Decomposition analysis of gas consumption in the residential sector in Ireland

To-date, decomposition analysis has been widely used at the macro-economic level and for in-depth analyses of the industry and transport sectors; however, its application in the residential sector has been rare. This paper uses the Log-Mean Divisia Index I (LMDI-I) methodology to decompose gas consumption trends in the gas-connected residential sector in Ireland from 1990 to 2008, which despite an increasing number of energy efficiency policies, experienced total final consumption growth of 470%. The analysis decomposes this change in gas consumption into a number of effects, examining the impact over time of market factors such as a growing customer base, varying mix of dwelling types, changing share of vacant dwellings, changing size of new dwellings, the impact of building regulations policy and other factors such as the weather. The analysis finds the most significant effects are changing customer numbers and changing intensity; the analysis also quantifies the impact of building regulations and compares it with other effects such as changing size of new dwellings. By comparing the historical impact on gas consumption of policy factors and non-policy factors, this paper highlights the challenge for policy-makers in achieving overall energy consumption reduction.

In Ireland from 1990 to 2008, natural gas total final consumption (TFC), for all sectors of the economy, grew by an average annual rate of 6.1%, the highest rate for any fuel in Ireland during that period; in 2008, natural gas accounted for 12.4% of Ireland's TFC. In the residential sector, natural gas TFC grew by an average annual rate of 10.2%, growing from a 5.2% share in 1990 to a 21% share in 2008 (Howley et al., 2009a). Ireland's natural gas import dependency went from 0% in 1990 to 92% in 2008 (Howley et al., 2009b). The EU has also experienced growth in natural gas TFC and import dependency: over the period 1990 to 2008, natural gas TFC in the EU12 grew at an average annual rate of 3% (Eurostat, 2010b); concurrently, the EU expanded to 27 member state countries and from a low of 45.6% in 1998, import dependency climbed to 62.3% in 2008 (Eurostat, 2010a), as illustrated in Fig. 1.

1. Download full-size image

Fig. 1Source: prepared by the authors based on Eurostat (2010b).

The negative consequences of Europe's import dependency became apparent during the Russia Ukraine gas crisis in 2008. The second EU strategic energy review was published following this crisis and focused on security of supply (European-Commission-Energy, 2009), proposing a significant upgrade to gas infrastructure in an attempt to address the volatility of EU gas supply. Other EU energy policies that also seek to reduce our dependence on imported gas include a focus on energy efficiency (notably the Energy Services Directive (ESD) (EC, 2006), Energy Performance in Buildings Directive1 (EC, 2010)) and renewable energy (Renewable Energy Directive (EC, 2009)). However, it remains the view of both the EU and IEA that natural gas will have a bridging role between the energy past and future (Vinois, 2010, Baroni, 2010).

1.1. Policy trends

The residential sector is a significant energy user and represents a significant opportunity for energy efficiency improvement in Ireland. It has been the focus of increased policy activity, as can be seen from the MURE2 database, a centralized storehouse listing energy policies for all sectors of the economy for all EU member countries.3 For 1990–1999, the MURE database lists five energy end-use policies in Ireland's residential sector; for 2000–2008, fourteen policies are listed (MURE-II, 2011).4 In 2006, as obliged by the Energy Services Directive (EC, 2006), Ireland prepared a National Energy Efficiency Action Plan (NEEAP), which contained many of the same policies. A particular focus has been on improving the energy performance of new homes. Ireland's NEEAP had building regulations (BR) contributing a 48% share of the energy savings target for 2020, making it the most significant type of policy measure in the residential sector. Until 2008, most of the other policies listed in MURE were either niche policies (e.g. fuel poor dwellings, appliance efficiency and renewable energy grants) or pilot policies (low carbon homes, smart metering). The only policy to address existing dwellings, focusing on the largest energy end-use category of space heating was an information campaign, promoting behavioral change towards efficiency in the home.

1.2. Lit review

Despite the increased number of policies in Ireland, there has been a limited number of empirical analyses of the energy performance of the dwelling stock (O'Doherty et al., 2008) or of the impact of energy efficiency policies generally (Hull et al., 2009, Rogan et al., 2011). It is particularly pertinent that such research be done since Ireland has one of the most inefficient housing stocks in Northern Europe (Healy, 2004) and in recent times has experienced an above average level of new builds (DEHLG, 2010). There have been two previous analyses on residential natural gas TFC in Ireland. Hull et al. (2009) used a 10% anonymized sample and a more detailed sample of 48 dwellings to conduct a preliminary analysis on the impact of building regulations, dwelling type, changing dwelling size and gas tariffs; the study produced results that were “not conclusive” but suggested a “substantial rebound effect and/or a degree of non-compliance with historic building regulations”; the authors recommended further work, particularly ex-post analyses of energy efficiency policies. In an examination of the impact of a national energy efficiency advertising campaign on natural gas TFC in 2006 and 2007, Diffney et al. (2009) found no significant impact on gas consumption.

A number of other analyses have been carried out on aspects of energy consumption in the Irish residential sector. Scott (1997) did a regression analysis on ownership of household energy saving appliances in Ireland with a focus on the reasons for their low take-up. Clinch et al. (2001) constructed a model of the Irish housing stock to model the energy and emissions impact of home insulation energy efficiency measures. O'Doherty et al. (2008)conducted a regression analysis on energy appliances and energy saving features in Irish dwellings. Lyons et al. (2009) calculated expenditure and own price elasticities for fuels in the household sector. Two Sustainable Energy Authority of Ireland (SEAI) reports, Energy in the Residential Sector (ÓLeary et al., 2008) and Energy Efficiency in Ireland (Dennehy et al., 2009), analyzed energy consumption in the residential sector and included a decomposition analysis of four effects (size, diffusion of central heating, technical efficiency and behavior), which impacted the energy consumption of the dwelling stock between 1995 and 2006.

A number of these analyses on gas and energy consumption in the residential sector have highlighted the need for further analysis of factors impacting energy consumption. This further analysis is important for a number of reasons: (i) measuring the efficacy of energy efficiency policies is important for designing better future policies, (ii) there is an obligation to report energy savings under the European Energy Services Directive (EC, 2006) and (iii) in view of the growing volume of energy efficiency policies, more ex-post analyses are required to sufficiently model current and future energy demand. It is in this context that an expanded decomposition analysis is seen as a powerful analytical tool.

1.3. Decomposition analysis

A collaborative report between the IEA and the IAEA recommended the use of decomposition to study changes in energy demand in all sectors, including the residential sector (IAEA, 2005) and the IEA has demonstrated the usefulness of decomposition analysis by comparing nine countries for five different effects (Taylor, 2009). This paper uses the Log-Mean Divisia Index (LMDI) decomposition analysis methodology. Decomposition analysis was first used in the late 1970s to examine the impact of changes in product mix on energy intensity and consumption. It has since broadened its scope to include analysis of energy supply and demand, energy-related emissions, material flow and dematerialization, monitoring of national energy efficiency trends and making cross-country comparisons of energy performance (Ang, 2004).

A survey of decomposition studies in 2000, which found 124 studies in the literature (Ang and Zhang, 2000), focused largely on industry, with some studies examining the economy as a whole, and a smaller number focusing on transport. Decomposition papers focused on the residential sector are rare, just two currently exist in the literature (Unander et al., 2004, Achao and Schaeffer, 2009), which provided a motivation for this paper.

Decomposition analysis in Ireland has been carried out on industry (Cahill et al., 2010, Cahill and Ó Gallachóir, 2010), on the economy as a whole (O'Mahony, 2010) and on the transport sector but this represents the first decomposition analysis on the residential sector and focuses on natural gas demand. The analysis carried out here is readily replicable in other countries to quantify the policy and non-policy factors affecting natural gas demand trends in the residential sector.

A number of papers have included the residential sector in broader analyses: Munksgaard et al. (2000) analyzed the impact of direct and indirect household consumption on emissions in Denmark, Wachsmann et al. (2009) did a structural decomposition analysis of energy demand in Brazil, which included the residential sector, Donglan et al. (2010) examined changing emissions in rural and urban household sectors in China and Kumbaroğlu (2011)examined emissions in Turkey.

1.4. Analysis objectives

Over the time scale 1990–2008, this decomposition analysis seeks to quantify the impact of a number of policy and non-policy factors on natural gas TFC. As the table in Appendix Ashows, there was a growing volume of energy efficiency policy measures over this period; however, for most of the period, the focus was on BR5 for new dwellings, with little emphasis on the existing dwellings stock. By appropriate manipulation of the available data, this paper provides a preliminary analysis of the dominant energy efficiency policy for the period: the 2002 BR. Other energy efficiency policies occur too late in the analysis period to be examined, do not address space heating of existing dwellings or are not readily quantifiable within a top-down methodology such as decomposition analysis, which is not typically used to evaluate specific energy efficiency policies. Although decomposition analysis is not an ex-post policy analysis tool, it can give an indicative quantification of the absolute and relative impact of the 2002 BR policy within a framework that also quantifies non-policy market effects (changing number of dwellings, changing share of vacant dwellings, dwelling size and type of dwellings) and intensity effects (weather effect and intensity effect).

This paper is focused on the natural gas section of the residential sector because in terms of a source of data, metered gas consumption is a uniquely excellent source (Ó Cléirigh and AEA Energy & Environment, 2008), providing accurate consumption data and many other details of dwelling type and location as described in Section 3. This paper seeks to incorporate this additional data to make a detailed analysis of the natural gas consuming residential sector. It is due to a lack of data for other fuels (oil, solid fuels and electricity), that these other sectors have not been included. The generality of the results will be seen within the context of the known differences between the national dwelling stock and the gas dwelling stock (Coniffe, 2000).

The paper is organized as follows: Section 2 discusses the trends from 1990 to 2008 of natural gas TFC, the gas network and the dwelling stock. Section 3 outlines the LMDI-I methodology, how the formula was developed to capture five effects and the data sources used. Sections 4 and 5 presents the results and discussion. Section 6 concludes and outlines potential for future analysis and research.

2. Residential natural gas sector: 1990–2008

2.1. Gas consumption trends

During the period 1990–2008, Bord Gáis Éireann (BGÉ) was the only retailer of piped natural gas to the residential sector in Ireland. In 1990, residential gas TFC was 142 ktoe, a 5.2% share of overall residential TFC; throughout the period that followed, 1990–2008, Ireland underwent a sustained economic boom during which GDP grew by 182% and TFC for all fuels in all sectors of the economy grew by 85% (Howley et al., 2009a). Residential natural gas TFC grew by an average annual rate of 10.2%, the highest rate for any fuel in the residential sector during that period and by 2008, natural gas had a 21% share of overall residential TFC (Howley et al., 2009a). Contained within these overall trends of growth in residential gas customers and natural gas TFC is a more complex interweave of trends that do not tally with the macro trend of constant growth. Starting in 2000, average annual consumption of the dwelling stock began to decline, see Fig. 2.

1. Download full-size image

Fig. 2Source: prepared by the authors based on BGÉ data set.

2.2. Growth of gas network

Starting in 1990, BGÉ had 139,000 residential customers, approximately 14% of all dwellings in Ireland; in this year, over 90% of all gas connections were in Cork and Dublin. In 1993, an interconnector with the UK made landfall on the east coast of Ireland and over the rest of the decade, the gas network expanded along the northeast coast, the southeast coast and west into the rapidly growing towns in the Greater Dublin Area.6 In the early 2000s there was continued expansion around the existing urban areas of Limerick and Cork and in the towns along the existing Cork–Dublin pipeline. By 2004, the Seven Heads gas field off the south coast had been opened, a second interconnector with the UK had been constructed and a pipeline between Galway and Dublin had been completed, see Fig. 3. By 2008, natural gas residential customer numbers had grown by 342%, at a constant annual growth rate of 9%, to reach 616,000, approximately 39% of all dwellings in Ireland.

If you are interested in this research article, you can download the full length article through the download page using the link below, both PDF file and Word file are provided.

https://www.sciencedirect.com/search/advanced?tak=LMDI&show=25&sortBy=relevance

The driving forces of change in energy-related CO2 emissions in Ireland: A multi-sectoral decomposition from 1990 to 2007

Ireland recorded significant growth in energy-related carbon emissions from 1990 to 2007 as the country underwent rapid economic development. Using the LMDI decomposition analysis method, this paper aims to identify and analyse the driving forces of CO₂ emissions in eleven final energy consuming sectors. This multi-sectoral analysis is based on four economic sectors, the residential sector and gives a detailed representation of transport in keeping with UNFCCC recommendations. Scale, structure and intensity effects are explored and substantial heterogeneity in sectoral performance is observed. Scale growth in economic and transport activity was considerable. Some improvements in energy intensity were recorded in the economic sectors. In transport, increases in intensity contributed to a significant increase in emissions, while energy intensity decreased in the residential sector. The declining emissions coefficient of electricity was important in limiting emissions but renewable energy has been slow to penetrate the demand side. The results have relevance in considering development paths and can aid in identifying policy measures required to address the key driving forces of emissions in the sectors. The rapid increase in transport emissions in particular raises concerns of future lock-in to a higher emissions trajectory.

Energy and CO2 emissions are integral to issues of development in all nations and have become a significant policy challenge in the Republic of Ireland. Through the Kyoto Protocol and the European Union (EU) Burden Sharing Agreement (Council Decision 2002/358/EC) the target agreed for Ireland was to limit the increase in greenhouse gas (GHG) emissions to +13% of 1990 levels by 2008–2012. This coincided with a period of rapid economic development, during which infrastructure was expanded, lifestyles changed and energy demand and CO2 emissions increased substantially. In 2007, total GHG emissions in Ireland were 25% higher than the 1990 level (McGettigan et al., 2009). Energy-related CO2emissions increased by 49.4% and accounted for two-thirds of all GHG emissions. Understanding driving forces of CO2 emissions is essential to formulating climate change mitigation policy and the fulfilment of applicable targets. Decomposition analysis of change in emissions provides a robust means of achieving this objective.

At the United Nations Framework Convention on Climate Change (UNFCCC) Dublin Workshop on Fourth National Communications from Annex I Parties (UNFCCC, 2004), index decomposition analysis (IDA) was recommended to quantify key drivers of emissions and separate effects such as energy efficiency and GDP growth. In the literature, applications of IDA have undergone substantial changes since the late 1970s. Recently, IDA methods particularly the logarithmic mean divisia index (LMDI) technique have been widely applied to track economy-wide energy efficiency trends by different countries/organisations (Ang et al., 2010). Sectoral analysis has been expanding from energy demand and CO2 emissions in industry and manufacturing sub-sectors to analysis such as UK road freight in Sorrell et al. (2009).

Insight into the driving forces underlying change in CO2 emissions in Irish sectors has been limited to the decomposition analysis of manufacturing. Given the large increase in emissions, particularly from the under-investigated transport sectors, the absence of appropriate enquiry could have consequences for policy. In the in-depth review (IDR) of Ireland’s third national communication to the UNFCCC, Rolle et al. (2005) highlighted this gap in knowledge. The reviewers recommended that changes in GHG emissions of Irish transport be linked to changes in modal split and changes in physical activity by passenger kilometres (p-km) and tonne kilometres (t-km). The objectives of this study were to identify and analyse trends in the historical driving forces of CO2 in all of the energy end-use sectors from 1990–2007. It also forms the first response to these UNFCCC recommendations. The sectors are analysed separately but inter-sectoral shifts in the shares of activity in the economic sectors and the transport modes are also analysed. Rather than develop policy recommendations per se, this study contributes to the discussion of appropriate mitigation measures by engaging with gaps in knowledge of trends and driving forces in the end-use sectors.

The results obtained are not only relevant to Irish policy-making but may provide useful insights for other countries experiencing a development transition. The analysis of the historical progression of key indicators, particularly energy intensity, also functioned as the first step in the development of scenarios of future CO2 emissions in O’ Mahony et al. (submitted). Similar to Wu et al. (2005) who analysed driving forces in China, the originality of this study lies in the use of a framework that gives a disaggregated multi-sectoral decomposition. Multi-sectoral decompositions in previous studies often deal with three or four sectors (Diakoulaki et al., 2006, Lise, 2006; Tunç et al., 2009) whereas this study disaggregates eleven final consumption sectors. Whereas the EU-ODEX has been used to track energy efficiency in some sectors (Dennehy et al., 2009), this is the first comprehensive index decomposition analysis of the Republic of Ireland, including of the non-manufacturing economic sectors, the residential sector and of particular importance, the transport modes. While this study decomposes all final consumption sectors, particular focus is accorded to the disaggregation of transport recognising both that it is under-investigated and also its importance in total emissions. This is as opposed to Oh et al. (2010) a multi-sectoral decomposition of South Korea, which concentrated on disaggregated manufacturing but aggregated transport. This multi-sectoral analysis may provide deeper insights than the macro level approach recommended by the UNFCCC (2004).

The rest of this paper is organised as follows. Section 2 presents sectoral emissions from 1990–2007 and the classification of the sectors in Ireland that leads to the decomposition scheme. Section 3 introduces our decomposition framework and Section 4 describes data sources. In Section 5, we present the decomposition analysis results from each sector and an aggregate analysis. Section 6 concludes this study.

2. Sectoral CO2 emissions in Ireland

Fig. 1 shows the evolution of the sectoral contribution to total energy CO2 emissions in Ireland from 1990 to 2007. The classification of final energy end-use sectors was established in the Ireland’s energy balance sheets communicated to the European Commission and the International Energy Agency. It can be observed from Fig. 1 that there was an increasing trend in the total CO2 emissions and a concentration of growth occurred from 1993–2001. The residential and industry sectors are the major contributors but their shares have declined over time. All transport modes experienced substantial growth in emissions with the exception of rail.1

1. Download full-size image

Fig. 1

The decomposition framework used in this study is based on the sectoral classification previously described. This sectoral disaggregation is similar to that used in Agnolucci et al. (2009), Oh et al. (2010) and to a lesser degree that of Wu et al. (2005). Full coverage of the main final consumption sectors is achieved using data by sector and fuel type. The four economic sectors include agriculture, industry, commercial services and public services, but does not separate construction as disaggregated was unavailable. Agnolucci et al. (2009)split industry into energy intensive/extensive branches but gross value added (GVA) data in Ireland is not disaggregated along these lines pre-1995.2 Industry remains aggregated to facilitate a full analysis from 1990 to 2007. Agriculture and public services are given their own characterisation as agriculture is unique in its use of energy and public services are unique in the instruments necessary to reduce CO2 emissions due to state control.

In our analysis, a methodological challenge arises in attempting to understand the increase in emissions from transport since 1990. A response is required to the recommendations of Rolle et al. (2005), that changes in emissions be linked to changes in modal split and activity by p-km and t-km. Ang and Zhang (2000) specifically suggested the use of LMDI I to measure the physical efficiency of transport. As discussed by Timilsina and Shrestha (2009)given a lack of data, a common approach in energy literature is to measure modal shift by changes in modal fuel consumption in total transport fuels (EIA, 2007, IEA, 2004). This proxy method assumes the same intensity across the different modes and weakens results. The approach adopted in this study retains the physical measure of modal shift and also intensity. As shown in Fig. 1, transport in the decomposition framework is split into six sub-sectors or modes corresponding to energy and CO2 data. This reflects the considerable modal differences in the provision of transport services while responding to the recommendations of Rolle et al. (2005). International aviation and maritime transport are not considered as both are memo items in national inventories and excluded from national totals and quantitative targets (IPCC, 1999). It should be pointed out that a similar limitation to Agnolucci et al. (2009) arose in the case of rail since the data on energy use by passenger and freight components cannot be separated. As such, we aggregate the total activities performed by rail passenger transport and rail freight using the approach suggested by Diakoulaki et al. (2006). Since the ratio of rail p-km to t-km was not stable, the results of aggregate intensity for rail are interpreted with this in mind. The road public passenger sector is also an aggregation of bus and taxi transport modes. It is worth noting that the approach of Timilsina and Shrestha (2009) aggregates road transport and also aggregates rail transport. The activity of the unspecified and fuel tourism sectors cannot be measured. These sectors are aggregated to complete the analysis but effects are not measured.

In keeping with Ekins and Barker (2001) it is recognised that energy demand is more related to energy services (heat, light, power, mobility, etc.) than for energy itself per se. This has implications for the drivers that lead to change in energy CO2 emissions, but it is difficult to accommodate given the huge variety of energy services required in each sector. In order to overcome this, energy services are represented by either monetary or physical indicators based on the characteristics of different sectors. Usually, the use of physical indicators is considered to be more accurate but this may only be applicable for particular sectors such as transport (Diakoulaki et al., 2006, Freeman et al., 1997). In the case of the economic sectors, both physical and monetary indicators of output can be used, but both present with potential limitations. Physical indicators can create difficulties in aggregating disparate physical outputs across different products, commodities or service groups, while monetary indicators can mislead due to changes in unit prices. Similar multi-sectoral studies have used monetary indicators of economic output (Wu et al., 2005, Oh et al., 2010, Lise, 2006, Diakoulaki et al., 2006). In this study, using the monetary indicator GVA facilitates a commonality in the method of analysis across the sectors and also the analysis of structural shifts. For the residential sector, the number of households is taken as the activity indicator. Transport activity is represented by mobility rather than vehicle distance as mobility is the primary energy service sought (Ekins and Barker, 2001). Therefore, p-km and t-km are, respectively, taken as the activity indicators for passenger and freight transport modes.

3. Methodology

Index decomposition analysis (IDA) has been widely accepted as an analytical tool for supporting policymaking on national energy and environmental issues (Ang, 2004b). The decomposition of the change in an aggregate indicator into a pre-defined set of factors helps to understand the progression of driving forces, the impact of major processes occurring and policy dimensions tied to these processes (Steenhof et al., 2006). The results of an IDA application study have direct policy implications such as evaluation of energy conservation programs (Ang, 2004b, Ang and Liu, 2007). They may also provide a basis for forecasting (Ang, 2004a) or scenario analysis of future evolution. The results of this study are used as the basis for scenarios presented in O’ Mahony et al. (submitted).

A range of techniques have been established under the umbrella of IDA, among which the LMDI I technique has been identified as the preferred approach by Ang (2004b). The mathematical properties of the technique suggest its suitability for this study including: perfect decomposition, consistency in aggregation and ability to handle zero values. In the methodological literature (Ang, 2004b) recommends the multiplicative and additive LMDI I methods for their theoretical foundation, adaptability, ease of use and ease of result interpretation. LMDI I has both additive and multiplicative forms. In this study, it is applied in multiplicative form chain-linked annually accommodating separate decomposition of the sectors and subsequent aggregation to total change. The basic mathematical formulae for IDA and LMDI I can be found in Ang (2004b) developed from work by Ang and Liu (2001). The work of Ang and Liu (2001) was extended by Wu et al. (2005) as a three-level decomposition for China disaggregated by sector and province. In contrast, this study applies a two-level decomposition for Ireland without provincial disaggregation but for a greater number of sectors. The approach used facilitates the elaboration of sector-specific insights. The decomposition schemes applied to each of the sectors are detailed in Eqs. (1), (2), (3) where index i=1, 2,…,6 respectively denote coal, oil, peat, gas, renewables and electricity and index t the year from 0 (base year) to t (target year). Eq. (1) is applied to each of the economic sectors for j=1,2,3,4 denoting industry, commercial services, public services and agriculture:

(1)$\frac{Ceco{n}_{j,t}}{Ceco{n}_{j,0}}=\sum _{i=1}^6\frac{C_{tij}}{F{F}_{tij}}\frac{F{F}_{tij}}{F{F}_{tj}}\frac{F{F}_{tj}}{E_{tj}}\frac{E_{tj}}{Y_{tj}}\frac{Y_{tj}}{Y_t}{Y}_t$

In Eq. (2) applied to each of the transport sectors, j indexes sector, for j=5,6,…,10 for private car transport, road public passenger transport (bus and taxi), road freight transport, rail transport (passenger and freight), domestic aviation and aggregated unspecified and fuel tourism:

(2)$\frac{Ctran{s}_{j,t}}{Ctran{s}_{j,0}}=\sum _{i=1}^6\frac{C_{tij}}{F{F}_{tij}}\frac{F{F}_{tij}}{F{F}_{tj}}\frac{F{F}_{tj}}{E_{tj}}\frac{E_{tj}}{T{D}_{tj}}\frac{T{D}_{tj}}{TT{D}_t}TT{D}_t$

Eq. (3) applies to j=11 the residential sector:

(3)$\frac{Cre{s}_{j,t}}{Cre{s}_{j,0}}=\sum _{i=1}^6\frac{C_{tij}}{F{F}_{tij}}\frac{F{F}_{tij}}{F{F}_{tj}}\frac{F{F}_{tj}}{E_{tj}}\frac{E_{tj}}{TH{N}_t}TH{N}_t$

The meanings of the variables in Eqs. (1), (2), (3) are described in Table 1.

Table 1

Item	Meaning	Item	Meaning
Ctij	CO2 emissions fossil fuel i sector j year t	Yt	Total economic output year t
FFtij	Consumption fossil fuel i sector j year t	TDtj	Passenger/ Freight Distance sector j year t
FFtj	Total consumption fossil fuels sector j year t	TTDt	Total Transport Distance year t
Etj	Total energy consumption sector j year t	THNt	Total Household Number year t
Ytj	Economic output sector j year t

Assume that CEtij=Ctij/FFtij is the carbon emissions coefficient for fuel i in sector j for year t; FStij=FFtij/FFtj is the ratio of fossil fuel i to total fossil fuels in sector j for year t; REtj=FFtj/Etj is the share of total fossil fuels in total energy consumption in sector j for year t; EIEtj=Etj/Ytj is the energy intensity of economic sector j ( j=1,2,3,4) for year t; EITtj=Etj/TDtj is the energy intensity of each transport sector (mode) j ( j=5,6,7,8,9,10) for year t; EIRtj=Etj/HNtj is the energy intensity of the residential sector for j=11 for year t; EStj=Ytj/Yt is the share of economic output in sector j (j=1,2,3,4) in total economic output for year t; ETt=Yt/Y0 is the change in total economic output for year t; TStj=TDtj/TTDt is the share of transport distance in sector (mode) j in total transport distance (j=5,6,7,8,9,10) for year t; TTt=TTDt/TTD0 is the change in total transport distance for year t; HNt=THNt/THN0 is the change in the total number of households for year t; where 0 is the base year and t the target year. Eqs. (1), (2), (3) can then be re-written as

(4)$\frac{Ceco{n}_{j,t}}{Ceco{n}_{j,0}}=\sum _{i=1}^{6}C{E}_{tij}F{S}_{tij}R{E}_{tj}EI{E}_{tj}E{S}_{tj}E{T}_t$

$\frac{Ctran{s}_{j,t}}{Ctran{s}_{j,0}}=\sum _{i=1}^{6}C{E}_{tij}F{S}_{tij}R{E}_{tj}EI{T}_{tj}T{S}_{tj}T{T}_t$$\frac{Cre{s}_{j,t}}{Cre{s}_{j,0}}=\sum _{i=1}^{6}C{E}_{tij}F{S}_{tij}R{E}_{tj}EI{R}_{tj}H{N}_t$

The steps required to develop Eqs. (4), (5), (6) as LMDI I are detailed in Ang and Liu (2001). The detailed decomposition formulae applied in this study are presented in the Appendix. These give the determinant effects in each of the sectors described in Table 2 along with the nomenclature used for results. These effects can be categorised into three groups: the intensity effects Cemc, Cinte, Cintt and Cintr, the structure effects Cffse, Crepe, Ces and Cts, and the scale effects Cet, Ctt and Chn. The Cemc is the ratio of CO2 per unit of energy for each fuel type in each sector. It analyses fuel quality and the installation of abatement technologies. As electricity is included as a fuel type in the consuming sectors, this effect also shows the change in the CO2 coefficient of electricity due to fuel switching and renewables in power generation. The Cinte, Cintt, Cintr effects measure the change in CO2from the change in the intensity of energy use in each sector and can represent the push and pull of both technological efficiency and socio-economic behaviour. They can also subsume intra-sectoral structural changes and energy price effects. In the economic sectors Cinte measures change based on the energy consumption per unit of GVA. Cintt measures change in CO2 based on the energy consumption per unit of travel activity (p-km and t-km), while Cintr measures change through the energy consumption per household unit. Cffse is a structural effect that represents the ratio of each fuel type in total fossil fuels. This effect measures the substitution of fossil fuels within each sector but not in electricity as this is a demand side analysis. Crepe shows the penetration of renewable energy into total final consumption under demand side control in each sector and not that in power generation. Cesmeasures the change in the structure of the economy, and Cts measures change in the structure of transport modes. The scale effects Cet, Ctt and Chn measure the changes in CO2emissions due to the changes in total economic output of the economic sectors, total transport work performed and total number of households respectively. Ctot indicates the aggregated change of all effects over time in each sector.

If you are interested in this research article, you can download the full length article through the download page using the link below, both PDF file and Word file are provided.

https://www.sciencedirect.com/search/advanced?tak=LMDI&show=25&sortBy=relevance