Über das Meerschwein

Decomposing Inequalities in CO2-Emissions
Michael Jakob and Nicole Grunewald
Entdeken Project Meeting, Göttingen, 5 April 2011
Motivation
 Distribution of global per-capita CO2-emissions is
highly unequal
 Roughly 1 bn people in industrialized countries are
responsible for about 50% of global emissions
(WEO, 2008)
 However, in recent years growth of global
emissions almost entirely due to DCs (Raupach et
al., 2007)
1971 - 75
Inequality in GDP, energy use, and emissions
90/10: 51.2
80/20: 20.8
2001 - 05
GDP
90/10: 60.8
80/20: 29.8
90/10: 18.6
80/20: 7.1
E
90/10: 10.4
80/20: 5.3
90/10: 41.3
80/20: 17.5
CO2
90/10: 40.8
80/20: 11.2
Previous Research
 Numerous studies have examined convergence of
CO2-emissions (e.g. Strazicich and List, 2003; Lee
and Chang, 2008; Romero-Ávila, 2008)
 Markandya et al. (2006) and Jakob et al. (2010)
relate convergence of CO2-emissions to economic
convergence
 Duro and Padilla (2006) decompose Theil-Index
along Kaya factors (GDP per capita, energy
intensity, carbon intensity)
Research Idea
 Perform analysis similar to Duro and Padilla, but
along different dimensions:
– energy carriers
– economic sectors
 Energy carriers:
– Coal products
– Oil Products
– Natural Gas
 Sectors:
– Industry
– Transport
– Other sectors (residential, service, government)
Decomposition
 Duro and Padilla: Theil-Index T   pi ln
i
c
ci
 Appropriate for decomposition along Kaya factors
(multiplicative), but not energy carriers or sectors
(additive)
 Variance decomposition:
Var (CO2 )  Var (CO2oil  CO2gas  CO2coal ) 
Var (CO2oil )  Var (CO2gas )  Var (CO2coal ) 
3 Variance terms
2Cov(CO2oil , CO2gas )  2Cov(CO2oil , CO2coal ) 
gas
2
2Cov(CO
coal
2
, CO
)
3 Covariance terms
Data
 IEA CO2 emissions from fuel combustion, 2007
edition
 Tentative exploration of the data (1990-2005):
– 70 countries with full information on energy carriers
(comprising >5.3 bn people in 2005)
 Report where emissions are physically generated
– Would have to divide emissions from the power sector to
economic sectors
– Could use information from IEA energy balances
Variance Decomposition
1990
1995
2000
2005
4.82
4.73
5.50
6.93
Var(coal)
10.16
5.03
4.80
5.18
Var(gas)
1.69
1.60
1.78
2.02
Cov(oil, coal)
3.62
2.18
1.86
1.63
Cov(oil, gas)
1.52
1.33
1.61
2.23
Cov(coal, gas)
1.37
0.99
0.79
0.67
Total Variance
29.69
20.36
20.60
23.18
Var(oil)
Share of total variance explained by…
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
Cov(coal, gas)
Cov(oil, gas)
Cov(oil,coal)
Var(gas)
Var(coal)
Var(oil)
1990
1995
2000
2005
Next steps
 Main issue: choice of appropriate metric to
measure inequality in global CO2-emissions
 Variance doesn‘t fulfill Pigou-Dalton requirement
 Most common indices (e.g. Theil, GINI etc.) are
not additively decomposable
 Multi-dimensional indices (e.g. Tsui, 1995) don‘t
seem to be appropriate in this context
 Herfindahl-Index might also be of interest
– Probably doesn‘t fulfill Pigou-Dalton either
Herfindahl-Index
 CO
H   
i  CO
i
2
w
2
2

 CO
  
i 

i ,coal
2
2
 CO
CO
i ,oil
2
w
2
 CO
2
i , gas
2



2
2
 CO

 CO 
 CO

   
   
 3 direct terms
  
i  CO
i  CO
i  CO



CO2i ,coalCO2i ,oil
CO2i ,coalCO2i , gas
CO2i ,oilCO2i , gas
 2
 2
 2
w2
w2
w2
i
i
i
CO2
CO2
CO2
i ,coal
2
w
2
i ,oil
2
w
2
3 interaction terms
i , gas
2
w
2