputs (i.e. industry, services, and agriculture output) and de-trended energy intensity is negative for the European countries. The relationship for the dominating services sector is positive for the sample of poor countries.
The impacts of some main economic indicators (i.e. GDP growth, growth of main sector output, growth in GDP per capita, and population growth) on de-trended energy intensity are large and energy saving for European countries. Unfortunately they are equally large but energy using in the poor countries. Hence, we found that the cost of converting energy into GDP is low in the European countries as compare to poor countries. Therefore, policy makers in both rich and poor countries should find methods and technology to improve energy efficiency in the poor countries. Poor developing countries can avoid the long lasting high energy intensity trap by improving their energy conversion and technology, with well managed family policy, and with sustained economic growth.
25
References ADEME (2003), ENERDATA, ISI-FhG, Energy efficiency in the European Union 19902001. SAVE-ODYSSEE Project on Energy Efficiency Indicators Alcantara, V., Duro, J.A., 2004. Inequality of energy intensities across OECD countries: a note. Energy Policy 32, 1257-1260. Arellano M. and Bond S. (1991) “Some tests of specification for panel data, Monte Carlo evidence and an application to employment equations”, RES 58, 277– 297. rd
Baltagi B. H. (2002) Econometric Analysis of Panel Data, 3 ed. John Wiley and Sons. Baumol, W. J., Batey, B. S. S., and Edward W. N. (1985), “Unbalanced Growth Revisited: Asymptotic Stagnancy and New Evidence” American Economic Review”, Vol. 75, No. 4, pgs. 806-17. Bjork, G. C. (1999), “The Way it Worked and Why it won’t: Structural Change and the Slowdown of U.S. Economic Growth” Westport: Praeger Publishers. Blanchard, O.J., Gali, J., 2008. The macroeconomic effects of oil price shocks: why are the 2000s so different from the 1970s? NBER Working Paper n 13368. Dinda, S., Goondoo, D., Pal, M., 2000. Air Quality and economic growth: an empirical study. Ecological Economics 34(3), 409-423. Dinda, S., 2004. Envirnmental Kuznets curve hypothesis: a survey. Ecological Economics 49(4), 431-455. EEA (2008), EN21 Final Energy Consumption Intensity, European Environmental Agency, EEA Report No 6/2008. Ezcurra, R., 2007. Distribution energy intensities: a cross-country analysis. Energy Policy 35, 5254-5259. Fisher, R.A. (1932). Statistical Methods for Research Workers, 4th Edition, Edinburgh: Oliver and Boyd. Grossman, G.M., Krueger, A., 1995. Economic growth and the envirnmen. Quarterly Journal of Economics 10(2), 353-377. Hannesson, R., 2002. Energy use and GDP growth, 1950-97. Organization of Petroleum Expoting Countries, OPEC Review. Huang, B.N., Hwang, M.J., Yang, C.W. (2008), Causal relationship between energy consumption and GDP growth revisited: Adynamic panel data approach, Journel of Ecological Economics 67(2008), 41-54. IEEA. 2009. Overall Energy Efficiency Trends and Policies in the EU 27. Lessons from the ODYSSEE MURE project, 2009. Kilian, L., 2008a. A comparison of the effects of exogenous oil supply shocks on output and inflation in the G7 countries. Journal of the European Economic Association 6, 78121. Kilian, L., 2008b. The economic effects of energy price shocks. Journal of Economic Literature 46, 871-909. Kilian, L., 2009. Not all oil price shocks are alike: disentangling demand and supply shocks in the crude oil market. American Economic Review 99, 1053-1069. Kaldor, N., (1966), “Causes of the Slow Rate of Growth of the United Kingdom”, Cambridge: Cambridge University Press Le Pen, Y., Sevi, B., 2007. On the non-convergence intensities: evidence from a pair wise econometric approach. Ecological Economics, XXXXX.
26
Levin, A., Lin, C.F. and C. Chu (2002). Unit Roots in Panel Data: Asymptotic and Finite Sample Properties, Journal of Econometrics 108:1-14. Maddala, G.S. and S. Wu (1999) A Comparative Study of Unit Root Tests with Panel Data and A New Simple Test. Oxford Bulletin of Economics and Statistics 61, 631-52. Markandya, A., Pedroso-Galinato, S., Streimikiene, D., 2006. Energy intensity in transition economies: is there convergence towards the EU average? Energy Economics 28, 121-145. Panayotou, T. (1993). Empirical test and policy analysis of environmental degradation at different stages of economic development. Working Paper WP238, Technology and Employment Programe, International Labour office, Geneva. Stern, D. (2003). Energy and Economic Growth, Working Paper, Department of Economics, Sage 3208, Rensselaer Polytechnic Institute, 110 8th Street Troy. Wolff, E. N., (1985b), “The Magnitude and Causes of the Recent Productivity Slowdown in the United States: A Survey of Recent Studies”, In Baumol, W. and McLennon, K. eds. Productivity Growth and U.S. Competitiveness. New York: Oxford University Press. 27-57. Wilber, S., (2002), “Are Services Bad for Growth? Evidence from a Panel of OECD Economies”, Unpublished Ph.D. Thesis, Georgetown University Washington, and DC. World Energy Council (1993), Energy for Tomorrow, s World, st. Martin, s Press, New York.
Appendix: Poor countries. Figure A1. Upward trend in Energy Intensity of poor Countries LNEI Afg hanistan
Bang lades
9.5
Bhutan
7.2
India
10
9.0
9.20 9.15
9
7.0 8.5
9.10 8
8.0
6.8
9.05 7
7.5
9.00 6.6
6
7.0 6.5 1980
1985
1990
1995
2000
2005
6.4 1980
1985
Indinesia
1990
1995
2000
5 1980
2005
8.95
1985
M aldevs
9.1 9.0
1995
2000
2005
9.5
9.3
9.0
9.2
1990
1995
2000
2005
2000
2005
7.6
7.2 9.1
8.0 6.8
7.5
8.8
8.9
7.0 8.7
1985
1990
1995
2000
2005
6.4
6.5
8.8
6.0 1980
8.7 1980
1985
Pakistan
1990
1995
2000
2005
2000
2005
Srilanka
8.98
8.0
8.96
7.9
8.94 7.8 8.92 7.7 8.90 7.6
8.88 8.86 1980
1985
Nepal
9.0
8.6 1980
8.90 1980
M alyisa
8.5 8.9
1990
1985
1990
1995
2000
2005
7.5 1980
1985
1990
1995
27
1985
1990
1995
2000
2005
6.0 1980
1985
1990
1995
Figure A2: De-trended Energy Intensity of Poor Countries DEN Afg hanistan
Bang lades
1.5
Bhutan
.15
India
2
.15
.10
1.0
.10 1
.05
.05
0.5 .00
0
.00
0.0 -.05 -1
-0.5 -1.0 1980
-.05
-.10
1985
1990
1995
2000
2005
-.15 1980
-.10
1985
Indinesia
1990
1995
2000
-2 1980
2005
1985
Maldevs
.2 .1
1990
1995
2000
2005
.8
.2
.4
.1
.0
.0
-.4
-.1
1990
1995
2000
2005
2000
2005
Nepal .3 .2 .1
-.1
.0
-.2
1985
1990
1995
2000
2005
-.8 1980
1985
Pakistan .2
.04
.1
.00
.0
-.04
-.1
1985
1990
1995
1990
1995
2000
2005
2000
2005
-.2 1980
-.1
1985
1990
1995
2000
2005
-.2 1980
1985
1990
1995
Srilanka
.08
-.08 1980
1985
Malyisa
.0
-.3 1980
-.15 1980
2000
2005
-.2 1980
1985
1990
1995
Table A1 Panel Unit Root Test Results for Poor countries
TEST LLC Fisher- ADF PP- test
ln( Et / Yt ) DT
(ln Yt )
Test value -2.86 30.80 30.12
Test value 5.028 83.30 129.84
p-value 0.002** 0.078* 0.080*
p-value 0.000*** 0.000*** 0.000***
Note: (1) Automatic selection of lags based on minimum AIC: 0-3. (2) ***, **, and * denote rejection of null hypothesis at 1%. 5% and 10% level of significance. Respectively (3) Deterministic components. PP, Breitung, and Fisher-ADF tests: fix cross section effects and Individual trends. IPS: fixed cross section effects.
28
Figure A3: Time cross plots of GDP and Energy/GDP of Poor Countries. Afg hanistan
Bang lades
12,000
1,200
10,000
1,100
8,000
1,000
Bhutan
India
16,000
10,000 9,500 9,000
900
4,000
800
2,000
700
0
600
EI
6,000
EI
EI
EI
12,000
8,000
8,500 8,000
4,000
0
10,000,000 20,000,000 30,000,000
7,500 0 0
PPP
100,000,000
300,000,000
7,000 0
PPP
Indinesia 12,000
8,500
10,000
8,000
3,000,000
0
PPP
Maldevs
9,000
1,000,000
2,000,000,000
6,000,000,000
PPP
Malyisa
Nepal
11,000
2,000
10,000
1,600
8,000
EI
EI
EI
6,000
7,000
EI
9,000
7,500
1,200
8,000 4,000
6,500
5,500
0 0
500,000,000
1,500,000,000
6,000 0
1,000,000
2,000,000
PPP
PPP
Pakistan
Srilanka
7,800
2,600
7,600
2,400
3,000,000
7,400
2,200
7,200
2,000
7,000
1,800 0
200,000,000 400,000,000 600,000,000 PPP
0
400 0
200,000,000 400,000,000 600,000,000 PPP
EI
2,800
EI
8,000
800
7,000
2,000
6,000
50,000,000 100,000,000 150,000,000 PPP
29
0
20,000,000 40,000,000 60,000,000 PPP
Figure. A4: Time cross plots of Population and Energy/GDP of Poor Countries. Afghanistan
Banglades
12,000
1,200
10,000
1,100
8,000
1,000
Bhutan
India
16,000
10,000 9,500 9,000
900
4,000
800
2,000
700
EI
6,000
EI
EI
EI
12,000
8,000
8,500 8,000
4,000 7,500
POP
0 440 480 520 560 600 640 680
16 0, 00 0
40,000
14 0, 00 0
30,000
12 0, 00 0
20,000
10 0, 00 0
600
80 ,0 00
0 10,000
7,000 600,000
POP
800,000
1,000,000 1,200,000
POP
POP
Indinesia
Maldevs
9,000
12,000
8,500
10,000
8,000
Malyisa
10,000
1,600
EI
EI
EI
6,000
EI
9,000
7,000
1,200
8,000 4,000
6,500
160,000
200,000
0 150
240,000
POP
200
250
300
350
POP
Pakistan
2,600
7,600
2,400
EI
7,800
EI
2,800
7,400
2,200
7,200
2,000
120,000
160,000
POP
200,000
6,000 12,000 16,000 20,000 24,000 28,000 POP
Srilanka
8,000
800
7,000
2,000
6,000
7,000 80,000
2,000
8,000
7,500
5,500 120,000
Nepal
11,000
1,800 14,000 16,000 18,000 20,000 22,000 POP
30
400 12,000 16,000 20,000 24,000 28,000 POP
Figure A5: Time cross plots of GDP per capita and Energy/GDP of poor countries.
Afghanistan
Banglades
12,000
1,200
10,000
1,100
8,000
1,000
Bhutan
India
16,000
10,000 9,500 9,000
900
4,000
800
2,000
700
EI
6,000
EI
EI
EI
12,000
8,000
8,500 8,000
4,000
0 200
400
600
7,500
600 500
800
0 1,000
CGDP
1,500
2,000
2,500
7,000 0
2,000
CGDP
Indinesia 12,000
8,500
10,000
8,000
6,000
0
CGDP
Maldevs
9,000
4,000
1,000
2,000
3,000
4,000
CGDP
Malyisa
Nepal
11,000
2,000
10,000
1,600
8,000
EI
EI
EI
7,500 6,000
7,000
EI
9,000 1,200
8,000 4,000
6,500
5,500
0 0
2,000
4,000
6,000
6,000 0
CGDP
2,000
4,000
6,000
8,000
CGDP
Pakistan
2,600
7,600
2,400
EI
7,800
EI
2,800
7,400
2,200
7,200
2,000
7,000
1,800 0
1,000
2,000 CGDP
3,000
4,000
0
2,000
0
5,000
10,000 15,000 20,000 CGDP
Srilanka
8,000
800
7,000
2,000
6,000
4,000
CGDP
31
6,000
400 500
1,000
1,500
CGDP
2,000
I. Introduction Energy is the main input for the economic growth. Sustainable development in economic growth underlines the importance of sustained growth of energy productivity. This is manifested by the declining energy intensity for industrial countries. Energy intensity is a measure of the energy efficiency of a nation’s economy. It is calculated as units of energy per unit of GDP. Higher energy intensity indicates a high price or cost of converting energy into GDP. Low energy intensity indicates a low price or cost of converting energy into GDP. In addition, national energy intensities change over time. Countries with higher GDP tend to have energy intensities that improve, helping to insulate them from some of the erosive effects of declining energy supplies. Countries at the bottom of the GDP scale tend to require more and more energy to produce the GDP. For the support of this argument a study by Huang (2008) found that in poor countries, 1% increase in economic growth requires more than 1% increase in energy use. On other hand rich Europen economies require less and less energy input in relative terms. For the 1
support of this argument a study by EEA (2008) found that over the period 1990-2002, European GDP grows at an annual average rate of 2.2 % and total energy consumption at annual average rate of 0.5%. As a result, total energy intensity in the EU fall as an average rate of 1.7% (see EEA 2008). In general the picture resembles closely to EKC phenomena. EKC is an “inverted U” relationship between the level of economic development and the pollution level. In poor countries, as economies improve, pollution gradually increases and as the industrial potential expands. Sooner or later, the pollution problem becomes a major concern which requires immediate actions. In general, as the income increases beyond some threshold, there is a tendency towards producing low pollution products. (See Haung et al. 2008, Grossman and Krueger 1995, Dinda 2004 and Dinda et al. 2000).
With current policies, European energy demand will grow yearly at 1.2 percent to 2020. Europe represents 17 percent of global energy consumption, less than the 22 percent of the United States, the world’s largest energy consumer, but more than China, which is at 14 percent. Yet energy intensity (the energy used to generate real GDP in PP-corrected US-dollars) varies significantly across the Europe. Northwestern Europe’s economies run at a relatively low level of energy intensity of 7,200 BTUs per GDP. The energy intensity of Northeastern Europe is almost twice as high, slightly above the global average of 12,600 BTUs per GDP. Meanwhile, Southern Europe falls between the two with an energy intensity of 8,300 BTUs per GDP, similar to that of the United States. Note that if Europe captures the full potential of increase in energy productivity, the continent could abate energy demand equivalent to 8 million barrels of oil per day, or double the final electricity consumption of all EU-25 countries today. Solely using the available techneologies offers positive returns (see EEA 2008).
Decrease in energy intensity in developed countries is a largely accepted hypothesis in energy literature. However the relationship between energy intensity and economic growth is not so clear. Typically it is assumed that for the developed countries a negative relationship between energy intensity and economic growth is valid. Thus the “energy2
saving” high GDP-level effect is extended to be valid also for economic growth. Main argument is the fact that service economy is less energy demanding than preceding economic epochs, and energy saving economy is only possible with high technology (Stern 2003). However if the technology and energy composition effects (i.e. declining trend of energy/GDP ratio) are removed, the growth effects may still be energy saving since non-trend effects are still affected by technology spill-over effects.
Interestingly, these hypotheses have not been formally tested. Today’s economic literature has focused on the issue of convergence of energy intensities. Some papers examined the issue by comparing energy intensities of developed and developing countries (Alcantra and Duro, 2004), and other conducted a convergence analysis of transitory countries (Markandya et al. 2006, Ezcurra 2007, Le Pen and Sevi 2007).
In this paper we explore the European energy intensity trends both for individual countries and for the panel of countries. The growth dependence of energy intensity and its main sector (i.e. industry, services and agriculture) outputs are studied in details. The role of population growth is also noticed when the analysis is cast in GDP per capita format. The results are compared to experience of some developing countries where the population growth has the major impact on energy consumption.
We found negative trend in energy intensity for European countries and positive trend for the poor countries. Therefore, we focus on the de-trended intensity response to economic growth and its main sectors. We de-trended the energy intensity series and found the negative impact of economic growth on de-trended energy intensity with European countries. However, we still found positive relationship between de-trend energy intensity and economic growth for the panel of poor countries. These findings support the hypothesis that energy intensity falls as a country’s economic growth takes place after a certain level of economic development. The high income group (i.e. rich countries) has low energy intensities because energy consumption in these countries grows more slowly than GDP and they have undertaken energy conservation polices. Hence, the role oil price shocks on the falling energy intestines of European economies are questioned. 3
Where as, in poor countries growth requires intensive energy use, and the business cycles and the price shocks affect energy use positively. Hence the cost of converting energy into GDP is high in poor countries. We also found that the services sector is negatively related with energy intensities in European countries but positive relationship was found in poor countries. A positive relationship was also found between population growth and de-trended energy intensity for both groups of countries.
The study is organized as follows. Section 2 presents data and variables. Section 3 reviews the theory of economic growth and energy intensity in Europe. Section 4 presents model and estimation procedures, and 5 gives the results and 6 present the conclusion.
2. Data and Variables Energy Intensity: The data on energy intensity is from the US Energy Information Administration (EIA Independent Statistic and Analysis). Data for this variable is calculated by dividing the data on total primary energy consumption in quadrillion British thermal by the gross domestic product using purchasing power parities in billions of 2000 U.S. dollars for each available country and year from Global Insight. Real GDP Per Capita: The data on Real GDP per Capita, Real GDP and Population is from Center for International Comparison at the University of Pennsylvania (Penn Tables). The definition of variables is following: Real Gross Domestic Product Per Capita and components for 1996 are obtained from an aggregation using price parties and domestic currency expenditures for consumption, investment and government of August 2001 vintage. Real Gross Domestic Product is calculated by multiplying the data on Real GDP Per Capita for each country by Population. Population: Population variable is from World Bank Development Indicators 2001, and United Nation Development Center sources. Industry Output: Industry corresponds to ISIC divisions 10-45 and includes manufacturing (ISIC divisions 15-37). It comprises value added in mining, manufacturing (also reported as a separate subgroup), construction, electricity, water, and gas. Value added is the net output of a sector after adding up all outputs and subtracting intermediate inputs. It is calculated without making deductions for depreciation of fabricated assets or 4
depletion and degradation of natural resources. The origin of value added is determined by the International Standard Industrial Classification (ISIC), revision 3. Data are in current U.S. dollars. Source: World Bank national accounts data, and OECD National Accounts data files. Service Output: Services correspond to ISIC divisions 50-99. They include value added in wholesale and retail trade (including hotels and restaurants), transport, and government, financial, professional, and personal services such as education, health care, and real estate services. Also included are imputed bank service charges, import duties, and any statistical discrepancies noted by national compilers as well as discrepancies arising from rescaling. Value added is the net output of a sector after adding up all outputs and subtracting intermediate inputs. It is calculated without making deductions for depreciation of fabricated assets or depletion and degradation of natural resources. The industrial origin of value added is determined by the International Standard Industrial Classification (ISIC), revision 3. Data are in current U.S. dollars. Source: World Bank national accounts data, and OECD National Accounts data files. Agriculture Output: Agriculture corresponds to ISIC divisions 1-5 and includes forestry, hunting, and fishing, as well as cultivation of crops and livestock production. Value added is the net output of a sector after adding up all outputs and subtracting intermediate inputs. It is calculated without making deductions for depreciation of fabricated assets or depletion and degradation of natural resources. The origin of value added is determined by the International Standard Industrial Classification (ISIC), revision 3. Data are in current U.S. dollars. Source: World Bank national accounts data, and OECD National Accounts data files. Time Period: 1980-2006, 27 years. Developed Countries: Austria, Belgium, Cyprus, Denmark, Finland, France, Greece, Iceland, Ireland, Italy, Luxembourg, Malta, Netherlands, Norway, Poland, Portugal, Spain, Sweden, and United Kingdom. Poor Developing Countries: We selected 8 SAARC (i.e. South Asian Association for Regional Cooperation) countries plus their two important bordering countries (Indonesia and Malaysia). The countries are following, Afghanistan, Bangladesh, Bhutan, India, Indonesia, Maldives, Malaysia, Nepal, Pakistan, and Srilanka. 5
3. Economic Growth and Energy Intensity in European Economies Energy intensity can be reduced by improving efficiency in the use of energy and by improving economic activities. Basically the economies with high GDP level manage to do this better than poor countries. This is due the (relative) less energy consuming service economy compared to agricultural or industry economy and efficiency of energy production. The level of economy’s technology, its energy base and conversion are key elements in the evolution of energy/GDP –ratio. Figure 1: Time cross plot of GDP –levels and Energy/GDP -ratio. BEL
8,400
11,000
8,000
FIN 14,000
PPP
GBR
8,500
13,000 8,000
0 00 ,0 0
0
PPP
FRA
6,400
5,000 40,000,000
6,200 120,000,000
200,000,000
0
500,000,000
PPP
20 ,0
0
0
00 ,0 0
PPP
00 ,0 0
6,800
6,600
15 ,0
300,000,000
8,000
7,000
6,000
5, 00
100,000,000
7,200
80 ,0 12 00, 0, 00 0 0 16 00, 0, 00 0 0 20 00, 0, 00 00 0 24 0, 0, 00 0 0 28 00, 0, 00 0 0 32 00, 0, 00 0 0 36 00, 0, 00 00 0 0, 00 0
0
9,000
7,000 6,800
10 ,0
6,000
7,600
0, 00 0
6,500
10,000
ESP 7,200
8,000
EI
12,000
DNK 9,000
EI
7,000
8,800
EI
EI
7,500
CYP
13,000
EI
AUT 8,000
1,500,000,000
PPP
GRC
IRL
10,000
7,000
9,000
9,000
6,500
8,000
8,000
6,000
7,000 EI
7,500
EI
11,000
EI
EI
EI
12,000
7,000
5,500
6,000
6,000
5,000
5,000
5,000 400,000,000
4,500
10,000 7,000 9,000 8,000 40,000,000
120,000,000
6,500 500,000,000
200,000,000
PPP
1,500,000,000
1,200,000,000
PPP
ISL
ITA
18,000
2,000,000,000
4,000 0
PPP
0 50,000,000
NLD
10,000
12,000
16,000 6,400
11,000
9,000
16,000
150,000,000 PPP
MLT
18,000
17,000
300,000,000
PPP
LUX
6,800
100,000,000
14,000 8,000
EI
12,000
EI
6,000
EI
EI
EI
10,000 15,000
9,000 14,000
10,000 5,600
7,000
13,000
8,000
8,000
0 00 ,0 0
0
7,000 0
4,000,000 8,000,000 12,000,000 PPP
40 ,0
00 ,0 0
0 00 ,0 0
POL
30 ,0
NOR
2,000,000,000
20 ,0
PPP
0
1,200,000,000
PPP
6,000
0
5,000,000 10,000,000
6,000 00 ,0 0
0
5,200 400,000,000
10 ,0
12,000
0
200,000,000
600,000,000
PPP
PPP
15,000
PRT
16,000
14,000
14,000
13,000
SWE
6,500
14,000
6,000
13,000
5,500
12,000 EI
12,000
EI
EI
EI
12,000 5,000
11,000
10,000 11,000 8,000
10,000 9,000
6,000 0
100,000,000 PPP
300,000,000
4,500
10,000
4,000
9,000
3,500 0
200,000,000
600,000,000
8,000 0
100,000,000
PPP
PPP
300,000,000
0
100,000,000
300,000,000
PPP
Average energy intensity is decreasing approximately for all 20 European countries except Greece, Iceland, and Portugal (see Fig 1). In Iceland this is caused by an increase 6
of energy intensive industries, like the aluminum industry. On other hand in Greece and Portugal, this is most likely caused by decline in economic growth (see Table 1). In general economic growth seems to have an important role in lowering energy intensity. Between 1990 and 2006, economic growth in Europe required increasing energy inputs but the growth rate was less than the GDP growth rate. Total energy consumption increased until 2004 and stabilized all the way through 2006. Over the period 1990-2006, average GDP grew and average energy intensity in the EU decrease (see Table 1).
Table 1: Economic Growth and Energy Intensity in Europe 1980 - 2006 Economic Growth
Energy Intensity
Country
1981- 90
1991-2000
2000-2006
1981-90
1991-2000
2000-2006
Austria
0.064
0.039
0.044
8.856
8.795
8.761
Belgium
0.062
0.037
0.045
9.367
9.274
9.160
Cyprus
0.101
0.059
0.059
8.894
8.963
8.893
Denmark
0.064
0.044
0.050
8.854
8.790
8.616
Finland
0.070
0.034
0.049
9.334
9.337
9.165
France
0.064
0.036
0.043
8.913
8.901
8.823
Greece
0.046
0.042
0.075
8.629
8.780
8.70
Iceland
0.064
0.045
0.073
9.479
9.605
9.730
Ireland
0.074
0.083
0.073
8.926
8.766
8.492
Italy
0.067
0.032
0.036
8.726
8.663
8.648
Luxembourg
0.087
0.066
0.077
9.387
9.022
8.860
Malta
0.082
0.069
0.041
9.079
9.040
8.883
Netherlands
0.061
0.048
0.040
9.231
9.108
9.011
Norway
0.052
0.063
0.068
9.528
9.355
9.214
Poland
0.041
0.055
0.063
-
9.430
9.029
Portugal
0.075
0.050
0.034
8.520
8.634
8.680
Spain
0.072
0.044
0.065
8.792
8.758
8.782
Sweden
0.062
0.034
0.050
9.430
9.340
9.109
United Kingdom
0.067
0.043
0.053
8.980
8.837
8.631
High average energy intensity found in era 1981-90 may still belong to the era of adaption to the high oil prices of the 1970’s. Average energy intensity and average economic growth decreased approximately for all 20 European countries in period 19902000. The period 2000-2006, should be characterized as well-managed era. Here, we observe a negative relationship between economic growth and energy intensity.
7
The negative relationship between energy intensities and economic growth of the panel of European countries indicates that the cost of converting energy into GDP is decreasing in the European nations (see Fig 2). The cost of converting energy into GDP is increasing in poor countries because the relationship between energy intensities and economic growth in the poor countries is positive (see, Appendix Fig. A3, and see also Hannesson 2002).
Figure 2: Time cross plot of Energy/GDP -ratio and dlnGDP 18,000 16,000 14,000
EI
12,000 10,000 8,000 6,000 4,000 2,000 -.10
-.05
.00
.05
.10
.15
.20
GDP
4. Model and Estimation Procedure Many results confirm that the following energy-output relationship or linear regression is valid for industrial countries: ln Et ln Yt t Where 0 1.
By adding on both side ln Yt gives the energy intensity form ln Et ln Yt ln Yt ln Yt t
8
ln( Et / Yt ) ( 1) ln Yt t .
Next we take a difference on both side of the last equation resulting in ln( Et / Yt ) ln( Et 1 / Yt 1 ) ( 1) ln Yt t .
Now if assume that past energy intensity can be approximated with an elementary trend function like ( Et 1 / Yt 1 ) exp(c1 c 2Trt 1 ) .
Thus we estimate model ln( Et 1 / Yt 1 ) c1 c 2Trt 1 t 1
to data and use the fit
ˆ 2t to ln( Et 1 / Yt 1) F cˆ 1 c Tr 1 de-trend the current energy
intensity. That is
ln( Et / Yt ) (cˆ1 cˆ2Trt ) ln( Et / Yt ) DT ( 1) ln Yt t . We expect the sign of parameter d ( 1) to be negative in following regression model
1)
ln( Et / Yt ) DT d ln Yt t .
The parameter d measures the short run dependency of energy-intensity on economic growth but it gives also an indirect estimate of long run energy output relationship, i.e.
d 1. Thus the small negative value d implies low energy intensity and large negative value of d implies high energy intensity.
Next we propose three steps of analysis in order to analyze the impact of economic growth on de-trended energy intensity.
9
Step. 1. Analyze the trends in energy intensity series Take the natural log of energy intensity (E/Y) series, and draw the graphs for each country in the panel. Figure 3. Trends in Energy Intensity in European Countries LNEI AUT
BEL
9.00
CYP
9.5
8.95
DNK
9.08
9.1
9.04
9.0
9.00
8.9
8.96
8.8
8.92
8.7
8.88
8.6
ESP 8.88
9.4
8.84
8.90 8.85
9.3
8.80
8.80
9.2
8.76
8.75 8.70 1980
1985
1990
1995
2000
2005
9.1 1980
1985
1990
FIN
1995
2000
2005
8.84 1980
1985
FRA
9.5
1990
1995
2000
2005
8.5 1980
1985
GBR
9.05
1995
2000
2005
8.72 1980
1985
GER
9.2
1995
2000
2005
2000
2005
2000
2005
2000
2005
8.9
8.95
9.0
1990
GRC
9.00
9.00 9.4
1990
8.8
8.95 8.90 9.3
8.90
8.8
8.7 8.85
8.85 9.2
8.6
9.1 1980
1985
1990
1995
2000
2005
8.75 1980
1985
1990
IRL
1995
2000
2005
8.4 1980
1985
1990
ISL
9.2
1995
2000
2005
8.75 1980
1985
1990
ITA
9.8
1995
2000
2005
8.5 1980
1985
1990
LUX
8.80
9.7
9.0
8.6
8.80
8.80
1995
MLT
9.8
9.2
9.6
9.1
8.75 9.4
9.6
9.0
8.8
8.70
9.2
9.5
8.9 9.0
8.6
8.65
9.4
8.4 1980
1985
1990
1995
2000
2005
9.3 1980
8.8
8.8
1985
NLD
1990
1995
2000
2005
8.60 1980
1985
1990
NOR
1995
2000
2005
8.6 1980
1985
1990
POL
1995
2000
2005
8.7 1980
1985
PRT
9.4
9.6
9.8
8.8
9.3
9.5
9.6
8.7
9.2
9.4
9.4
9.1
9.3
9.2
1990
1995
SWE 9.6
9.4 8.6 8.5
9.2
8.4 9.0
9.2
8.9 1980
1985
1990
1995
2000
2005
9.1 1980
9.0
9.0
1985
1990
1995
2000
2005
8.8 1980
8.3
1985
1990
1995
2000
2005
8.2 1980
1985
1990
1995
2000
2005
8.8 1980
1985
1990
1995
Fig, 3. shows clear downward trends in energy intensity for most of series. The data for some countries is very noisy with large swings over the years. Some do not show clear downward trends (Greece, Iceland, and Portugal). In all other countries energy intensity falls. The existence of a trend in energy intensity has significant implications for feasible target energy consumption (see Le Pen and Sevi 2007).
10
Note that from macroeconomic point of view the declining energy share started in the late 1980’s can be caused by oil shocks (see Blanchard and Gal 2008, Killian 2008a,b and 2009). Thus, energy intensity decline is due to these business cycle effects. However we argue that declining trend is due to the technology and better energy conversion since if the business cycle view is valid there should be a positive correlation between de-trended energy intensity and GDP growth rate. In general oil price shocks may have long run energy intensity effects but typically business cycle theories imply short run non-trend effects to be pro-cyclical. Our hypothesis is contrary to this as we argue that GDP growth effects are still energy saving, i.e. they have negative effects on the de-trended energy intensity.
Step. 2. De-trending energy intensity ln( Et / Yt ) DT series for each sample country. 2006
E Suppose a sample it of T observations on energy intensity series for each cross Yit t 1980
section country i = 1, 2, …,N.
Consider the model where ln( Eit / Yit ) is a sum of deterministic and stochastic components:
2)
3)
ln( Eit / Yit ) ci1 ci 2Trt it
it i it 1 it .
c2 is the slope of the deterministic trend measure, t is an AR(1) random variable whose with first-order autocorrelation coefficient , t is IID random variable. An important issue is whether t is stationary or not. If the series is I(0) this means that stochastic shocks have only a transitory effect on intensity series, and then t series can be used for further empirical analysis. But when | | 1 , t is a random walk type series,
11
i.e. I (1) series, and the regression model with GDP growth rate is unbalanced since ln Yt is expected to be stationary in regression model 1).
Table 2 Results from the Estimation of Model 1 (HAC t-values in parenthesis) _______________________________________________ c1 c2 Country Austria 8.92 [488.88]*** -0.006[-6.24]** Belgium 9.43[322.0]*** -0.011[-13.31]** -0.0004[-0.28] Cyprus 8.93[334.59]*** Denmark 8.96[303.87]*** -0.013[-7.44]** Finland 9.92[310.38]*** -0.009[-5.509]** France 8.96[624.91]*** -0.005[-6.64]** Greece 8.59[288.09]*** 0.007[3.83]* Iceland 9.40[417.55]*** 0.013[8.032]** 9.07[320.32] *** -0.022[-11.25]*** Ireland Italy
8.74[614.27]***
-0.004[-5.07]**
Luxembourg
9.54[290.41]***
-0.032[-10.20]***
Malta
9.12[222.82]***
-0.007[-3.35]
Netherlands
9.31[613.96]***
Norway
9.60[353.26]***
Poland
10.15[184.27]***
Portugal
8.45[180.61]***
Spain
8.79[464.07]***
-0.001[-1.03]
Sweden
9.54[331.75]***
-0.016[-7.58]**
United Kingdom
9.10[715.10]***
-0.0192[-22.74]***
Panel Regression
9.12[933.98]***
-0.008[-13.78]***
* -0.012[-11.65]*** -0.015[-11.23]*** -0.047[-15.45]*** 0.011[4.05]**
(1) *** significant at 1 % level of significance (2) ** significant at 5 % level of significance (3) * significant at 10% level of significance
Now, we eliminated the estimated trend functions from the energy intensity series, i.e. we obtained following series
ln( Eit / Yit ) cˆi1 cˆ2Trt 1 ˆit ln( Eit / Yit ) DT
Graphs of de-trended energy intensity ln( Eit / Yit ) DT series are shown in Fig. 4.
12
Figure 4. De-trended Energy Intensities in European Countries DEN AUT
BEL
.08
.08
.04
.04
.00
CYP
DNK
.15
.15
.10
.10
.05
.05
.00
.00
-.05
-.05
ESP .08
.04
.00
-.04
.00
-.04
-.08 1980
1985
1990
1995
2000
2005
-.08 1980
1985
1990
FIN
1995
2000
2005
-.10 1980
1985
FRA
.20
1990
1995
2000
2005
-.10 1980
-.04
1985
GBR
.08
.08
.04
.04
1990
1995
2000
2005
-.08 1980
1985
1990
GRC
1995
2000
2005
2000
2005
2000
2005
IRL
.12
.10
.08
.15
.05
.04
.10
.00 .05
.00
.00
.00 -.04
.00 -.04
-.08
-.04
-.05
-.05
-.12
-.10 1980
1985
1990
1995
2000
2005
-.08 1980
1985
1990
ISL
1995
2000
2005
-.08 1980
1985
1990
ITA
.10
2000
2005
-.16 1980
1985
1990
LUX
.06
1995
2000
2005
-.10 1980
1985
1990
MLT
.15
1995
NLD
.2
.10
.04
.05
1995
.08
.1
.04
.05 .02
.0
.00
.00
.00
.00
-.1 -.05
-.05
-.02
-.10 1980
1985
1990
1995
2000
2005
-.04 1980
1985
1990
NOR
1995
2000
2005
-.15 1980
-.04
-.2
-.10
1985
1990
POL
1995
2000
2005
-.3 1980
1985
PRT
.10
.10
.1
.05
.05
.0
.00
.00
-.1
1990
1995
2000
2005
2000
2005
-.08 1980
1985
1990
1995
SWE .10 .05 .00 -.05
-.05
-.05
-.2
-.10 1980
-.10 1980
-.3 1980
1985
1990
1995
2000
2005
1985
1990
1995
2000
2005
-.10
1985
1990
1995
2000
2005
-.15 1980
1985
1990
1995
Step. 3. Unit Root Tests on De-trended Energy Intensity and Economic Growth In order to find unbiased relationship between energy intensity and economic growth, we apply following tests. a. Apply unit root test on de-trend energy intensity ln( Et / Yt ) DT series. b. Apply unit root test on real GDP growth (ln Yt ) series. The analysis of non-stationary panels is similar to analysis of non-stationary time series of the 1980s. However non-stationary panels include some unique issue such as crosssectional heterogeneity and correlation. We use LLC test (see Levin, Lin, and Chu 2002), Fisher-ADF and Fisher –PP test (see Maddala & Wu 1999). We test unit roots in panel of 13
series, i.e. energy intensity ln( Et / Yt ) DT series and real GDP growth (ln Yt ) series for 19 cross sections with 27 time periods. Table 3. Summary of five tests Test
Null
Alternative
LLC Fisher-ADF
Unit Root Unit Root
Possible Deterministic Components* None. F. T None. F. T
Autocorrelation Correction Method
No Unit Root Lags Some cross- sections Lags with out Unit Root Fisher PP-test Unit Root Some cross- sections None. F. T Kernel with out Unit Root *) None: no deterministic components. F: fixed cross section effects. T: individual trend effects
Table 4. Panel unit root test results
Test LLC Fisher- ADF PP- test
ln( Et / Yt ) DT Test value p-value -2.86 0.002** 54.30 0.078* 53.12 0.080*
(ln Yt )
Test value -6.01 121.64 141.36
p-value 0.000*** 0.000*** 0.000***
Note: (1) Automatic selection of lags based on minimum AIC: 0-3. (2) ***, **, and * denote rejection of null hypothesis. At 1%. 5% and 10% level of Significance, respectively (3) Deterministic components. PP, Breitung, and Fisher-ADF tests: fix cross section effects and Individual trends. IPS: fixed cross section effects.
Table 4 present the panel unit root test results of series. Table indicates that de-trended energy intensity series ln( Et / Yt ) DT and real GDP growth (lnGDP) series are stationary with all three tests. Therefore, both series are integrated of order zero, i.e. I (0) -series.
14
5. The Relationship between Economic Growth and De-trended Energy Intensity 5.1. Rich countries We use following M1 and M2 panel regression models in order to find the impact of economic growth on de-trended energy intensity
M1)
ln( E / Y )it DT i d ln Yit it
M2)
ln( E / Y )it DT i ln( E / Y ) DT it 1 d ln Yit 'it
The OLS estimation of error component panel data model with lagged dependent variable in the set of regressors produces biased coefficient estimates. The basic problem of using OLS is that the lagged dependent variable is correlated with the error term as the dependent variable ln( E / Y ) DT it is a function of 'it and it immediately follows that ln( E / Y ) DT it 1 is also a function of 'it . The fixed effect (FEM) estimators are also biased
and inconsistent unless the number of time periods is large (for details, see e.g. Baltagi 2002, pp. 129-131).
In order to cope with these problems estimators based on General Method of Moments (GMM) are employed which are consistent for with fixed T. We exploit the GMM-DIFF procedure of Arellano and Bond (1991), which suggests first to difference the model and then to use lags of the dependent and explanatory variables as instruments for the lagged dependent variable. We expect the sign of parameter d ( 1) to be negative in regression model for European countries. We use panel model for 19 European countries. The results in Table 5 show negative value of d which implies that relationship between de-trend energy intensity and real GDP growth in European countries is negative. Hence, hypothesis that the energy intensity falls as a country’s economic growth increase at certain level of development is meaning full. Also the decreasing energy intensity is not necessary a
15
result of business cycles and energy prices shocks since this hypothesis demands that coefficient d gets a positive value.
Table 5 Panel Results (N = 19, T =26, 1980-2006) Dependent variable ln( Et / Yt ) DT
M1 (LSDV)
M2 (LSDV)
M2 ( 7 ) (GMM)
0.54[5.05]
-0.13[-1.80]
(0.002)***
(0.07)*
dSR
-0.062[-1.76]
-0.17[-2.15]
-0.24[-1.79]
(0.09)*
(0.012)**
(0.08)*
SR
0.92
d LR d SR / (1 )
-0.36
LR
0.64
R2
0.30
0.50
No of observation DW-statistic Hansen test (p-Val) ( 4)
483 2.54**
483 2.15**
-0.21 0.79
400 0.90
AR1 (p-Val) (5)
0.06*
AR2 (p-Val) ( 6)
0.85
(1) *** significant at 1 % level of significance (2) ** significant at 5 % level of significance (3) * significant at 10% level of significance (4) Hansen test for over identified restrictions: H 0 :instruments do not correlate withResiduals (5) Arellano – Bond test of first-order autocorrelation, H : There is no first order-autocorrelation. 0
(6) Arellano– Bond test of second-order autocorrelation ,H :There is no second order-autocorrelation. 0
(7) Instrument: Dependent variable lagged 2 periods. Explanatory variables in current period.
Many studies such as Hannesson (2002) found that relationship between growth in energy use and GDP became weaker after the first oil shock especially in the rich countries. The relationship between economic growth and energy consumption growth will also become weaker, in the sense that any given growth rate will require less growth in energy use. Hence, in European economies falling energy intensities implies energy 16
use tend to grow more slowly than GDP. For the support of this argument an other study by Huang et al (2008) applied the dynamic panel data model in order to find causality relationship between energy consumption growth and economic growth, They found also that in rich countries income change leads energy consumption change and the overall effect of economic growth on energy consumption is negative.
A conclusion drawn in a study by the World Economic Council that the energy intensity falls as a country’s economic growth increases, and that this does not, however, occur until a country has reached a certain level of development (see WEC 1993, p.49) is only partly supported by our analysis. However it is not straightforward to conclude that the energy intensity is always higher in poor countries than in rich ones. Much depends also on how GDP is measured (see Hannesson 2002). Hence we try find out also the impact of economic growth on de-trended energy intensity for poor countries that have low GDP per capita with high population. Next we conduct the same three steps of analysis which we did for the European countries above.
5.2. Poor Countries First we analyze energy intensity trend in the sample of poor countries, i.e. we estimate Eq. 1). The relationships of energy intensity with GDP, GDP per capita and population growth in poor country is found Figures in Appendix (see Fig. A1, Fig.A2, and Table. A1). The Table 6 shows positive and significant c2 for all poor countries except Afghanistan, India, and Sri Lanka. Thus, we found positive trend in energy intensity
ln( Et / Yt ) series for majority of sample countries.
In order to get the relationship between energy intensity and economic growth in poor countries, we regressed de-trended energy intensity on the growth rate of real GDP, in panel context. Here, we used M1 and M2 (GMM) regression (Table 7).
17
Table.6
Results from the Estimation of Equation (1) (HAC t-values in parenthesis)
Poor Countries
c1
Afghanistan
c2
8.34[19.06]**
-0.04[-1.87]*
Bangladesh
6.47[255.88]***
0.02[14.50]***
Bhutan
7.35[12.23]***
India
9.09[250.09]***
0.01[2.97]** -0.001[-0.96]
Indonesia
8.71[379.27]***
0.01[4.03]***
Maldives
7.12 [7.80]***
0.03[0.041]***
Malaysia
6.83[34.22]***
0.08[6.92]***
Nepal
6.29[207.55]***
0.047[5.48]***
Pakistan
8.88[747.61]***
Sri Lanka
7.75[140.75]***
0.001[2.50]* -0.002[-1.00]
Panel Regression
7.80[135.25]***
0.023[6.14]***
(1) *** significant at 1% level of significance (2) ** significant at 5% level of significance (3) * significant at 10% level of significance
The results in Table 7 show the positive value of d. Note also that the relationship between energy intensity and real GDP in poor countries is above one ( d 1) . Hence, the increasing energy intensity is a result of economic growth that does not induce energy saving. The increasing energy intensities in poor countries may also be the result of business cycles and oil price shocks, since this hypothesis demands that coefficient of d gets a positive values. Thus the energy intensities of GDP increase in poor countries. The energy consumption has, in most case, grown more rapidly than GDP. For example in India energy consumption grown more rapidly than GDP (see Hannesson 2002). Huang et al. (2008) found that in poor countries, 1% increase in economic growth requires more than 1% increase in energy use. The cost of converting energy into GDP is high in poor countries.
18
Table 7 Panel Results (N = 10, T = 26, 1980-2006) Dependent ln( Et / Yt ) DT
M2 ( 7 ) (GMM)
variable M1 (LSDV)
0.77[43.40]
dSR
SR
(1) (2) (3) (4)
(0.000)***
0.10[1.30](0.13)
0.36[13.34] (0.000)***
1.10
d LR d SR / (1 )
1.56
LR
2.56
R2
0.15
No of observation DW-statistic Hansen test (p-Val) ( 4)
260 1.80
250 0.99
AR1 (p-Val) (5)
0.49
AR2 (p-Val) ( 6)
0.48
*** significant at 1 % level of significance, ** significant at 5 % level of significance * significant at 10% level of significance Hansen test for over identified restrictions: H 0 :instruments do not correlate with residuals
(5) Arellano – Bond test of first-order autocorrelation, H : There is no first order-autocorrelation. 0
(6) Arellano– Bond test of second-order autocorrelation ,H :There is no second order autocorrelation. 0
(7) Instrument: Dependent variable lagged 2 periods. Explanatory variables in current period.
5.3. Energy Intensity and the Main Sectors Contributing to Economic Growth Typically we can divide an economy into three main sectors that is agricultural, industrial, and services sectors1. Theoretically there exist some views how these sectors are related with economic growth. The relation of industrial sector with economic growth 1
We have divided the economy in to these three main sectors since the separate data is available on these different sectors.
19
has it roots in Kaldor views about manufacturing sector. Kaldor (1966) argued that an industrial sector is the “engine of growth”. It is widely believed that an expansion of the service sector relative to the rest of the economy leads to a reduction in the long run rate of growth of output per capita (see for example Baumol et at. 1985, Bjork 1999, Wolff 1985b, Wilber 2002, among others). Considered following relationship between GDP and its main sector outputs:
Y (t ) Ae t AG (t ) I (t ) c S (t ) | ln( ) ln Y (t ) A ' t ln AG (t ) c ln I (t ) ln S (t )
| dt
dY (t ) / dt dAG (t ) / dt dI (t ) / dt dS (t ) / dt c . Y (t ) AG (t ) I (t ) S (t )
4)
ln Yit i ln Agricultureit c ln Industryit ln Servicesit it
In order to find the impact of sectors on de-trended energy intensity we recall that
(M1)
ln( E / Y ) it
DT
d ln Yit it
By plugging the Eq.4) into model M1), we get the relationship between de-trended energy
intensity and economy sectors. We use following fixed effects and dynamic panel data models.
Fixed Effect Model: (M3) ln( Eit / Yut ) DT i 1 ln Agricultureit 2 ln Industryit 3 ln Servicesit it
20
Dynamic Panel Data Model:
ln( E / Y )it DT i ln( E / Y ) DT it 1 1 ln Agricultureit (M4)
2 ln Industryit 3 ln Servicesit it
In the earlier phases of development there is a shift away from agriculture towards heavy industry, while in the later stages of development there is a shift from the more resource intensive extractive and heavy industrial sectors towards services. It is often argued that this will result in an increase in energy used per unit of output in the early stages of economic development and a reduction in energy used per unit output in the later stages of economic development (Panayotou, 1993).
However, service sector still need large energy and resource inputs (Stern 2003). The energy consumption of the service sector comprises the energy used in buildings of the public and private service sector. This sector is also often referred to as tertiary sector. The share of the sector in the final energy consumption has increased slightly in the EU25 (13 % in 2004 vs. 12% in 1990). The energy consumption of this sector is often calculated as a residual as the balance between the total final energy consumption and the energy consumption of industry and agriculture. The economic growth is faster for services than for industry sector or the whole economy: 3.2 %/year on average from 1990 to 2004 as compared to 1.4 % for industry and 2 % for the GDP. As a consequence, the service sector has increased its contribution to the GDP: in 2004, 64 % of the GDP was generated by services in the EU-25 up from 54% in 1990 (59% in 1996). ( see IEEA 2009).
21
Table. 8 Sector Output Growth and De-trended Energy Intensity European
ln( Et / Yt ) DT
M3 (OLS)
1
2 3
[-1.68]*
(7)
M3 (OLS)
(GMM)
M4
(7)
(GMM)
- 0.55
- 0.25
[-2.90]**
[-1.60]*
-0.005
-0.002
-0.0001
[-1.80]*
[-1.10]
[1.45]*
-0.003
-0.006
-0.10
-0.01
[-1.05]
[1.30]
[1.40]
-0.12
-0.10
0.19
0.029
[-1.85]*
[-2.10]**
[1.70]*
[6.34]**
0.40
No of observation
420
DW-statistic
1.60
AR2 (p-Val) ( 6)
M4
[-1.40]
R2
Hansen test (pVal) ( 4) AR1 (p-Val) (5)
(1) (2) (3) (4)
-0.004
Poor
0.30 400
255
200
2.55* 0.34
0.90
0.05*
0.69
0.63
0.61
*** significant at 1 % level of significance ** significant at 5 % level of significance * significant at 10% level of significance Hansen test for over identified restrictions: H 0 :instruments do not correlate with residuals
(5) Arellano – Bond test of first-order autocorrelation, H : There is no first order-autocorrelation. 0
(6) Arellano– Bond test of second-order autocorrelation ,H :There is no second order-autocorrelation. 0
(7) Instrument: Dependent variable lagged 2 periods. Explanatory variables in current period.
The results in Table 8 show for European panel that the relationship between services sector and de-trended energy intensity is negative and significant ( 3 ) . The services sector has a relatively low level of final energy intensity albeit the sector is fast growing fastest and the share of services sector is highest in European countries. Drivers impacting on services final energy intensity include: improvements in energy efficiency, use of information and communication technology in offices, the average office or floor 22
space per unit of added value, climatic conditions, and insulation. These factors have much lower energy scale than the industrial processing has. Therefore, the cost of converting energy in to GDP lowest in services sector because services sector has increased its efficiency in Europe. On other hand for the panel of poor economies, the relationship between services sector and de-trend energy intensity is positive. The positive value for the services sector is highest. This is may be due to inefficiencies in services sector.
Final sector energy intensities are influenced by structural changes in the economy, i.e. shifts in the GDP structure among economic or industrial branches. For instance, an increasing share of services in the GDP, all other things being equal, results in a decrease of the final energy intensity because it requires much less energy to create one unit of GDP in the services sector than in the manufacturing industry For the same reason, a falling contribution of energy-intensive branches to the industry value added also results in a decrease of the final energy intensity (see IEEA 2009)
5.2. Energy Intensity, GDP per capita and Population Growth Two main indicators can be considered to characterize the level of activity in the service sector: GDP per capita, which is increased by almost 30 % from 1996 to 2004 and employment by 15 %. The labour productivity has been rising steadily since 1996 in the EU-25. Therefore, next we study the role of population growth and GDP per capita.
Fixed Effect Model: (M5)
ln( E / Y )it DT i 1 ln POPit 2 ln GDPcit it
Dynamic Panel Data Model:
(M6)
ln( E / Y )it DT i ln( E / Y ) DT it 1 1 ln POPit 2 ln GDPcit it
23
European countries have high and rising income levels. They have also stable or declining populations and constantly improving energy intensities. The result of all this is the high GDP per capita level. Their (active) population and energy intensity move in same directions that help insulate them from the worst effects of energy declines. The results in Table 9 support this argument. We found negative coefficient on real GDP per capita growth for European panel ( 2 ) but the pure population growth effect ( 1 ) is positive. Hannesson (2008) describes also that there is some indication that growth in energy use declines with GDP per capita, for any give growth rate of GDP.
Table. 9 GDP per capita growth, Population growth, and De-trended Energy Intensity European
ln( Et / Yt ) DT
Poor
M5 (OLS)
M6 ( 7 ) (GMM)
M5 (OLS)
M6 ( 7 ) (GMM)
0.93 [1.70]* -0.22 [-2.95]**
-0.003 [-2.02]** 0.0001 [2.05]** -0.19 [-25.68]***
1.99 [1.30] 0.10 [1.10]
0.12 [23.80]*** 0.99 [7.54]*** 0.24 [1.34]
483
464
255
200
1
2 R2 No of observation DW-statistic Hansen test (pVal) ( 4) M1(p-Val) (5) M2(p-Val) ( 6) (1) (2) (3) (4)
1.20
1.15 0.34
0.82
0.99
0.31
0.35
0.18
*** significant at 1 % level of significance ** significant at 5 % level of significance * significant at 10% level of significance Hansen test for over identified restrictions: H 0 :instruments do not correlate with residuals
(5) Arellano – Bond test of first-order autocorrelation, H : There is no first order-autocorrelation. 0
(6) Arellano– Bond test of second-order autocorrelation ,H :There is no second order-autocorrelation. 0
(7) Instrument: Dependent variable lagged 2 periods. Explanatory variables in current period.
On other hand poor countries have few energy options, their population growth is high, and energy intensities are high. Therefore, their economies tend to show worsening 24
energy intensities over time. We obtained positive relationship between energy intensity and population growth for poor countries, whereas the relationship between GDP per capita growth and de-trended energy intensities are positive for poor countries.
6. Conclusion We found negative energy intensity trend for the European countries and positive energy intensity trend for a sample of poor countries. Therefore, we de-trended the energy intensity series and analyzed the impact of economic growth, sector outputs, population, and GDP per capita on de-trended energy intensities. We found negative relationship between economic growth and de-trended energy intensity for the European countries, and positive relationship for the poor countries. The rich countries have undertaken energy conservation polices. In poor countries growth requires intensive energy use. Energy intensity correlations in poor countries can be also a result of business cycles and oil prices shocks. The relationship between sector outputs (i.e. industry, services, and agriculture output) and de-trended energy intensity is negative for the European countries. The relationship for the dominating services sector is positive for the sample of poor countries.
The impacts of some main economic indicators (i.e. GDP growth, growth of main sector output, growth in GDP per capita, and population growth) on de-trended energy intensity are large and energy saving for European countries. Unfortunately they are equally large but energy using in the poor countries. Hence, we found that the cost of converting energy into GDP is low in the European countries as compare to poor countries. Therefore, policy makers in both rich and poor countries should find methods and technology to improve energy efficiency in the poor countries. Poor developing countries can avoid the long lasting high energy intensity trap by improving their energy conversion and technology, with well managed family policy, and with sustained economic growth.
25
References ADEME (2003), ENERDATA, ISI-FhG, Energy efficiency in the European Union 19902001. SAVE-ODYSSEE Project on Energy Efficiency Indicators Alcantara, V., Duro, J.A., 2004. Inequality of energy intensities across OECD countries: a note. Energy Policy 32, 1257-1260. Arellano M. and Bond S. (1991) “Some tests of specification for panel data, Monte Carlo evidence and an application to employment equations”, RES 58, 277– 297. rd
Baltagi B. H. (2002) Econometric Analysis of Panel Data, 3 ed. John Wiley and Sons. Baumol, W. J., Batey, B. S. S., and Edward W. N. (1985), “Unbalanced Growth Revisited: Asymptotic Stagnancy and New Evidence” American Economic Review”, Vol. 75, No. 4, pgs. 806-17. Bjork, G. C. (1999), “The Way it Worked and Why it won’t: Structural Change and the Slowdown of U.S. Economic Growth” Westport: Praeger Publishers. Blanchard, O.J., Gali, J., 2008. The macroeconomic effects of oil price shocks: why are the 2000s so different from the 1970s? NBER Working Paper n 13368. Dinda, S., Goondoo, D., Pal, M., 2000. Air Quality and economic growth: an empirical study. Ecological Economics 34(3), 409-423. Dinda, S., 2004. Envirnmental Kuznets curve hypothesis: a survey. Ecological Economics 49(4), 431-455. EEA (2008), EN21 Final Energy Consumption Intensity, European Environmental Agency, EEA Report No 6/2008. Ezcurra, R., 2007. Distribution energy intensities: a cross-country analysis. Energy Policy 35, 5254-5259. Fisher, R.A. (1932). Statistical Methods for Research Workers, 4th Edition, Edinburgh: Oliver and Boyd. Grossman, G.M., Krueger, A., 1995. Economic growth and the envirnmen. Quarterly Journal of Economics 10(2), 353-377. Hannesson, R., 2002. Energy use and GDP growth, 1950-97. Organization of Petroleum Expoting Countries, OPEC Review. Huang, B.N., Hwang, M.J., Yang, C.W. (2008), Causal relationship between energy consumption and GDP growth revisited: Adynamic panel data approach, Journel of Ecological Economics 67(2008), 41-54. IEEA. 2009. Overall Energy Efficiency Trends and Policies in the EU 27. Lessons from the ODYSSEE MURE project, 2009. Kilian, L., 2008a. A comparison of the effects of exogenous oil supply shocks on output and inflation in the G7 countries. Journal of the European Economic Association 6, 78121. Kilian, L., 2008b. The economic effects of energy price shocks. Journal of Economic Literature 46, 871-909. Kilian, L., 2009. Not all oil price shocks are alike: disentangling demand and supply shocks in the crude oil market. American Economic Review 99, 1053-1069. Kaldor, N., (1966), “Causes of the Slow Rate of Growth of the United Kingdom”, Cambridge: Cambridge University Press Le Pen, Y., Sevi, B., 2007. On the non-convergence intensities: evidence from a pair wise econometric approach. Ecological Economics, XXXXX.
26
Levin, A., Lin, C.F. and C. Chu (2002). Unit Roots in Panel Data: Asymptotic and Finite Sample Properties, Journal of Econometrics 108:1-14. Maddala, G.S. and S. Wu (1999) A Comparative Study of Unit Root Tests with Panel Data and A New Simple Test. Oxford Bulletin of Economics and Statistics 61, 631-52. Markandya, A., Pedroso-Galinato, S., Streimikiene, D., 2006. Energy intensity in transition economies: is there convergence towards the EU average? Energy Economics 28, 121-145. Panayotou, T. (1993). Empirical test and policy analysis of environmental degradation at different stages of economic development. Working Paper WP238, Technology and Employment Programe, International Labour office, Geneva. Stern, D. (2003). Energy and Economic Growth, Working Paper, Department of Economics, Sage 3208, Rensselaer Polytechnic Institute, 110 8th Street Troy. Wolff, E. N., (1985b), “The Magnitude and Causes of the Recent Productivity Slowdown in the United States: A Survey of Recent Studies”, In Baumol, W. and McLennon, K. eds. Productivity Growth and U.S. Competitiveness. New York: Oxford University Press. 27-57. Wilber, S., (2002), “Are Services Bad for Growth? Evidence from a Panel of OECD Economies”, Unpublished Ph.D. Thesis, Georgetown University Washington, and DC. World Energy Council (1993), Energy for Tomorrow, s World, st. Martin, s Press, New York.
Appendix: Poor countries. Figure A1. Upward trend in Energy Intensity of poor Countries LNEI Afg hanistan
Bang lades
9.5
Bhutan
7.2
India
10
9.0
9.20 9.15
9
7.0 8.5
9.10 8
8.0
6.8
9.05 7
7.5
9.00 6.6
6
7.0 6.5 1980
1985
1990
1995
2000
2005
6.4 1980
1985
Indinesia
1990
1995
2000
5 1980
2005
8.95
1985
M aldevs
9.1 9.0
1995
2000
2005
9.5
9.3
9.0
9.2
1990
1995
2000
2005
2000
2005
7.6
7.2 9.1
8.0 6.8
7.5
8.8
8.9
7.0 8.7
1985
1990
1995
2000
2005
6.4
6.5
8.8
6.0 1980
8.7 1980
1985
Pakistan
1990
1995
2000
2005
2000
2005
Srilanka
8.98
8.0
8.96
7.9
8.94 7.8 8.92 7.7 8.90 7.6
8.88 8.86 1980
1985
Nepal
9.0
8.6 1980
8.90 1980
M alyisa
8.5 8.9
1990
1985
1990
1995
2000
2005
7.5 1980
1985
1990
1995
27
1985
1990
1995
2000
2005
6.0 1980
1985
1990
1995
Figure A2: De-trended Energy Intensity of Poor Countries DEN Afg hanistan
Bang lades
1.5
Bhutan
.15
India
2
.15
.10
1.0
.10 1
.05
.05
0.5 .00
0
.00
0.0 -.05 -1
-0.5 -1.0 1980
-.05
-.10
1985
1990
1995
2000
2005
-.15 1980
-.10
1985
Indinesia
1990
1995
2000
-2 1980
2005
1985
Maldevs
.2 .1
1990
1995
2000
2005
.8
.2
.4
.1
.0
.0
-.4
-.1
1990
1995
2000
2005
2000
2005
Nepal .3 .2 .1
-.1
.0
-.2
1985
1990
1995
2000
2005
-.8 1980
1985
Pakistan .2
.04
.1
.00
.0
-.04
-.1
1985
1990
1995
1990
1995
2000
2005
2000
2005
-.2 1980
-.1
1985
1990
1995
2000
2005
-.2 1980
1985
1990
1995
Srilanka
.08
-.08 1980
1985
Malyisa
.0
-.3 1980
-.15 1980
2000
2005
-.2 1980
1985
1990
1995
Table A1 Panel Unit Root Test Results for Poor countries
TEST LLC Fisher- ADF PP- test
ln( Et / Yt ) DT
(ln Yt )
Test value -2.86 30.80 30.12
Test value 5.028 83.30 129.84
p-value 0.002** 0.078* 0.080*
p-value 0.000*** 0.000*** 0.000***
Note: (1) Automatic selection of lags based on minimum AIC: 0-3. (2) ***, **, and * denote rejection of null hypothesis at 1%. 5% and 10% level of significance. Respectively (3) Deterministic components. PP, Breitung, and Fisher-ADF tests: fix cross section effects and Individual trends. IPS: fixed cross section effects.
28
Figure A3: Time cross plots of GDP and Energy/GDP of Poor Countries. Afg hanistan
Bang lades
12,000
1,200
10,000
1,100
8,000
1,000
Bhutan
India
16,000
10,000 9,500 9,000
900
4,000
800
2,000
700
0
600
EI
6,000
EI
EI
EI
12,000
8,000
8,500 8,000
4,000
0
10,000,000 20,000,000 30,000,000
7,500 0 0
PPP
100,000,000
300,000,000
7,000 0
PPP
Indinesia 12,000
8,500
10,000
8,000
3,000,000
0
PPP
Maldevs
9,000
1,000,000
2,000,000,000
6,000,000,000
PPP
Malyisa
Nepal
11,000
2,000
10,000
1,600
8,000
EI
EI
EI
6,000
7,000
EI
9,000
7,500
1,200
8,000 4,000
6,500
5,500
0 0
500,000,000
1,500,000,000
6,000 0
1,000,000
2,000,000
PPP
PPP
Pakistan
Srilanka
7,800
2,600
7,600
2,400
3,000,000
7,400
2,200
7,200
2,000
7,000
1,800 0
200,000,000 400,000,000 600,000,000 PPP
0
400 0
200,000,000 400,000,000 600,000,000 PPP
EI
2,800
EI
8,000
800
7,000
2,000
6,000
50,000,000 100,000,000 150,000,000 PPP
29
0
20,000,000 40,000,000 60,000,000 PPP
Figure. A4: Time cross plots of Population and Energy/GDP of Poor Countries. Afghanistan
Banglades
12,000
1,200
10,000
1,100
8,000
1,000
Bhutan
India
16,000
10,000 9,500 9,000
900
4,000
800
2,000
700
EI
6,000
EI
EI
EI
12,000
8,000
8,500 8,000
4,000 7,500
POP
0 440 480 520 560 600 640 680
16 0, 00 0
40,000
14 0, 00 0
30,000
12 0, 00 0
20,000
10 0, 00 0
600
80 ,0 00
0 10,000
7,000 600,000
POP
800,000
1,000,000 1,200,000
POP
POP
Indinesia
Maldevs
9,000
12,000
8,500
10,000
8,000
Malyisa
10,000
1,600
EI
EI
EI
6,000
EI
9,000
7,000
1,200
8,000 4,000
6,500
160,000
200,000
0 150
240,000
POP
200
250
300
350
POP
Pakistan
2,600
7,600
2,400
EI
7,800
EI
2,800
7,400
2,200
7,200
2,000
120,000
160,000
POP
200,000
6,000 12,000 16,000 20,000 24,000 28,000 POP
Srilanka
8,000
800
7,000
2,000
6,000
7,000 80,000
2,000
8,000
7,500
5,500 120,000
Nepal
11,000
1,800 14,000 16,000 18,000 20,000 22,000 POP
30
400 12,000 16,000 20,000 24,000 28,000 POP
Figure A5: Time cross plots of GDP per capita and Energy/GDP of poor countries.
Afghanistan
Banglades
12,000
1,200
10,000
1,100
8,000
1,000
Bhutan
India
16,000
10,000 9,500 9,000
900
4,000
800
2,000
700
EI
6,000
EI
EI
EI
12,000
8,000
8,500 8,000
4,000
0 200
400
600
7,500
600 500
800
0 1,000
CGDP
1,500
2,000
2,500
7,000 0
2,000
CGDP
Indinesia 12,000
8,500
10,000
8,000
6,000
0
CGDP
Maldevs
9,000
4,000
1,000
2,000
3,000
4,000
CGDP
Malyisa
Nepal
11,000
2,000
10,000
1,600
8,000
EI
EI
EI
7,500 6,000
7,000
EI
9,000 1,200
8,000 4,000
6,500
5,500
0 0
2,000
4,000
6,000
6,000 0
CGDP
2,000
4,000
6,000
8,000
CGDP
Pakistan
2,600
7,600
2,400
EI
7,800
EI
2,800
7,400
2,200
7,200
2,000
7,000
1,800 0
1,000
2,000 CGDP
3,000
4,000
0
2,000
0
5,000
10,000 15,000 20,000 CGDP
Srilanka
8,000
800
7,000
2,000
6,000
4,000
CGDP
31
6,000
400 500
1,000
1,500
CGDP
2,000
