Understanding high-emitting households in the UK through a cluster analysis

Xinfang Wang; Ming Meng

doi:10.1007/s11708-019-0647-6

Understanding high-emitting households in the UK through a cluster analysis

Wang, Xinfang; Meng, Ming 2019-12-15 00:00:00 Front. Energy 2019, 13(4): 612–625 https://doi.org/10.1007/s11708-019-0647-6 RESEARCH ARTICLE Xinfang WANG, Ming MENG Understanding high-emitting households in the UK through a cluster analysis © The Authors (2019). This article is published with open access at link.springer.com and journal.hep.com.cn Abstract Anthropogenic climate change is a global 1 Introduction problem that affects every country and each individual. It is largely caused by human beings emitting greenhouse Climate change has various impacts on water, food, gases into the atmosphere. In general, a small percentage of industry, health, ecosystems and coastal systems [1]. To the population is responsible for a large amount of avoid the ‘dangerous’ impacts of climate change, parties to emissions. This paper focuses on high emitters and their the United Nations Framework Convention on Climate CO emissions from energy use in UK homes. It applies a Change (UNFCCC) came to an international agreement— cluster approach, aiming to identify whether the high the Paris Agreement— on ‘holding the increase in global emitters comprise clusters where households in each average temperature to well below 2°C above pre- cluster share similar characteristics but are different from industrial levels’ [2]. The UK introduced its own carbon the others. The data are mainly based on the Living Cost budgets and stated its commitment in the Climate Change and Food survey in the UK. The results show that after Act 2008 to achieve an 80% emission reduction by 2050 equivalising both household emissions and income, the compared to the 1990 baseline [3]. To achieve these high emitters can be clustered into six groups which share emission reduction targets, the ﬁrst ﬁve UK carbon budgets similar characteristics within each group, but are different covering the period from 2008 to 2032 have been set in law from the others in terms of income, age, household [4–6]. Anderson [7] and Pye et al. [8] have argued that the composition, category and size of the dwelling, and tenure Paris Agreement requires more radical and rapid emission type. The clustering results indicate that various combina- reductions than the UK targets. tions of socioeconomic factors, such as low-income single The energy used in homes accounted for around 27% of female living in an at least six-room property, or high- territorial-based CO emissions in the UK in 2016 income retired couple owning a large detached house, categorised by end users, while the industry and transport could all lead to high CO emissions from energy use at sectors were responsible for 30% and 36% of the total UK home. Policymakers should target each high-emitter CO emissions respectively [9]. Due to the switch from cluster differently to reduce CO emissions from energy solid fuel to gas in electricity generation and reduced solid consumption at home more effectively. fuel use in homes, the CO emissions from UK household energy use decreased by 35% between 1990 and 2016 [9]. Keywords cluster analysis, emissions reduction, energy However, the total energy consumption by households use, high emitters, household energy consumption, socio- increased by 1% in 2016 compared to the 1990 level, economic factors measured by tonnes of oil equivalent [10]. To reduce CO emissions from energy use at home, the UK launched Feed-in Tariffs (FIT) in April 2010, which supported households, businesses and other organisations to generate electricity from renewable sources. Households who Received Dec. 29, 2018; accepted Jul. 15, 2019; online Dec. 15, 2019 participate in the FIT are paid by their energy suppliers Xinfang WANG with both a generation tariff for each unit of energy they School of Chemical Engineering, University of Birmingham, Edgbaston, generate from renewable source and an export tariff for Birmingham B15 2TT, UK E-mail: x.wang.10@bham.ac.uk each unit that feeds back into the grid [11]. Furthermore, for improving energy efﬁciency at home, the UK Ming MENG government launched the Green Deal and Energy Department of Economics and Management, North China Electric Company Obligation (ECO) that required energy compa- Power University, Baoding 071003, China Xinfang WANG et al. Understanding high-emitting households in the UK 613 nies to provide heating and insulation improvements to capita and household size could be caused by the households [12–14]. The Green Deal allowed households economies of scale at the household level, as members of to repay the cost of installed insulation system through the same household could share gas and electricity use for their saved energy bills [13,14], although participation was space heating, cooking, lighting and the utilization of low partly because of the uncertainties on energy savings various appliances at home most of the time [24]. The age that could be achieved and the house resale value due to of household members also matters [20]. For example, attached Green Deal loan to the property [15,16]. The ECO older people might require more energy for space heating was also launched early in 2013 to provide additional at home due to their poor health [25]. support, especially for vulnerable households and hard-to- Household income is another key variable discussed in treat homes, which placed legal obligations on energy the literature as it inﬂuences both emissions from direct suppliers to deliver energy efﬁciency measures to energy use at home, and emissions embedded in consumed residential energy users [12]. As discrepancies in terms products and services [26–28,37]. Household composition of wealth, well being and emissions exist at an intra- inﬂuences the economic resources available to a household national level, it is important to focus on the high-emitting measured by income, due to the economies of scale [44]. households who could have a larger potential for reducing The reason for this is that the members of the same their energy consumption and CO emissions than others. household not only share the energy used for performing Previous research did not explore different groups among different practices, but also share other products and high emitters, or how emission reduction policies could services such as furniture, cookware, cars, Internet service, target each group differently [11–42]. TV license, and so on. To address the inﬂuences from the This paper aims to identify particular groups within the economies of scale, the disposable household income society who are likely to be high emitters from their energy (gross income after taxes and beneﬁts) is equivalised in this consumption at home. This will facilitate more targeted paper with the commonly used Organisation for Economic policies for reducing household CO emissions in the UK. Co-operation and Development (OECD) [44] modiﬁed Therefore, the cluster analysis is applied to explore the scales (Table 1). make-up of the high-emitting household group in terms of Table 1 Household income equivalisation factors socioeconomic factors and dwelling-related characteris- Member of the household Modiﬁed OECD for tics. After the introduction in Section 1, Section 2 draws on disposable income the literature related to the distribution of UK household First adult 1 emissions, and outlines the research gap that this paper Second and subsequent adults 0.5 addresses. Section 3 explains the data used to identify and Children aged 14–18 0.5 explore high emitters, as well as the cluster method that is used for the analysis. After presenting the clustering results Children under 14 0.3 in Section 4 with further discussions in Section 5, it concludes by considering the importance of the analysis for targeted emission reduction policies among households Studies have identiﬁed that in general the larger the ﬂoor in the UK in Section 6. area of the dwelling, the more energy will be required for space heating and other energy use, assuming other conditions of the dwelling are similar [25,26,30,31,36]. 2 Previous work on socioeconomic factors Space and water heating accounted for around 81% of total and household energy consumption energy use at home in the UK between 1990 and 2013 [10]. For this reason, the size of dwelling plays an important role The CO emissions produced by households could vary in overall household energy consumption and related CO 2 2 signiﬁcantly [32,33,43]. Chancel and Piketty [32] have emissions, especially during the winter, due to the energy estimated that the top 10% of emitters account for around required for space heating. The electricity use for lights and 45% of direct and embedded greenhouse gas (GHG) appliances is also positively correlated with the size of the emissions globally. Likewise, Oxfam [33] has estimated dwelling [29,30,45], as there may be more appliances in that the total share of CO emissions from the top 10% of larger houses that are used in everyday life, than in an high-income people is approximately 50%. Previous average-sized dwelling. literature for the UK [17–20] and other countries [21– The type of dwelling, whether it is a detached, semi- 27,35–38] has shown that household emissions are related detached or terraced house, or a bungalow or ﬂat, also to a variety of socioeconomic factors and dwelling-related inﬂuences energy use and the related CO emissions, characteristics. For example, there exist a negative link largely due to the diverse amount of energy required for between CO emissions per capita and household size (the space heating across these different types of dwellings number of people in a household), and a positive link [46,47]. In addition to energy use for space heating, between CO emissions per household and household size electricity use for lighting and appliances is also inﬂuenced [21–23]. This negative link between CO emissions per by the type of dwelling [20,45]. Büchs and Schnepf [20] 2 614 Front. Energy 2019, 13(4): 612–625 have identiﬁed that the CO emissions from total energy electricity consumption among households excluding use at home are the highest among households living in electric heating and electric showers, where it is found detached houses, followed by the semi-detached occupiers, that 85% of the high electricity-consuming households in and then those in terraced houses. Households living in their study (39 out of 46 households that belong to the top ﬂats have the lowest energy emissions overall. This could 20% high-electricity users) have at least one key be partly because detached houses are likely to have a contributing factor that could lead to their higher electricity greater ﬂoor area, while ﬂats are likely to have a smaller consumption. The key contributing factors are: at least one in general in the UK [46]. The shared insulation and three people living in the dwelling, the dwelling size being larger than 130 m heat between dwellings also reduce the heat loss among , the age of the Household Reference 1) semi-detached houses, terraced houses and ﬂats, compared Person (HRP) being between 45 and 54, the HRP being with detached dwellings [46]. The mean heat loss for unemployed but not retired, and householders belonging to 2) different dwelling types in the UK is presented in Table 2, the professional and managerial socioeconomic group . which shows the mean heat loss ranges from 167 W/°C for While Palmer et al. [29] have estimated the average value a ﬂat, up to 342 W/°C for a detached house [46]. of each of the socioeconomic factors and dwelling-related characteristics among high electricity-consuming house- Table 2 Mean heat loss of dwelling types holds, they have not identiﬁed whether different combina- –1 Type of dwelling Heat loss/(W$°C ) tions of these factors are more likely to lead to high Detached 342 electricity use collectively. Overall, previous studies have identiﬁed socioeconomic factors and dwelling-related Semi-detached 264 characteristics that inﬂuence energy consumption and Terraced 235 related CO emissions. However, they have not explored Bungalow 225 whether various combinations of the factors are all likely to Flat 167 link with particularly high CO emissions, which is crucial to understand if the aim of reducing household energy UK mean 247 consumption and related CO emissions is to be achieved. For example, studies may ﬁnd that, in general, high- Tenure type could also inﬂuence the energy consump- income households are likely to have more emissions than tion at home [36]. In general, private rented domestic households with an average income. However, such buildings have relatively low thermal insulation installed studies do not show that if a household does not belong and require more energy for space heating, due largely to to a high-income group, it could be a high emitter due to a the ‘tenant-landlord problem’ [17]. The ‘tenant-landlord combination of other socioeconomic factors or dwelling- problem’ refers to the mismatch between landlords who related characteristics, such as household composition, size pay the cost of insulation and tenants who receive the of dwelling and tenure type. To address this gap in the beneﬁts [17]. As a ﬁnancial incentive, the UK government literature, this paper undertakes cluster analysis within introduced the Landlord’s Energy Saving Allowance high-emitting households in the UK. (LESA) between April 2004 and April 2015, which The clustering technique has been used in several other provided grants to landlords for upfront payments of energy and emission related studies, both to cluster various energy efﬁciency measures such as loft and cavity countries and households [34,48,49]. The aim of cluster wall insulation, solid wall insulation, draft-prooﬁng and analysis is to classify the whole sample into distinguished ﬂoor insulation [39]. Although results from previous clusters, where it is relatively homogeneous within each studies [17,20,30,45] indicate that high energy users and single cluster and heterogeneous across different clusters. high emitters are likely to own a house that is not only On a country level, Lamb at al. [49] have applied the large, but also detached, these studies have not explicitly clustering technique to identify the similarity and diversity explored whether some high emitters do not own large of human development and CO emissions between detached houses; and if this is the case, what other factors developing and developed nations. On a household level, could collectively lead to their high energy use level. Pullinger et al. [48] have used the clustering approach to In particular, the household income, age of house- identify distinct household groups according to their water holders, and category of dwelling have been identiﬁed as using practices. Likewise, Element Energy [34] have the main inﬂuencing factors on household energy investigated household electricity use in the UK using expenditure in North Carolina in the US [38]. Likewise cluster analysis. It conducted cluster analysis among in the UK, Palmer et al. [29] have focused only on householders in their sample based on their annual 1) The HRP is the person who owns the dwelling or is responsible for renting it. If the dwelling is joint owned or rented, the person with the highest income would be the HRP. If two or more householders have the same highest income or they all have zero income, the oldest should be identiﬁed as the HRP. 2) Palmer et al. [29] have divided all households into three socioeconomic groups: professional and managerial; supervisory, clerical and skilled manual; and semi-skilled, unskilled, pensioner and non-working group. Xinfang WANG et al. Understanding high-emitting households in the UK 615 electricity consumption, 6–7 pm peak-time electricity use, from energy consumption data. Household expenditure on socioeconomic factors and dwelling-related characteris- gas, electricity, and oil are collected from the 2012 Living tics, number and energy efﬁciency level of appliances, as Cost and Food (LCF) survey, which is a household survey well as their climate change attitudes and electricity carried out by the Ofﬁce for National Statistics (ONS) in conservation behaviors [34]. Compared to the traditional the UK [40]. The survey covers the whole UK, including regression analysis, the clustering technique is especially England, Scotland, Wales, and Northern Ireland [40]. beneﬁcial for studies exploring whether there are different Households in Northern Ireland are excluded in the combinations of independent variables and corresponding analysis presented here, due to the much higher level of values that lead to a similar value for the dependent oil use at home than other regions. Using oil leads to variable (for example, high CO emissions from energy around 38% higher CO emission per kWh than natural gas 2 2 use at home) [48]. The analysis presented in this paper [52]. 5593 households in total were selected using a multi- applies the cluster method to classify the high-emitting stage stratiﬁed random sample method from approximately households in order to identify whether the high emitters 26.4 million UK households in the 2012 LCF survey [41]. comprise several groups which are more homogeneous in Initially, the ﬁrst stratum in the sample selection was terms of socioeconomic factors and dwelling-related deﬁned by the Government Ofﬁce Regions (GORs) and characteristics within each group but heterogeneous across two variables, which were social class of the HRP and different groups. The homogeneity within a cluster means ownership of cars. Then, 638 out of 1.8 million postal that the households within one cluster are grouped together sectors were randomly selected from the ﬁrst stratum. All by well-deﬁned similarities. On the contrary, the hetero- households in each selected postal sector were accessed for geneity across clusters means that households in one the survey. As a result, 52% of the selected households cluster are separated from those in other clusters by well- responded to the survey, which constituted the 5593 deﬁned dissimilarities [50,51]. This clustering method has households in the data set. not been used to classify high emitters in other studies, and Less than 1% of the households use other fuels, such as will be an important original contribution to knowledge solid fuel or Calor gas, in the 2012 LCF survey. For this through this research. reason, the other fuels are not considered in estimating total CO emissions from energy use at home. This analysis estimates the energy used by each household in the 3 Material and methods selected survey sample by dividing the household energy bills by corresponding energy unit prices as shown in 3.1 Material Table 3 [53,54]. The price for domestic oil in 2012 was also obtained from Department of Energy and Climate There are currently no data sets in the UK that provide both Change (DECC) [55], with no regional prices available. household CO emissions and socioeconomic factors [20]. After calculation, gas and electricity use are measured in This research thus estimates household CO emissions kWh and oil consumption is measured in liters, instead of Table 3 Gas and electricity unit prices across UK regions for different payment methods Gas unit price by payment method/£ Electricity unit price by payment method/£ Government ofﬁce regions Credit Direct Prepayment Overall Credit Direct Prepayment Overall debit meters debit meters North East 4.58 4.21 4.54 4.37 14.89 13.63 14.87 14.18 North West and Merseyside 4.61 4.24 4.59 4.41 15.14 13.85 15.19 14.46 Yorkshire and the Humber 4.61 4.21 4.60 4.38 14.85 13.55 14.81 14.16 East Midlands 4.57 4.23 4.61 4.38 14.78 13.63 14.86 14.14 West Midlands 4.72 4.30 4.62 4.48 15.10 13.75 15.04 14.38 Eastern 4.62 4.28 4.59 4.42 14.83 13.64 14.80 14.16 London 4.69 4.37 4.62 4.53 14.82 13.73 14.86 14.38 South East 4.7 4.32 4.59 4.47 14.70 13.60 14.72 14.04 South West 4.67 4.33 4.59 4.45 15.66 14.53 15.69 15.03 Wales 4.65 4.32 4.61 4.47 16.01 14.62 15.84 15.25 Scotland 4.59 4.20 4.54 4.36 15.58 14.28 15.32 14.84 Northern Ireland –– – 4.47 17.07 16.44 16.65 16.72 UK (excluded 4.65 4.28 4.59 4.43 15.13 13.90 15.20 14.48 Northern Ireland for gas unit price) 616 Front. Energy 2019, 13(4): 612–625 British pounds. For this reason, there is uncertainty about households with more members than average. Thus the using a monetary value to estimate energy consumption analysis presented in this paper applies DECC’s [56] Low volume, because the actual volume of energy measured in Income High Cost (LIHC) equivalisation factors (Table 4) kWh or liters that each household used is not available. to equivalise household CO emissions estimates before The gas, electricity, and oil use are converted into CO deﬁning, clustering and identifying the high-emitting emissions with factors obtained from the AEA [52]. The households. The LIHC equivalisation scale is based on conversion factors for electricity in 2012 are the average the energy requirement of the households, which was used grid conversion factors over the previous ﬁve years, which by DECC [56] to identify households in fuel poverty. are updated annually [52]. The calculations result in an Table 4 Equivalisation factors for fuel bills under the LIHC deﬁnition estimation of 5.3 tonnes CO /year for average UK of fuel poverty household emissions from energy use at home, based on Number of people in the household Equivalisation factor the 2012 LCF survey. This estimation is consistent with the results in Büchs and Schnepf [20], which shows that One 0.82 average household emissions from energy use at home are Two 1.00 about 5.1 tonnes CO /year during 2006–2009. Three 1.07 In addition to household energy expenditure, the 2012 Four 1.21 LCF survey also provides socioeconomic factors and Five or more 1.32 dwelling-related information for households. The analysis aims to include as many socioeconomic factors and dwelling-related characteristics as possible from the 2012 After equivalising the household CO emission esti- LCF survey for data analysis in order to provide a fuller mates from energy use at home, the top 10% of emitting picture of who the high emitters are and why they emit households are deﬁned as high emitters for the cluster more than others. Based on the studies associated with analysis, which constitute 510 households. The 10% range household CO emissions from energy use at home and is selected to be consistent with relevant studies conducted socioeconomic factors that are introduced in Section 2, the by Brand [43], Chancel and Piketty [32], and Oxfam [33]. household variables included in the analysis are household composition, tenure type, category of dwelling, number of rooms in the accommodation, equivalised disposable 3.2.2 Cluster method household income, age of the oldest person, sex of HRP, GORs, as well as ownership of cars and second dwelling in There are three principal clustering methods: hierarchical the UK. The ownership of cars and second dwellings in the clustering, k-means clustering and Two-step clustering UK are included in the analysis as an additional indicator approaches. Among these three clustering methods, only of the wealth level of the households. The sex of HRP is the Two-step cluster ﬁts with the mixed data of continuous included in the analysis to complement the information on and categorical variables [51]. Therefore, the Two-step household composition. Education level data were also cluster method is selected for the analysis. The continuous collected in the 2012 LCF survey, but 32% of household variables are standardised using the Standard Score (also members did not provide this information, thus the named as Z Score). The categorical variables are variables cannot be used dependably in the analysis. All manipulated as dummy variables, with a numerical value other available socioeconomic factors and dwelling-related of 0 or 1. In other words, if the answer is ‘yes’ for the characteristics in the 2012 LCF survey are covered by the dummy variable (For example, do the household occupants selected variables in the cluster analysis to cluster the high- live in a detached house?), the variable has a numerical emitting households. value 1. If the answer is ‘no’, a numerical value 0 is allocated to the variable. The Pearson correlation tests are 3.2 Methods of analysis then undertaken to check the correlation between any two of the selected continuous variables, and the Pearson’s Chi-square tests are used to check the correlation between 3.2.1 Equivalising household CO emissions any two categorical variables [57]. As explained in Section 2, household composition The ﬁrst step of the Two-step cluster analysis is called inﬂuences energy requirements at home as members in pre-cluster, where the data are scanned one-by-one to the same household are able to share the energy used for decide whether to merge the data with the previously space heating, lighting, cooking, and appliance use most of formed clusters or start a new cluster, according to the log- the time. The analysis aims to identify who the high likelihood distance criterion [58]. The second step of the emitters are based on CO emissions estimates from their Two-step cluster analysis merges the sub-clusters identi- energy consumption. Not equivalising the household CO ﬁed in the ﬁrst step, where the ﬁnal number of clusters is emissions estimates is likely to result in the deﬁned high- decided through two stages. At stage one, the initial emitting households comprising a larger percentage of estimate of the number of clusters is computed using the Xinfang WANG et al. Understanding high-emitting households in the UK 617 Bayesian information criterion (BIC) criterion, which is genous each individual cluster is and the more hetero- commonly used as an objective selection criteria to avoid geneity exists across different clusters. The cluster quality arbitrariness in deciding the number of clusters [59]. At is treated as ‘poor’ if the Silhouette measure is between –1 stage two, in order to decide the ﬁnal number of clusters, and 0.2, while it is ‘fair’ if it is between 0.2 and 0.5 and the largest relative increase in distance between the two ‘good’ if it is larger than 0.5 [58]. closest clusters is identiﬁed using the ratio calculation, shown in Eq. (1) [58]. 4 Results d ðC Þ min k RðkÞ¼ , (1) d ðC Þ min kþ1 As mentioned in Section 3, the Pearson correlation tests are undertaken among the selected continuous variables, and where C is the cluster model containing k clusters and the Pearson’s Chi-square tests are used to check the d (C ) is the minimum cluster distance for cluster model min k correlations between any two categorical variables [57]. C . The larger the absolute value derived from the Pearson The ﬁnal number of clusters is decided by comparing the correlation is, the more correlated the two continuous two largest R ratios. If the largest is 1.15 times greater than variables are. Likewise, the larger the Cramer’s V for the second largest, the model with the largest R ratio is Pearson’s Chi-square is, the more correlated the two selected as the optimal number of clusters; alternatively, categorical variables are. If the absolute value derived from from those two models with the largest R ratio, the one the Pearson correlation or the Cramer’s V for Pearson’s with the larger number of clusters is selected as the optimal Chi-square is close to 1, it may inﬂuence the cluster results. number of clusters [58]. The cluster quality is measured by This is because that in this case, the inﬂuence of the two the ‘Silhouette measure of cohesion and separation’, which related variables on the clustering results would be similar; is calculated by using Eq. (2). including both variables means that the inﬂuence is bðxÞ– aðxÞ counted twice during the clustering procedure. According sðxÞ¼ , (2) to the correlation test results in Table 5, the number of cars maxfaðxÞ,bðxÞg with the number of rooms, and the number of rooms with where s(x) is the ‘Silhouette measure of cohesion and the household income are more correlated continuous separation’, a(x) is the average distance of x to all other variables than others. For categorical variables, the sex of cases in the same cluster, and b(x) is the minimum average the HRP and the composition of the household are more distance of x to cases in any of the other clusters. correlated than others (Table 6). The analysis has included The larger the Silhouette measure, the more homo- a relatively large sample (510 high-emitting households) to Table 5 Correlations between continuous variables Pearson Second dwelling Cars and vans in Weekly disposable Rooms in Age of the correlation in the UK household household income accommodation oldest person Second dwelling 1 0.133** 0.266** 0.232** 0.014 in the UK Cars and vans in household 0.133** 1 0.341** 0.485** 0.094* Weekly 0.266** 0.341** 1 0.480** 0.121** disposable household income Rooms in accommodation 0.232** 0.485** 0.480** 1 0.233** Age of the oldest person 0.014 0.094* 0.121** 0.233** 1 Notes: ** indicates that the correlation is signiﬁcant at the 0.01 level (2-tailed) while * indicates that the correlation is signiﬁcant at the 0.05 level (2-tailed). Table 6 Correlations between categorical variables Composition of household Category of dwelling Tenure type Sex of HRP GORs Cramer’s V for Pearson’s Chi-square 1 0.237* 0.320** 0.574** 0.207 Composition of household 0.237* 1 0.263** 0.213** 0.181** Category of dwelling 0.320** 0.263** 1 0.294** 0.166* Tenure type 0.574** 0.213** 0.294** 1 0.180 Sex of HRP 0.207 0.181** 0.166* 0.180 1 GORs Notes: ** indicates that the correlation is signiﬁcant at the 0.01 level (2-tailed) while * indicates that the correlation is signiﬁcant at the 0.05 level (2-tailed). 618 Front. Energy 2019, 13(4): 612–625 reduce the risk of clustering results being inﬂuenced by socioeconomic factors and dwelling-related characteristics correlations between variables. Furthermore, the value for all six identiﬁed high-emitter clusters. distribution of each variable, comparing between the high- As shown in Table 7, the government ofﬁce region and emitting households and the remaining 90% households in ownership of second dwelling in the UK are not the 2012 LCF survey sample, are drawn in Fig. 1. Figure 1 distinguishable among the high-emitting households. The shows that none of the variables would dominate the household composition, income, category of dwelling, cluster results, because the values of each variable of high- tenure type, age of the oldest person, sex of HRP, number emitting households are distributed across all ranges. of vehicles owned and rooms in accommodation collec- Likewise, the values of each variable of the remaining 90% tively inﬂuence CO emissions from energy use at home. households are also distributed across all ranges. There- For example, if a two-adult household does not belong to fore, all the continuous and categorical variables are any high-income clusters (Clusters A and B), but rents a included for the clustering process. dwelling that is poorly insulated without gas central As a result, six high-emitter clusters are identiﬁed, with a heating, they can require more energy for space heating ‘fair’ quality being achieved for the cluster results which would result in high CO emissions from energy use measured with the ‘Silhouette measure of cohesion and at home. On the other hand, if the households are high- separation’ (Section 3.2.2). Table 7 lists the selected income ones who own a ﬂat outright and work outside the Fig. 1 Relationship between CO emissions and different variables—high emitters (in red) versus the remaining 90% households (in blue) (The y axes are equivalised annual household CO emissions. The unit for equivalised weekly disposable household income in (g) is ‘£’. The household income, the age of the oldest person, the number of second dwelling in the UK, as well as the number of rooms and cars are capped at the highest value in the 2012 LCF survey, as shown in the relevant diagrams, for the purpose of anonymisation. The representation of each value for all categorical variables is given in Electronic Supplementary Material.) Xinfang WANG et al. Understanding high-emitting households in the UK 619 Table 7 Key distinguishable variables for identiﬁed high-emitter clusters Variables Cluster A Cluster B Cluster C Cluster D Cluster E Cluster F Number of households 104 107 75 56 120 48 Composition of household 71% two adults, 28% two adults, 23% one adult with children; 34% single female; 18% two adults, no children; 56% two adults, no children no children; 39% two adults with children; 30% two adults, 51% two adults with children; no children; 46% two adults 12% two adults, no children no children; 21% at least three adults 38% at least three adults with children; 25% at least three adults 23% at least three adults Category of dwelling 95% detached 99% detached 61% semi-detached 38% detached; 91% semi-detached 81% semi-detached or terraced; 12% ﬂat 57% semi-detached or terraced or terraced or terraced Tenure type 98% own outright 91% own with a 64% rent; 37%own 64% own outright; 33% rent; 63% own 92% own outright mortgage with a mortgage 30% own with a mortgage with a mortgage Second dwelling in the UK – – ––– - Sex of HRP 91% male 99% male 96% female 100% female 100% male 100% male Mean equivalised disposable household 703 758 291 448 396 391 –1 Income/(£$week ) Mean number of cars and vans 2 2 1 2 1 1 Mean number of rooms 8 9 6 7 6 7 Age of the oldest person 73% over 60 82% under 59 96% under 59 57% over 60; 87% under 59 71% over 60 29% between 50 and 59 Government ofﬁce region – – ––– – Mean equivalised household CO 14.45 11.77 9.97 12.88 10.23 11.54 emissions from energy use at home (tonnes CO /year) Notes: —If no data are presented for the variable or cluster, it indicates this variable is relative evenly spread across all values under this cluster. 620 Front. Energy 2019, 13(4): 612–625 home on weekdays, they are less likely to be high emitters, these criteria. In contrast, among the remaining 90% as they are not typical households in any of the identiﬁed households, only 24% (402 out of 1660) of households clusters. with a male HRP and the oldest person under 59 meet these According to the cluster results, typical socioeconomic criteria. characteristics for each cluster are selected to compare the The comparison between high-emitter clusters and the remaining 90% households shows that among all the households in each high-emitter cluster with the remaining clusters, the households in cluster A are the most 90% households in the 2012 LCF survey sample. The distinguishable ones, followed by those in Cluster B, and combination of typical socioeconomic characteristics of then Cluster D. The households in Clusters C, E, and F are each cluster shows that: less distinguishable from the remaining 90% households, (1) If the HRP is female, the household is likely to be a but still have some of the typical characteristics that high high emitter if 1) the age of the oldest person is under 49; they live in a emitters in these clusters share. Although the households in non-detached property; rent it or own it with a mortgage; Cluster C have lower incomes than the other high-emitter and they own no more than one car and no more than seven clusters, they may rent a dwelling that is poorly insulated rooms at home. Among the high-emitter cluster C, 55% without gas central heating. Therefore, they could require (41 out of 75) households meet these criteria. In contrast, more energy for space heating which would result in high among the remaining 90% households, only 28% (496 out CO emissions from energy use at home. They may also be of 1787) of households with a female HRP meet all these part-time employed or unemployed who spend more time criteria. at home during the day compared to full-time employed 2) the age of the oldest person is over 50; the household people; therefore more energy would be consumed during has at least one car; there are at least seven rooms at home; the day for space heating, cooking, and entertaining. This and the householder owns their property either with a is consistent with the ﬁndings from Büchs and Schnepf mortgage or outright. Among the high-emitter cluster D, [20], where the households with female HRPs are likely to 55% (31 out of 56) households meet these criteria. In have higher CO emissions from direct energy use at contrast, among the remaining 90% households, only 10% home, which could relate to a workless status and along (181 out of 1787) of households with a female HRP meet time spent at home. The clustering results (Table 7) also all these criteria. show that the households that do not own a car or have less (2) If the HRP is male, and the age of the oldest person is than six rooms in their accommodation, the retired over 60, the household is likely to be a high emitter if households that do not own their accommodation, the 1) they own a detached house outright; have no children; households with a male HRP and an average equivalised and have at least two cars and eight rooms at home. Among disposable income less than £390, and the households with the high-emitter cluster A, 72% (75 out of 1104) house- a female HRP and an average equivalised disposable holds meet these criteria. In contrast, among the remaining income less than £290 are less likely to be high emitters 90% households, only 6% (69 out of 1147) of households compared with other households. with a male HRP and the oldest person over 60 meet these criteria. 2) they own a semi-detached house outright, have at 5 Discussion least two adults; and at least one car. Among the high- emitter cluster F, 56% (27 out of 48) households meet these The identiﬁed six high-emitter clusters support ﬁndings in criteria. In contrast, among the remaining 90% households, the existing literature that household energy consumption only 21% (239 out of 1147) of households with a male and CO emissions are inﬂuenced by various socio- HRP and the oldest person over 60 meet these criteria. economic factors and dwelling-related characteristics. (3) If the HRP is male, and the age of the oldest person is Moreover, they also show that in addition to each of the under 59, the household is likely to be a high emitter if socioeconomic factors and dwelling-related characteristics 1) they own a detached house with a mortgage; have at identiﬁed as inﬂuential in the literature, various combina- least two adults; and at least two cars and eight rooms at tions of these characteristics can jointly lead to high CO home. Among the high-emitter cluster B, 62% (66 out of emissions from energy use at home. Previous studies have 107) households meet these criteria. In contrast, among the mainly used regression analysis to investigate the relation- remaining 90% households, only 7% (111 out of 1660) of ships between household CO emissions (the dependent households with a male HRP and the oldest person under variable) and socioeconomic or dwelling characteristics (independent variables) [20,28]. Through regression 59 meet these criteria. 2) they live in a non-detached house, either renting or analysis, these studies identiﬁed some correlations owning with a mortgage; have at least two adults; and at between household emissions and socioeconomic or least one car and six rooms at home. Among the high- dwelling factors. For example, the type of dwelling, tenure emitter cluster E, 53% (63 out of 120) households meet type, as well as the age and income levels of householders Xinfang WANG et al. Understanding high-emitting households in the UK 621 are all correlated with household CO emissions urban location is included in the input variables and the [20,28,31]. Moreover, the regression model can be used ﬂoor area is included instead of the number of rooms. In to estimate the likely amount of emissions for a particular spite of the data limitations, the cluster analysis results household, giving the household’s values for all indepen- based on the LCF still show that the high emitters comprise dent variables in the model. However, due to the limitation different clusters of the households who share similar of the regression technique, it cannot provide insights into socioeconomic factors within each cluster but are different whether and what different combinations of independent from others, which provide useful information for more variables indicate particularly high levels of household targeted emission reduction policies on the different high- emissions. For example, through regression analysis, emitter clusters. In addition to socioeconomic factors and dwelling- Büchs and Schnepf [20] show that the age of the HRP related characteristics, other factors, such as the energy positively correlates with the emissions from direct energy use at home. However, they do not specify that younger efﬁciency of the dwelling and appliances, householders’ families may also more likely be high emitters if they are daily routines, and their use of the home may also lead to renting an old house that is energy inefﬁcient. Likewise, different energy consumption and CO emission levels the regression analysis can show that the size of the [63–65]. For example, a middle-aged couple who rent their dwelling is positively associated with the household CO accommodation can be high emitters due to the ‘tenant- emissions from energy use at home; but it may not disclose landlord problem’ discussed in Section 2. They can live in that householders living in smaller dwellings (for example, less insulated dwellings with less efﬁcient appliances, one or two-bedroom ﬂats) can be high emitters if they have which require more energy to deliver the same energy no access to gas at home and mainly use electricity for services for heating, cooking and cleaning. High-income space heating. In contrast, cluster analysis, which is families with younger children can be high emitters due to applied to the analysis presented in this paper, can identify their separate cooking for children [66]. Retired house- all these possible combinations. Of the identiﬁed high- holds that own their dwellings outright can be high emitter clusters, the socioeconomic factors and dwelling- emitters because of their more vulnerable health conditions related factors are more homogeneous within one cluster and longer time spent at home in general, where more while heterogeneous compared with other clusters. Using energy can be used for space heating and entertaining [20]. the clustering technique to classify high emitters addresses Some of the high-emitting retired households can also live a gap in the literature around exploring the various in larger houses with more additional appliances, which combinations of socioeconomic factors and dwelling- they had been using before their child(ren) moved out [67]. related factors that are most likely to link to high household Further research on people’s routines and use of home are CO emissions. necessary to provide a fuller picture of why these clusters The LCF survey was selected as the most appropriate of households are more likely to be high emitters than others. survey to identify the high emitters through cluster Rebound effects have been discussed widely in relation analysis, because the LCF data set covers information not only on household gas and electricity bills separately, to the emission reduction achievement focusing on the but also a variety of socioeconomic factors and dwelling- households [68,69]. Rebound effects refer to people related characteristics required to identify their inﬂuences consuming the money saved on energy bills from on energy consumption and CO emissions. However, improved energy efﬁciency or behavior change in a there are limitations to using the LCF data set, as some of particular energy service (e.g. lighting, cooking, space the variables are measured more indicatively than others. heating and cleaning) on using more energy for that service For example, the size of the dwellings is measured by the (also known as the direct rebound effect), or on other number of rooms at home, because the data on ﬂoor area is products and services that have direct or embedded CO not available. The rural or urban location of the household emissions (also known as the indirect rebound effect) is not available from the LCF survey either, which could [68,69]. As clariﬁed in Section 1, the analysis aims to also affect the level of energy use at home, especially for identify high-emitter clusters and the potential opportu- space heating due to the lack of access to gas in some rural nities to reduce household CO emissions from higher area and the urban heat island effect [30,60]. The urban emitting households. The emission reductions from high heat island effect means that the temperature in urban areas emitters’ energy use are likely to lead to rebound effects. is generally higher than that in surrounding rural areas, The range of rebound effects may vary signiﬁcantly among largely due to deforestation, the replacement of the land different high-emitter clusters and across various carbon surface by non-evaporating and non-porous materials such mitigation policies. The cluster analysis results in Section 4 as asphalt and concrete, and the more intensive layout of show that some identiﬁed high-emitter clusters (such as buildings and streets within an urban landscape [61,62]. Clusters A and B) share an average household income The identiﬁed high-emitter clusters might be different if about twice as high as other clusters (such as Clusters C, E the input variables are changed, for example if the rural or and F). The high-income high energy users are more likely 622 Front. Energy 2019, 13(4): 612–625 to already be able to afford as much gas and electricity they through general income tax and government spending, require as possible. They are less likely to spend the cost rather than from the energy market or energy suppliers savings on more direct energy consumption at home, but where costs are passed on to all customers but only beneﬁt are more likely to spend them on other products and those households that have renewable energy systems services (e.g., purchasing more expensive cars or ﬂying installed. abroad for holidays). On the other hand, some high-emitter As introduced in Section 1, for energy efﬁciency clusters are lower-income ones. If the higher-energy- improvements, the UK government mandates energy consuming households have not been able to afford as companies to provide heating and insulation improvements much energy as they need or have tight budgets, they can to lower-income and vulnerable households, for example, through the ECO and previous Green Deal. Due to supplier spend the energy payments saved from efﬁciency improve- obligations, they are ﬁnanced by raising overall energy ments on more gas and/or electricity use at home. For example, some householders may leave more lights on prices for customers [42]. The impact is highly regressive, while away, after swapping them for efﬁcient LED lights, because the high-income households pay a much smaller because the total payments for lighting would not increase share of their income on home energy compared to the or would still be reduced compared with previous low-income households in general [42]. When energy inefﬁcient lights. Policies focusing on energy and emission prices increase, the share of income spent on home energy reductions from higher-income higher-energy users may bills may increase much more among the low-income lead to smaller rebound effects and achieve more net households than high-income households if the energy emission reductions than others [60,69,70]. In contrast, savings from efﬁciency improvements are not sufﬁcient energy and emission reductions from lower-income enough to offset increased energy prices. This can lead to higher-energy users can involve higher rebound effects, more serious fuel poverty issues among the low-income which offset the emission reduction effort to a larger extent high emitters, especially retired low-income households [60,69,70]. Future research on reducing energy consump- living in large houses after their children have moved out. tion and CO emissions needs to consider the different size Retired or older people could require more energy for of the likely rebound effect for each high-emitter cluster. space heating, in part due to health conditions. In addition, The estimate could provide evidence on whether and how low-income high emitters may also rent poorly insulated much net CO emissions can be reduced from the high- dwellings and are constrained from insulting it due to their emitter clusters after taking into account rebound effects. tenure type. Cluster C comprises 64% households who rent Policy measures on promoting renewably-generated their properties. This category of householders generally electricity (e.g., the FIT) may achieve more net emission receives few beneﬁts from energy efﬁciency schemes because of the ‘tenant-landlord problem’ discussed in reductions from the low-income high emitters than other Section 2 [17]. As a ﬁnancial incentive, the UK emission reduction policy instruments. This is due to the government introduced the LESA program [39]. However, increased share of total energy use provided by renewably- generated electricity that reduces the CO the program was not widely known and the amount of intensity of energy use. For example, both improving energy efﬁciency grant provided was insufﬁcient [71]. The research and increasing renewably-generated electricity use at home presented here suggests that more policies like the LESA may lead to reduced household energy bills and cause should be initiated with an increased level of ﬁnancial similar direct rebound effects on energy use. If house- incentive supported by government spending, and be holders rebound into using more electricity, they will have widely publicised among landlords, for example, through less impact on CO emissions if they use renewably- the media or letting agents. Policymakers should continue generated electricity. This can be especially valuable to to assist the private rented sector as well as low-income low-income high emitters who are likely to have larger households with older people, and ensure that emission direct rebound effects than other high emitters. The CO reduction policies do not result in more serious fuel emissions caused by the rebound effects can be reduced poverty issues among the low-income high emitters due to when a larger percentage of electricity is generated from increased energy prices as a result of policy interventions. renewable sources. For this reason, policies such as the FIT Financial grants, such as the Winter Fuel Payment subsidy targeted at low-income high-electricity users would be in the UK, could target low-income high energy users attractive for improving carbon mitigation. Furthermore, rather than the current arrangement where people born on the cost of the FIT scheme is shared by all electricity or before 5 May 1953 are eligible to apply for the subsidy regardless of income [72]. customers, which is likely to result in households that do The ﬁndings of this paper not only apply to households not participate in the FIT scheme paying for those who are in the UK, but also other countries where high emitters in the scheme. This could lead to a larger gap between the could comprise clusters of households whose socio- rich and the poor, as there is no provision in the FIT scheme economic characteristics are homogeneous within one to ensure its uptake by low-income households. Therefore, cluster but heterogeneous compared with other clusters. this paper suggests that incentives could be ﬁnanced Xinfang WANG et al. Understanding high-emitting households in the UK 623 Future research can identify the drivers of high energy high-emitter clusters. As reducing energy consumption at consumption at a larger scale through comparing the UK home could lead to rebound effects, it is also important to with other countries. The comparison of drivers of high understand that the range of rebound effects could vary energy consumption across countries would partly depend signiﬁcantly among different high-emitter clusters and on the availability of household survey data in those across various policy measures. More targeted policies countries, which are expected to cover both energy would facilitate a greater amount of emission reductions in consumption and socioeconomic factors at home. Interna- the short to medium term. tional comparison on whether and how the drivers of high While the results indicate that different combinations of socioeconomic factors and dwelling-related characteristics energy consumption differ across countries would con- could all link with high energy consumption and resulting tribute to the global emission reductions by focusing on CO these drivers. It could also offer insight on supra-national emissions, these combinations only explain partly policy making and collaborations for reducing household why some householders are responsible for more CO energy use and CO emissions. emissions than others. The data on energy efﬁciency of the dwelling and appliances are not available for this cluster analysis, and there is no information on high emitters’ 6 Conclusions daily routines and their use of home that could require energy to complete. Further research could be conducted to Household energy consumption accounts for almost a third explore the routines and daily practices of the households of total UK territorial-based CO emissions. It is important who belong to different high-emitting clusters, in order to to reduce emissions from energy use at home in the short to provide a fuller explanation of why these households are medium term for achieving the climate mitigation targets more likely to be high emitters than the others. in the UK and globally. In this paper, attention has been Acknowledgements This work was funded by the School of Mechanical, paid to the high-emitting households and their socio- Aerospace and Civil Engineering and the Sustainable Consumption Institute economic factors, as high emitters could have a larger at the University of Manchester. potential to reduce their CO emissions than the others. Through cluster analysis, the study identiﬁes six different Electronic Supplementary Material Supplementary material is available combinations of socioeconomic factors and dwelling- in the online version of this article at https://doi.org/10.1007/s11708-019- 0647-6 and is accessible for authorized users. related characteristics that can lead to overall high CO emissions from energy use at home. The results show that Open Access This article is licensed under a Creative Commons the high-emitting households belong to several typical Attribution 4.0 International License, which permits use, sharing, adaptation, clusters sharing similar socioeconomic factors and dwell- distribution and reproduction in any medium or format, as long as you give ing-related characteristics within each cluster, but different appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. from other clusters. According to the typical characteristics The images or other third party material in this article are included in the of households in each cluster, households with a male article’s Creative Commons licence, unless indicated otherwise in a credit line HRP, the oldest person over 60, own a detached house to the material. If material is not included in the article’s Creative Commons outright, at least two cars and eight rooms with no children licence and your intended use is not permitted by statutory regulation or at home (Cluster A) are most likely to be high emitters exceeds the permitted use, you will need to obtain permission directly from the copyright holder. among the clusters. The next group of households who are To view a copy of this licence, visit http://creativecommons.org/licenses/ also likely to be high emitters are those who have a male by/4.0/. HRP, oldest person under 59, at least two adults, own a detached house with a mortgage, and at least two cars and eight rooms (Cluster B). Households with a female HRP Notations are also likely to be high emitters if the oldest person is ECO Energy company obligation over 50, they own their property either with a mortgage or outright, and have at least one car and seven rooms FIT Feed-in Tariffs (Cluster D). High emitters in Clusters C, E and F are less GORs Government ofﬁce regions distinguishable from the remaining 90% households, GHG Greenhouse gas compared with Clusters A, B and D, but still shows HRP Household reference person some typical characteristics that high emitters in these clusters share. LCF Living cost and food This paper is of high signiﬁcance not only in the UK, but LESA Landlord’s energy saving allowance also in other countries. Different from the main stream LIHC Low income high cost regression analysis, the cluster study within high emitters OECD Organisation for Economic Co-operation and Development is an innovative approach developed and presented in this ONS Ofﬁce for National Statistics paper, which provides useful information for more targeted policies in the UK and other countries focusing on different UNFCCC United Nations Framework Convention on Climate Change 624 Front. Energy 2019, 13(4): 612–625 14: 50–72 References 20. Büchs M, Schnepf S. Who emits most? Association between socio- 1. Intergovernmental Panel on Climate Change (IPCC). AR5 climate economic factors and UK households’ home energy, transport, change 2014: impacts, adaptation and vulnerability. 2017–01–09, emissions. Ecological Economics, 2013, 90: indirect and total CO available at IPCC website 114–123 2. United Nations. Paris Agreement. 2017–04–28, available at unfccc 21. Lenzen M, Murray S A. The ecological footprint–issues and trends. website ISA Research Paper 01–03. Sydney: University of Sydney, 2003 3. Her Majesty’s Stationary Ofﬁce (HMSO). Climate Change Act 22. Kenny T, Gray N F. A preliminary survey of household and 2008. London: TSO, 2008 personal carbon dioxide emissions in Ireland. Environment Inter- 4. Her Majesty’s Stationary Ofﬁce (HMSO). The Carbon Budget national, 2009, 35(2): 259–272 Order 2009. Statutory instruments Number 1259. London: HMSO, 23. Jones C M, Kammen D M. Quantifying carbon footprint reduction opportunities for US households and communities. Environmental 5. Her Majesty’s Stationary Ofﬁce (HMSO). The Carbon Budget Science & Technology, 2011, 45(9): 4088–4095 Order 2011. Statutory instruments Number 1603. London: HMSO, 24. Vringer K, Blok K. The direct and indirect energy requirements of households in the Netherlands. Energy Policy, 1995, 23(10): 893– 6. Her Majesty’s Stationary Ofﬁce (HMSO). The Carbon Budget Order 2016. Statutory instruments Number 785. London: HMSO, 25. Bhattacharjee S, Reichard G. Socio-economic factors affecting individual household energy consumption: a systematic review. In: 7. Anderson K. Talks in the city of light generate more heat. Nature, The ASME 5th International Conference on Energy Sustainability, 2015, 528(7583): 437 Washington, D.C., USA, 2011 8. Pye S, Li F G N, Price J, Fais B. Achieving net-zero emissions 26. Pachauri S. An analysis of cross-sectional variations in total through the reframing of UK national targets in the post-Paris household energy requirements in India using micro survey data. Agreement era. Nature Energy, 2017, 2(3): 17024 Energy Policy, 2004, 32(15): 1723–1735 9. Department for Business, Energy and Industrial Strategy (BEIS). 27. Kadian R, Dahiya R P, Garg H P. Energy-related emissions and Final UK greenhouse gas emissions national statistics: ﬁnal mitigation opportunities from the household sector in Delhi. Energy estimates of UK greenhouse gas emissions from 1990. London: Policy, 2007, 35(12): 6195–6211 BEIS, 2018 28. Gough I, Abdallah S, Johnson V, Ryan-Collins J, Smith C. The 10. Department for Business, Energy and Industrial Strategy (BEIS). distribution of total greenhouse gas emissions by households in the Energy consumption in the UK. London: BEIS, 2016 UK, and some implications for social policy. London: Centre for 11. Ares E. Feed-in Tariffs. London: House of Commons. 2012 Analysis of Social Exclusion, 2012 12. Her Majesty’s Stationary Ofﬁce (HMSO). The Electricity and Gas 29. Palmer J, Terry N, Armitage P, Godoy-Shimizu D. Savings, beliefs (Energy Companies Obligation) (Amendment) Order 2014. Statu- and demographic change. Cambridge Architectural Research tory instruments Number 1131. London: HMSO, 2014 Limited, Element Energy, and Loughborough University, 2014 13. Department of Energy and Climate Change (DECC). The Green 30. Thumim J, White V. Distributional impacts of personal carbon Deal: A Summary of the Government’s Proposals. URN: 10D/996. trading: a report to the department for environment, food and rural London: DECC, 2010 affairs. London: Department for Environment, Food and Rural 14. Foulds C, Powell J. Using the Homes Energy Efﬁciency Database as Affairs, 2008 a research resource for residential insulation improvements. Energy 31. Fahmy E, Thumim J, White V. The distribution of UK household Policy, 2014, 69: 57–72 CO emissions. Interim Report. JRF programme paper: Climate 15. Dowson M, Poole A, Harrison D, Susman G. Domestic UK retroﬁt change and social justice, York, 2011 challenge: barriers, incentives and current performance leading into 32. Chancel L, Piketty T. Carbon and inequality: from Kyoto to Paris. the Green Deal. Energy Policy, 2012, 50: 294–305 Paris: Paris School of Economics, 2015 16. Balcombe P, Rigby D, Azapagic A. Investigating the importance of 33. Oxfam. Carbon emissions and income inequality. Technical Note. motivations and barriers related to micro-generation uptake in the Oxford, 2015 UK. Applied Energy, 2014, 130: 403–418 34. Element Energy. Further analysis of data from the household 17. Druckman A, Jackson T. Household energy consumption in the UK: electricity usage study: consumer archetypes. Final Report for a highly geographically and socio-economically disaggregated DECC and Defra. Cambridge: Element Energy limited, 2013 model. Energy Policy, 2008, 36(8): 3177–3192 35. Morrison B M, Gladhart P M. Energy and families: the crisis and the 18. Druckman A, Jackson T. The carbon footprint of UK households response. Journal of Home Economics, 1976, 68: 15–18 1990–2004: a socio-economically disaggregated, quasi-multi-regio- 36. Kaza N. Understanding the spectrum of residential energy nal input-output model. Ecological Economics, 2009, 68(7): 2066– consumption: a quantile regression approach. Energy Policy, 2010, 38(11): 6574–6585 19. Baiocchi G, Minx J, Hubacek L. The impact of social factors and 37. Lenzen M. Energy and greenhouse gas cost of living for Australia consumer behaviour on carbon dioxide emissions in the United during 1993/94. Energy, 1998, 23(6): 497–516 Kingdom: a regression based on input-output and geo-demographic 38. Sultana S, Pourebrahim N, Kim H. Household energy expenditures consumer segmentation data. Journal of Industrial Ecology, 2010, in North Carolina: a geographically weighted regression approach. Xinfang WANG et al. Understanding high-emitting households in the UK 625 Sustainability, 2018, 10(5): 1511 56. Department of Energy and Climate Change (DECC). Annual fuel 39. HM Revenue and Customs. Deductions: General rules: main types poverty statistics report. URN: 14D/146 . London: DECC, 2014 of expenses: Landlord’s Energy Savings Allowance (LESA). 2017– 57. Mooi E, Sarstedt M. A concise guide to market research: the 04–28, available at gov.uk website process, data and methods using IBM SPSS statistics. Berlin 40. Ofﬁce for National Statistics (ONS). Living cost and food survey Heidelberg: Springer-Verlag, 2011 volume A: introduction January-December 2012. London: ONS, 58. IBM Knowledge Centre. SPSS statistics two-step cluster algorithms. 2017–04–28, available at ibm.com website 41. Ofﬁce for National Statistics (ONS). Input-output supply and use 59. Okazaki S. What do we know about mobile Internet adopters? tables for 1997 to 2013. London: ONS, 2013 A cluster analysis. Information & Management, 2006, 43(2): 127– 42. Gough I. Carbon mitigation policies, distributional dilemmas and social policies. Journal of Social Policy, 2013, 42(2): 191–213 60. Druckman A, Chitnis M, Sorrell S, Jackson T. Missing carbon 43. Brand C. Transport and carbon emissions analysis. Project Paper reductions? Exploring rebound and backﬁre effects in UK house- Number 2. Oxford: Environmental Change Institute, University of holds. Energy Policy, 2011, 39(6): 3572–3581 Oxford, 2013 61. Oke T R. Inadvertent climate modiﬁcation. In: Boundary Layer 44. OECD. OECD framework for statistics on the distribution of Climates. London: Routledge, 1987 household income, consumption and wealth. OECD publishing. 62. Lo C P, Quattrochi D A, Luvall J C. Application of high-resolution 2017–04–28, available at keepeek.com website thermal infrared remote sensing and GIS to assess the urban heat 45. Yohanis Y G, Mondol J D, Wright A, Norton B. Real-life energy use island effect. International Journal of Remote Sensing, 1997, 18(2): in the UK: how occupancy and dwelling characteristics affect 287–304 domestic electricity use. Energy and Building, 2008, 40(6): 1053– 63. Shove E. Comfort, Cleanliness and Convenience: the Organisation of Normality. Oxford and New York: Berg, 2003 46. Utley J I, Shorrock L D. Domestic energy fact ﬁle 2008. BRE 64. Gram-Hanssen K. Standby consumption in households analyzed Environment, DECC and EST, 2008 with a practice theory approach. Journal of Industrial Ecology, 2010, 47. The Scottish Government. Scottish house condition survey: key 14(1): 150–165 ﬁndings 2010. Edinburgh: The Scottish Government, 2010 65. Maller C, Horne R, Dalton T. Green renovations: Intersections of 48. Pullinger M, Browne A, Anderson B, Medd W. Patterns of water: daily routines, housing aspirations and narratives of environmental the water related practices of households in southern England, and sustainability. Theory and Society, 2012, 29: 255–275 their inﬂuence on water consumption and demand management. 66. Strengers Y, Nicholls L, Maller C. Curious energy consumers: Final Report of the ARCC-Water/ SPRG Patterns of Water Projects. humans and nonhumans in assemblages of household practice. Lancaster: Lancaster University, 2013 Journal of Consumer Culture, 2016, 16(3): 761–780 49. Lamb W F, Steinberger J K, Bows-Larkin A, Peters G P, Roberts J 67. Butler C, Parkhill K A, Shirani F, Henwood K, Pidgeon N. T, Wood F R. Transitions in pathways of human development and Exploring the dynamics of energy demand through a biographical carbon emissions. Environmental Research Letters, 2014, 9(1): lens. Nature and Culture, 2014, 9(2): 164–182 68. Sorrell S, Dimitropoulos J. The rebound effect: microeconomic 50. Field A. Cluster analysis. 2017–04–28, available at website of deﬁnitions, limitations and extensions. Ecological Economics, statisticshell.com 2008, 65(3): 636–649 51. Everitt B S, Landau S, Leese M, Stahl D. Cluster Analysis. 5th ed. 69. Chitnis M, Sorrell S, Druckman A, Firth S K, Jackson T. Who Chichester: John Wiley & Sons Ltd., 2011 rebounds most? Estimating direct and indirect rebound effects for 52. AEA. 2012 Guidelines to Defra/DECC’s GHG conversion factors different UK socioeconomic groups. Ecological Economics, 2014, for company reporting. Produced for the Department for Environ- 106: 12–32 ment, Food and Rural Affairs (Defra) and the Department of Energy 70. Sorrell S, Dimitropoulos J, Sommerville M. Empirical estimates of and Climate Change (DECC) (Version 1.0), London, 2012 the direct rebound effect: a review. Energy Policy, 2009, 37(4): 53. Department of Energy and Climate Change (DECC). Quarterly 1356–1371 energy prices. URN: 12D/276D. London: DECC, 2012 71. Boardman B. Home truths: a low-carbon strategy to reduce UK 54. The Consumer Council. Comparative domestic cost of gas voil housing emissions by 80% by 2050. ECI Research Report 34. report. Belfast: The Consumer Council, 2013 Oxford: Environmental Change Unit, University of Oxford, 2007 55. Department of Energy and Climate Change (DECC). Quarterly 72. UK Government. Winter fuel payment. 2019–03–01, available at energy prices. URN: 14D/276C. London: DECC, 2014 website of gov.ukeligibility http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Frontiers in Energy Springer Journals http://www.deepdyve.com/lp/springer-journals/understanding-high-emitting-households-in-the-uk-through-a-cluster-K3LsESvF2e

Loading next page...

References (59)

(2009)
Statutory instruments Number 1259
Y Strengers, L Nicholls, C Maller (2016)
Curious energy consumers: humans and nonhumans in assemblages of household practice
Journal of Consumer Culture, 16
M Büchs, S Schnepf (2013)
Who emits most? Association between socioeconomic factors and UK households’ home energy, transport, indirect and total CO2 emissions
Ecological Economics, 90
B M Morrison, P M Gladhart (1976)
Energy and families: the crisis and the response
Journal of Home Economics, 68
T R Oke (1987)
Boundary Layer Climates
(2011)
Statutory instruments Number 1603
E Ares (2012)
Feed-in Tariffs
W F Lamb, J K Steinberger, A Bows-Larkin, G P Peters, J T Roberts, F R Wood (2014)
Transitions in pathways of human development and carbon emissions
Environmental Research Letters, 9
(2013)
Living cost and food survey volume A: introduction January-December 2012
T Kenny, N F Gray (2009)
A preliminary survey of household and personal carbon dioxide emissions in Ireland
Environment International, 35
(2016)
Statutory instruments Number 785
(2010)
URN: 10D/996
I Gough (2013)
Carbon mitigation policies, distributional dilemmas and social policies
Journal of Social Policy, 42
S Sultana, N Pourebrahim, H Kim (2018)
Household energy expenditures in North Carolina: a geographically weighted regression approach
Sustainability, 10
E Shove (2003)
Comfort, Cleanliness and Convenience: the Organisation of Normality
(2014)
URN: 14D/146
(2018)
Final UK greenhouse gas emissions national statistics: final estimates of UK greenhouse gas emissions from 1990
N Kaza (2010)
Understanding the spectrum of residential energy consumption: a quantile regression approach
Energy Policy, 38
(2014)
URN: 14D/276C
K Gram-Hanssen (2010)
Standby consumption in households analyzed with a practice theory approach
Journal of Industrial Ecology, 14
C Butler, K A Parkhill, F Shirani, K Henwood, N Pidgeon (2014)
Exploring the dynamics of energy demand through a biographical lens
Nature and Culture, 9
(2013)
Comparative domestic cost of gas voil report
G Baiocchi, J Minx, L Hubacek (2010)
The impact of social factors and consumer behaviour on carbon dioxide emissions in the United Kingdom: a regression based on input-output and geo-demographic consumer segmentation data
Journal of Industrial Ecology, 14
(2010)
Scottish house condition survey: key findings 2010
C M Jones, D M Kammen (2011)
Quantifying carbon footprint reduction opportunities for US households and communities
Environmental Science & Technology, 45
M Pullinger, A Browne, B Anderson, W Medd (2013)
Final Report of the ARCC-Water/SPRG Patterns of Water Projects
J Thumim, V White (2008)
Distributional impacts of personal carbon trading: a report to the department for environment, food and rural affairs
C P Lo, D A Quattrochi, J C Luvall (1997)
Application of high-resolution thermal infrared remote sensing and GIS to assess the urban heat island effect
International Journal of Remote Sensing, 18
K Anderson (2015)
Talks in the city of light generate more heat
Nature, 528
B S Everitt, S Landau, M Leese, D Stahl (2011)
Cluster Analysis
S Sorrell, J Dimitropoulos (2008)
The rebound effect: microeconomic definitions, limitations and extensions
Ecological Economics, 65
M Chitnis, S Sorrell, A Druckman, S K Firth, T Jackson (2014)
Who rebounds most? Estimating direct and indirect rebound effects for different UK socioeconomic groups
Ecological Economics, 106
(2013)
Input-output supply and use tables for 1997 to 2013
S Pachauri (2004)
An analysis of cross-sectional variations in total household energy requirements in India using micro survey data
Energy Policy, 32
S Okazaki (2006)
What do we know about mobile Internet adopters? A cluster analysis
Information & Management, 43
R Kadian, R P Dahiya, H P Garg (2007)
Energy-related emissions and mitigation opportunities from the household sector in Delhi
Energy Policy, 35
C Foulds, J Powell (2014)
Using the Homes Energy Efficiency Database as a research resource for residential insulation improvements
Energy Policy, 69
(2012)
URN: 12D/276D
(2016)
Energy consumption in the UK
P Balcombe, D Rigby, A Azapagic (2014)
Investigating the importance of motivations and barriers related to micro-generation uptake in the UK
Applied Energy, 130
A Druckman, M Chitnis, S Sorrell, T Jackson (2011)
Missing carbon reductions? Exploring rebound and backfire effects in UK households
Energy Policy, 39
A Druckman, T Jackson (2009)
The carbon footprint of UK households 1990–2004: a socio-economically disaggregated, quasi-multi-regional input-output model
Ecological Economics, 68
I Gough, S Abdallah, V Johnson, J Ryan-Collins, C Smith (2012)
The distribution of total greenhouse gas emissions by households in the UK, and some implications for social policy
B Boardman (2007)
ECI Research Report 34
L Chancel, T Piketty (2015)
Carbon and inequality: from Kyoto to Paris
Y G Yohanis, J D Mondol, A Wright, B Norton (2008)
Real-life energy use in the UK: how occupancy and dwelling characteristics affect domestic electricity use
Energy and Building, 40
M Lenzen (1998)
Energy and greenhouse gas cost of living for Australia during 1993/94
Energy, 23
S Pye, F G N Li, J Price, B Fais (2017)
Achieving net-zero emissions through the reframing of UK national targets in the post-Paris Agreement era
Nature Energy, 2
(2014)
Statutory instruments Number 1131
A Druckman, T Jackson (2008)
Household energy consumption in the UK: a highly geographically and socio-economically disaggregated model
Energy Policy, 36
K Vringer, K Blok (1995)
The direct and indirect energy requirements of households in the Netherlands
Energy Policy, 23
C Brand (2013)
Project Paper Number 2
(2013)
Final Report for DECC and Defra
(2008)
Climate Change Act 2008
M Lenzen, S A Murray (2003)
ISA Research Paper 01-03
C Maller, R Horne, T Dalton (2012)
Green renovations: Intersections of daily routines, housing aspirations and narratives of environmental sustainability
Theory and Society, 29
E Mooi, M Sarstedt (2011)
A concise guide to market research: the process, data and methods using IBM SPSS statistics
M Dowson, A Poole, D Harrison, G Susman (2012)
Domestic UK retrofit challenge: barriers, incentives and current performance leading into the Green Deal
Energy Policy, 50
S Sorrell, J Dimitropoulos, M Sommerville (2009)
Empirical estimates of the direct rebound effect: a review
Energy Policy, 37

Publisher: Springer Journals
Copyright: Copyright © 2019 by The Authors
Subject: Energy; Energy Systems; Energy, general
ISSN: 2095-1701
eISSN: 2095-1698
DOI: 10.1007/s11708-019-0647-6
Publisher site: See Article on Publisher Site

Abstract

Front. Energy 2019, 13(4): 612–625 https://doi.org/10.1007/s11708-019-0647-6 RESEARCH ARTICLE Xinfang WANG, Ming MENG Understanding high-emitting households in the UK through a cluster analysis © The Authors (2019). This article is published with open access at link.springer.com and journal.hep.com.cn Abstract Anthropogenic climate change is a global 1 Introduction problem that affects every country and each individual. It is largely caused by human beings emitting greenhouse Climate change has various impacts on water, food, gases into the atmosphere. In general, a small percentage of industry, health, ecosystems and coastal systems [1]. To the population is responsible for a large amount of avoid the ‘dangerous’ impacts of climate change, parties to emissions. This paper focuses on high emitters and their the United Nations Framework Convention on Climate CO emissions from energy use in UK homes. It applies a Change (UNFCCC) came to an international agreement— cluster approach, aiming to identify whether the high the Paris Agreement— on ‘holding the increase in global emitters comprise clusters where households in each average temperature to well below 2°C above pre- cluster share similar characteristics but are different from industrial levels’ [2]. The UK introduced its own carbon the others. The data are mainly based on the Living Cost budgets and stated its commitment in the Climate Change and Food survey in the UK. The results show that after Act 2008 to achieve an 80% emission reduction by 2050 equivalising both household emissions and income, the compared to the 1990 baseline [3]. To achieve these high emitters can be clustered into six groups which share emission reduction targets, the ﬁrst ﬁve UK carbon budgets similar characteristics within each group, but are different covering the period from 2008 to 2032 have been set in law from the others in terms of income, age, household [4–6]. Anderson [7] and Pye et al. [8] have argued that the composition, category and size of the dwelling, and tenure Paris Agreement requires more radical and rapid emission type. The clustering results indicate that various combina- reductions than the UK targets. tions of socioeconomic factors, such as low-income single The energy used in homes accounted for around 27% of female living in an at least six-room property, or high- territorial-based CO emissions in the UK in 2016 income retired couple owning a large detached house, categorised by end users, while the industry and transport could all lead to high CO emissions from energy use at sectors were responsible for 30% and 36% of the total UK home. Policymakers should target each high-emitter CO emissions respectively [9]. Due to the switch from cluster differently to reduce CO emissions from energy solid fuel to gas in electricity generation and reduced solid consumption at home more effectively. fuel use in homes, the CO emissions from UK household energy use decreased by 35% between 1990 and 2016 [9]. Keywords cluster analysis, emissions reduction, energy However, the total energy consumption by households use, high emitters, household energy consumption, socio- increased by 1% in 2016 compared to the 1990 level, economic factors measured by tonnes of oil equivalent [10]. To reduce CO emissions from energy use at home, the UK launched Feed-in Tariffs (FIT) in April 2010, which supported households, businesses and other organisations to generate electricity from renewable sources. Households who Received Dec. 29, 2018; accepted Jul. 15, 2019; online Dec. 15, 2019 participate in the FIT are paid by their energy suppliers Xinfang WANG with both a generation tariff for each unit of energy they School of Chemical Engineering, University of Birmingham, Edgbaston, generate from renewable source and an export tariff for Birmingham B15 2TT, UK E-mail: x.wang.10@bham.ac.uk each unit that feeds back into the grid [11]. Furthermore, for improving energy efﬁciency at home, the UK Ming MENG government launched the Green Deal and Energy Department of Economics and Management, North China Electric Company Obligation (ECO) that required energy compa- Power University, Baoding 071003, China Xinfang WANG et al. Understanding high-emitting households in the UK 613 nies to provide heating and insulation improvements to capita and household size could be caused by the households [12–14]. The Green Deal allowed households economies of scale at the household level, as members of to repay the cost of installed insulation system through the same household could share gas and electricity use for their saved energy bills [13,14], although participation was space heating, cooking, lighting and the utilization of low partly because of the uncertainties on energy savings various appliances at home most of the time [24]. The age that could be achieved and the house resale value due to of household members also matters [20]. For example, attached Green Deal loan to the property [15,16]. The ECO older people might require more energy for space heating was also launched early in 2013 to provide additional at home due to their poor health [25]. support, especially for vulnerable households and hard-to- Household income is another key variable discussed in treat homes, which placed legal obligations on energy the literature as it inﬂuences both emissions from direct suppliers to deliver energy efﬁciency measures to energy use at home, and emissions embedded in consumed residential energy users [12]. As discrepancies in terms products and services [26–28,37]. Household composition of wealth, well being and emissions exist at an intra- inﬂuences the economic resources available to a household national level, it is important to focus on the high-emitting measured by income, due to the economies of scale [44]. households who could have a larger potential for reducing The reason for this is that the members of the same their energy consumption and CO emissions than others. household not only share the energy used for performing Previous research did not explore different groups among different practices, but also share other products and high emitters, or how emission reduction policies could services such as furniture, cookware, cars, Internet service, target each group differently [11–42]. TV license, and so on. To address the inﬂuences from the This paper aims to identify particular groups within the economies of scale, the disposable household income society who are likely to be high emitters from their energy (gross income after taxes and beneﬁts) is equivalised in this consumption at home. This will facilitate more targeted paper with the commonly used Organisation for Economic policies for reducing household CO emissions in the UK. Co-operation and Development (OECD) [44] modiﬁed Therefore, the cluster analysis is applied to explore the scales (Table 1). make-up of the high-emitting household group in terms of Table 1 Household income equivalisation factors socioeconomic factors and dwelling-related characteris- Member of the household Modiﬁed OECD for tics. After the introduction in Section 1, Section 2 draws on disposable income the literature related to the distribution of UK household First adult 1 emissions, and outlines the research gap that this paper Second and subsequent adults 0.5 addresses. Section 3 explains the data used to identify and Children aged 14–18 0.5 explore high emitters, as well as the cluster method that is used for the analysis. After presenting the clustering results Children under 14 0.3 in Section 4 with further discussions in Section 5, it concludes by considering the importance of the analysis for targeted emission reduction policies among households Studies have identiﬁed that in general the larger the ﬂoor in the UK in Section 6. area of the dwelling, the more energy will be required for space heating and other energy use, assuming other conditions of the dwelling are similar [25,26,30,31,36]. 2 Previous work on socioeconomic factors Space and water heating accounted for around 81% of total and household energy consumption energy use at home in the UK between 1990 and 2013 [10]. For this reason, the size of dwelling plays an important role The CO emissions produced by households could vary in overall household energy consumption and related CO 2 2 signiﬁcantly [32,33,43]. Chancel and Piketty [32] have emissions, especially during the winter, due to the energy estimated that the top 10% of emitters account for around required for space heating. The electricity use for lights and 45% of direct and embedded greenhouse gas (GHG) appliances is also positively correlated with the size of the emissions globally. Likewise, Oxfam [33] has estimated dwelling [29,30,45], as there may be more appliances in that the total share of CO emissions from the top 10% of larger houses that are used in everyday life, than in an high-income people is approximately 50%. Previous average-sized dwelling. literature for the UK [17–20] and other countries [21– The type of dwelling, whether it is a detached, semi- 27,35–38] has shown that household emissions are related detached or terraced house, or a bungalow or ﬂat, also to a variety of socioeconomic factors and dwelling-related inﬂuences energy use and the related CO emissions, characteristics. For example, there exist a negative link largely due to the diverse amount of energy required for between CO emissions per capita and household size (the space heating across these different types of dwellings number of people in a household), and a positive link [46,47]. In addition to energy use for space heating, between CO emissions per household and household size electricity use for lighting and appliances is also inﬂuenced [21–23]. This negative link between CO emissions per by the type of dwelling [20,45]. Büchs and Schnepf [20] 2 614 Front. Energy 2019, 13(4): 612–625 have identiﬁed that the CO emissions from total energy electricity consumption among households excluding use at home are the highest among households living in electric heating and electric showers, where it is found detached houses, followed by the semi-detached occupiers, that 85% of the high electricity-consuming households in and then those in terraced houses. Households living in their study (39 out of 46 households that belong to the top ﬂats have the lowest energy emissions overall. This could 20% high-electricity users) have at least one key be partly because detached houses are likely to have a contributing factor that could lead to their higher electricity greater ﬂoor area, while ﬂats are likely to have a smaller consumption. The key contributing factors are: at least one in general in the UK [46]. The shared insulation and three people living in the dwelling, the dwelling size being larger than 130 m heat between dwellings also reduce the heat loss among , the age of the Household Reference 1) semi-detached houses, terraced houses and ﬂats, compared Person (HRP) being between 45 and 54, the HRP being with detached dwellings [46]. The mean heat loss for unemployed but not retired, and householders belonging to 2) different dwelling types in the UK is presented in Table 2, the professional and managerial socioeconomic group . which shows the mean heat loss ranges from 167 W/°C for While Palmer et al. [29] have estimated the average value a ﬂat, up to 342 W/°C for a detached house [46]. of each of the socioeconomic factors and dwelling-related characteristics among high electricity-consuming house- Table 2 Mean heat loss of dwelling types holds, they have not identiﬁed whether different combina- –1 Type of dwelling Heat loss/(W$°C ) tions of these factors are more likely to lead to high Detached 342 electricity use collectively. Overall, previous studies have identiﬁed socioeconomic factors and dwelling-related Semi-detached 264 characteristics that inﬂuence energy consumption and Terraced 235 related CO emissions. However, they have not explored Bungalow 225 whether various combinations of the factors are all likely to Flat 167 link with particularly high CO emissions, which is crucial to understand if the aim of reducing household energy UK mean 247 consumption and related CO emissions is to be achieved. For example, studies may ﬁnd that, in general, high- Tenure type could also inﬂuence the energy consump- income households are likely to have more emissions than tion at home [36]. In general, private rented domestic households with an average income. However, such buildings have relatively low thermal insulation installed studies do not show that if a household does not belong and require more energy for space heating, due largely to to a high-income group, it could be a high emitter due to a the ‘tenant-landlord problem’ [17]. The ‘tenant-landlord combination of other socioeconomic factors or dwelling- problem’ refers to the mismatch between landlords who related characteristics, such as household composition, size pay the cost of insulation and tenants who receive the of dwelling and tenure type. To address this gap in the beneﬁts [17]. As a ﬁnancial incentive, the UK government literature, this paper undertakes cluster analysis within introduced the Landlord’s Energy Saving Allowance high-emitting households in the UK. (LESA) between April 2004 and April 2015, which The clustering technique has been used in several other provided grants to landlords for upfront payments of energy and emission related studies, both to cluster various energy efﬁciency measures such as loft and cavity countries and households [34,48,49]. The aim of cluster wall insulation, solid wall insulation, draft-prooﬁng and analysis is to classify the whole sample into distinguished ﬂoor insulation [39]. Although results from previous clusters, where it is relatively homogeneous within each studies [17,20,30,45] indicate that high energy users and single cluster and heterogeneous across different clusters. high emitters are likely to own a house that is not only On a country level, Lamb at al. [49] have applied the large, but also detached, these studies have not explicitly clustering technique to identify the similarity and diversity explored whether some high emitters do not own large of human development and CO emissions between detached houses; and if this is the case, what other factors developing and developed nations. On a household level, could collectively lead to their high energy use level. Pullinger et al. [48] have used the clustering approach to In particular, the household income, age of house- identify distinct household groups according to their water holders, and category of dwelling have been identiﬁed as using practices. Likewise, Element Energy [34] have the main inﬂuencing factors on household energy investigated household electricity use in the UK using expenditure in North Carolina in the US [38]. Likewise cluster analysis. It conducted cluster analysis among in the UK, Palmer et al. [29] have focused only on householders in their sample based on their annual 1) The HRP is the person who owns the dwelling or is responsible for renting it. If the dwelling is joint owned or rented, the person with the highest income would be the HRP. If two or more householders have the same highest income or they all have zero income, the oldest should be identiﬁed as the HRP. 2) Palmer et al. [29] have divided all households into three socioeconomic groups: professional and managerial; supervisory, clerical and skilled manual; and semi-skilled, unskilled, pensioner and non-working group. Xinfang WANG et al. Understanding high-emitting households in the UK 615 electricity consumption, 6–7 pm peak-time electricity use, from energy consumption data. Household expenditure on socioeconomic factors and dwelling-related characteris- gas, electricity, and oil are collected from the 2012 Living tics, number and energy efﬁciency level of appliances, as Cost and Food (LCF) survey, which is a household survey well as their climate change attitudes and electricity carried out by the Ofﬁce for National Statistics (ONS) in conservation behaviors [34]. Compared to the traditional the UK [40]. The survey covers the whole UK, including regression analysis, the clustering technique is especially England, Scotland, Wales, and Northern Ireland [40]. beneﬁcial for studies exploring whether there are different Households in Northern Ireland are excluded in the combinations of independent variables and corresponding analysis presented here, due to the much higher level of values that lead to a similar value for the dependent oil use at home than other regions. Using oil leads to variable (for example, high CO emissions from energy around 38% higher CO emission per kWh than natural gas 2 2 use at home) [48]. The analysis presented in this paper [52]. 5593 households in total were selected using a multi- applies the cluster method to classify the high-emitting stage stratiﬁed random sample method from approximately households in order to identify whether the high emitters 26.4 million UK households in the 2012 LCF survey [41]. comprise several groups which are more homogeneous in Initially, the ﬁrst stratum in the sample selection was terms of socioeconomic factors and dwelling-related deﬁned by the Government Ofﬁce Regions (GORs) and characteristics within each group but heterogeneous across two variables, which were social class of the HRP and different groups. The homogeneity within a cluster means ownership of cars. Then, 638 out of 1.8 million postal that the households within one cluster are grouped together sectors were randomly selected from the ﬁrst stratum. All by well-deﬁned similarities. On the contrary, the hetero- households in each selected postal sector were accessed for geneity across clusters means that households in one the survey. As a result, 52% of the selected households cluster are separated from those in other clusters by well- responded to the survey, which constituted the 5593 deﬁned dissimilarities [50,51]. This clustering method has households in the data set. not been used to classify high emitters in other studies, and Less than 1% of the households use other fuels, such as will be an important original contribution to knowledge solid fuel or Calor gas, in the 2012 LCF survey. For this through this research. reason, the other fuels are not considered in estimating total CO emissions from energy use at home. This analysis estimates the energy used by each household in the 3 Material and methods selected survey sample by dividing the household energy bills by corresponding energy unit prices as shown in 3.1 Material Table 3 [53,54]. The price for domestic oil in 2012 was also obtained from Department of Energy and Climate There are currently no data sets in the UK that provide both Change (DECC) [55], with no regional prices available. household CO emissions and socioeconomic factors [20]. After calculation, gas and electricity use are measured in This research thus estimates household CO emissions kWh and oil consumption is measured in liters, instead of Table 3 Gas and electricity unit prices across UK regions for different payment methods Gas unit price by payment method/£ Electricity unit price by payment method/£ Government ofﬁce regions Credit Direct Prepayment Overall Credit Direct Prepayment Overall debit meters debit meters North East 4.58 4.21 4.54 4.37 14.89 13.63 14.87 14.18 North West and Merseyside 4.61 4.24 4.59 4.41 15.14 13.85 15.19 14.46 Yorkshire and the Humber 4.61 4.21 4.60 4.38 14.85 13.55 14.81 14.16 East Midlands 4.57 4.23 4.61 4.38 14.78 13.63 14.86 14.14 West Midlands 4.72 4.30 4.62 4.48 15.10 13.75 15.04 14.38 Eastern 4.62 4.28 4.59 4.42 14.83 13.64 14.80 14.16 London 4.69 4.37 4.62 4.53 14.82 13.73 14.86 14.38 South East 4.7 4.32 4.59 4.47 14.70 13.60 14.72 14.04 South West 4.67 4.33 4.59 4.45 15.66 14.53 15.69 15.03 Wales 4.65 4.32 4.61 4.47 16.01 14.62 15.84 15.25 Scotland 4.59 4.20 4.54 4.36 15.58 14.28 15.32 14.84 Northern Ireland –– – 4.47 17.07 16.44 16.65 16.72 UK (excluded 4.65 4.28 4.59 4.43 15.13 13.90 15.20 14.48 Northern Ireland for gas unit price) 616 Front. Energy 2019, 13(4): 612–625 British pounds. For this reason, there is uncertainty about households with more members than average. Thus the using a monetary value to estimate energy consumption analysis presented in this paper applies DECC’s [56] Low volume, because the actual volume of energy measured in Income High Cost (LIHC) equivalisation factors (Table 4) kWh or liters that each household used is not available. to equivalise household CO emissions estimates before The gas, electricity, and oil use are converted into CO deﬁning, clustering and identifying the high-emitting emissions with factors obtained from the AEA [52]. The households. The LIHC equivalisation scale is based on conversion factors for electricity in 2012 are the average the energy requirement of the households, which was used grid conversion factors over the previous ﬁve years, which by DECC [56] to identify households in fuel poverty. are updated annually [52]. The calculations result in an Table 4 Equivalisation factors for fuel bills under the LIHC deﬁnition estimation of 5.3 tonnes CO /year for average UK of fuel poverty household emissions from energy use at home, based on Number of people in the household Equivalisation factor the 2012 LCF survey. This estimation is consistent with the results in Büchs and Schnepf [20], which shows that One 0.82 average household emissions from energy use at home are Two 1.00 about 5.1 tonnes CO /year during 2006–2009. Three 1.07 In addition to household energy expenditure, the 2012 Four 1.21 LCF survey also provides socioeconomic factors and Five or more 1.32 dwelling-related information for households. The analysis aims to include as many socioeconomic factors and dwelling-related characteristics as possible from the 2012 After equivalising the household CO emission esti- LCF survey for data analysis in order to provide a fuller mates from energy use at home, the top 10% of emitting picture of who the high emitters are and why they emit households are deﬁned as high emitters for the cluster more than others. Based on the studies associated with analysis, which constitute 510 households. The 10% range household CO emissions from energy use at home and is selected to be consistent with relevant studies conducted socioeconomic factors that are introduced in Section 2, the by Brand [43], Chancel and Piketty [32], and Oxfam [33]. household variables included in the analysis are household composition, tenure type, category of dwelling, number of rooms in the accommodation, equivalised disposable 3.2.2 Cluster method household income, age of the oldest person, sex of HRP, GORs, as well as ownership of cars and second dwelling in There are three principal clustering methods: hierarchical the UK. The ownership of cars and second dwellings in the clustering, k-means clustering and Two-step clustering UK are included in the analysis as an additional indicator approaches. Among these three clustering methods, only of the wealth level of the households. The sex of HRP is the Two-step cluster ﬁts with the mixed data of continuous included in the analysis to complement the information on and categorical variables [51]. Therefore, the Two-step household composition. Education level data were also cluster method is selected for the analysis. The continuous collected in the 2012 LCF survey, but 32% of household variables are standardised using the Standard Score (also members did not provide this information, thus the named as Z Score). The categorical variables are variables cannot be used dependably in the analysis. All manipulated as dummy variables, with a numerical value other available socioeconomic factors and dwelling-related of 0 or 1. In other words, if the answer is ‘yes’ for the characteristics in the 2012 LCF survey are covered by the dummy variable (For example, do the household occupants selected variables in the cluster analysis to cluster the high- live in a detached house?), the variable has a numerical emitting households. value 1. If the answer is ‘no’, a numerical value 0 is allocated to the variable. The Pearson correlation tests are 3.2 Methods of analysis then undertaken to check the correlation between any two of the selected continuous variables, and the Pearson’s Chi-square tests are used to check the correlation between 3.2.1 Equivalising household CO emissions any two categorical variables [57]. As explained in Section 2, household composition The ﬁrst step of the Two-step cluster analysis is called inﬂuences energy requirements at home as members in pre-cluster, where the data are scanned one-by-one to the same household are able to share the energy used for decide whether to merge the data with the previously space heating, lighting, cooking, and appliance use most of formed clusters or start a new cluster, according to the log- the time. The analysis aims to identify who the high likelihood distance criterion [58]. The second step of the emitters are based on CO emissions estimates from their Two-step cluster analysis merges the sub-clusters identi- energy consumption. Not equivalising the household CO ﬁed in the ﬁrst step, where the ﬁnal number of clusters is emissions estimates is likely to result in the deﬁned high- decided through two stages. At stage one, the initial emitting households comprising a larger percentage of estimate of the number of clusters is computed using the Xinfang WANG et al. Understanding high-emitting households in the UK 617 Bayesian information criterion (BIC) criterion, which is genous each individual cluster is and the more hetero- commonly used as an objective selection criteria to avoid geneity exists across different clusters. The cluster quality arbitrariness in deciding the number of clusters [59]. At is treated as ‘poor’ if the Silhouette measure is between –1 stage two, in order to decide the ﬁnal number of clusters, and 0.2, while it is ‘fair’ if it is between 0.2 and 0.5 and the largest relative increase in distance between the two ‘good’ if it is larger than 0.5 [58]. closest clusters is identiﬁed using the ratio calculation, shown in Eq. (1) [58]. 4 Results d ðC Þ min k RðkÞ¼ , (1) d ðC Þ min kþ1 As mentioned in Section 3, the Pearson correlation tests are undertaken among the selected continuous variables, and where C is the cluster model containing k clusters and the Pearson’s Chi-square tests are used to check the d (C ) is the minimum cluster distance for cluster model min k correlations between any two categorical variables [57]. C . The larger the absolute value derived from the Pearson The ﬁnal number of clusters is decided by comparing the correlation is, the more correlated the two continuous two largest R ratios. If the largest is 1.15 times greater than variables are. Likewise, the larger the Cramer’s V for the second largest, the model with the largest R ratio is Pearson’s Chi-square is, the more correlated the two selected as the optimal number of clusters; alternatively, categorical variables are. If the absolute value derived from from those two models with the largest R ratio, the one the Pearson correlation or the Cramer’s V for Pearson’s with the larger number of clusters is selected as the optimal Chi-square is close to 1, it may inﬂuence the cluster results. number of clusters [58]. The cluster quality is measured by This is because that in this case, the inﬂuence of the two the ‘Silhouette measure of cohesion and separation’, which related variables on the clustering results would be similar; is calculated by using Eq. (2). including both variables means that the inﬂuence is bðxÞ– aðxÞ counted twice during the clustering procedure. According sðxÞ¼ , (2) to the correlation test results in Table 5, the number of cars maxfaðxÞ,bðxÞg with the number of rooms, and the number of rooms with where s(x) is the ‘Silhouette measure of cohesion and the household income are more correlated continuous separation’, a(x) is the average distance of x to all other variables than others. For categorical variables, the sex of cases in the same cluster, and b(x) is the minimum average the HRP and the composition of the household are more distance of x to cases in any of the other clusters. correlated than others (Table 6). The analysis has included The larger the Silhouette measure, the more homo- a relatively large sample (510 high-emitting households) to Table 5 Correlations between continuous variables Pearson Second dwelling Cars and vans in Weekly disposable Rooms in Age of the correlation in the UK household household income accommodation oldest person Second dwelling 1 0.133** 0.266** 0.232** 0.014 in the UK Cars and vans in household 0.133** 1 0.341** 0.485** 0.094* Weekly 0.266** 0.341** 1 0.480** 0.121** disposable household income Rooms in accommodation 0.232** 0.485** 0.480** 1 0.233** Age of the oldest person 0.014 0.094* 0.121** 0.233** 1 Notes: ** indicates that the correlation is signiﬁcant at the 0.01 level (2-tailed) while * indicates that the correlation is signiﬁcant at the 0.05 level (2-tailed). Table 6 Correlations between categorical variables Composition of household Category of dwelling Tenure type Sex of HRP GORs Cramer’s V for Pearson’s Chi-square 1 0.237* 0.320** 0.574** 0.207 Composition of household 0.237* 1 0.263** 0.213** 0.181** Category of dwelling 0.320** 0.263** 1 0.294** 0.166* Tenure type 0.574** 0.213** 0.294** 1 0.180 Sex of HRP 0.207 0.181** 0.166* 0.180 1 GORs Notes: ** indicates that the correlation is signiﬁcant at the 0.01 level (2-tailed) while * indicates that the correlation is signiﬁcant at the 0.05 level (2-tailed). 618 Front. Energy 2019, 13(4): 612–625 reduce the risk of clustering results being inﬂuenced by socioeconomic factors and dwelling-related characteristics correlations between variables. Furthermore, the value for all six identiﬁed high-emitter clusters. distribution of each variable, comparing between the high- As shown in Table 7, the government ofﬁce region and emitting households and the remaining 90% households in ownership of second dwelling in the UK are not the 2012 LCF survey sample, are drawn in Fig. 1. Figure 1 distinguishable among the high-emitting households. The shows that none of the variables would dominate the household composition, income, category of dwelling, cluster results, because the values of each variable of high- tenure type, age of the oldest person, sex of HRP, number emitting households are distributed across all ranges. of vehicles owned and rooms in accommodation collec- Likewise, the values of each variable of the remaining 90% tively inﬂuence CO emissions from energy use at home. households are also distributed across all ranges. There- For example, if a two-adult household does not belong to fore, all the continuous and categorical variables are any high-income clusters (Clusters A and B), but rents a included for the clustering process. dwelling that is poorly insulated without gas central As a result, six high-emitter clusters are identiﬁed, with a heating, they can require more energy for space heating ‘fair’ quality being achieved for the cluster results which would result in high CO emissions from energy use measured with the ‘Silhouette measure of cohesion and at home. On the other hand, if the households are high- separation’ (Section 3.2.2). Table 7 lists the selected income ones who own a ﬂat outright and work outside the Fig. 1 Relationship between CO emissions and different variables—high emitters (in red) versus the remaining 90% households (in blue) (The y axes are equivalised annual household CO emissions. The unit for equivalised weekly disposable household income in (g) is ‘£’. The household income, the age of the oldest person, the number of second dwelling in the UK, as well as the number of rooms and cars are capped at the highest value in the 2012 LCF survey, as shown in the relevant diagrams, for the purpose of anonymisation. The representation of each value for all categorical variables is given in Electronic Supplementary Material.) Xinfang WANG et al. Understanding high-emitting households in the UK 619 Table 7 Key distinguishable variables for identiﬁed high-emitter clusters Variables Cluster A Cluster B Cluster C Cluster D Cluster E Cluster F Number of households 104 107 75 56 120 48 Composition of household 71% two adults, 28% two adults, 23% one adult with children; 34% single female; 18% two adults, no children; 56% two adults, no children no children; 39% two adults with children; 30% two adults, 51% two adults with children; no children; 46% two adults 12% two adults, no children no children; 21% at least three adults 38% at least three adults with children; 25% at least three adults 23% at least three adults Category of dwelling 95% detached 99% detached 61% semi-detached 38% detached; 91% semi-detached 81% semi-detached or terraced; 12% ﬂat 57% semi-detached or terraced or terraced or terraced Tenure type 98% own outright 91% own with a 64% rent; 37%own 64% own outright; 33% rent; 63% own 92% own outright mortgage with a mortgage 30% own with a mortgage with a mortgage Second dwelling in the UK – – ––– - Sex of HRP 91% male 99% male 96% female 100% female 100% male 100% male Mean equivalised disposable household 703 758 291 448 396 391 –1 Income/(£$week ) Mean number of cars and vans 2 2 1 2 1 1 Mean number of rooms 8 9 6 7 6 7 Age of the oldest person 73% over 60 82% under 59 96% under 59 57% over 60; 87% under 59 71% over 60 29% between 50 and 59 Government ofﬁce region – – ––– – Mean equivalised household CO 14.45 11.77 9.97 12.88 10.23 11.54 emissions from energy use at home (tonnes CO /year) Notes: —If no data are presented for the variable or cluster, it indicates this variable is relative evenly spread across all values under this cluster. 620 Front. Energy 2019, 13(4): 612–625 home on weekdays, they are less likely to be high emitters, these criteria. In contrast, among the remaining 90% as they are not typical households in any of the identiﬁed households, only 24% (402 out of 1660) of households clusters. with a male HRP and the oldest person under 59 meet these According to the cluster results, typical socioeconomic criteria. characteristics for each cluster are selected to compare the The comparison between high-emitter clusters and the remaining 90% households shows that among all the households in each high-emitter cluster with the remaining clusters, the households in cluster A are the most 90% households in the 2012 LCF survey sample. The distinguishable ones, followed by those in Cluster B, and combination of typical socioeconomic characteristics of then Cluster D. The households in Clusters C, E, and F are each cluster shows that: less distinguishable from the remaining 90% households, (1) If the HRP is female, the household is likely to be a but still have some of the typical characteristics that high high emitter if 1) the age of the oldest person is under 49; they live in a emitters in these clusters share. Although the households in non-detached property; rent it or own it with a mortgage; Cluster C have lower incomes than the other high-emitter and they own no more than one car and no more than seven clusters, they may rent a dwelling that is poorly insulated rooms at home. Among the high-emitter cluster C, 55% without gas central heating. Therefore, they could require (41 out of 75) households meet these criteria. In contrast, more energy for space heating which would result in high among the remaining 90% households, only 28% (496 out CO emissions from energy use at home. They may also be of 1787) of households with a female HRP meet all these part-time employed or unemployed who spend more time criteria. at home during the day compared to full-time employed 2) the age of the oldest person is over 50; the household people; therefore more energy would be consumed during has at least one car; there are at least seven rooms at home; the day for space heating, cooking, and entertaining. This and the householder owns their property either with a is consistent with the ﬁndings from Büchs and Schnepf mortgage or outright. Among the high-emitter cluster D, [20], where the households with female HRPs are likely to 55% (31 out of 56) households meet these criteria. In have higher CO emissions from direct energy use at contrast, among the remaining 90% households, only 10% home, which could relate to a workless status and along (181 out of 1787) of households with a female HRP meet time spent at home. The clustering results (Table 7) also all these criteria. show that the households that do not own a car or have less (2) If the HRP is male, and the age of the oldest person is than six rooms in their accommodation, the retired over 60, the household is likely to be a high emitter if households that do not own their accommodation, the 1) they own a detached house outright; have no children; households with a male HRP and an average equivalised and have at least two cars and eight rooms at home. Among disposable income less than £390, and the households with the high-emitter cluster A, 72% (75 out of 1104) house- a female HRP and an average equivalised disposable holds meet these criteria. In contrast, among the remaining income less than £290 are less likely to be high emitters 90% households, only 6% (69 out of 1147) of households compared with other households. with a male HRP and the oldest person over 60 meet these criteria. 2) they own a semi-detached house outright, have at 5 Discussion least two adults; and at least one car. Among the high- emitter cluster F, 56% (27 out of 48) households meet these The identiﬁed six high-emitter clusters support ﬁndings in criteria. In contrast, among the remaining 90% households, the existing literature that household energy consumption only 21% (239 out of 1147) of households with a male and CO emissions are inﬂuenced by various socio- HRP and the oldest person over 60 meet these criteria. economic factors and dwelling-related characteristics. (3) If the HRP is male, and the age of the oldest person is Moreover, they also show that in addition to each of the under 59, the household is likely to be a high emitter if socioeconomic factors and dwelling-related characteristics 1) they own a detached house with a mortgage; have at identiﬁed as inﬂuential in the literature, various combina- least two adults; and at least two cars and eight rooms at tions of these characteristics can jointly lead to high CO home. Among the high-emitter cluster B, 62% (66 out of emissions from energy use at home. Previous studies have 107) households meet these criteria. In contrast, among the mainly used regression analysis to investigate the relation- remaining 90% households, only 7% (111 out of 1660) of ships between household CO emissions (the dependent households with a male HRP and the oldest person under variable) and socioeconomic or dwelling characteristics (independent variables) [20,28]. Through regression 59 meet these criteria. 2) they live in a non-detached house, either renting or analysis, these studies identiﬁed some correlations owning with a mortgage; have at least two adults; and at between household emissions and socioeconomic or least one car and six rooms at home. Among the high- dwelling factors. For example, the type of dwelling, tenure emitter cluster E, 53% (63 out of 120) households meet type, as well as the age and income levels of householders Xinfang WANG et al. Understanding high-emitting households in the UK 621 are all correlated with household CO emissions urban location is included in the input variables and the [20,28,31]. Moreover, the regression model can be used ﬂoor area is included instead of the number of rooms. In to estimate the likely amount of emissions for a particular spite of the data limitations, the cluster analysis results household, giving the household’s values for all indepen- based on the LCF still show that the high emitters comprise dent variables in the model. However, due to the limitation different clusters of the households who share similar of the regression technique, it cannot provide insights into socioeconomic factors within each cluster but are different whether and what different combinations of independent from others, which provide useful information for more variables indicate particularly high levels of household targeted emission reduction policies on the different high- emissions. For example, through regression analysis, emitter clusters. In addition to socioeconomic factors and dwelling- Büchs and Schnepf [20] show that the age of the HRP related characteristics, other factors, such as the energy positively correlates with the emissions from direct energy use at home. However, they do not specify that younger efﬁciency of the dwelling and appliances, householders’ families may also more likely be high emitters if they are daily routines, and their use of the home may also lead to renting an old house that is energy inefﬁcient. Likewise, different energy consumption and CO emission levels the regression analysis can show that the size of the [63–65]. For example, a middle-aged couple who rent their dwelling is positively associated with the household CO accommodation can be high emitters due to the ‘tenant- emissions from energy use at home; but it may not disclose landlord problem’ discussed in Section 2. They can live in that householders living in smaller dwellings (for example, less insulated dwellings with less efﬁcient appliances, one or two-bedroom ﬂats) can be high emitters if they have which require more energy to deliver the same energy no access to gas at home and mainly use electricity for services for heating, cooking and cleaning. High-income space heating. In contrast, cluster analysis, which is families with younger children can be high emitters due to applied to the analysis presented in this paper, can identify their separate cooking for children [66]. Retired house- all these possible combinations. Of the identiﬁed high- holds that own their dwellings outright can be high emitter clusters, the socioeconomic factors and dwelling- emitters because of their more vulnerable health conditions related factors are more homogeneous within one cluster and longer time spent at home in general, where more while heterogeneous compared with other clusters. Using energy can be used for space heating and entertaining [20]. the clustering technique to classify high emitters addresses Some of the high-emitting retired households can also live a gap in the literature around exploring the various in larger houses with more additional appliances, which combinations of socioeconomic factors and dwelling- they had been using before their child(ren) moved out [67]. related factors that are most likely to link to high household Further research on people’s routines and use of home are CO emissions. necessary to provide a fuller picture of why these clusters The LCF survey was selected as the most appropriate of households are more likely to be high emitters than others. survey to identify the high emitters through cluster Rebound effects have been discussed widely in relation analysis, because the LCF data set covers information not only on household gas and electricity bills separately, to the emission reduction achievement focusing on the but also a variety of socioeconomic factors and dwelling- households [68,69]. Rebound effects refer to people related characteristics required to identify their inﬂuences consuming the money saved on energy bills from on energy consumption and CO emissions. However, improved energy efﬁciency or behavior change in a there are limitations to using the LCF data set, as some of particular energy service (e.g. lighting, cooking, space the variables are measured more indicatively than others. heating and cleaning) on using more energy for that service For example, the size of the dwellings is measured by the (also known as the direct rebound effect), or on other number of rooms at home, because the data on ﬂoor area is products and services that have direct or embedded CO not available. The rural or urban location of the household emissions (also known as the indirect rebound effect) is not available from the LCF survey either, which could [68,69]. As clariﬁed in Section 1, the analysis aims to also affect the level of energy use at home, especially for identify high-emitter clusters and the potential opportu- space heating due to the lack of access to gas in some rural nities to reduce household CO emissions from higher area and the urban heat island effect [30,60]. The urban emitting households. The emission reductions from high heat island effect means that the temperature in urban areas emitters’ energy use are likely to lead to rebound effects. is generally higher than that in surrounding rural areas, The range of rebound effects may vary signiﬁcantly among largely due to deforestation, the replacement of the land different high-emitter clusters and across various carbon surface by non-evaporating and non-porous materials such mitigation policies. The cluster analysis results in Section 4 as asphalt and concrete, and the more intensive layout of show that some identiﬁed high-emitter clusters (such as buildings and streets within an urban landscape [61,62]. Clusters A and B) share an average household income The identiﬁed high-emitter clusters might be different if about twice as high as other clusters (such as Clusters C, E the input variables are changed, for example if the rural or and F). The high-income high energy users are more likely 622 Front. Energy 2019, 13(4): 612–625 to already be able to afford as much gas and electricity they through general income tax and government spending, require as possible. They are less likely to spend the cost rather than from the energy market or energy suppliers savings on more direct energy consumption at home, but where costs are passed on to all customers but only beneﬁt are more likely to spend them on other products and those households that have renewable energy systems services (e.g., purchasing more expensive cars or ﬂying installed. abroad for holidays). On the other hand, some high-emitter As introduced in Section 1, for energy efﬁciency clusters are lower-income ones. If the higher-energy- improvements, the UK government mandates energy consuming households have not been able to afford as companies to provide heating and insulation improvements much energy as they need or have tight budgets, they can to lower-income and vulnerable households, for example, through the ECO and previous Green Deal. Due to supplier spend the energy payments saved from efﬁciency improve- obligations, they are ﬁnanced by raising overall energy ments on more gas and/or electricity use at home. For example, some householders may leave more lights on prices for customers [42]. The impact is highly regressive, while away, after swapping them for efﬁcient LED lights, because the high-income households pay a much smaller because the total payments for lighting would not increase share of their income on home energy compared to the or would still be reduced compared with previous low-income households in general [42]. When energy inefﬁcient lights. Policies focusing on energy and emission prices increase, the share of income spent on home energy reductions from higher-income higher-energy users may bills may increase much more among the low-income lead to smaller rebound effects and achieve more net households than high-income households if the energy emission reductions than others [60,69,70]. In contrast, savings from efﬁciency improvements are not sufﬁcient energy and emission reductions from lower-income enough to offset increased energy prices. This can lead to higher-energy users can involve higher rebound effects, more serious fuel poverty issues among the low-income which offset the emission reduction effort to a larger extent high emitters, especially retired low-income households [60,69,70]. Future research on reducing energy consump- living in large houses after their children have moved out. tion and CO emissions needs to consider the different size Retired or older people could require more energy for of the likely rebound effect for each high-emitter cluster. space heating, in part due to health conditions. In addition, The estimate could provide evidence on whether and how low-income high emitters may also rent poorly insulated much net CO emissions can be reduced from the high- dwellings and are constrained from insulting it due to their emitter clusters after taking into account rebound effects. tenure type. Cluster C comprises 64% households who rent Policy measures on promoting renewably-generated their properties. This category of householders generally electricity (e.g., the FIT) may achieve more net emission receives few beneﬁts from energy efﬁciency schemes because of the ‘tenant-landlord problem’ discussed in reductions from the low-income high emitters than other Section 2 [17]. As a ﬁnancial incentive, the UK emission reduction policy instruments. This is due to the government introduced the LESA program [39]. However, increased share of total energy use provided by renewably- generated electricity that reduces the CO the program was not widely known and the amount of intensity of energy use. For example, both improving energy efﬁciency grant provided was insufﬁcient [71]. The research and increasing renewably-generated electricity use at home presented here suggests that more policies like the LESA may lead to reduced household energy bills and cause should be initiated with an increased level of ﬁnancial similar direct rebound effects on energy use. If house- incentive supported by government spending, and be holders rebound into using more electricity, they will have widely publicised among landlords, for example, through less impact on CO emissions if they use renewably- the media or letting agents. Policymakers should continue generated electricity. This can be especially valuable to to assist the private rented sector as well as low-income low-income high emitters who are likely to have larger households with older people, and ensure that emission direct rebound effects than other high emitters. The CO reduction policies do not result in more serious fuel emissions caused by the rebound effects can be reduced poverty issues among the low-income high emitters due to when a larger percentage of electricity is generated from increased energy prices as a result of policy interventions. renewable sources. For this reason, policies such as the FIT Financial grants, such as the Winter Fuel Payment subsidy targeted at low-income high-electricity users would be in the UK, could target low-income high energy users attractive for improving carbon mitigation. Furthermore, rather than the current arrangement where people born on the cost of the FIT scheme is shared by all electricity or before 5 May 1953 are eligible to apply for the subsidy regardless of income [72]. customers, which is likely to result in households that do The ﬁndings of this paper not only apply to households not participate in the FIT scheme paying for those who are in the UK, but also other countries where high emitters in the scheme. This could lead to a larger gap between the could comprise clusters of households whose socio- rich and the poor, as there is no provision in the FIT scheme economic characteristics are homogeneous within one to ensure its uptake by low-income households. Therefore, cluster but heterogeneous compared with other clusters. this paper suggests that incentives could be ﬁnanced Xinfang WANG et al. Understanding high-emitting households in the UK 623 Future research can identify the drivers of high energy high-emitter clusters. As reducing energy consumption at consumption at a larger scale through comparing the UK home could lead to rebound effects, it is also important to with other countries. The comparison of drivers of high understand that the range of rebound effects could vary energy consumption across countries would partly depend signiﬁcantly among different high-emitter clusters and on the availability of household survey data in those across various policy measures. More targeted policies countries, which are expected to cover both energy would facilitate a greater amount of emission reductions in consumption and socioeconomic factors at home. Interna- the short to medium term. tional comparison on whether and how the drivers of high While the results indicate that different combinations of socioeconomic factors and dwelling-related characteristics energy consumption differ across countries would con- could all link with high energy consumption and resulting tribute to the global emission reductions by focusing on CO these drivers. It could also offer insight on supra-national emissions, these combinations only explain partly policy making and collaborations for reducing household why some householders are responsible for more CO energy use and CO emissions. emissions than others. The data on energy efﬁciency of the dwelling and appliances are not available for this cluster analysis, and there is no information on high emitters’ 6 Conclusions daily routines and their use of home that could require energy to complete. Further research could be conducted to Household energy consumption accounts for almost a third explore the routines and daily practices of the households of total UK territorial-based CO emissions. It is important who belong to different high-emitting clusters, in order to to reduce emissions from energy use at home in the short to provide a fuller explanation of why these households are medium term for achieving the climate mitigation targets more likely to be high emitters than the others. in the UK and globally. In this paper, attention has been Acknowledgements This work was funded by the School of Mechanical, paid to the high-emitting households and their socio- Aerospace and Civil Engineering and the Sustainable Consumption Institute economic factors, as high emitters could have a larger at the University of Manchester. potential to reduce their CO emissions than the others. Through cluster analysis, the study identiﬁes six different Electronic Supplementary Material Supplementary material is available combinations of socioeconomic factors and dwelling- in the online version of this article at https://doi.org/10.1007/s11708-019- 0647-6 and is accessible for authorized users. related characteristics that can lead to overall high CO emissions from energy use at home. The results show that Open Access This article is licensed under a Creative Commons the high-emitting households belong to several typical Attribution 4.0 International License, which permits use, sharing, adaptation, clusters sharing similar socioeconomic factors and dwell- distribution and reproduction in any medium or format, as long as you give ing-related characteristics within each cluster, but different appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. from other clusters. According to the typical characteristics The images or other third party material in this article are included in the of households in each cluster, households with a male article’s Creative Commons licence, unless indicated otherwise in a credit line HRP, the oldest person over 60, own a detached house to the material. If material is not included in the article’s Creative Commons outright, at least two cars and eight rooms with no children licence and your intended use is not permitted by statutory regulation or at home (Cluster A) are most likely to be high emitters exceeds the permitted use, you will need to obtain permission directly from the copyright holder. among the clusters. The next group of households who are To view a copy of this licence, visit http://creativecommons.org/licenses/ also likely to be high emitters are those who have a male by/4.0/. HRP, oldest person under 59, at least two adults, own a detached house with a mortgage, and at least two cars and eight rooms (Cluster B). Households with a female HRP Notations are also likely to be high emitters if the oldest person is ECO Energy company obligation over 50, they own their property either with a mortgage or outright, and have at least one car and seven rooms FIT Feed-in Tariffs (Cluster D). High emitters in Clusters C, E and F are less GORs Government ofﬁce regions distinguishable from the remaining 90% households, GHG Greenhouse gas compared with Clusters A, B and D, but still shows HRP Household reference person some typical characteristics that high emitters in these clusters share. LCF Living cost and food This paper is of high signiﬁcance not only in the UK, but LESA Landlord’s energy saving allowance also in other countries. Different from the main stream LIHC Low income high cost regression analysis, the cluster study within high emitters OECD Organisation for Economic Co-operation and Development is an innovative approach developed and presented in this ONS Ofﬁce for National Statistics paper, which provides useful information for more targeted policies in the UK and other countries focusing on different UNFCCC United Nations Framework Convention on Climate Change 624 Front. Energy 2019, 13(4): 612–625 14: 50–72 References 20. Büchs M, Schnepf S. Who emits most? Association between socio- 1. Intergovernmental Panel on Climate Change (IPCC). AR5 climate economic factors and UK households’ home energy, transport, change 2014: impacts, adaptation and vulnerability. 2017–01–09, emissions. Ecological Economics, 2013, 90: indirect and total CO available at IPCC website 114–123 2. United Nations. Paris Agreement. 2017–04–28, available at unfccc 21. Lenzen M, Murray S A. The ecological footprint–issues and trends. website ISA Research Paper 01–03. Sydney: University of Sydney, 2003 3. Her Majesty’s Stationary Ofﬁce (HMSO). Climate Change Act 22. Kenny T, Gray N F. A preliminary survey of household and 2008. London: TSO, 2008 personal carbon dioxide emissions in Ireland. Environment Inter- 4. Her Majesty’s Stationary Ofﬁce (HMSO). The Carbon Budget national, 2009, 35(2): 259–272 Order 2009. Statutory instruments Number 1259. London: HMSO, 23. Jones C M, Kammen D M. Quantifying carbon footprint reduction opportunities for US households and communities. Environmental 5. Her Majesty’s Stationary Ofﬁce (HMSO). The Carbon Budget Science & Technology, 2011, 45(9): 4088–4095 Order 2011. Statutory instruments Number 1603. London: HMSO, 24. Vringer K, Blok K. The direct and indirect energy requirements of households in the Netherlands. Energy Policy, 1995, 23(10): 893– 6. Her Majesty’s Stationary Ofﬁce (HMSO). The Carbon Budget Order 2016. Statutory instruments Number 785. London: HMSO, 25. Bhattacharjee S, Reichard G. Socio-economic factors affecting individual household energy consumption: a systematic review. In: 7. Anderson K. Talks in the city of light generate more heat. Nature, The ASME 5th International Conference on Energy Sustainability, 2015, 528(7583): 437 Washington, D.C., USA, 2011 8. Pye S, Li F G N, Price J, Fais B. Achieving net-zero emissions 26. Pachauri S. An analysis of cross-sectional variations in total through the reframing of UK national targets in the post-Paris household energy requirements in India using micro survey data. Agreement era. Nature Energy, 2017, 2(3): 17024 Energy Policy, 2004, 32(15): 1723–1735 9. Department for Business, Energy and Industrial Strategy (BEIS). 27. Kadian R, Dahiya R P, Garg H P. Energy-related emissions and Final UK greenhouse gas emissions national statistics: ﬁnal mitigation opportunities from the household sector in Delhi. Energy estimates of UK greenhouse gas emissions from 1990. London: Policy, 2007, 35(12): 6195–6211 BEIS, 2018 28. Gough I, Abdallah S, Johnson V, Ryan-Collins J, Smith C. The 10. Department for Business, Energy and Industrial Strategy (BEIS). distribution of total greenhouse gas emissions by households in the Energy consumption in the UK. London: BEIS, 2016 UK, and some implications for social policy. London: Centre for 11. Ares E. Feed-in Tariffs. London: House of Commons. 2012 Analysis of Social Exclusion, 2012 12. Her Majesty’s Stationary Ofﬁce (HMSO). The Electricity and Gas 29. Palmer J, Terry N, Armitage P, Godoy-Shimizu D. Savings, beliefs (Energy Companies Obligation) (Amendment) Order 2014. Statu- and demographic change. Cambridge Architectural Research tory instruments Number 1131. London: HMSO, 2014 Limited, Element Energy, and Loughborough University, 2014 13. Department of Energy and Climate Change (DECC). The Green 30. Thumim J, White V. Distributional impacts of personal carbon Deal: A Summary of the Government’s Proposals. URN: 10D/996. trading: a report to the department for environment, food and rural London: DECC, 2010 affairs. London: Department for Environment, Food and Rural 14. Foulds C, Powell J. Using the Homes Energy Efﬁciency Database as Affairs, 2008 a research resource for residential insulation improvements. Energy 31. Fahmy E, Thumim J, White V. The distribution of UK household Policy, 2014, 69: 57–72 CO emissions. Interim Report. JRF programme paper: Climate 15. Dowson M, Poole A, Harrison D, Susman G. Domestic UK retroﬁt change and social justice, York, 2011 challenge: barriers, incentives and current performance leading into 32. Chancel L, Piketty T. Carbon and inequality: from Kyoto to Paris. the Green Deal. Energy Policy, 2012, 50: 294–305 Paris: Paris School of Economics, 2015 16. Balcombe P, Rigby D, Azapagic A. Investigating the importance of 33. Oxfam. Carbon emissions and income inequality. Technical Note. motivations and barriers related to micro-generation uptake in the Oxford, 2015 UK. Applied Energy, 2014, 130: 403–418 34. Element Energy. Further analysis of data from the household 17. Druckman A, Jackson T. Household energy consumption in the UK: electricity usage study: consumer archetypes. Final Report for a highly geographically and socio-economically disaggregated DECC and Defra. Cambridge: Element Energy limited, 2013 model. Energy Policy, 2008, 36(8): 3177–3192 35. Morrison B M, Gladhart P M. Energy and families: the crisis and the 18. Druckman A, Jackson T. The carbon footprint of UK households response. Journal of Home Economics, 1976, 68: 15–18 1990–2004: a socio-economically disaggregated, quasi-multi-regio- 36. Kaza N. Understanding the spectrum of residential energy nal input-output model. Ecological Economics, 2009, 68(7): 2066– consumption: a quantile regression approach. Energy Policy, 2010, 38(11): 6574–6585 19. Baiocchi G, Minx J, Hubacek L. The impact of social factors and 37. Lenzen M. Energy and greenhouse gas cost of living for Australia consumer behaviour on carbon dioxide emissions in the United during 1993/94. Energy, 1998, 23(6): 497–516 Kingdom: a regression based on input-output and geo-demographic 38. Sultana S, Pourebrahim N, Kim H. Household energy expenditures consumer segmentation data. Journal of Industrial Ecology, 2010, in North Carolina: a geographically weighted regression approach. Xinfang WANG et al. Understanding high-emitting households in the UK 625 Sustainability, 2018, 10(5): 1511 56. Department of Energy and Climate Change (DECC). Annual fuel 39. HM Revenue and Customs. Deductions: General rules: main types poverty statistics report. URN: 14D/146 . London: DECC, 2014 of expenses: Landlord’s Energy Savings Allowance (LESA). 2017– 57. Mooi E, Sarstedt M. A concise guide to market research: the 04–28, available at gov.uk website process, data and methods using IBM SPSS statistics. Berlin 40. Ofﬁce for National Statistics (ONS). Living cost and food survey Heidelberg: Springer-Verlag, 2011 volume A: introduction January-December 2012. London: ONS, 58. IBM Knowledge Centre. SPSS statistics two-step cluster algorithms. 2017–04–28, available at ibm.com website 41. Ofﬁce for National Statistics (ONS). Input-output supply and use 59. Okazaki S. What do we know about mobile Internet adopters? tables for 1997 to 2013. London: ONS, 2013 A cluster analysis. Information & Management, 2006, 43(2): 127– 42. Gough I. Carbon mitigation policies, distributional dilemmas and social policies. Journal of Social Policy, 2013, 42(2): 191–213 60. Druckman A, Chitnis M, Sorrell S, Jackson T. Missing carbon 43. Brand C. Transport and carbon emissions analysis. Project Paper reductions? Exploring rebound and backﬁre effects in UK house- Number 2. Oxford: Environmental Change Institute, University of holds. Energy Policy, 2011, 39(6): 3572–3581 Oxford, 2013 61. Oke T R. Inadvertent climate modiﬁcation. In: Boundary Layer 44. OECD. OECD framework for statistics on the distribution of Climates. London: Routledge, 1987 household income, consumption and wealth. OECD publishing. 62. Lo C P, Quattrochi D A, Luvall J C. Application of high-resolution 2017–04–28, available at keepeek.com website thermal infrared remote sensing and GIS to assess the urban heat 45. Yohanis Y G, Mondol J D, Wright A, Norton B. Real-life energy use island effect. International Journal of Remote Sensing, 1997, 18(2): in the UK: how occupancy and dwelling characteristics affect 287–304 domestic electricity use. Energy and Building, 2008, 40(6): 1053– 63. Shove E. Comfort, Cleanliness and Convenience: the Organisation of Normality. Oxford and New York: Berg, 2003 46. Utley J I, Shorrock L D. Domestic energy fact ﬁle 2008. BRE 64. Gram-Hanssen K. Standby consumption in households analyzed Environment, DECC and EST, 2008 with a practice theory approach. Journal of Industrial Ecology, 2010, 47. The Scottish Government. Scottish house condition survey: key 14(1): 150–165 ﬁndings 2010. Edinburgh: The Scottish Government, 2010 65. Maller C, Horne R, Dalton T. Green renovations: Intersections of 48. Pullinger M, Browne A, Anderson B, Medd W. Patterns of water: daily routines, housing aspirations and narratives of environmental the water related practices of households in southern England, and sustainability. Theory and Society, 2012, 29: 255–275 their inﬂuence on water consumption and demand management. 66. Strengers Y, Nicholls L, Maller C. Curious energy consumers: Final Report of the ARCC-Water/ SPRG Patterns of Water Projects. humans and nonhumans in assemblages of household practice. Lancaster: Lancaster University, 2013 Journal of Consumer Culture, 2016, 16(3): 761–780 49. Lamb W F, Steinberger J K, Bows-Larkin A, Peters G P, Roberts J 67. Butler C, Parkhill K A, Shirani F, Henwood K, Pidgeon N. T, Wood F R. Transitions in pathways of human development and Exploring the dynamics of energy demand through a biographical carbon emissions. Environmental Research Letters, 2014, 9(1): lens. Nature and Culture, 2014, 9(2): 164–182 68. Sorrell S, Dimitropoulos J. The rebound effect: microeconomic 50. Field A. Cluster analysis. 2017–04–28, available at website of deﬁnitions, limitations and extensions. Ecological Economics, statisticshell.com 2008, 65(3): 636–649 51. Everitt B S, Landau S, Leese M, Stahl D. Cluster Analysis. 5th ed. 69. Chitnis M, Sorrell S, Druckman A, Firth S K, Jackson T. Who Chichester: John Wiley & Sons Ltd., 2011 rebounds most? Estimating direct and indirect rebound effects for 52. AEA. 2012 Guidelines to Defra/DECC’s GHG conversion factors different UK socioeconomic groups. Ecological Economics, 2014, for company reporting. Produced for the Department for Environ- 106: 12–32 ment, Food and Rural Affairs (Defra) and the Department of Energy 70. Sorrell S, Dimitropoulos J, Sommerville M. Empirical estimates of and Climate Change (DECC) (Version 1.0), London, 2012 the direct rebound effect: a review. Energy Policy, 2009, 37(4): 53. Department of Energy and Climate Change (DECC). Quarterly 1356–1371 energy prices. URN: 12D/276D. London: DECC, 2012 71. Boardman B. Home truths: a low-carbon strategy to reduce UK 54. The Consumer Council. Comparative domestic cost of gas voil housing emissions by 80% by 2050. ECI Research Report 34. report. Belfast: The Consumer Council, 2013 Oxford: Environmental Change Unit, University of Oxford, 2007 55. Department of Energy and Climate Change (DECC). Quarterly 72. UK Government. Winter fuel payment. 2019–03–01, available at energy prices. URN: 14D/276C. London: DECC, 2014 website of gov.ukeligibility

Journal

Frontiers in Energy – Springer Journals

Published: Dec 15, 2019

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Understanding high-emitting households in the UK through a cluster analysis

Understanding high-emitting households in the UK through a cluster analysis

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Understanding high-emitting households in the UK through a cluster analysis

Understanding high-emitting households in the UK through a cluster analysis

References (59)

Abstract

Journal

Recommended Articles

There are no references for this article.

Our policy towards the use of cookies