1. CO2 Emissions and Income Inequality
By Nicole Grunewald, Stephan Klasen, Inmaculada Martínez-Zarzoso and Chris Muris
Georg-August University of Göttingen

2. Outline Motivation Underlying Theories Related Literature
Data and Empirical Model
Main Results
Conclusions
11/24/2018
Grunewald, Klasen, Martinez-Zarzoso and Muris (2012) CO2 and Income Inequality

3. Motivation
Can we wait for a development path that takes us on a sustainable future? EKC versus IPAT
Who are the winners and the losers of environmental degradation?
Changes in income distribution also affect pollution levels
11/24/2018
Grunewald, Klasen, Martinez-Zarzoso and Muris (2013) CO2 and Income Inequality

4. Underlying Theories Political economy (Boyce, 1994, 1998)
Greater power inequality  more pollution
Aggregation bias (Heerink, Mulato and Bulte, 2001)
Lower inequality  higher emissions
Emulation theory (Vleben, 1919)
Higher inequality  higher emissions
Rate of environmental trade preference: willingness to trade present benefits for future benefits. Higher rate: more weight on the present
11/24/2018
Grunewald, Klasen, Martinez-Zarzoso and Muris (2013) CO2 and Income Inequality

5. Underlying Theories(cont)
Political economy (Boyce, 1994, 1998)
Via cost-benefit analysis and impacts on the rate of environmental time preference (short-run benefits, long-run costs)
Normative Cost-Benefit Rule versus Positive Power-Weighted Social Decision Rule
max i bi
where i = power of ith individual
More unequal distributions of wealth and power tend to yield worse environmental outcomes
Rate of environmental trade preference: willingness to trade present benefits for future benefits. Higher rate: more weight on the present. But In a society characterized by substantial inequalities of wealth and power, these considerations imply that the level of environmental degradation prescribed by BCA may be “too much.”
11/24/2018
Grunewald, Klasen, Martinez-Zarzoso and Muris (2012) CO2 and Income Inequality

6. Underlying Theories (cont)
Political economy (Boyce, 1994, 1998)
Incidence: Social decisions on environmental protection systematically favor those with more wealth and power over those with less
II. Magnitude: More unequal distributions of wealth and power tend to yield worse environmental outcomes
Rate of environmental trade preference: willingness to trade present benefits for future benefits. Higher rate: more weight on the present
11/24/2018
Grunewald, Klasen, Martinez-Zarzoso and Muris (2012) CO2 and Income Inequality

7. Political Economy (Boyce, 1994, 1998) II
Political Economy (Boyce, 1994, 1998) II. Magnitude: Two Types of Inefficiency
losers’ marginal cost
winners’ marginal benefit $
Type-I inefficiency
Type-II inefficiency
E" E* E'
Level of environmentally degrading economic activity
Key: E* = “optimal” level prescribed by cost-benefit analysis; E' = level under the power-weighted social decision rule when winners are more powerful than losers; E" = level under the power-weighted social decision rule when losers are more powerful than winners
11/24/2018
Grunewald, Klasen, Martinez-Zarzoso and Muris (2012) CO2 and Income Inequality

8. Underlying Theories (cont)
Aggregation Bias: For concave EKC at household level Redistribution  increase pollution
Vf(Y)
Vf(Y)
Rate of environmental trade preference: willingness to trade present benefits for future benefits. Higher rate: more weight on the present Y1 Y1 Y2 Y2 Ym Ym
11/24/2018
Grunewald, Klasen, Martinez-Zarzoso and Muris (2012) CO2 and Income Inequality

9. Underlying theories (cont)
Emulation Theory (Vleben, 1919)
Higher inequality  increase conspicuous consumption  higher emissions
But, the rich are believed to have a lower marginal propensity to consume (emit) than the poor! (Ravallion et al., 2000)
Opposite argument from the Comparative propensity to consume theory
Rate of environmental trade preference: willingness to trade present benefits for future benefits. Higher rate: more weight on the present
11/24/2018
Grunewald, Klasen, Martinez-Zarzoso and Muris (2012) CO2 and Income Inequality

10. Related Literature Authors Dependent Variable EKC Other variables
Income Inequality
Sample and model
Torras and Boyce (1998)
Sulfur dioxide, Smoke, Heavy particles etc
Yes
GDP cubed
Literacy
Political rights
Gini (+), low-,
(-),high- income
19-42 countries
OLS
Ravallion et al. (2000)
CO2 per capita
Population and Gini interactions with all variables
Time trend
Gini (+),
Interaction with GDP (+)
42 countries,
Average Gini 1980s,
OLS and FE
Borghesi (2000)
GDP cubed, Population density, Industry share,
Interaction
OLS: Gini (-),
Interaction (+),
FE: Gini not statist. Signinficant
37 countries,
OLS , RE and FE
Heerink et al. (2001) No
Gini (-)
64 countries, 1985
11/24/2018
Grunewald, Klasen, Martinez-Zarzoso and Muris (2012) CO2 and Income Inequality

11. Novelties in this Study
Adjusted Gini index by Grün &Klasen and SWIID by Frederick Solt
This allows us to analyze a larger sample of countries over a longer time period
Gini Dataset
Observations
Countries
Years
WDI
515
142
WIDER
1448
146
Gruen and Klasen
SWIID_net
4403
165
11/24/2018

12. Data description
Unbalanced panel dataset: 165 countries from 1980 to 2008:
CO2 from Oak Ridge Center
GNP from Penn World Tables
GINI from Gruen and Klasen as well as Standardized World Income Inequality Database
Other variables from WDI
11/24/2018
Grunewald, Klasen, Martinez-Zarzoso and Muris (2012) CO2 and Income Inequality

13. Empirical model The EKC model augmented with inequality is given by,
ln denotes natural logs
CO2 denotes CO2 emissions per capita
GDPpc is GDP per capita
Gini is the adjusted Gini index
GDPGini is an interaction term between lnGPDpc and lGini
A composite error term and time dummies are added
11/24/2018
Grunewald, Klasen, Martinez-Zarzoso and Muris (2012) CO2 and Income Inequality

14. Empirical Model Grouped Fixed Effects Estimator (preferred model)
Determines groups endogenously
Controls for time invariant group effects
Allows for time varying group effects
Instrumental variable approach (alternative technique), Instruments:
Corruption
Political orientation
Education level
Labour regulations
11/24/2018
Grunewald, Klasen, Martinez-Zarzoso and Muris (2012) CO2 and Income Inequality

15. Results FE Estimator Dep. Var. lCO2_pc WDI Gruen and Klasen SWIID_net
FE Estimator
Dep. Var. lCO2_pc
WDI
Gruen and Klasen
SWIID_net
lGDP_pc
2.681
2.492**
2.302***
(2.106)
(0.978)
(0.690)
lGDP_pc_sq
-0.120***
-0.126***
(0.0794)
(0.0420)
(0.0302)
lGINI
1.051
-0.788
-1.403*
(1.894)
(1.039)
(0.759)
lGDP_pclGINI
-0.178
0.0611
0.154*
(0.216)
(0.118)
(0.0886)
Constant
-22.93*
-18.67***
-17.63***
(12.76)
(5.766)
(4.143)
Observations
493
1,325
3,966
R-squared
0.309
0.532
0.444
Number of cid
140
139
158
Note:Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1; Year Dummies included
11/24/2018
Grunewald, Klasen, Martinez-Zarzoso and Muris (2012) CO2 and Income Inequality

16. Results Group FE Estimator Nr. of groups 6 5 4 3 2 Dep. Var. lCO2_pc
Group FE Estimator
Nr. of groups 6 5 4 3 2
Dep. Var. lCO2_pc
SWIID_market
lGDP_pc
4.147***
4.133***
3.729***
2.985***
3.296***
(0.252)
(0.251)
(0.345)
(0.494)
(0.668)
lGDP_pc_sq
-0.205***
-0.204***
-0.149***
-0.134***
-0.151***
( )
( )
(0.0107)
(0.0130)
(0.0188)
lGINI
-1.241**
-1.273**
-0.706
-2.092**
-2.030
(0.532)
(0.528)
(0.580)
(0.989)
(1.319)
lGDP_pclGINI
0.127**
0.144
0.124
(0.0615)
(0.0609)
(0.0666)
(0.113)
(0.150)
Iassignment 2
-1.555***
2.177***
-0.699***
0.809**
0.859***
(0.101)
(0.106)
(0.0945)
(0.310)
(0.136)
Iassignment 3
-0.925***
-1.580***
0.528***
-0.779***
(0.0810)
(0.132)
(0.0781)
(0.0814)
Iassignment 4
-0.454***
-0.484***
1.631***
(0.0777)
(0.110)
(0.394)
Iassignment 5
0.0313
-0.951***
(0.0977)
Iassignment 6
2.223***
(0.0387)
Constant
-26.44***
-26.22***
-25.01***
-19.58***
-21.01***
(1.980)
(2.002)
(2.523)
(3.914)
(5.272)
Observations
2,933
R-squared
0.977
0.973
0.966
0.955
0.925
Robust standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1
11/24/2018

17. Results Group FE Estimator Nr. of groups 7 6 5 4 3 Dep. Var. lCO2_pc
Group FE Estimator
Nr. of groups 7 6 5 4 3
Dep. Var. lCO2_pc
SWIID_Net
lGDP_pc
3.729***
3.230***
3.286***
3.602***
2.242***
(0.258)
(0.269)
(0.272)
(0.366)
(0.570)
lGDP_pc_sq
-0.182***
-0.169***
-0.177***
-0.154***
-0.156***
( )
( )
( )
(0.0122)
(0.0157)
lGINInet
-1.237***
-1.994***
-2.203***
-1.071**
-4.033***
(0.379)
(0.432)
(0.426)
(0.524)
(0.907)
GDPGINIn
0.141***
0.221***
0.241***
0.0439
0.419***
(0.0428)
(0.0476)
(0.0474)
(0.0580)
(0.101)
_Iassignmen_2
3.119***
1.520***
-3.601***
0.517***
-1.587***
(0.0810)
(0.127)
(0.104)
(0.0820)
(0.168)
_Iassignmen_3
0.984***
0.616***
-2.966***
-0.626***
-0.749***
(0.112)
(0.0932)
(0.0858)
(0.158)
_Iassignmen_4
0.462***
0.621***
-2.046***
1.598***
(0.0730)
(0.0734)
(0.107)
(0.457)
_Iassignmen_5
-0.636***
3.589***
-2.497***
(0.0945)
(0.103)
(0.0894)
_Iassignmen_6
0.507***
1.074***
(0.0526)
(0.0966)
_Iassignmen_7
(0.0448)
Observations
2,937
R-squared
0.981
0.977
0.973
0.966
0.956
Robust standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1
11/24/2018

18. Results
We find that the relationship between inequality and CO2 per capita varies with GDP per capita:
The elasticities for the minimum income level is
-1% and for the maximum income level it is 0.5%
Once a country passes the threshold of a GDP per capita of 9330 US$ (PPP), redistribution measures will be good for the environment (decrease CO2 per capita)
11/24/2018
Grunewald, Klasen, Martinez-Zarzoso and Muris (2012) CO2 and Income Inequality

19. Results
Above average inequality countries beyond the income threshold are:
Argentina, Austria, Brazil, Chile, Hong Kong, Israel, Kazakhstan, Latvia, Mauritius, Mexico, Panama, Russian Federation, Singapore, South Africa, Trinidad and Tobago, Turkey, United States and Uruguay
11/24/2018
Grunewald, Klasen, Martinez-Zarzoso and Muris (2012) CO2 and Income Inequality

20. Grouped Fixed Effects (5 groups) G1 Blue G2 Red G3 Black G4 Grun G5
Yelow
11/24/2018
11/24/2018
Grunewald, Klasen, Martinez-Zarzoso and Muris (2012) CO2 and Income Inequality
Grunewald, Klasen, Martinez-Zarzoso and Muris (2012) CO2 and Income Inequality 20

21. Group Specific Time Effects1
Australia
Belarus
Bosnia and Herzegovina
Bulgaria
Canada
Czech Republic
Estonia
Guinea-Bissau
Guyana
India
Jordan
Kyrgyz Republic
Luxembourg
Macedonia, FYR
Mauritania
Poland
Russian Federation
Sierra Leone
South Africa
Suriname
Trinidad and Tobago
United States 3
Albania
Botswana
Brazil
Burkina Faso
Cameroon
Cape Verde
Chad
Colombia
Costa Rica
El Salvador
Gabon
Guatemala
Haiti
Lao PDR
Madagascar
Mali
Mauritius
Namibia
Nepal
Panama
Paraguay
Peru
Sri Lanka
St. Lucia
Swaziland
Tanzania
Uganda
Uruguay 4
Algeria
Armenia
Bangladesh
Belgium
Benin
Burundi
Cyprus
Denmark
Egypt, Arab Rep.
Finland
Georgia
Germany
Ghana
Greece
Hungary
Indonesia
Iran, Islamic Rep.
Ireland
Israel
Jamaica
Japan
Korea, Rep.
Malaysia
Mozambique
Netherlands
Nicaragua
Niger
Nigeria
Pakistan
Romania
Senegal
Singapore
Slovak Republic
Slovenia
Tajikistan
Thailand
United Kingdom
Venezuela, RB
Vietnam
Yemen, Rep.
Zambia 5
Angola
Argentina
Austria
Belize
Bolivia
Cambodia
Central African Republic
Chile
Comoros
Cote d'Ivoire
Croatia
Djibouti
Dominican Republic
Ecuador
Ethiopia
Fiji
France
Gambia, The
Guinea
Honduras
Hong Kong, China
Iceland
Italy
Kenya
Latvia
Lithuania
Malawi
Malta
Mexico
Morocco
New Zealand
Norway
Papua New Guinea
Philippines
Portugal
Rwanda
Spain
Sweden
Switzerland
Tunisia
Turkey 2
Azerbaijan
China
Kazakhstan
Moldova
Mongolia
Turkmenistan
Ukraine
Uzbekistan
11/24/2018

22. Main Conclusions
Declining inequality can lead to declining CO2 per capita emissions if a country has passed an average income level of 9330 US$ (PPP)
Some high income inequality countries such as Brazil, Chile, Mexico or South Africa are in that group
Still many other high inequality countries are not in this group such as China or Indonesia, there redistribution policies may lead to increasing emissions
11/24/2018
Grunewald, Klasen, Martinez-Zarzoso and Muris (2012) CO2 and Income Inequality

23. Thank you for your attention
11/24/2018
Grunewald, Klasen, Martinez-Zarzoso and Muris (2012) CO2 and Income Inequality

24. Grouped Fixed Effects (7 groups)6
Sierra Leone 7
Chad
Gabon
Namibia
11/24/2018
11/24/2018
11/24/2018
Grunewald, Klasen, Martinez-Zarzoso and Muris (2012) CO2 and Income Inequality
Grunewald, Klasen, Martinez-Zarzoso and Muris (2012) CO2 and Income Inequality
Grunewald, Klasen, Martinez-Zarzoso and Muris (2012) CO2 and Income Inequality 24 24

25. Descriptive statistics
Variable
Obs
Mean
Std. Dev.
Min
Max
lCO2_pc
4146
-7.87
1.70
-13.47
-4.57
lGDP_pc
4064
8.40
1.25
5.02
11.20
lGDP_pc_sq
72.14
20.95
25.16
125.33
lGINInet
3292
3.60
0.28
2.71
4.27
GDPGINIn
3087
30.89
3.91
20.49
40.97
11/24/2018
Grunewald, Klasen, Martinez-Zarzoso and Muris (2012) CO2 and Income Inequality