Carbon Reduction Policy Full

1
JRF programme paper:

Climate change and social justice

The distribution of UK household CO
2

emissions: Interim report

Dr Eldin Fahmy, Joshua Thumim and Vicki White

November 2011

This paper:

presents initial findings from a quantitative study into the likely social
distributional impacts of UK Government energy and climate change
policies;

aims to further the development of socially just and environmentally
effective carbon reduction policies, by:
revealing the distributional consequences of current and possible future
policies to reduce carbon emissions from UK households.
enhancing understanding of these social aspects of climate policy within
energy, climate change and social policy arenas.

The Joseph Rowntree Foundation (JRF) commissioned this paper to
inform its work on climate change and social justice, a programme which
supports the development of sociall y just responses to climate change in
the UK.

ISBN 9781859358955
University of Bristol and Centre for Sustainable Energy

www.jrf.org.uk

2

J RF commissioned this paper to inform its work on climate change and social
justice, a programme which supports the development of socially just
responses to climate change in the UK.

The J oseph Rowntree Foundation has supported this project as part of its
programme of research and innovative development projects, which it hopes
will be of value to policy-makers, practitioners and service users. The facts
presented and views expressed in this report are, however, those of the authors
and not necessarily those of J RF.

J oseph Rowntree Foundation
The Homestead
40 Water End
York YO30 6WP

www.jrf.org.uk

This report, or any other J RF publication, can be downloaded free from the J RF
website (www.jrf.org.uk/publications/).

University of Bristol and Centre for Sustainable Energy

First published 2011 by the J oseph Rowntree Foundation

All rights reserved. Reproduction of this report by photocopying or electronic
means for non-commercial purposes is permitted. Otherwise, no part of this
report may be reproduced, adapted, stored in a retrieval system or transmitted
by any means, electronic, mechanical, photocopying, or otherwise without the
prior written permission of the J oseph Rowntree Foundation.

ISBN: 9781859358955 (pdf)
Ref: 2737

Contact:

J osh Stott (josh.stott@jrf.org.uk)

3

Contents Page
Executive Summary 4

Section 1 Introduction 14

Section 2 Data and methods 15

Section 3 The composition of mean total CO
2
emissions from all sources 16

Income 17
Household type 20
Housing tenure 22
Number of workers in household 24
Employment status of HRP 26
Age of HRP 28
Socio-Economic Group of HRP 30
Settlement type 32
Number of cars in household 34
Domestic fuel type 36
Government Office Region 38

Section 4 Summarising the social distribution of emissions estimated effect
sizes 40

Section 5 Conclusions 43

References 45

Appendices 46

About the authors 60

4

Executive Summary
Introduction and aims
This report presents initial findings from a quantitative study, undertaken by the
Centre for Sustainable Energy and the Universities of Bristol and Oxford, to
explore the likely social distributional impacts of UK Government energy and
climate change policies. Presented here is an analysis of the distribution of
household carbon emissions drawing on a new and comprehensive dataset
developed in phase 1 of the project. As far as we are aware, this is the first
integrated analysis of emissions based entirely and directly on nationally
representative survey data on: consumption of household fuels; private road
travel; public transport usage; and domestic and international aviation. The
analysis provides new evidence and insight into who is responsible for emitting
how much carbon dioxide, and identifies the relative contributions of different
aspects of consumption to household carbon emissions.

The research study Understanding the Social Impacts of UK Climate
Policies aims to further the development of socially just and environmentally
effective carbon reduction policies, by:

revealing the distributional consequences of current and possible future
policies to reduce carbon emissions from UK households;

enhancing understanding of these social aspects of climate policy within
energy, climate change and social policy arenas.

Context
UK Government policies to reduce CO
2
emissions do not impact UK
households uniformly. The majority of these policies are funded by consumers
via energy bills that is, all customers pay a set amount on each unit of energy
consumed. Household characteristics interact with various aspects of the
design, implementation and uptake of such policies to determine the way
individual households, and groups of similar households, benefit. For example
the Feed-in Tariff (FiT) generates a revenue stream for households able to
overcome the capital barriers to taking advantage of the opportunity presented
by the policy; however this revenue is raised from the electricity bills of all
households. Consequently the FIT can be expected to have a regressive
distributional impact across UK households.

The Climate Change Act 2008 created a legally binding target to reduce the
UKs emissions of greenhouse gases (GHGs) to at least 80 per cent below
1990 levels by 2050. The Government set out three carbon budgets for a
phased reduction in emissions to the 2050 target. The Climate Change
Commission (CCC) then recommended that the Government establish a fourth
budget for 20232027 which set a limit of 1,950 MtCO2e (a cut of 50 per cent
on 1990). The Government accepted this ambitious target in May 2011 (DECC
2011: Implementing the Climate Change Act 2008: the Government's proposal
for setting the fourth carbon budget).

These targets reflect the increasingly urgent need to reduce emissions. The UK
carbon reduction policy framework is likely to have to become increasingly
5

aggressive if we are to progress toward these targets. It is therefore essential
that we understand the social distributional impacts of both existing and
proposed policies, so that we can feed this understanding back into the policy
design process. This is a requirement if we are to implement policies which:

(i) minimise (or at the very least, avoid exacerbating) the hardship faced by
vulnerable households;

(ii) are fair, and are seen to be fair: a likely precondition for successful
carbon reduction policies.

Methods
Perhaps surprisingly given the urgency of the situation, there is at present is no
unified dataset representing household carbon emissions from all direct
sources (By direct sources we refer to the emissions of carbon dioxide
associated with the consumption of household fuels and transport energy
services this excludes emissions embodied in the production and distribution
of other goods and services). This research project seeks to address this
fundamental gap. While a detailed report on the research methodology
employed to develop this data is beyond the scope of this paper, the key inputs
required to meet the aims of the project are summarised below.

To understand the distribution of household carbon emissions and model the
likely impacts of Government policies, a quantitative representation is needed
of the following:

Energy use and CO
2
emissions from UK households (covering
household fuels, private road transport, public transport, and aviation).

Housing condition and characteristics, and hence carbon reduction
opportunities.

Access to alternative transport modes and services.

The following data sources have been used to construct these representations:

Category Element Survey dataset Year Source
Energy
and CO
2

Household fuels Living Costs and Food
Survey (LCF) (formerly
The Expenditure and Food
Survey or EFS)
20042007 ONS
Private road
transport
National Travel Survey (NTS) 20022006 DfT
Public transport National Travel Survey 20022006 DfT
Domestic leisure
flights
National Travel Survey 20022006 DfT
International
leisure flights
Air Passenger Survey (APS) 19992008 CAA
Housing condition and English House Condition 2007 CLG
6

characteristics Survey (EHCS)
Transport Mode Accessibility National Travel Survey 20022006 DfT

Energy consumption and emissions are first derived from survey data in each of
the above nationally representative datasets (LCF for household fuels; NTS for
private road travel, public transport and domestic aviation; and the APS for
international air travel). The resulting emissions estimates are then combined
using advanced statistical methods (Multiple imputation using Markov-chain
Monte Carlo see Section 2: Data and Methods) to produce a single synthetic
dataset of household carbon emissions from all sources listed above.

Two distinct phases have been applied in the methodology. The first phase
involves creating an LCF-based dataset of carbon emissions from all the above
sources representative of all households within Britain. This can then be
analysed to further understanding of the social distribution of emissions from all
direct sources across GB households. This initial phase is the focus of the
analysis presented in this paper.

The second phase in the project is to develop an EHCS-based dataset
containing the same suite of household fuel, personal travel and international
air travel carbon emissions data (albeit limited to England only) as described
above (this time imputed from the LCF, NTS and APS respectively), complete
with the detailed housing condition data contained within the EHCS. The latter
is required both for a detailed assessment of household carbon abatement
opportunities, and to model the impacts of energy efficiency and carbon
emission reduction policies. This second phase of research is underway and
will be published in 2012.

The distribution of GB household emissions
The key points to note from the analysis of the distribution of household carbon
emissions by socio-demographic variables are summarised below.

Mean average CO
2
emissions are strongl y correlated with income:
households within the highest equivalised income decile have mean total
CO
2
emissions more than twice that of households within the lowest
equivalised income decile. Emissions from pri vate road travel and
aviation account for a high proportion of this differential: aviation
emissions of the highest income decile are more than six times that of the
lowest income decile.

7

Fig. A: Mean annual total CO
2
emissions from all sources by equivalised
household income decile (metric tons)

Multi-adult households and couples (with or without children) have
significantly higher CO2 emissions on average compared with other
household types. Mean CO2 emissions are lowest in single pensioner
households, who have notabl y low transport-related emissions compared
to other household types.

Fig. B: Mean annual total CO
2
emissions from all sources by household
type (metric tons)

Mean household total CO2 emissions have a parabolic relationship with
age, with a peak in the middle-years (HRP aged 35-60years). This trend in
emissions across life course is likel y to reflect underlying differences in
income and command over resources associated with age, as well as
social differences in household size and composition.

8

Fig. C: Mean annual total CO
2
emissions from all sources by age of HRP
(rescaled)

There is a strong, linear relationship between car ownership and average
household carbon emissions. As expected, the difference in mean total
emissions is mainly attributable to emissions from private vehicles.
However, further anal ysis shows car ownership is a strong predictor of
emissions from other sources (notabl y avi ation and domestic fuel). These
variations in other emissions sources associated with levels of car
ownership is likel y to reflect the indirect impacts of other socio-economic
differences (and especiall y inequalities in equivalised household income)
which are also associated with car ownership, as discussed further
below.

Fig. D: Mean annual CO
2
emissions from all sources by number of cars in
household (metric tons)

9

Anal ysis by domestic heating fuel shows average household emissions
are significantl y higher for households using oil to heat their home and
lowest for electricall y heated households. This can be primarily
associated with the variation in carbon emissions of different household
fuels (i.e. oil is far more carbon intensive), but it is also likely that
domestic fuel type is a proxy for a wide range of socio-economic
inequalities within the population, including income, property size and
type, which will affect patterns of household energy consumption.

Fig. E: Mean annual total CO
2
emissions from all sources by domestic fuel
type (rescaled)

Differences in mean household CO2 emissions by settlement type appear
modest relative to other socio-demographic variations. These differences
are nonetheless significant, with rural household CO2 emissions being
one-fifth higher than urban households. Emissions from domestic fuel
use appear to vary more substantially by settlement type compared to
other social dimensions, with rural household fuel CO2 being around 25
per cent higher than in urban dwellings.

10

Fig. F: Mean annual CO
2
emissions from all sources by settlement type
(England and Wales only) (metric tons)

Estimated effect sizes
Analysis of variance (ANOVA) of mean total household CO
2
emissions
presented in this report has shown significant differences between groups as
defined by: income; tenure; number of workers in the household; employment
status; age; socio-economic group; settlement type; car ownership and
domestic heating fuel.

The estimated effect size that is, the degree to which the different variables
listed above can be said to explain or predict variations in CO2 emissions
from one-way ANOVA is a useful way of summarising the overall pattern of
social variation in household CO
2
emissions. The estimated effect size, which
varies between 0 and 1, describes the proportion of overall variation in CO
2

emissions attributable to each predictor variable (e.g. income, tenure etc.).
Predictor variables with an effect size of 0.1 and above for each CO
2
emissions
source are shown below. With regard to public transport emissions, none of the
predictors analysed here accounts individually for more than 10 per cent of
variation in emissions from this source. Government Office Region and car
ownership appeared the strongest predictors of public transport emissions (both
having an effect size of 0.06).

11

Table A: Estimates of univariate effect size for selected respondent and
household predictors of mean annual CO
2
emissions (predictors with an
effect size of >= 0.1 shown)
Domestic fuel Aviation Private vehicles Total CO2 (All
sources)
Number of bedrooms (0.17) Equiv. household income
(0.15)
Number of cars in household (0.9) Number of cars in
household (0.21)
Household type
(composition) (0.12)
NS-Sec of HRP (0.1) Equiv. household income (0.19) Household type
(0.17)
Household size (number of
occupants) (0.12)
Household type (0.18) Number of bedrooms
(0.15)
Dwelling type (0.1) Tenure status (0.17) Number of workers in
hhld. (0.14)
Number of workers in hhld. (0.16) Household size (0.14)
NS-Sec of HRP (0.15) Equiv. household
income (0.12)
Household size (0.13) NS-Sec of HRP (0.1)
Employment status of HRP (0.12) Tenure status (0.1)
Number of bedrooms (0.11)
Dwelling type (0.11)

Table A above suggests the social patterning of variation in household CO
2

emissions arising from domestic fuel usage is primarily attributable to
differences in the size (number of bedrooms and number of occupants),
composition and physical construction (dwelling type) of the household. By
contrast, only two variables appear to account for over 10 per cent of variation
in household aviation emissions, income appearing the strongest individual
predictor.

The social patterning of private vehicle emissions appears very similar to that
pertaining to total emissions from all sources (although the magnitude of effects
sizes varies somewhat). Level of car ownership not surprisingly shows a strong
association with emissions from private vehicles, but also individually accounts
for over 20 per cent of the variation in total CO
2
emissions from all sources. In
addition, household type (composition), dwelling size (number of bedrooms),
the number of workers in the household, household size (number of
occupants), equivalised income, occupational class (NS-SEC), and housing
tenure appear strong determinants of variations in emissions from all sources
and private vehicles.

The effect sizes shown above are based on univariate analysis only. The earlier
discussion of distributional analysis of emissions by socio-demographics
showed that while some factors do not necessarily appear strong predictors in
themselves (i.e. do not appear in table A above) there are still significant
differences in average household emissions between groups. For example,
rural households have significantly higher emissions on average that their
urban counterparts (Fig. G), but settlement type appears to account for only 1
per cent of the overall variation in total CO
2
emissions. It is likely that a number
of factors influence this phenomenon for example, reliance on more carbon-
intensive heating fuel in rural areas; increased private road travel; and housing
stock characteristics e.g. typically older properties. Thus the effect of settlement
type is diluted in a univariate analysis where such higher-order interactions are
12

not accounted for. Multi-factorial analysis will help in understanding these
interrelated factors.

Key implications of the findings to date
The analysis presented in this report has some important implications in the
context of designing and delivering socially equitable carbon emission reduction
policies, summarised below.

There are significant differences in mean total household emissions along the
following dimensions:

Income;

household type;

Age of Household Reference Person;

Employment status of Household Reference Person;

Dwelling size (number of bedrooms);

Main heating fuel;

Tenure;

Car ownership.

These patterns tend to support the observation that, in general, those groups
most likely to be vulnerable and/or socially excluded tend to have lower than
average CO
2
emissions.

The relative inelasticity of domestic fuel use (cf transport) is likely to mean that
policies which increase the cost of domestic fuels will be regressive.

Domestic fuel accounts for approximately 60 per cent of the total direct
household emissions, but transport (specifically private cars and international
aviation) accounts for the majority of the social variation in these emissions.

The increasingly strong association between household emissions and income
when transport emissions are included means that treating total direct carbon
emissions in an integrated way may offer the best approach to minimising
regressive impacts from carbon reduction policies. Put another way if
household carbon reduction policies addressed all transport emissions as well
as those from household fuel use, there would be far fewer low-income/high-
carbon households, and policies which placed a cost on carbon itself would be
likely to be more progressive.

Although beyond the scope of this report, it would also be interesting to
investigate the use of Minimum Income Standards approaches to equivalising
carbon emissions. This could enable an understanding of the levels of carbon
emissions required to meet acceptable living standards.
13

Next steps
The analysis presented in this report has focused on the distribution of average
CO
2
emissions by different socio-demographics. While ANOVA has shown
significant differences between groups (based on income, etc.), the issue of
variation in emissions within e.g. income groups still needs to be considered.
For example, previous research has suggested a subset of low income
households have above average energy consumption. Further analysis is
required to establish that such policies can be designed to avoid penalising this
minority of vulnerable households who may have higher than average
emissions.

While this report shows clearly discernable and comprehensible patterns (not
least that affluent households have higher consumption lifestyles and thus
higher emissions) what is not clear from this LCF-based analysis is the extent
to which these consumption patterns are driven by necessity over lifestyle
choices. For example, in the context of a householders response to climate
change policies, to what extent do the high emitting rural households have
access to alternative transport measures; opportunities to improve the thermal
efficiency of their home; or to switch to less carbon-intensive energy sources?
Similarly, how much of the apparent low consumption amongst low income
households is actually poverty driven, rather than associated with actual lower
levels of energy need e.g. smaller housing?

The next phase of the project will build on the analysis presented in this paper
to explore:

1. the social distributional impacts of existing and potential new UK
Government climate change policies;

2. the distribution of carbon emission abatement opportunities in English
households;

The work will assess the distributional impacts, and attempt to answer
questions of the following nature:

Feed-in Tariffs: do higher income households tend to benefit at the
expense of lower income households?

The Renewable Heat Incentive (RHI): is the policy more progressive
as a result of being funded from general taxation?

Would a carbon tax be progressive, and is it possible to avoid
penalising low income/high consumers?

Personal carbon trading: are the abatement opportunities sufficient to
allow low income households to benefit from a downstream trading
mechanism?

Green Deal: are the actual energy savings high enough to repay the
finance, and does this vary with income?

14

1. Introduction
While several studies examine the distributional impacts of carbon reduction
policies at an international scale, far fewer have focused on the equity of
distributional impacts within countries. Of these, most consider the impact of a
carbon tax on fuels, rather than the more complex range of policies that already
exist or have been proposed for the UK, and few have investigated the
distributional effects of such policies between households.

In recent years analysts have begun to model the potential distributional
implications for households of carbon taxes, whether direct (Speck, 1999,
Tiezzi, 2005) or indirect (Symons, et al., 1994; Gough, et al., 2011). Other
researchers have developed this approach to examine the potential for
redistribution in mitigating the regressivity of potential carbon reduction policies
in China (Brenner, et al., 2007) and the United States (DeCanio, 2007). Work
by Callan et al. (2008) in Ireland and Dresner and Ekins (2004) in the UK are
especially relevant in examining the potential distributional impacts of carbon
mitigation policies. However, Dresner and Ekins analyses are not based on
housing condition data and as a result focus on the potential for aggregate
carbon savings. Opportunities for modelling the distributional impacts for
individual households were restricted to carbon emissions from household fuel
consumption.

Druckman and J ackson (2007, 2008) use expenditure and 2001 UK census
data to estimate spatially disaggregated models of carbon emissions. The
synthetic estimation approach taken by Druckman and J ackson provides very
useful small-area estimates of carbon emissions, although improvements in the
reliability of small-area models are possible as a result of further model
validation and data harmonisation. Nevertheless, few existing studies have
focused specifically upon mitigation opportunities at a household level in
examining distributional impacts of climate change policies, nor has existing
work sought to encompass the range of sources necessary to understanding
households carbon footprint.

Gough et al.s (2011) innovative work is therefore especially significant in this
respect. These authors analyses model the distribution of total embodied
greenhouse gas emissions using UK Expenditure and Food Survey data. These
authors analyses identify equivalised household income as the main driver of
household emissions along with household composition and employment
status. However, they conclude that the effects of current policy in raising
domestic energy process are highly regressive and call for further research to
model the potential distributional effects of alternative policies such as personal
carbon allowances.

This project builds upon this earlier work. It defines a narrower scope of
emissions excluding carbon embodied in goods and services but uses more
precise approaches to estimating direct emissions from household fuel and
transport energy services. Precision is important in this context, because the
dataset is being developed specifically for the purpose of modelling the social
distribution impacts of UK energy and climate policy, with the objective of
influencing future policy development.
15

Consumption and emissions are all calculated directly in nationally
representative surveys, and these surveys are then combined to create a
synthetic but representative dataset covering emissions from: household fuel
consumption; private road travel; public transport usage; and domestic and
international aviation. In the absence of a comprehensive national carbon
emissions survey, this approach gives us the best available representation of
the distribution of the emissions covered by the work.

The project aims to fill in some key research gaps in advancing our
understanding of the way that CO
2
emissions are distributed across UK
households, and in using these data to inform analysis of the likely social
distributional consequences of possible UK energy and climate policies. In this
interim report we summarise the social distribution of direct household carbon
emissions from private usage of domestic fuel, private vehicles, public transport
and aviation. This report summarises some initial findings from the analysis of
this dataset with regard to the relationship between emissions and selected
household, respondent and area characteristics.

2. Data and methods
In this section we summarise the data and methods used in generating UK
household level CO
2
estimates. Perhaps surprisingly, there remains no single
representative dataset describing the distribution of domestic emissions in the
UK. As a result, it was necessary to create such a dataset synthetically, by
using multiple imputation methods in order to merge data for the 20022008
period drawn from the Expenditure and Food Survey, the English House
Condition Survey, the Annual Passenger Survey, and the National Travel
Survey. In order to do so, extensive harmonisation of sources was first
necessary to ensure that data from different sources use a comparable
measurement framework prior to data merging (see Patsios, et al., 2011).

The resultant dataset comprises observations for 24,207 private households
drawn from the Expenditure and Food Survey (EFS) for the period 20042007.
The data itself comprises original EFS data for the 20042007 period (the host
survey) as well as additional data drawn from a range of donor surveys,
namely the Annual Passenger Survey, and the National Travel Survey. For
each respondent, the resultant dataset comprises the original source data and
a series of estimates based upon multiple imputation. Multiple imputation is a
technique for replacing missing data values with m>1 simulated versions using
the Markov chain Monte Carlo method in order to derive estimates and
confidence intervals which incorporate missing-data uncertainty (see Rubin,
1988; Schafer, 1999; Little and Rubin, 2002).

In most cases the relative efficiency of estimates does not increase
substantially beyond five imputations (Rubin, 1987), and these analyses are
therefore based upon the original data and five separate imputations derived
from multiple imputation. The results described below refer to the pooled
estimates derived by multiple imputation unless stated otherwise. In this interim
report we summarise the social distribution of emissions from all sources with
regard to selected household, respondent and area characteristics as detailed
in Table 1 (below). We also briefly summarise the pattern of association for
16

emissions from aviation, private vehicles, public transport, and domestic fuel
with regard to selected household, respondent and area characteristics. The
estimated marginal means together with their standard errors and 95 per cent
confidence intervals for all emissions sources are tabulated in full in the
Appendix (Tables A1-A5).

Table 1: Summary of CO
2
measurement and analysis variables
Dependent
measures
Domestic fuel emissions
Private vehicle emissions
Public transport emissions
Aviation emissions
Total household CO
2
emissions
Dwelling and
area variables
Government Office Region
Urban-rural classification (England & Wales only)
Dwelling type
Heating fuel type
Number of bedrooms
Household
variables
Household size
Household composition
Number of dependent children in household
Number of workers in household
Net annual disposable household income (decile)
OECD equivalised net annual disposable household
income (decile)
Tenure status
Number of cars in household
Household
Respondent
variables
Age of Household Respondent Person (HRP)
Employment status of Household Respondent Person
NS-Sec of Household Respondent Person
SEG of Household Respondent Person

Section 2 of this report describes the composition of total mean CO
2
emissions
by source and selected household, respondent and area characteristics for the
sample. It also illustrates the estimated marginal means and confidence
intervals for total household CO
2
emissions from all sources. Section 3 of this
report summarises the pattern of association between total direct household
CO
2
emissions by source for selected household, respondent and area
characteristics based upon one-way analysis of variance. The potential
implications of these analyses for policies to mitigate carbon emissions in
Britain are briefly discussed in Section 4.

3. The composition of mean total CO
2
emissions from all
sources
In this section, we describe the composition of total mean direct household CO
2

emissions from all sources amongst resident private households in Britain. To
do so, we first compare total CO
2
utilisation by source for different sample
groups in order to establish which sample groups have the highest and lowest
17

total emissions by source (Fig. 1a to 12a, below). We also consider the extent
to which differences in sample mean total emissions from all sources reflect
underlying differences in the wider GB population. Figures 1b to 12b (below)
illustrate estimated marginal means for total CO
2
emissions from all sources for
different sample sub-groups based upon pooled estimates obtained across all
five multiple imputations data sets. In the interests of clarity of presentation, we
do not illustrate the social distribution of mean total CO
2
emissions arising from
aviation, private vehicles, public transport, and domestic fuel but estimated
marginal means together with their standard errors and associated confidence
intervals for the individual components of total household CO
2
emissions are
tabulated in the Appendix (see Tables A1-A5).

Since we are interested here in the social distribution of emissions (rather than
in their absolute values) these estimates have been rescaled so that the sample
mean equals 1. This makes it easier to compare sample means across groups.
For example, a mean estimate of 1.5 for group x indicates that average
household emissions for this group are 50 per cent higher than for all sample
households. Similarly, a mean estimate of 0.4 for group y indicates average
household emissions for this group are only 40 per cent of those found in the
sample as a whole.

Figures 1b to 11b (below) show sample means along with confidence intervals
for the pooled estimates based on multiple imputation. Confidence intervals
indicate the range of plausible values for group means within the wider
population of interest, in this case all private resident households in Britain. A
95 per cent confidence interval denotes the estimated range of predicted values
for the population mean which can be specified with 95 per cent certainty.
Where the confidence interval for groups x and y do not overlap we can
therefore be at least 95 per cent confident that a real difference in means also
exists in the wider population.

Income
Figure 2.1a (below) describes the distribution of mean CO
2
emissions from all
sources in metric tons by net household disposable income decile. These data
show that mean total emissions from all sources increase with income decile.
For all income deciles, domestic fuel accounts for the majority of mean total
emissions. Nevertheless, the income differential in mean total emissions arises
primarily as a result of increasing aviation and private vehicles associated with
rising disposable incomes. For example, in comparison with the lowest income
decile, mean aviation emissions are three and a half times higher, and mean
private vehicle emissions twice as high, as those for households in the highest
income decile.

18

Fig. 2.1a: Mean annual total CO
2
emissions from all sources by net
household disposable income decile (metric tons)

However, net disposable income is a relatively imprecise measure of
households command over resources since this will also depend upon the size
and composition of households. For example, households comprising two
adults and two children will clearly require a higher level of income in order to
achieve the same command over resources as households comprising a single
working age adult. It is therefore best practice in income measurement to adjust
incomes to take account of differences in household size and composition in
order to provide a more precise estimate of households command over
resources a technique known as income equivalisation (e.g. Canberra Group,
2001).

Figure 2.2a (below) therefore describes the distribution of mean CO
2
emissions
from all sources in metric tons by OECD equivalised net household disposable
income decile. After adjusting for differences in household size and
composition, the relationship between mean CO
2
emissions and household
income is especially strong. A clear, positive and consistent relationship is
evident between equivalised income and total mean CO
2
emissions
emissions increase dramatically with rising household incomes. For example,
households within the highest income decile have mean total CO
2
emissions
which are more than twice as high as households within the lowest decile of the
equivalised income distribution (14.4 tons compared with 6.7 tons). As with net
disposable income (Fig. 2.1a, above), the income differential in mean total CO
2

emissions is mostly accounted for by dramatic differences in emissions from
aviation and private vehicles. Compared with households in the lowest income
decile, households in the highest income decile have aviation emissions which
are more than six times higher and private vehicle emissions which are more
than four times higher.

19

2
emissions from all sources by
equivalised household income decile (metric tons)

Figure 2.2b (below) shows estimated marginal means and confidence intervals
for total emissions from all sources by net disposable equivalised household
income decile. These data confirm that social differences in mean total CO
2

emissions from all sources by equivalised income exist not only for the sample
under consideration but also for the wider population of interest. While little is
known about the social distribution of emission at a household level, these
results confirm the findings of earlier work conducted by Gough et al. (2011)
which suggest a strong and broadly linear relationship between embodied
greenhouse gas emissions and (equivalised) income. These income differences
in emissions are likely to reflect social differences associated with income in
households command over resources and the patterns of consumption
associated with income differences. It is also possible that such trends may
reflect compositional differences between households (for example, associated
with household type) as well as other lifestyle differences (for example,
associated with differences in employment patterns and household work
intensity).

20

Fig. 2.2b: Mean annual CO
2
emissions from all sources by equivalised
household income decile (rescaled)

NOTE: Central estimates are means rescaled to a sample mean of 1. A group mean of
1.5 indicates that household emissions for this sample group are 50 per cent higher than
for the sample as a whole. Error bars show the range of plausible population means at
the 95 per cent confidence interval (95 per cent CI).

Household type
Figure 2.3a (below) describes the distribution of sample mean CO
2
emissions
from all sources in metric tons by household type. These data show that mean
total CO
2
emissions are highest amongst households comprising three or more
adults and amongst couple households (with or without dependent children).
Total mean CO
2
emissions are substantially lower amongst single person
households (single pensioners and single working age people), and to a lesser
extent amongst pensioner couples and single parents with dependent children.
As with income, these differences are largely associated with social variations
in aviation and private vehicle emissions. For example, compared with single
pensioners, households comprising three or more adults have mean total
aviation and private vehicle emissions which are more than five times higher.

21

2
household type (metric tons)

Figure 2.3b (below) shows estimated marginal means and 95 per cent
confidence intervals for total emissions from all sources by household type.
These data confirm that social differences in mean total CO
2
emissions from all
sources by household type exist not only for the sample under consideration
but also for the wider population of interest. Variations in emissions by
household type are likely to reflect variations in the size and age profile of
household and associated differences in levels and patterns of consumption.
To this extent, it would in future be desirable in principle to take account of such
differences in household size and composition in estimating household CO
2

emissions. However, since household type is also known to vary consistently
with income and standard of living, it may also be that these differences reflect
underlying differences in income and income adequacy across households of
different types given the strong association between equivalised income and
emissions reviewed above.

22

Fig. 2.3b: Mean annual total CO
2
emissions from all sources by household
type (rescaled)


Housing tenure
2
emissions
from all sources in metric tons by housing tenure. These data show that mean
total CO
2
emissions are highest amongst mortgage holders and lowest amongst
renters. In comparison with renters, mean total CO
2
emissions amongst
mortgage holders are two thirds higher. Again, these social variations are
primarily a result of differences in mean total emissions from aviation and
private vehicle transport. In comparison with renters, mean totals for mortgage
holders are approximately two and a half times higher for both aviation and
private vehicle emissions.

23

Fig. 2.4a: Mean annual CO
2
emissions from all sources by housing tenure
(metric tons)

confidence intervals for total emissions from all sources by tenure status.
2
emissions from all
sources by housing tenure exist not only for the sample under consideration but
also in most cases for the wider population of interest. Further work is needed
to better understand the underlying drivers of differences in household CO
2

emissions by housing tenure. However, it is likely that such effects reflect both
underlying socio-economic differences between households in their command
over resources, as well as (associated) differences in the size, location and built
structure of dwellings occupied on a rental basis in comparison with the owner-
occupier sector. At the same time it should also be acknowledged that the
private rental sector is itself highly heterogeneous in its social composition and
further disaggregation would be desirable in this respect.

24

2
emissions from all sources by tenure
(rescaled)


Number of workers in household
2
emissions
from all sources in metric tons by the number of workers in the household.
These data show a strong, positive and consistent relationship between mean
total CO
2
emissions from all sources and the number of workers in the
household. In comparison with households where no-one is in paid work, mean
totals CO
2
emissions from all sources for households comprising four or more
workers are approximately two and a half times higher. Again, differences in
mean total emissions from all sources are largely attributable to social
variations in emissions from aviation and private vehicles. Compared with
workless households, mean total emissions from aviation are nearly four times
higher, and mean total emissions from private vehicles are nearly six times
higher for households comprising four or more workers.

25

2
emissions from all sources by number of
workers in household (metric tons)

confidence intervals for total emissions from all sources by the number of
workers resident in the household. These data confirm that social differences in
mean total CO
2
emissions from all sources by number of workers resident in the
household exist not only for the sample under consideration but also for the
wider population of interest. Substantial variations in mean household
emissions according to the number of workers within the household are likely to
reflect differences in household size. To this extent, as noted above, it would in
future be desirable in principle to take account of such differences in household
size and composition in estimating household CO
2
emissions. Nevertheless, it
is also likely that such variation is partly accounted for indirectly by differences
in total household income arising from different levels of participation in paid
work and associated differences in consumption patterns including but by no
means not limited to the emissions directly associated with labour market
participation, for example, as a result of commuting to work.

26

2
workers in household (rescaled)


Employment status of HRP
Figure 2.6a (below) describes the distribution of sample mean total CO
2

emissions from all sources in metric tons by the employment status of the
Household Reference Person (HRP The Household Reference Person
typically responds to household survey interview questions on behalf of the
household as a whole defined. In the EFS the HRP is defined as the
householder with the highest income or the oldest of two or more householders
with the same income). These data show that mean total CO
2
emissions are
highest amongst households where the HRP is in employment and substantially
lower amongst households where the HRP is economically inactive or
unemployed. Again social variations in mean total emissions are primarily
attributable to variation in mean total emissions from private vehicles and to a
lesser extent as a result of aviation emissions. For example, in comparison
with households where the HRP is unemployed or retired, mean total emissions
from private vehicles are at least two and a half times higher for households
where the HRP is a full-time employee.

27

2
emissions from all sources by employment
status of HRP (metric tons)

confidence intervals for total emissions from all sources by the employment
status of the HRP. These data confirm that social differences in mean total CO
2

emissions from all sources by employment status exist not only for the sample
under consideration but also for the wider population of interest. These trends
are likely to reflect the effects of substantial inequalities in household
disposable income arising from different sources of income and therefore
inequalities in command over resources and levels of consumption to which this
gives rise. However, it is likely that such differences are also at least partly
explained by differences in household composition and size associated with
HRP employment status for example in comparing the retired population and
those in full-time employment.

28

2
employment status of HRP (rescaled)


Age of HRP
2

emissions from all sources in metric tons by the age of the HRP. These data
reveal a curvilinear relationship between mean total emissions and the age of
the HRP household mean total emissions rise with age peaking in the middle
years (35-60 years) before declining quite steeply in older age. For example,
mean total emissions for the 45-55 age group are approximately 20 per cent
higher than those for the sample as a whole. In contrast, for those HRPs aged
75 or more mean total emissions are less than 70 per cent of those for the
sample as a whole.

29

2
emissions from all sources by age of HRP
(metric tons)

confidence intervals for total emissions from all sources by the age group of the
HRP. These data confirm that social differences in mean total CO
2
emissions
from all sources by the age of the HRP exist not only for the sample under
consideration but also for the wider population of interest. These trends in
household CO
2
emissions across the life course are likely to reflect underlying
differences in income and command over resources associated with age, with
average (equivalised) incomes for the young and older people typically being
substantially lower than those for people in their middle years. At the same
time, such trends are also likely to reflect social differences in household size
and composition across the life course (for example, associated with child
rearing) and the associated additional consumption of these households.

30

2
emissions from all sources by age of
HRP (rescaled)


Socio-Economic Group of HRP
2

emissions from all sources in metric tons by the Socio-Economic Group (SEG)
of the HRP. These data show that mean total CO
2
emissions vary with the
Socio-Economic Group of the HRP with higher status groups having
substantially higher mean total CO
2
emissions than lower status groups. Again,
these social variations in mean total CO
2
emissions are primarily associated
with social differences in mean total aviation and private vehicle emissions. For
example, compared with semi and unskilled manual workers (SEG DE), mean
total aviation emissions are approximately two and a half times higher, and
private vehicle emissions 50 per cent higher, for higher professional and
managerial groups (SEG A).

31

2
emissions from all sources by Socio-
Economic Group of HRP (metric tons)

confidence intervals for total emissions from all sources by the Socio-Economic
Group (SEG) of the HRP. These data confirm that social differences in mean
total CO
2
emissions from all sources by Socio-Economic Group of the HRP
exist not only for the sample under consideration but also for the wider
population of interest. Occupational class differences in total household
emissions from all sources are likely to reflect underlying inequalities in
(equivalised) household incomes between occupational groups and related
inequalities in command over resources and associated consumption levels.

32

2
emissions from all sources by Socio-
Economic Group of HRP (rescaled)


Settlement type
2

emissions from all sources in metric tons by the settlement type. In comparison
with other social variations in mean total emissions, variation by settlement type
is comparatively modest. Nevertheless, these data show that mean total
emissions from all sources are highest amongst households living in rural
areas (villages, hamlets and isolated locations). Thus, compared with
households in urban areas, households living in rural areas have mean total
emissions which are approximately one fifth higher. In contrast with other social
differences in emissions, social variations in emissions by settlement type are
primarily a consequence of social differences in domestic dwelling emissions.
Households living in rural areas have mean total emissions from domestic fuel
use which are approximately one quarter higher than for households living in
more urban environments.

33

2
emissions from all sources by settlement type
(England and Wales only) (metric tons)

confidence intervals for total emissions from all sources by settlement type.
2
emissions from all
sources by settlement type exist not only for the sample under consideration
but also for the wider population of interest. Differences in total household
emissions from all sources between urban and rural areas are likely to reflect a
wide variety of compositional differences in the populations of urban and rural
areas. However, inequalities in (equivalised) income and associated command
over resources between urban and rural areas are likely to be a key driver. It is
possible that differences in domestic fuel emissions between urban and rural
areas may reflect differences in required heating regimes and climatic
conditions, though it is equally likely that such differences are at least partially
accounted for by differences in dwelling construction and size. Similarly, higher
private vehicle emissions in rural areas are likely to reflect both higher rates of
vehicle ownership and reduced availability of accessible public transport in
many rural communities.

34

2
settlement type (England and Wales only) (rescaled)


Number of cars in household
2

emissions from all sources in metric tons by the number of cars available to the
household. These data reveal a strong and linear relationship between mean
total CO
2
emissions from all sources and the number of cars available to the
household. Mean total CO
2
emissions from all sources rise with rising levels of
car ownership with households owning three or more vehicles having mean
total emissions which are nearly three times those of households who do not
own a private vehicle. As one might expect these social differences in mean
total emissions are mainly attributable to emissions from private vehicles.
Nevertheless, these data also show that car ownership is also associated with
social differences in emissions from other sources, notably aviation and
domestic fuel.

35

2
cars in household (metric tons)

confidence intervals for total emissions from all sources by the number of cars
to which the household has access. These data confirm that social differences
in mean total CO
2
emissions from all sources by levels of household car
ownership exist not only for the sample under consideration but also for the
wider population of interest. As discussed above, variations in private vehicle
emissions are a key driver of variation in total household emissions from all
sources and a direct association between levels of car ownership and private
vehicle emissions is therefore to be expected. However, variation in other
emissions sources associated with levels of car ownership is likely to reflect the
indirect impacts of other socio-economic differences (and especially inequalities
in equivalised household income) which are also associated with car
ownership.

36

Fig. 2.10b: Mean annual aggregate CO
2
number of cars in household (rescaled)


Domestic fuel type
2

emissions from all sources in metric tons by domestic fuel type. These data
show that mean total CO
2
emissions are highest amongst households using oil-
based domestic fuel systems and lowest amongst households using electric-
based heating systems, with mean total CO
2
emissions from all sources being
approximately twice as high amongst the households using oil-based systems
in comparison with those using electric based systems. As might be expected,
these differences are primarily associated with social variations in mean total
CO
2
emissions from domestic fuel itself. Nevertheless, these data also show
that fuel type is also associated with other social differences in emissions from
other sources, notably from aviation and private vehicles.

37

2
emissions from all sources by domestic fuel
type (metric tons)

confidence intervals for total emissions from all sources by domestic fuel type.
2
emissions from all
sources by domestic fuel type exist not only for this sample but also for the
wider population of interest. The direct effects of domestic fuel type upon
household emissions are of course limited to domestic fuel emissions and are
likely to reflect differences in the energy efficiency of different sources and,
more importantly, wider social differences in the size and type of private
dwellings. However, it is also likely that domestic fuel type is a proxy for a wide
range of socio-economic inequalities within the population, including income
inequalities, which affect households command over resources and therefore
their patterns of consumption.

38

2
domestic fuel type (rescaled)


Government Office Region
2
emissions
from all sources in metric tons by Government Office Region. These data show
that there is comparatively little social variation in mean total CO
2
emissions by
region. Nevertheless, there are some interesting regional differences in the
composition of total emissions. For example, compared with most other regions
mean total emissions from public transport and aviation are higher in London
and emissions from private vehicles and domestic fuel are lower.

39

2
emissions from all sources by Government
Office Region (metric tons)

confidence intervals for total emissions from all sources by Government Office
Region. These data confirm that social differences in mean total CO
2
emissions
from all sources by region exist not only for this sample but also for the wider
population of interest. The drivers of regional differences in total household
emissions are complex and likely to reflect the influence of a wide range of
compositional effects with regard to demographic and socio-economic
distinctions within the British population. Considerable further work would be
needed to disentangle these complex effects but it worth noting that in most
cases there is no statistically significant difference in total emissions from all
sources.

40

2
Government Office Region (rescaled)


4. Summarising the social distribution of emissions
estimated effect sizes
Table 2 (below) summarises the pattern of association for each predictor
reviewed in Section 2 with regard to mean annual total CO
2
emissions from all
sources based on analysis of variance (ANOVA) methods. Analysis of variance
is a parametric statistical method which can be used to compare within and
between group variation in CO
2
estimates in order to estimate the extent to
which variation in scores arises as a result of differences between groups (e.g.
by tenure, income decile, etc.) rather than as a result of within group
differences. The resulting test statistic (F ratio) offers a formal test of the null
hypothesis of no between-group difference within the wider population.

In this section we report estimated effect sizes (partial eta squared) derived
from one-way analysis of variance. The estimated effect size varies between 0
and 1 and describes the proportion of total variation in CO
2
emissions scores
attributable to the predictor variable. As such it is useful way of summarising the
overall pattern of social variation in our measures of interest attributable to a
range of household, respondent and area characteristics as detailed in Table 2
(below).

41

Table 2: Estimates of univariate effect size for selected respondent and
household predictors of mean annual CO
2
emissions
Predictor Aviation
Private
vehicles
Public
transport
Domestic
fuel
ALL
SOURCES
Number of cars in
household .08 .90 .06 .09 .21
Household type .06 .18 .04 .12 .17
Number of bedrooms .04 .11 <.01 .17 .15
Household size .02 .13 .02 .12 .14
Number of workers in hhld. .08 .16 .05 .06 .14
Equiv. household income .15 .19 <.01 .05 .12
NS-Sec of HRP .10 .15 .03 .04 .10
Tenure status .05 .17 .03 .06 .10
Dwelling type .02 .11 .03 .10 .09
Employment status of
HRP .07 .12 .03 .02 .07
Age of HRP (banded) .04 .08 .03 .04 .06
Heating fuel type .01 .03 .01 .04 .04
SEG of HRP .04 .04 <.01 .02 .03
Number of dep. children <.01 .02 .01 .04 .03
Net household income .09 .05 <.01 <.01 .02
Settlement type <.01 .02 .03 .01 .01
Government Office Region .01 .02 .06 <.01 <.01
NOTE: Effect estimates are one-way ANOVA partial eta squared values (based upon data iteration 1).
Estimates vary between 0 and 1 and indicate the proportion of total variation in total CO2 emissions
attributable to the predictor variable. All effects are significant at the 95 per cent confidence level. All
analyses performed on the natural log of the dependent variables.
For ease of interpretation, effect sizes greater than or equal to 0.1 are
highlighted in Table 2 (above) in order to indicate those variables which have
the strongest direct association with variation in total CO
2
emissions. In effect,
(and leaving aside the influence of other variables) an effect size of 0.1
indicates that 10 per cent of the variation in our dependent measure can be
explained (or accounted for) by the predictor of interest. Considering firstly
total CO
2
emissions from all sources (data column 5), levels of car ownership,
household type, dwelling size (number of bedrooms), household size (number
of occupants), the number of workers in the household, equivalised income,
occupational class (NS-SEC), and housing tenure appear to be the most
important determinants of variations in emissions from all sources. Each of
these variables accounts individually for between 10 per cent and 21 per cent of
the variation in total emissions from all sources.

Considering the individual components of total household CO
2
emissions (data
columns 1-4), it is evident that the extent and nature of social patterning of
emissions varies considerably by source. With regard to public transport
emissions (data column 3), none of the predictors analysed here accounts
individually for more than 10 per cent of the variation in total household
emissions from this source, though levels of car ownership and the number of
workers in a household account for 6 per cent and 5 per cent respectively of the
total variance in public transport emissions.
42

With regard to private vehicle emissions (data column 2), the social structure
of emissions is very similar to that pertaining to total emissions though the
magnitude of effects sizes varies somewhat. Again, household type, dwelling
size (number of bedrooms), household size, the number of workers in the
household, equivalised income, occupational class (NS-SEC), housing tenure
along with dwelling type and HRP employment status appear to be the primary
determinants of variations in household emissions attributable to private
vehicles. Naturally, we would also expect to find a very strong association
between levels of household car ownership and household emissions
attributable to private vehicles.

The social patterning of households emissions attributable to aviation is in
some respects similar to that associated with total household CO
2
emissions
and emissions attributable to private vehicles. However, in most cases the
pattern of association appears to be considerably weaker. With the exceptions
of equivalised income and occupational class (NS-SEC) none of the predictors
analysed here individually account for more than 10 per cent of the total
variance in household aviation emissions. Nevertheless, equivalised household
income and occupational class (NS-SEC) explain respectively 15 per cent and
10 per cent of the total variance in households emissions from aviation (for
non-business purposes).

As one might expect, the social patterning of variation in households CO
2

emissions arising from domestic fuel usage is primarily attributable to
differences in the size and composition of households, and the size and
construction of dwellings themselves. Dwelling size (number of bedrooms),
household size and type, and dwelling type each individually account for at
least 10 per cent of the total variance in households domestic fuel emissions.
In comparison, other socio-demographic and area-based differences within the
population appear to explain relatively little of the variation in household CO
2

emissions associated with domestic fuel use.

In summary, it is evident that a wide range of demographic and socio-economic
characteristics are associated with variation in total household CO
2
emissions.
In general terms, these reflect underlying differences in households size and
structure, their social status and command over resources, and (associated
with these) differences in the size and built structure of the dwellings
households occupy. As one would expect given the implications of dwelling size
and structure for thermal efficiency and heat requirement, CO
2
utilisation as a
result of domestic fuel use is most strongly associated with the size and built
structure of dwellings. In particular, it is noticeable that regional differences in
household emissions attributable to domestic fuel usage, and those associated
with settlement type, are trivial in comparison. The findings suggest that
variation in emissions associated with domestic fuel use is driven by socio-
economic and demographic factors rather than reflecting climatic conditions
and the physical environment.

In comparison, emissions from aviation and private vehicle use reflect the
influence of a much wider range of socio-economic factors associated with
social status and command over resources (i.e. income-related). This is to be
expected since consumption patterns relating to aviation and private vehicle
use (and therefore also the associated CO
2
utilisation) are inherently more
43

elastic with regard to the resources available to households in comparison with
emissions associated with domestic fuel. Since the above analyses describe
only the direct relationship between each predictor and the outcome of interest
it is also likely that many of the predictors of interest will overlap in their
effects. For example, it may be that the strong association between private
vehicle emissions and household type may be attributable to other socio-
economic differences between households of different types.

Finally, there is little evidence of any substantial social patterning of household
emissions attributable to public transport. This is not entirely surprising since
the availability of adequate public transport is likely to be a primary driver of
trends in usage and this is largely independent of the characteristics of
households themselves. The relationship between region and public transport
usage is a case in point here since much of the explanatory power of region in
predicting public transport emissions is likely to be an effect of the elevated
levels of emissions from this source in London. Unsurprisingly, as the above
analyses show, the volume of households public transport emissions is
inversely proportional to levels of household car ownership. That these
emissions are also positively associated with the number of workers in a
household is likely to be at least partly attributable to the commuting
requirements of these households.

5. Conclusions
The findings described above demonstrate the extent to which households
domestic CO
2
emissions reflect the influence of a wide range of demographic
and socio-economic distinctions. Many of these phenomena are of course
highly complex and inter-correlated and further work using multivariate
modelling approaches will therefore be needed in order to disentangle these
effects and to better understand the underlying drivers of social variation in
households domestic CO
2
emissions. As a result it would be premature at this
stage to seek to draw any firm conclusions regarding the potential distributional
impacts of proposed or possible carbon mitigation policies.

However, a number of observations are warranted at this stage which may
potentially have a bearing on future carbon mitigation policies. Firstly, while
domestic fuel emissions accounts for the majority (approximately 60 per cent)
of total household emissions from all sources, as described above much of the
social variation in households domestic carbon emissions arises from private
vehicle and (to a lesser extent) aviation emissions. Social variation in domestic
energy consumption accounts for rather less of the total variation in household
emissions. Emissions associated with public transport usage are also negligible
in comparison with these sources and the social patterning of emissions from
this source is much less pronounced.

Further improvements in domestic energy efficiency and reduced energy
consumption will depend to a significant extent upon relatively long term
improvements in housing stock. Moreover, since the energy demand
associated with domestic fuel use is relatively inelastic with regard to
demographic and socio-economic differences within the population (at least in
comparison with aviation and private vehicle usage), the distributional
44

consequences of current energy policies focused upon raising domestic energy
prices are likely to be highly regressive. More imaginative policy solutions will
certainly be required to ensure that the required reductions in CO
2
emissions
associated with domestic fuel use are delivered in ways which are socially
equitable. More emphasis will also need to be placed on carbon mitigation
policies associated with aviation and private vehicle usage if reductions in total
carbon emissions are to be achieved in ways which do not exacerbate existing
socio-economic inequalities.

Secondly, the above analyses demonstrate the centrality of demographic and
socio-economic distinctions in accounting for variation in household CO
2

emissions. In comparison, geo-spatial variations in households domestic
emissions are relatively trivial. However, any assessment of the distributive
equity of current patterns of household energy consumption and associated
emissions clearly needs to take account of variations in household size and
composition. Determining the energy and associated CO
2
emissions required
by households of different types in order to meet their material and social needs
should therefore be a priority for future research in the area.

Thirdly, these data demonstrate the centrality of socio-economic inequalities,
including income inequalities, in determining households command over
resources and therefore their patterns of consumption and associated
emissions. In general terms, these data suggest that substantial reductions in
carbon emissions could be achieved by reducing energy consumption amongst
those groups currently over-consuming relative to the population as a whole.
Such a strategy would also appear to be consistent with the pursuit of social
equity in the distribution and utilisation of societal resources. However, as noted
above the socio-economic and demographic correlates of household carbon
emissions are often inter-correlated and therefore require much more detailed
analysis of the potential distributive impacts of potential carbon mitigation
policies at the household level given the potential for complex, non-linear
effects and interactions within these data. Phase Two of this project will
therefore seek to provide a detailed assessment of the distributional
consequences of existing and potential carbon mitigation policies for
households in different circumstances.
45

6. References
Brenner, M., Riddle, M. and Boyce, J . (2007) A Chinese sky trust? Distributional
impacts of carbon charges and revenue recycling in China. Energy Policy,
35(3): 1771-1784
Callan, T., Lyons, S., Scott, S., Tol, R. and Verde, R. (2008) The distributional
implications of a carbon tax in Ireland, ESRI Working Paper 250
Canberra Group (2001) United Nations Expert Group on Household Income
Statistics, 2001: Final Report and Recommendations. Ottawa: Canberra Group
DeCanio, S., (2007) Distribution of emissions allowances as an opportunity.
Climate Policy, 7(2): 91-103
Dresner, S. and Ekins, P. (2004) Economic Instruments for a Socially Neutral
National Home Energy Efficiency Programme. Policy Studies Institute,
Research discussion paper 19
Druckman, A. and J ackson, T. (2007). The Local Area Resource Analysis
(LARA) Model: Concepts, Methodology and Applications. RESOLVE Working
Paper Series 02-07, University of Surrey
http://www.surrey.ac.uk/resolve/Docs/WorkingPapers/RESOLVE_WP_02-
07.pdf
Druckman, A. and J ackson, T. (2008) Household energy consumption in the
UK: A highly geographically and socio-economically disaggregated model.
Energy Policy, 36(8): 3177-3192
Gough, I., Abdallah, S., J ohnson, V., Ryan-Collins, J . and Smith, C. (2011) The
distribution of total greenhouse gas emissions by households in the UK, and
some implications for social policy. CASE Working Paper152. London: London
School of Economics. http://sticerd.lse.ac.uk/dps/case/cp/CASEpaper152.pdf
Little, R. and Rubin, D. (2002) Statistical Analysis with Missing Data. (2
nd
Ed.)
New J ersey: J ohn Wiley and Sons
Patsios, D., White, V. and Brand, C. (2011) Summary Report on Survey
harmonisation. JRF Understanding the social impacts of UK climate policies
project paper
Rubin, D. (1988) Multiple Imputation for Non-response in Surveys. New J ersey:
J ohn Wiley and Sons
Schafer, J . (1999) Multiple Imputation: A primer. Statistical Methods in Medical
Research, 8(3): 3-15
Speck, S. (1999) Energy and carbon taxes and their distributional implications.
Energy Policy, 27(11): 659-667
Symons, E., Proops, J . and Gay, P. (1994) Carbon Taxes, Consumer Demand
and Carbon Dioxide Emissions: A Simulation Analysis for the UK. Fiscal
Studies, 15(2): 1943
Tiezzi, S. (2005) The welfare effects and the distributive impact of carbon
taxation on Italian households. Energy Policy, 33(11): 1597-1612.
46

7. Appendices
Table A1: The distribution of total household CO
2
emissions from all
sources means, standard errors and confidence intervals (metric tons)

Mean SE 95%CI lo 95%CI hi
OECD
equivilised net
household
disposable
income
1 .674 .020 .633 .714
2 .679 .019 .641 .718
3 .789 .021 .747 .831
4 .866 .025 .811 .922
5 .956 .025 .901 1.012
6 1.033 .018 .997 1.069
7 1.133 .020 1.091 1.175
8 1.220 .031 1.147 1.293
9 1.275 .033 1.196 1.353
10 1.440 .026 1.384 1.496
Net household
disposable
income
1 .847 .022 .801 .892
2 .906 .023 .858 .954
3 .919 .018 .883 .954
4 .932 .024 .882 .983
5 .994 .018 .958 1.029
6 1.016 .023 .968 1.064
7 1.026 .019 .989 1.063
8 1.084 .022 1.037 1.130
9 1.111 .028 1.049 1.173
10 1.224 .025 1.170 1.279
Number of
children
None .928 .008 .912 .945
One 1.121 .021 1.077 1.166
Two 1.221 .016 1.190 1.253
Three 1.210 .041 1.120 1.300
Four or more 1.183 .077 1.011 1.355
Household size One person .641 .011 .619 .663
Two persons 1.020 .010 .999 1.041
Three persons 1.167 .016 1.134 1.200
Four persons 1.325 .022 1.277 1.372
Five persons 1.337 .035 1.261 1.412
Six or more persons 1.425 .080 1.233 1.617
Age of HRP Under 25 .785 .045 .685 .885
25-35 .979 .022 .930 1.028
35-45 1.133 .015 1.102 1.164
45-55 1.194 .020 1.149 1.239
55-60 1.120 .023 1.072 1.168
60-65 1.019 .023 .972 1.065
65-70 .856 .016 .825 .887
75+ .665 .016 .634 .695
Household type Single pensioner .597 .016 .565 .629
Pensioner couple .906 .019 .868 .944
Single working age adult .687 .017 .653 .720
Couple, no dependent children 1.112 .012 1.087 1.136
Couple with dependent children 1.238 .020 1.193 1.284
Single parent with dependent children .786 .023 .741 .832
Three or more adults 1.416 .027 1.354 1.478

47

2
emissions from all
(continued)

Number of
bedrooms in
dwelling
One .556 .019 .519 .593
Two .810 .012 .787 .833
Three 1.043 .010 1.022 1.063
Four 1.359 .024 1.303 1.415
Five or more 1.700 .033 1.633 1.768
Accommodation
type
Detached 1.280 .023 1.225 1.336
Semi-detached 1.042 .013 1.014 1.070
Terraced .923 .015 .890 .956
Flat .672 .023 .620 .724
Other .879 .029 .820 .939
Tenure Owned Outright .983 .014 .953 1.013
Mortgage 1.221 .014 1.189 1.253
Renting .734 .011 .712 .756
Other .934 .044 .848 1.020
Number of cars
in household
None .575 .012 .550 .600
One .947 .009 .929 .965
Two 1.382 .018 1.340 1.424
Three or more 1.616 .049 1.497 1.736
Domestic fuel
type
Gas 1.032 .008 1.015 1.049
Electric .747 .016 .715 .779
Oil 1.314 .035 1.242 1.386
Other .923 .053 .809 1.037
Government
Office Region
N East .913 .027 .861 .966
N West & Mersey .999 .025 .945 1.053
Yorks and Humber .994 .023 .948 1.041
E Midlands 1.021 .028 .960 1.081
W Midlands 1.009 .025 .957 1.062
Eastern 1.043 .025 .990 1.097
London .984 .024 .933 1.035
S East 1.045 .016 1.013 1.078
S West .966 .029 .902 1.031
Wales .946 .029 .886 1.006
Scotland 1.032 .021 .989 1.076
Settlement type Urban >10k .978 .008 .961 .994
Town and fringe .996 .019 .960 1.033
Village, hamlet & isolated 1.178 .023 1.130 1.226
Number of
(paid) workers
in household
None .710 .010 .691 .729
One .918 .014 .890 .947
Two 1.251 .020 1.204 1.298
Three 1.479 .035 1.403 1.556
Four or more 1.768 .078 1.580 1.956
Employment
status of HRP
Self-employed 1.118 .032 1.046 1.190
FT employee 1.202 .012 1.176 1.229
PT employee 1.098 .021 1.054 1.142
ILO unemployed .796 .047 .698 .893
Retired .751 .011 .729 .772
Other econ inactive .853 .016 .820 .886

48

2
emissions from all
(continued)

NS-SEC of HRP Large employers/higher managerial 1.452 .038 1.369 1.535
Higher professionals 1.326 .029 1.264 1.388
Lower managerial/professionals 1.254 .026 1.192 1.315
Intermediate 1.038 .043 .936 1.140
Small employers/own account workers 1.071 .029 1.011 1.131
Lower supervisory/technical 1.092 .029 1.030 1.154
Semi-Routine .966 .028 .906 1.026
Routine .972 .024 .925 1.018
HRP retired .751 .011 .729 .772
Other econ inactive .788 .022 .743 .833
Not classifiable .770 .036 .697 .843
Socio-Economic
Group of HRP
A 1.342 .041 1.253 1.431
B 1.284 .018 1.244 1.324
C1 1.054 .033 .981 1.127
C2 1.079 .022 1.032 1.127
DE .947 .024 .899 .994

49

2
emissions from aviation
means, standard errors and confidence intervals (metric tons)

OECD
equivilised net
household
disposable
income
1 0.47 0.09 0.28 0.65
2 0.42 0.07 0.28 0.55
3 0.59 0.07 0.46 0.72
4 0.75 0.07 0.61 0.88
5 0.82 0.07 0.69 0.95
6 1.13 0.08 0.96 1.31
7 1.39 0.08 1.23 1.56
8 1.72 0.10 1.49 1.94
9 1.99 0.12 1.71 2.28
10 2.94 0.12 2.67 3.21
Net household
disposable
income
1 0.68 0.08 0.50 0.85
2 0.71 0.07 0.57 0.84
3 0.77 0.08 0.61 0.93
4 0.96 0.09 0.78 1.15
5 1.00 0.07 0.86 1.14
6 1.31 0.08 1.15 1.47
7 1.17 0.09 0.98 1.36
8 1.44 0.10 1.22 1.66
9 1.74 0.11 1.50 1.98
10 2.44 0.09 2.24 2.65
Number of
children
None 1.13 0.04 1.05 1.21
One 1.31 0.06 1.20 1.42
Two 1.55 0.07 1.41 1.69
Three 1.31 0.13 1.05 1.57
Four or more 1.13 0.29 0.49 1.77
Household size One person 0.53 0.04 0.45 0.61
Two persons 1.37 0.04 1.29 1.45
Three persons 1.42 0.07 1.27 1.57
Four persons 1.75 0.06 1.63 1.86
Five persons 1.58 0.12 1.32 1.85
Six or more persons 1.50 0.21 1.05 1.94
Age of HRP Under 25 0.80 0.13 0.54 1.07
25-35 1.29 0.09 1.08 1.50
35-45 1.49 0.06 1.38 1.61
45-55 1.48 0.08 1.31 1.64
55-60 1.44 0.08 1.29 1.59
60-65 1.45 0.10 1.23 1.67
65-70 0.96 0.06 0.85 1.07
75+ 0.34 0.07 0.20 0.47
Household type Single pensioner 0.34 0.05 0.23 0.44
Pensioner couple 1.02 0.07 0.87 1.16
Single working age adult 0.73 0.06 0.62 0.85
Couple, no dependent children 1.64 0.05 1.53 1.74
Couple with dependent children 1.55 0.05 1.46 1.65
Single parent with dependent children 0.63 0.09 0.46 0.79
Three or more adults 1.84 0.10 1.63 2.06

50

2
(continued)

Number of
bedrooms in
dwelling
One 0.54 0.07 0.40 0.68
Two 0.85 0.04 0.77 0.92
Three 1.18 0.03 1.11 1.25
Four 2.05 0.08 1.88 2.22
Five or more 2.79 0.23 2.24 3.33
Accommodation
type
Detached 1.82 0.05 1.71 1.93
Semi-detached 1.18 0.05 1.07 1.28
Terraced 0.98 0.05 0.88 1.08
Flat 0.79 0.05 0.69 0.90
Other 1.15 0.09 0.97 1.33
Tenure Owned Outright 1.21 0.07 1.06 1.37
Mortgage 1.66 0.05 1.53 1.78
Renting 0.62 0.04 0.55 0.70
Other 0.91 0.20 0.49 1.33
Number of cars
in household
None 0.44 0.05 0.34 0.54
One 1.10 0.04 1.02 1.18
Two 1.92 0.09 1.71 2.12
Three or more 2.31 0.13 2.02 2.61
Domestic fuel
type
Gas 1.29 0.03 1.22 1.36
Electric 0.67 0.06 0.55 0.80
Oil 1.51 0.11 1.29 1.73
Other 1.04 0.19 0.64 1.43
Government
Office Region
N East 0.73 0.10 0.53 0.92
N West & Mersey 1.23 0.07 1.10 1.36
Yorks and Humber 1.09 0.08 0.94 1.25
E Midlands 1.27 0.09 1.09 1.45
W Midlands 1.07 0.11 0.83 1.32
Eastern 1.14 0.07 1.01 1.27
London 1.56 0.08 1.38 1.74
S East 1.46 0.09 1.26 1.65
S West 1.06 0.08 0.90 1.22
Wales 0.73 0.10 0.53 0.92
Scotland 1.33 0.08 1.16 1.50
Settlement type Urban >10k 1.19 0.03 1.12 1.25
Town and fringe 1.13 0.08 0.97 1.29
Village, hamlet & isolated 1.37 0.07 1.23 1.52
Number of
(paid) workers
in household
None 0.58 0.04 0.49 0.67
One 1.11 0.05 1.01 1.21
Two 1.78 0.06 1.64 1.92
Three 1.95 0.14 1.65 2.26
Four or more 2.03 0.24 1.47 2.58
Employment
status of HRP
Self-employed 1.69 0.10 1.49 1.89
FT employee 1.64 0.05 1.53 1.76
PT employee 1.24 0.07 1.10 1.39
ILO unemployed 0.70 0.15 0.40 1.01
Retired 0.64 0.04 0.55 0.73
Other econ inactive 0.94 0.07 0.80 1.09

51

2
(continued)

NS-SEC of HRP Large employers/higher managerial 2.38 0.12 2.14 2.61
Higher professionals 2.23 0.11 2.00 2.46
Lower managerial/professionals 1.79 0.07 1.64 1.94
Intermediate 1.24 0.10 1.05 1.43
Small employers/own account workers 1.56 0.08 1.40 1.73
Lower supervisory/technical 1.18 0.08 1.03 1.34
Semi-Routine 0.90 0.08 0.74 1.07
Routine 0.94 0.09 0.75 1.13
HRP retired 0.64 0.04 0.55 0.73
Not classifiable 0.90 0.15 0.60 1.20
Socio-Economic
Group of HRP
A 2.27 0.19 1.84 2.71
B 1.91 0.06 1.78 2.04
C1 1.31 0.08 1.14 1.48
C2 1.28 0.06 1.17 1.40
DE 0.88 0.08 0.71 1.04

52

2
emissions from private
vehicles means, standard errors and confidence intervals (metric tons)

OECD
equivilised net
household
disposable
income
1 0.95 0.15 0.64 1.26
2 1.09 0.14 0.82 1.36
3 1.69 0.18 1.32 2.07
4 2.08 0.20 1.63 2.52
5 2.66 0.22 2.17 3.14
6 3.01 0.16 2.68 3.34
7 3.44 0.19 3.02 3.85
8 3.81 0.20 3.38 4.25
9 4.03 0.21 3.56 4.49
10 3.92 0.18 3.54 4.31
Net household
disposable
income
1 1.59 0.16 1.27 1.90
2 2.14 0.18 1.76 2.51
3 2.39 0.15 2.09 2.69
4 2.35 0.20 1.90 2.80
5 2.74 0.14 2.46 3.01
6 2.76 0.18 2.39 3.13
7 3.04 0.15 2.75 3.34
8 3.20 0.15 2.90 3.51
9 3.24 0.19 2.83 3.65
10 3.21 0.22 2.70 3.71
Number of
children
None 2.36 0.06 2.25 2.48
One 3.19 0.14 2.90 3.47
Two 3.57 0.13 3.32 3.82
Three 3.22 0.39 2.33 4.11
Four or more 2.57 0.45 1.64 3.50
Two persons 2.71 0.08 2.56 2.87
Three persons 3.46 0.12 3.20 3.71
Four persons 4.04 0.18 3.63 4.44
Five persons 3.73 0.30 3.05 4.40
Age of HRP Under 25 1.82 0.35 1.05 2.59
25-35 2.89 0.16 2.53 3.24
35-45 3.24 0.13 2.98 3.51
45-55 3.47 0.15 3.13 3.81
55-60 3.20 0.23 2.70 3.71
60-65 2.52 0.20 2.11 2.93
65-70 1.82 0.12 1.58 2.06
75+ 1.05 0.13 0.79 1.30

53

2
(continued)

Number of
bedrooms in
dwelling
One 0.98 0.15 0.68 1.27
Two 2.04 0.10 1.85 2.24
Three 2.88 0.08 2.70 3.06
Four 3.80 0.22 3.29 4.31
Five or more 4.17 0.25 3.67 4.66
Accommodation
type
Detached 3.56 0.20 3.07 4.05
Semi-detached 2.96 0.10 2.75 3.17
Terraced 2.38 0.13 2.08 2.69
Flat 1.27 0.17 0.91 1.64
Other 2.01 0.25 1.48 2.55
Mortgage 3.76 0.08 3.58 3.93
Renting 1.44 0.10 1.24 1.64
Other 2.25 0.42 1.38 3.13
Number of cars
in household
None 0.00 0.08 0.00 0.00
One 2.50 0.07 2.37 2.64
Two 4.76 0.15 4.40 5.11
Three or more 5.97 0.39 5.02 6.91
Domestic fuel
type
Gas 2.77 0.05 2.67 2.87
Electric 1.76 0.15 1.44 2.08
Oil 3.33 0.32 2.60 4.05
Other 2.83 0.50 1.68 3.98
Government
Office Region
N East 2.28 0.20 1.87 2.68
N West & Mersey 2.66 0.20 2.21 3.12
Yorks and Humber 2.62 0.16 2.29 2.94
E Midlands 2.92 0.27 2.31 3.53
W Midlands 2.82 0.20 2.39 3.24
Eastern 3.07 0.23 2.55 3.60
London 1.85 0.16 1.52 2.18
S East 2.90 0.12 2.65 3.14
S West 2.88 0.26 2.27 3.49
Wales 2.48 0.22 2.04 2.93
Scotland 2.53 0.15 2.22 2.83
Town and fringe 2.81 0.17 2.46 3.16
Number of
(paid) workers
in household
None 1.17 0.07 1.02 1.31
One 2.22 0.13 1.92 2.53
Two 3.97 0.14 3.64 4.30
Three 4.94 0.31 4.21 5.66
Four or more 6.03 0.73 4.21 7.85
Employment
status of HRP
Self-employed 2.53 0.26 1.95 3.12
FT employee 3.91 0.08 3.75 4.07
PT employee 3.30 0.13 3.03 3.56
ILO unemployed 1.39 0.36 0.64 2.13
Retired 1.36 0.09 1.19 1.53

54

2
(continued)

Intermediate 3.12 0.28 2.48 3.76
Semi-Routine 2.82 0.20 2.40 3.25
Routine 2.71 0.20 2.30 3.12
HRP retired 1.36 0.09 1.19 1.53
Socio-Economic
Group of HRP
A 3.97 0.39 3.07 4.87
B 4.04 0.13 3.77 4.32
C1 3.23 0.26 2.63 3.82
C2 3.08 0.18 2.68 3.48
DE 2.60 0.19 2.23 2.98

55

2
emissions from public
transport means, standard errors and confidence intervals (metric tons)

OECD
equivilised net
household
disposable
income
1 0.38 0.07 0.21 0.54
2 0.26 0.05 0.16 0.35
3 0.28 0.05 0.16 0.39
4 0.23 0.04 0.15 0.32
5 0.28 0.07 0.12 0.45
6 0.26 0.06 0.13 0.40
7 0.33 0.06 0.19 0.47
8 0.29 0.06 0.15 0.43
9 0.29 0.08 0.11 0.47
10 0.34 0.08 0.14 0.54
Net household
disposable
income
1 0.44 0.09 0.22 0.66
2 0.35 0.04 0.26 0.44
3 0.34 0.05 0.23 0.45
4 0.30 0.05 0.20 0.40
5 0.26 0.05 0.16 0.36
6 0.26 0.08 0.07 0.44
7 0.23 0.05 0.14 0.33
8 0.24 0.05 0.12 0.35
9 0.24 0.07 0.09 0.39
10 0.28 0.08 0.08 0.48
Number of
children
None 0.27 0.02 0.23 0.30
One 0.40 0.08 0.21 0.59
Two 0.30 0.05 0.19 0.41
Three 0.32 0.10 0.09 0.56
Four or more 0.42 0.20 -0.05 0.88
Two persons 0.23 0.03 0.17 0.28
Three persons 0.32 0.04 0.22 0.42
Four persons 0.40 0.06 0.27 0.53
Five persons 0.52 0.09 0.33 0.71
Age of HRP Under 25 0.58 0.15 0.21 0.95
25-35 0.36 0.05 0.24 0.48
35-45 0.34 0.05 0.23 0.44
45-55 0.38 0.06 0.23 0.53
55-60 0.24 0.06 0.10 0.38
60-65 0.19 0.04 0.11 0.28
65-70 0.18 0.04 0.11 0.25
75+ 0.18 0.04 0.10 0.26

56

2
transport means, standard errors and confidence intervals (continued)

Number of
bedrooms in
dwelling
One 0.36 0.05 0.27 0.45
Two 0.28 0.03 0.21 0.35
Three 0.29 0.02 0.24 0.34
Four 0.25 0.05 0.13 0.36
Five or more 0.47 0.15 0.11 0.84
Accommodation
type
Detached 0.15 0.03 0.10 0.20
Semi-detached 0.26 0.02 0.21 0.30
Terraced 0.34 0.03 0.28 0.40
Flat 0.45 0.08 0.24 0.65
Other 0.49 0.11 0.23 0.76
Mortgage 0.28 0.04 0.19 0.38
Renting 0.44 0.03 0.38 0.50
Other 0.29 0.15 -0.05 0.64
Number of cars
in household
None 0.55 0.05 0.43 0.66
One 0.24 0.02 0.19 0.29
Two 0.15 0.03 0.08 0.22
Three or more 0.26 0.08 0.10 0.43
Domestic fuel
type
Gas 0.31 0.02 0.27 0.34
Electric 0.30 0.05 0.20 0.40
Oil 0.09 0.06 -0.03 0.21
Other 0.20 0.09 0.02 0.37
Government
Office Region
N East 0.36 0.09 0.16 0.55
N West & Mersey 0.24 0.05 0.15 0.33
Yorks and Humber 0.24 0.05 0.13 0.35
E Midlands 0.20 0.07 0.06 0.35
W Midlands 0.23 0.06 0.11 0.36
Eastern 0.20 0.04 0.12 0.28
London 0.79 0.06 0.66 0.92
S East 0.23 0.05 0.10 0.35
S West 0.14 0.04 0.06 0.21
Wales 0.19 0.07 0.04 0.35
Scotland 0.34 0.07 0.16 0.51
Town and fringe 0.14 0.05 0.03 0.25
Number of
(paid) workers
in household
None 0.20 0.02 0.16 0.25
One 0.22 0.03 0.17 0.28
Two 0.30 0.04 0.21 0.38
Three 0.60 0.09 0.39 0.81
Four or more 1.77 0.42 0.62 2.91
Employment
status of HRP
Self-employed 0.22 0.06 0.08 0.35
FT employee 0.34 0.04 0.23 0.44
PT employee 0.35 0.05 0.25 0.45
ILO unemployed 0.76 0.19 0.29 1.24
Retired 0.18 0.03 0.13 0.23

57

2
transport means, standard errors and confidence intervals (continued)

Intermediate 0.39 0.13 0.06 0.71
Semi-Routine 0.44 0.11 0.18 0.70
Routine 0.41 0.11 0.16 0.67
HRP retired 0.18 0.03 0.13 0.23
Socio-Economic
Group of HRP
A 0.34 0.07 0.20 0.49
B 0.29 0.04 0.20 0.38
C1 0.37 0.08 0.17 0.57
C2 0.29 0.06 0.17 0.42
DE 0.41 0.06 0.29 0.53

58

2
emissions from
domestic fuel means, standard errors and confidence intervals (metric
tons)

OECD
equivilised net
household
disposable
income
1 4.98 0.08 4.81 5.14
2 5.06 0.07 4.92 5.20
3 5.37 0.07 5.23 5.51
4 5.65 0.07 5.52 5.78
5 5.86 0.07 5.73 5.99
6 5.98 0.07 5.85 6.11
7 6.23 0.07 6.10 6.36
8 6.44 0.07 6.31 6.57
9 6.50 0.07 6.37 6.63
10 7.26 0.07 7.13 7.40
Net household
disposable
income
1 5.80 0.08 5.63 5.98
2 5.92 0.07 5.79 6.05
3 5.74 0.07 5.61 5.87
4 5.75 0.07 5.62 5.88
5 5.99 0.07 5.86 6.12
6 5.89 0.07 5.75 6.03
7 5.87 0.07 5.74 5.99
8 6.01 0.07 5.88 6.14
9 5.95 0.07 5.82 6.08
10 6.38 0.07 6.25 6.51
Number of
children
None 5.57 0.03 5.51 5.62
One 6.37 0.06 6.26 6.48
Two 6.86 0.06 6.74 6.98
Three 7.31 0.12 7.07 7.55
Four or more 7.78 0.18 7.42 8.14
Two persons 5.95 0.03 5.88 6.01
Three persons 6.53 0.05 6.43 6.64
Four persons 7.13 0.05 7.03 7.24
Five persons 7.61 0.10 7.40 7.81
Age of HRP Under 25 4.69 0.13 4.43 4.95
25-35 5.30 0.05 5.19 5.41
35-45 6.31 0.05 6.22 6.40
45-55 6.68 0.05 6.58 6.78
55-60 6.38 0.07 6.24 6.51
60-65 6.08 0.07 5.94 6.22
65-70 5.64 0.06 5.54 5.75
75+ 5.12 0.06 5.00 5.23

59

2
emissions from
domestic fuel means, standard errors and confidence intervals
(continued)

Number of
bedrooms in
dwelling
One 3.71 0.07 3.58 3.84
Two 4.97 0.04 4.90 5.05
Three 6.13 0.03 6.07 6.19
Four 7.56 0.05 7.45 7.67
Five or more 9.66 0.12 9.43 9.90
Accommodation
type
Detached 7.34 0.04 7.25 7.42
Semi-detached 6.08 0.04 6.01 6.15
Terraced 5.58 0.04 5.50 5.65
Flat 4.24 0.06 4.13 4.36
Other 5.18 0.09 5.01 5.35
Mortgage 6.58 0.03 6.51 6.64
Renting 4.87 0.05 4.78 4.96
Other 5.93 0.17 5.61 6.26
Number of cars
in household
None 4.80 0.05 4.71 4.89
One 5.67 0.03 5.62 5.73
Two 7.07 0.04 6.99 7.15
Three or more 7.70 0.09 7.53 7.87
Domestic fuel
type
Gas 6.01 0.02 5.96 6.06
Electric 4.77 0.06 4.66 4.89
Oil 8.28 0.11 8.07 8.50
Other 5.22 0.15 4.93 5.50
Government
Office Region
N East 5.82 0.10 5.62 6.02
N West & Mersey 5.91 0.06 5.78 6.03
Yorks and Humber 6.05 0.07 5.90 6.19
E Midlands 5.86 0.08 5.71 6.01
W Midlands 6.02 0.07 5.88 6.17
Eastern 6.07 0.07 5.93 6.21
London 5.69 0.07 5.55 5.82
S East 5.93 0.06 5.82 6.04
S West 5.63 0.07 5.50 5.77
Wales 6.11 0.09 5.93 6.29
Scotland 6.18 0.07 6.04 6.32
Town and fringe 5.93 0.07 5.80 6.07
Number of
(paid) workers
in household
None 5.18 0.04 5.11 5.26
One 5.68 0.04 5.60 5.75
Two 6.53 0.04 6.46 6.60
Three 7.37 0.09 7.19 7.55
Four or more 7.95 0.17 7.60 8.29
Employment
status of HRP
Self-employed 6.80 0.08 6.64 6.96
FT employee 6.20 0.03 6.13 6.26
PT employee 6.14 0.07 6.01 6.28
ILO unemployed 5.15 0.14 4.87 5.43
Retired 5.36 0.04 5.28 5.43

60

2
emissions from
domestic fuel means, standard errors and confidence intervals
(continued)

Intermediate 5.68 0.09 5.51 5.86
Semi-Routine 5.54 0.09 5.37 5.71
Routine 5.71 0.09 5.53 5.88
HRP retired 5.36 0.04 5.28 5.43
Socio-Economic
Group of HRP
A 6.90 0.10 6.71 7.10
B 6.66 0.04 6.57 6.74
C1 5.69 0.07 5.56 5.83
C2 6.20 0.06 6.09 6.31
DE 5.63 0.08 5.47 5.78

8. About the authors
Dr Eldin Fahmy
Research Fellow, Centre for the Study of Poverty and Social J ustice, University
of Bristol

Joshua Thumim
Head of Research and Analysis, Centre for Sustainable Energy, Bristol

Vicki White
Researcher, Centre for Sustainable Energy, Bristol

Carbon Reduction Policy Full

Uploaded by

Copyright:

Available Formats

Carbon Reduction Policy Full

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Carbon Reduction Policy Full

Uploaded by

Copyright:

Available Formats

1

JRF programme paper:

You might also like

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.