The work of the Shift Project focuses on the link between energy consumption and the GDP growth. The econometrical research confirms the standpoint defended by ecological economists and introducing primary energy as a key factor that drives GDP growth. The results show that an increase of 10% of energy use per capita induces, on average, an increase (resp. decrease) of about 6 to 7% of GDP per capita. The research also concludes that the causality relation goes from the consumption of energy to growth in both the short and long-run. These findings sharply contrast with the custom, popular in macroeconomics, that consists in calibrating the output elasticity of energy according to the cost share of energy. In most countries, this practice leads to the postulate that energy elasticity should be close to 0.08% on average.

1. How Dependent Is Growth From
Primary Energy?
An Empirical Answer on 33 Countries
www.theshiftproject.org
Gael Giraud
CNRS, PSE, CES, Labex REFI
Zeynep Kahraman
The Shift Project

2. Introduction
Why is this relationship important ?
• Mainstream economic models do not include energy as a factor
that could foster economic growth.
• Ecological economists, often ascribe to energy the central role
in economic growth.
• Is energy an important driver of economic growth ?
• If so, what is the magnitude of the dependency of growth from
energy ?

3. Introduction
Why is this relationship important ?
Sources: BP statistical Review, 2012, Shilling et al. 1977, EIA, 2012, and World Bank (GDP), 2012.
0
10000
20000
30000
40000
50000
60000
0 20 40 60 80 100 120
GDPbillionconstant$
Oil Price per barrel in constant 2011 $
World Oil prices and GDP (1965 – 2011)

4. Introduction
Why is this relationship important ?
Source : BP statistical review, 2012, Shilling et al. 1977, EIA, 2012, and World Bank (GDP), 2012.
y = 0.6548x + 0.0103
-4%
-3%
-2%
-1%
0%
1%
2%
3%
4%
5%
-4% -3% -2% -1% 0% 1% 2% 3% 4% 5%
WorldGrossDomesticProductGrowth
Primary Energy Consumption Growth
Comparison of the World Gross Domestic Product growth with the World Primary Energy
Consumption Growth

5. Introduction
Why is this relationship important ?
Source: EIA, http://www.eia.gov/totalenergy/data/annual/pdf/sec1_13.pdf
0%
2%
4%
6%
8%
10%
12%
14%
16%
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
U.S Energy Expenditures as Share of GDP

6. Empirical methodology
Estimation of the long run relation
lnYi,t = βi,0+ βi,1 lnCi,t+ βi,2 lnEi,t-1+ βi,3 lnKi,t+εi,t
*All the variables are per capita
The main equation:

7. Empirical methodology
Estimation of the long run relation
An ECM approach:
∆𝑦it= ϕi𝑦i,t−1 +β′1 𝑐it +β′2 𝑒it−1 +β′3 𝑘it+
𝑗=1
𝑝−1
𝜆∗
ij∆𝑦i,t−j +
𝑗=0
𝑞−1
𝛿 ∗′
1j∆𝑐i,t−j
+
𝑗=1
𝑞−1
𝛿 ∗′
2j∆𝑒i,t−j +
𝑗=0
𝑞−1
𝛿 ∗′
3j∆𝑘i,t−j + μi+ αit+ε
"ϕi" is the error correction term, "βi" is long-run coefficients

8. Empirical methodology
The Data
Variables under scrutiny is:
- Primary energy consumption (million tons of oil equivalents)
- GDP (in 2000 U.S dollars)
- Gross Fixed Capital Formation (in 2000 U.S dollars)
- Population (millions)
World Bank, World
Development Indicators

9. Estimation of the long run relation
The Data
The analysis is based on a panel data covering the period from 1970 to 2011 for
33 countries.
Algeria France Netherlands
Argentina Germany Norway
Australia Greece Philippines
Austria Hungary Portugal
Belgium Iran South Korea
Brazil Ireland Spain
Canada Italy Sweden
Chile Japan Thailand
China Luxembourg United States
Costa Rica Malaysia United Kingdom
Denmark Mexico Venezuela

10. Time series properties of the data
Cross section dependence, Unit Root and Co-integration tests
1. Cross Section Dependence Test of Pesaran
2. Unit Root Tests:
• First Generation:
• Levin, Lin and Chu test
• Breitung
• Im, Pesaran and Shin
• ADF-Fisher
• Philips Perron – Fisher
• Second Generation:
• CIPS test
3. Co-integration Tests:
• Pedroni’s residual co-integration tests
• Westerlund test
common
unit root process
Individual
unit root process

11. Emprical Results
Co-integration tests results
Deterministic intercept and trend No deterministic intercept and trend
Alternative hypothesis: common AR coefs. (within-dimension)
Statistic Prob.
Panel v-Statistic 19.10098 0.0000
Panel rho-Statistic -5.165067 0.0000
Panel PP-Statistic -10.56038 0.0000
Panel ADF-Statistic -9.640764 0.0000
Statistic Prob.
Panel v-Statistic 12.12852 0.0000
Panel rho-Statistic -12.66436 0.0000
Panel PP-Statistic -17.26987 0.0000
Panel ADF-Statistic -16.24284 0.0000
Alternative hypothesis: individual AR coefs. (between-dimension)
Statistic Prob.
Group rho-Statistic -2.675141 0.0037
Group PP-Statistic -9.576716 0.0000
Group ADF-Statistic -8.976859 0.0000
Statistic Prob.
Group rho-Statistic -12.03752 0.0000
Group PP-Statistic -20.42889 0.0000
Group ADF-Statistic -18.09532 0.0000
Pedroni Residual Cointegration Test
Value Z value P value Robust p value
Gt -4.130 -13.580 0.000 0.000
Ga -18.174 -9.531 0.000 0.000
Pt -22.424 -11.338 0.000 0.000
Pa -18.275 -12.740 0.000 0.000
Westerlund panel cointegration test results

12. Emprical Results
Estimation of the long run relation
Model: PMG MG CCEP
Dependent variable: ∆Yit
Energy consumption per capita
(Cit)
0.6543
(0.053)***
0.8083
(0.105)***
0.5195
(0.213)***
Energy efficiency
(Eit-1)
0.5860
(0.064)***
0.8090
(0.164)***
0.5164
(0.214)***
Capital formation per capita
(Kit)
0.1018
(0.016)***
0.0716
(0.016)
0.269
(0.016)***
Convergence coefficient
(Yit-1)
-0.5540
(0.085)***
-0.8433
(0.085)**
-0.5724
(0.214)***
Hausman test p value 0.2304
Results of long-run estimations

13. Emprical Results
Granger Causality
Dependent
Variable
Sources of causation (independent
variables)
Short run Long run
ΔY ΔE ΔC ΔK ECT
ΔY - 10.93** 26.38*** 299.26*** -0.554***
ΔE 1754.6*** - 9526.42*** 8.37** -1.196***
ΔC 4.07 3.20 - 1.59 -0.533
ΔK 5.14 4.90 63.35*** - -0.273***
Panel causality test results

14. Conclusion
• Primary energy is a key factor that drives GDP growth: its long-run output
elasticity evolved around 0.6.
• Capital accumulation has played a minor role compared to energy : long-run
elasticity for capital around 0.2.
• These estimations are also robust to the choice of various sub periods of time
and subsamples of countries.
• There are good reasons to believe that, the output elasticity of energy is
decoupled from its GDP share.
• Our inquiry does not suggest that energy use be the sole first-order factor
driving growth. Efficiency plays a dual, almost comparable role.
• Energy and GDP cointegrate and energy use univocally Granger causes GDP in
the long-run

15. The Shift Project
Redesigning the Economy to Achieve Carbon Transition
Thank you for your attention