Who emits most? An analysis of UK households’ CO2 emissions

and their association with socio-economic factors

Milena Büchs & Sylke V. Schnepf

with Nick Bardsley

RSS Workshop, 5 July 2012

ESRC grant RES-000-22-4083

Motivation• Consensus on the need to implement environmental

policies. Less known on the distributional impact of these policies

• There is an emerging literature that examines the role of socio-economic factors (SEF) for emissions – but there is a lack of research comparing the association between SEF and CO2 emissions between different emission areas

– total hh emissions (Baiocchi 2010); – hh emissions from transport (Brand et al 2008,

2010)– direct hh emissions (Fahmy et al 2011)– per capita CO2 emissions (DEFRA 2008)– Per capita GHG emissions (Gough et al. 2011)– CO2 emissions at output area level and 7 OAC

groups (Druckman eta l 2008, 2009) 2

Research question• Which role do household characteristics play for

household CO2 emissions, separately for

– Home energy emissions (gas, electricity)– Transport emissions (motor fuels, public transport,

flights)– Other indirect emissions from food and other

consumption items– Total emissions

• Which areas of emissions should be targeted such that low income / disadvantaged households are least affected?

3

Data recap

• Merged the Expenditure and Food Survey 2006 and 2007 with the Living Cost and Food Survey 2008 and 2009; total household sample size 24,446

• Conversion for expenditure to CO2 emissions used (‘mixed’)

– Home energy & transport emissions: Exploit as much information as possible from LCF/EFS that can be merged with external sources (i.e. external price statistics (home energy, motor fuels); estimated passenger km (public transport) to estimate units of consumption. Apply DEFRA conversion factors to estimate CO2.)

– Indirect emissions: use REAP to estimate CO2/£ expenditure for 56 COICOP consumption categories 4

Structure talk

1 Annual average household emissions by emission area

2 Association of socio-economic factors with emissions

• Household size• Income• Age• Education

3 Which characteristics still matter conditional on income?

– OLS regression results– Quantile regression results 5

1 Annual mean hh CO2 emissions are 21.1 tonnes, with 5.1 t home energy, 5.3 t transport and 10.7 t indirect emissions

6

t% of total

  t

% of

total

Gas 2.511

Indirect he and mf 2.6 12

Electricity 2.1 10 Food 1.5 7Other home energy

0.52

Catering/hotels1.1 5

Total Home energy

5.1 24 Recreation0.8 4

    Clothing 0.7 3

Motor fuels 2.411

Furniture, appliances, tools 0.7 3

Flights 2.0 9 Cars 0.4 2Public transport 1.0 4 Other indirect 3.7 17Total Transport 5.3 25 Total Indirect 10.7 51

2 Association of SES with CO2 emissions

7

The role of household size & compositionAverage % increase in CO2 emissions by each additional household member compared to single adult household

8

Note: all figures significant at 1%. Ns denotes not significant. Results derive from OLS regressions with dependent variable type of emission. Sample size 21920 for total CO2, home energy and indirect emissions, and 18764 for transport. Model fit is 0.36 for total Co2, 0.37 for indirect emissions, 0.14 for home energy and 0.16 for transport

    Total Co2 Home energy

Indirect Transport

Adults   

2nd 90 37 97 131

3rd 29 21 31 35

4th 18 9 18 27

5th + 14 18 16 ns

Children  

1st 18 12 24 ns

2nd 14 12 15 18

3rd ns 10 ns -11

Annual hh CO2 emissions (tonnes) and income deciles

9

1 2 3 4 5 6 7 8 9 100

5

10

15

20

25

30

35

40

Totalin-di-recttransporthome en-ergy

Income deciles, equivalised OECD scale

Annual

hh

CO

2,

tonnes

10th, 50th, 90th CO2 emissions percentiles over income deciles

10

1 2 3 4 5 6 7 8 9 100

10

20

30

40

50

60

70

90th per-centileMedian

10th per-centile

Income deciles, equivalised OECD scale

Annual

hh C

O2,

tonnes

Percentage increase of CO2 emissions if income increases by 1% (log log OLS regression)

11

All coefficients significant at the 1% level; households with 0 emissions in area excluded

 Total CO2

IndirectHome energy

Transport

Income 0.6 0.7 0.3 0.9Constant -1.1 -2.1 -0.2 -4.2

Observations

21914

21914 21914 18761

R-squared 0.50 0.51 0.11 0.28

Change elasticity once focus onCO2 distribution (quantile regressions)

12

0.4

0.5

0.6

0.7

0.8

0.9

1

1.1

0.10.20.30.40.50.60.70.80.90.6

0.7

0.8

0.9

1

1.1

1.2

1.3Total CO2 emissions Indirect emissions

Quantile Quantile

Inco

me

Ela

stic

ity

The role of age

13

18-29

30-39

40-49

50-59

60-69

70-79

80+0

5

10

15

20

25

30

totalidhetrans

Age group

Annual

house

hold

CO

2,

tonnes

Low total CO2

High total CO2l

Low home energy

High home

energyLow

transportHigh

transportLow income 52.6 6.9 39.1 16.7 50.6 7.2High income 4.1 51.8 15.1 35.7 6.7 48.6Age<35 22.5 20.8 32.9 16.0 21.2 24.2Age35-64 17.1 33.5 20.9 30.9 17.2 32.1Age>65 43.6 10.3 27.6 19.5 44.4 10.8High education 9.2 42.6 19.0 32.5 9.1 40.4Low education 35.4 14.3 30.7 19.1 35.0 16.4Rural 19.2 32.5 22.4 32.0 19.9 29.8Urban 26.8 22.5 25.6 22.3 26.5 23.6Workless hh 45.5 12.4 39.5 19.8 44.5 12.5Female head 34.4 16.9 28.9 20.8 34.6 17.4Male head 19.1 30.1 22.5 27.6 19.0 29.8Ethnic 27.3 21.1 27.3 25.9 22.8 23.8

14

Row percent of hh in low (<=25th percentile) and high (75th percentile +) CO2 emission groups by hh characteristic

Note: The table provides row percentages. I.e. 0 values for home energy and transport are included.

3 Log CO2 emissions and socio-economic factors; OLS

15Bold printed coefficients significant at 1 % level, results conditional on household composition

VARIABLES Ln CO2 LN home energy

Ln indirect emissions

Ln transport

Lnincome 0.367 0.122 0.408 0.529Age 0.011 0.014 0.008 0.020age2_100 -0.011 -0.009 -0.009 -0.023Agetop -0.052 0.052 -0.093 -0.158Female hh 0.045 0.052 0.053 -0.033Education 1215

0.051 0.014 0.065 0.090

Education 16 0.063 -0.008 0.087 0.109Workless hh 0.023 0.052 0.019 -0.091Ethnicity -0.057 0.012 -0.135 0.066Rural hh 0.050 0.033 0.037 0.088No vehicle -0.270 -0.044 -0.267 -0.822# bedroom 0.106 0.166 0.089 0.053Constant -0.081 -0.309 -0.926 -2.657Observations 21908 21908 21908 18963R-squared 0.614 0.251 0.624 0.355

Tax burden expressed as proportion of disposable equivalised hh income assuming £100/ tonne CO2 tax

16

1 2 3 4 5 6 7 8 9 100

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18totalindirecttransporthome energy

Decile of Equivalised Household Income (modified OECD scale)

Prop

ortio

n in

com

e fo

r tax

Conclusions• Our research examines the role of socio-economic

factors for different areas of emissions – something that has not yet been directly compared using the same dataset

• Household size impacts differently in areas of transport, energy and indirect emissions. While a second adult living in a household doubles indirect emissions he/she only increases home energy CO2 emissions by 30%

• Surprisingly, high education still significant positive influence even after controlling for income for indirect and transport

• Taxes on home energy are likely to affect disadvantaged households most (including older and workless households)

• Whilst taxes on transport are still regressive overall, they are less regressive than all other forms of taxes. But will hit households in rural areas (even conditional on their income)

17