The role of energy-service demand reduction in global climate change mitigation: Combining energy modelling and decomposition analysis

Energy Policy 39 (2011) 7224–7233

Contents lists available at SciVerse ScienceDirect

Energy Policy journal homepage: www.elsevier.com/locate/enpol

The role of energy-service demand reduction in global climate change mitigation: Combining energy modelling and decomposition analysis Fabian Kesicki n, Gabrial Anandarajah UCL Energy Institute, University College London, 14 Upper Woburn Place, London, WC1H 0NN, United Kingdom

a r t i c l e i n f o

a b s t r a c t

Article history: Received 14 January 2011 Accepted 19 August 2011 Available online 3 September 2011

In order to reduce energy-related CO2 emissions different options have been considered: energy efficiency improvements, structural changes to low carbon or zero carbon fuel/technologies, carbon sequestration, and reduction in energy-service demands (useful energy). While efficiency and technology options have been extensively studied within the context of climate change mitigation, this paper addresses the possible role of price-related energy-service demand reduction. For this analysis, the elastic demand version of the TIAM–UCL global energy system model is used in combination with decomposition analysis. The results of the CO2 emission decomposition indicate that a reduction in energy-service demand can play a limited role, contributing around 5% to global emission reduction in the 21st century. A look at the sectoral level reveals that the demand reduction can play a greater role in selected sectors like transport contributing around 16% at a global level. The societal welfare loss is found to be high when the price elasticity of demand is low. & 2011 Elsevier Ltd. All rights reserved.

Keywords: CO2 emission reduction Energy-service demand Energy system modelling

1. Introduction Human beings demand energy in order to meet energy services such as heating, cooling, lighting, cooking, transport, or machine drive, via end-use devices, which consume final energy. In order to provide energy services (useful energy), conversion technologies (for example, power plants or refineries) and corresponding infrastructure are used to transform primary energy (for example coal) into final energy (for example electricity). End-use devices, such as light bulbs, refrigerators, or electric radiators, transform final energy into useful energy-services. Currently most of the primary energy sources used in the energy system are fossil fuels (hydrocarbons). Technologies are involved all the way from upstream to end-use sectors emitting CO2 and other greenhouse gases by burning fossil fuels. As a response to the challenge of climate change mitigation, there have been several research and modelling exercises carried out mostly discussing options like shifting to low carbon fuels, renewable energy and improving the efficiency of energy transformation and end-use technologies. Getting engineering right is not always enough, however; cultural and behavioural change affecting demand for energy services is necessary in order to meet the stringent climate change mitigation targets (Webler and Tuler, 2010). As human beings demand energy services, reducing energy-

n

Corresponding author. Tel.: þ44 203 108 5973; fax: þ 44 203 108 5986. E-mail address: [email protected] (F. Kesicki).

0301-4215/$ - see front matter & 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.enpol.2011.08.043

service demands will, ceteris paribus, result in lower energy consumption and consequently reduce CO2 emissions in the whole energy system. Moreover, demand reacts to price changes: an increased energy service price will result in a decreased demand for energy services and vice versa. This paper quantifies the role of energy-service demand reduction in meeting global CO2 mitigation targets by decomposing the results of the TIMES Integrated Assessment Model (TIAM)–UCL global energy system model using the Logarithmic Mean Divisia Index (LMDI) as a decomposition technique. The term ‘demand reduction’ generally means reduction in final energy consumption such as gasoline, diesel or electricity. In this paper, demand reduction means reduction in energy-service demands due to increased prices for meeting those services. This paper does not include price-independent and income-dependent behavioural adaptations, such as a non-price-related reduction in thermal comfort or the reduction of speed limits on motorways. Neither does this paper study the price response of final energy consumption, such as gasoline, to price changes. The overall objective of this paper is to examine the role of price-sensitive energy-service demand reduction in meeting global CO2 reduction. We estimate the level of CO2 emission reduction due to demand reduction at a regional and sectoral level and analyse the implications of demand reduction on societal welfare and regional emissions. The remaining paper is structured as follows: Section 2 reviews the literature on CO2 emission reduction and in particular on the role of energy-service demand. Section 3 presents the employed methodology, i.e. the TIAM–UCL model, demand reduction theory,

F. Kesicki, G. Anandarajah / Energy Policy 39 (2011) 7224–7233

and the decomposition technique used in this paper, while Section 4 defines the scenarios for the analysis. Section 5 discusses the results concerning the level of energy-service demand reduction in different end-use sectors, the contribution of demand reduction to CO2 mitigation, and the impacts of demand reduction in terms of welfare cost and regional emission mix. Section 6 concludes the paper.

2. Literature review In the context of global climate change mitigation a significant amount of literature has focused on possible ways to reduce carbon-intensive final energy consumption. Different ways to tackle rising carbon emissions include an increased energy efficiency, structural changes to low or zero carbon technologies, fuel switching, carbon capture and storage (CCS) and behavioural change. The last term can again be divided into price-related demand changes and those that are non-price related. Several studies have looked at the possible contribution of energy efficiency to limit climate change (see e.g. Hanaoka et al., ¨ 2009; Urge-Vorsatz and Metz, 2009). Further studies have reviewed the role of possible technology options, such as CCS, for the reduction of CO2 emissions (den Elzen et al., 2008; Fisher et al., 2007; Gerlagh, 2006). Moreover, there are studies that discuss ways to reduce energy demand, such as electricity or gas consumption, i.e. on the final energy level. Toke and Taylor (2007) discuss the introduction of a demand reduction obligation in the UK non-domestic sector, Hinnells (2008) looks at technologies in residential and non-residential buildings to reduce final energy demand, while Sartori et al. (2009) have considered technological options to reduce final energy in the Norwegian building stock. But there are limited studies available that look at energyservice demand reduction instead of final energy demand, particularly at a global level. Most of the studies on energy-service demand reduction used bottom-up, cost optimisation energy models such as MARKet ALlocation (MARKAL) or The Integrated MARKAL TIMES EFOM System (TIMES). Examples for studies relying on MARKAL are Anandarajah et al. (2008), Chen et al. (2007), and Kanudia and Shukla (1998). The TIMES model generator has been used in global studies by Vaillancourt et al. (2008), Syri et al. (2008), Loulou et al. (2009), and Ekholm et al. (2010). Next to model-based studies, there are other approaches that rely on the expert assessment of the emissions reduction potential. Pacala and Socolow (2004) considered reduced demand for car travel within the scope of their mitigation wedges, while Dietz et al. (2009) included demand reduction options for different energy service demands in household energy consumption as one possible option to reduce carbon emissions. A review of those studies permits to conclude that demand reduction is an important factor to be considered as the level of demand reduction seems to be significant in low carbon scenarios. According to Kanudia and Shukla (1998) demand reduction contributes up to 10% of the CO2 reduction in India in 2020. At a global level, Vaillancourt et al. (2008) found demand reduction in a climate change mitigation scenario to vary from 3% (for car travel) to 23% (for international aviation) in 2100. At a national level, demand reduction can be as high as 25% for the UK according to Anandarajah et al. (2008) and 30% for China according to Chen et al. (2007). In summary, there have been many studies that examine possible contributions of technological and efficiency options in order to reduce carbon emissions. In addition, studies have focused on the reduction of final energy consumption and based on bottom–up energy models occasionally on the reduction of energy-service demands. To the best knowledge of the authors, no

7225

study has so far translated a reduction of energy-service demands into a reduction of carbon emissions.1 The present study fills this gap by combining energy systems modelling and decomposition analysis. Research involving decomposition studies in the energy/emissions field has so far focused on the decomposition of historical data concerning energy consumption or CO2 emissions. In this context, CO2 emissions as an aggregate variable are decomposed into factors such as structural effects, carbon intensity or energy intensity (Diakoulaki et al., 2006; Zhang and Ang, 2001). Recently some studies have considered the development of underlying drivers of future emissions (Agnolucci et al., 2009; Kawase et al., 2006). To the knowledge of the authors, no study has so far analysed the role of demand reduction in meeting long-term CO2 mitigation scenarios using decomposition techniques. A merit of using decomposition technique is that the contribution of demand reduction in meeting CO2 emissions reduction target can be quantified in absolute or relative terms.

3. Methods 3.1. TIAM–UCL The 16-region TIAM–UCL model has been developed under the UK Energy Research Centre (UKERC) Phase II project (Anandarajah et al., 2010b) by breaking out the United Kingdom (UK) from the Western Europe Region in the 15 Region Energy Technology Systems Analysis Programme (ETSAP)–TIAM model (Loulou and Labriet, 2008).2 TIAM, an acronym for the TIMES Integrated Assessment Model, is a cost optimisation partial equilibrium model that minimises total discounted energy system cost in the standard version and maximises total societal welfare in the elastic demand version. Results of the optimisation include type and capacity of energy technologies, energy consumption by fuel, energy trade flows between world regions, energy system costs, the long-term prices for the energy carriers as well as the marginal costs of environmental measures. The model covers the time horizon from 2005 to 2100. The model horizon considered in TIAM–UCL is divided into periods of 10 year duration being represented by an average year. In each region, TIAM–UCL describes the entire energy system with all essential current and future energy technologies from the primary energy supply over the processing, conversion, transport, distribution of energy carriers to the end-use sectors and the energy-service demands. These demands are linked to exogenous underlying drivers like population growth or GDP development via demand elasticities. The primary energy resources and the petroleum processing sector are further divided in OPEC and non-OPEC sub-regions. The world regions are linked through trade in crude oil, hard coal, pipeline gas, liquefied natural gas (LNG), petroleum products (diesel, gasoline, naphtha and heavy fuel oil) and emission permits. On the resource side, the conventional and unconventional oil and gas reserves and resources in the different regions as well as various enhanced recovery methods are included in the model. Conventional oil is again divided into reserves, enhanced oil recovery reserves and yet-to-find resources. All different reserve and resource types of crude oil and natural gas are classified according to the regional structure of TIAM–UCL and divided into cost categories to account for varying supply costs.

1 Kanudia and Shukla (1998) are an exception to this, but it is not clear how the estimation of the reduction in carbon emissions is derived. 2 ETSAP–TIAM, originally developed by KanLo (www.kanors.com/DCM/TIAM).

7226

F. Kesicki, G. Anandarajah / Energy Policy 39 (2011) 7224–7233

In addition, renewable energy sources and their potentials as well as alternative technologies for synthetic fuels (e.g. coal-to-liquid, gas-to-liquid) and different pathways for hydrogen production are considered in the supply side of the model. TIAM–UCL has been further developed with respect to the original ETSAP–TIAM version. The most significant change has been the extraction of the United Kingdom (UK) out of the model region ‘Western Europe’ as an explicit region. Therefore, several model characteristics, such as trade patterns and resource endowments, have been updated and the model has been calibrated to the base year 2005. In addition, the drivers of the model, including population, household number, GDP and industrial output, have been revised for the whole model horizon and been calibrated to current data for the base year (see Table 1). Updates also include adding new drivers for demand projections and the revision of the uranium fuel cycle. Further details of the TIAM–UCL development are available in the model documentation (Anandarajah et al., 2010b). The TIAM–UCL model results have been presented at different international conferences (Anandarajah et al., 2010a; Pye et al., 2010). The elastic demand version of TIAM–UCL with emission trading has been used for the analysis. A simplified representation of energy supply and elastic demands is given in Fig. 1. The objective function of TIAM minimises the discounted total energy system cost in the standard (least-cost) version, where energy-services demand is fixed, i.e., they are a straight vertical line on the horizontal axes. In the elastic demand (partial equilibrium) version of the model, where the demands respond to supply price

Table 1 Main assumptions in TIAM–UCL for example regions (annual growth rates). Category Global Population (%) Household (%) GDP (%) USA Population (%) Household (%) GDP (%) Western Europe Population (%) Household (%) GDP (%) China Population (%) Household (%) GDP (%) India Population (%) Household (%) GDP (%)

Price

2010–2020

2020–2030

2030–2050

2050–2100

1.0 1.6 3.1

0.8 1.2 2.9

0.5 0.9 2.6

0.1 0.6 1.7

0.9 1.1 2.1

0.7 1.0 1.7

0.4 0.7 1.5

0.2 0.5 1.2

0.3 0.6 2.0

0.1 0.3 1.8

0.0 0.2 1.4

 0.1 0.0 0.8

0.6 1.0 6.6

0.2 0.7 4.7

 0.2 0.3 3.3

 0.4 0.2 1.6

1.2 1.7 7.1

0.8 1.3 5.9

0.4 0.9 4.1

0.1 0.8 2.3

(Max) Total surplus = consumer surplus + producer surplus Supply curve Policy scenarios

Consumer surplus

Supply curve Reference scenario

P1 Producer surplus Po

Demand curve Cost of demand reduction

Total energy system cost D1

Do

Quantity

Fig. 1. Representation of elastic demand version of TIAM.

changes (Loulou and Labriet, 2008), the objective function maximises total surplus (consumer surplus þproducer surplus) by minimising the total discounted system cost and the cost of demand reduction (as shown in Fig. 1). In the elastic demand version, these exogenously defined energy-service demands have been replaced with demand curves (actually implemented in a series of small steps). TIAM implicitly constructs demand curves using the supply prices generated in the reference case (in the standard version) and the price elasticity of demand. In the elastic demand version, demand functions determine how each energy-service demand varies as a function of the market price of that energy service. The demand function has the following functional form: D=D0 ¼ ðP=P0 ÞE where D is a demand for an energy-service in the policy scenario; D0 is the demand in the reference case from the standard TIAM– UCL version; P is the price of each energy-service demand in the policy scenario; P0 is the price of each energy-service demand in the reference case; E is the (negative) own-price elasticity of the demand. A combination of the change in prices (P/P0) and the elasticity parameter (E) determines the energy-service demand changes. The variation parameter is an additional parameter that sets the ultimate limit (reduction floor) to the demand change and the step parameter determines the size of the increment the model can select for that variation. Note that changes in energy-service demand also depend on the availability and costs of technological conservation, efficiency and fuel switching options as they influence the energy-service price. Under fixed energy-service demands in standard TIAM–UCL, CO2 reduction is achieved by energy efficiency improvements, shifting to low or zero carbon fuels/technologies and sequestration. In the elastic demand version of TIAM–UCL, demand reduction also plays a role in reducing CO2 emissions. Demand reduction depends on the price elasticity of demand and incremental costs of alternative options available to meet the energyservice demand. As there are no empirical measures of the relation between the level of energy-service demands and its price, it is very difficult to estimate the price elasticity of energyservice demands. There are no comprehensive studies on the price elasticities of energy-service demands, except for transport and residential energy-service demands, but then they do not cover the whole world (see e.g. Dreher et al., 1999). The elasticities (presented in Table 2) used in the TIAM–UCL model are long-run elasticities (price elasticity of energy-service demand such as heating, cooling, lighting) and most of them are same as in the ETSAP-TIAM model. Guertin et al. (2003) showed a clear variation in behavioural responses to changes in price across the income groups and energy services. Low-income households are more responsive to price changes than higher-income household, i.e., price elasticity of energy-service demand is high for low income groups as compared to that in the high income groups. People in developing countries receive a lower income compared to industrialised countries. Since they spend a relatively high share of their income on energy services, their response to price increases is relatively high compared to that of people in developed countries. This leads to regional differences in the value for elasticity; for example, elasticity for car demand is low for developed countries (Region 1) and high for developing countries (Region 2). Elasticity values also vary among energy services demand so that the elasticity of demand for international aviation is higher than that for international shipping as one would expect more reduction in passenger transport compared with freight transport when the price increases. It is important to stress the aggregate nature and sparse empirical basis for the estimation of

F. Kesicki, G. Anandarajah / Energy Policy 39 (2011) 7224–7233

Table 2 Price elasticity of demand for increasing prices used in TIAM–UCL LC-MED scenario (constant during 2020–2100). Energy-service demand

Region 1

Region 2

Region 3

Commercial—Space Cooling Commercial—Cooking Commercial—Space Heating Commercial—Hot Water Heating Commercial—Lighting Commercial—Electric Equipment Commercial—Refrigeration Industry—All Branches Residential—Space Cooling Residential—Cloth Dryers Residential—Cloth Washers Residential—Dish Washers Residential—Electric Appliances Residential—Space Heating Residential—Hot Water Heating Residential—Cooking Residential—Lighting Residential—Refrigeration Transport—Domestic aviation Transport—International aviation Transport—Bus Transport—Commercial Truck Transport—Three Wheeler Transport—Heavy Truck Transport—Light Truck Transport—Medium Truck Transport—Car Transport—Two Wheeler Transport—Freight Rail Transport—Passenger Rail Transport—Domestic Navigation Transport—Int. Navigation

 0.05  0.01  0.01  0.01  0.01  0.01  0.01  0.10  0.05  0.01  0.01  0.03  0.05  0.01  0.01  0.01  0.01  0.03  0.20  0.20  0.05  0.05  0.05  0.05  0.05  0.05  0.05  0.05  0.05  0.10  0.15  0.15

 0.25  0.05  0.10  0.10  0.15  0.20  0.15  0.10  0.10  0.05  0.05  0.05  0.30  0.05  0.15  0.01  0.10  0.30  0.20  0.30  0.15  0.10  0.10  0.10  0.40  0.10  0.40  0.10  0.10  0.10  0.10  0.15

 0.25  0.05  0.15  0.15  0.15  0.20  0.15  0.10  0.15  0.15  0.15  0.15  0.30  0.10  0.10  0.01  0.15  0.30  0.20  0.30  0.15  0.10  0.10  0.10  0.30  0.10  0.30  0.10  0.10  0.10  0.10  0.15

price elasticities of energy-service demands, so that sensitivity analysis around the elasticities is of particular importance. 3.2. Decomposition technique This paper explicitly investigates the role of demand reduction by decomposing CO2 reduction, resulting from imposed constraints onto its drivers. Index decomposition analysis represents in this context an appropriate tool to set forth the contribution of driving factors behind the change of an aggregate variable. The focus here is on the contribution of demand reduction towards CO2 mitigation. Therefore, the decomposition formula is kept very simple in the following way: CO2 ¼ Demand 

CO2 Demand

Usually, decomposition analysis uses final energy, but this energy form is not used in the present study as the focus is on price-sensitive energy-service demand (useful energy). In the standard version of TIAM, demand levels for all demand types and regions are an input into the energy system model based on underlying drivers, such as population or GDP. The results of the elastic demand version of TIAM include CO2 emissions and energy-service demand levels (adapted for changing prices) for all regions and sectors. These results are given over the entire model horizon in steps of ten years, which are used as an input for the decomposition analysis. As we are not interested in the absolute level of CO2 emissions, but rather in the changes of CO2 emissions between the reference and the mitigation scenario, a decomposition of the change has the following form:

DCO2 ¼ DDemandþ D

Region 1: Australia (AUS), Canada (CAN), Japan (JPN), USA, Western Europe (WEU), United Kingdom (UK). Region 2: Africa (AFR), China (CHI), India (IND), Middle-east (MEA), Other developing Asia (ODA). Region 3: Central and South America (CSA), Eastern Europe (EEU), Former Soviet Union (FSU), Mexico (MEX), South Korea (SKO).

ð1Þ

where CO2 represents total CO2 emissions, Demand stands for the energy-service demand level, and the last factor stands for carbon intensity of demand. The first factor on the right hand side of the equation is the energy-service demand, while the second can be interpreted as the CO2 intensity of demand. This last term is an aggregated variable that incorporates structural effects, efficiency changes and fuel switches. Since this paper discusses only the influence of demand changes in CO2 reduction, the latter factor is not further detailed. However, it should be noted that the contribution of the demand factor will remain the same independent of what other factors are included in the decomposition formula (see Sun and Malaska, 1998).

7227

CO2 þ residual Demand

ð2Þ

where DCO2 is the change in CO2 emissions and DDemand is the change in the level of energy-service demand. Consequently, the first summand represents a change in CO2 emissions due to a change in energy-service demand and the second summand stands for changes in CO2 emissions due to changes in the broadly-defined CO2 intensity of demand. Once demand-related CO2 emissions changes are calculated, the contribution of demand reduction to overall CO2 emission reduction is the result of the ratio: DDemand/DCO2. In this paper, the additive variant of the logarithmic mean Divisia index (LMDI) (Ang et al., 1998) is used in order to eliminate the residual. Decomposition analysis is a series expansion, which is truncated at first order and therefore possesses a residual. Nevertheless, the redistribution of this residual in the case of the LMDI simplifies the interpretation for decision makers. The LMDI method was chosen as it is judged easy to calculate and does not differ significantly from other methods that do not leave a residual. Thus, the exact decomposition formula looks as follows: ! M 0 0 COM Demand COM 2 CO2 2 CO2 DCO2 ¼  ln þ M 0 0 M lnCO2 lnCO2 lnCO2 lnCO02 Demand ! M COM 2 =Demand ln ð3Þ 0 CO02 =Demand where CO2 represents CO2 emissions and Demand stands for the energy-service demand level. The superscript M indicates a climate change mitigation scenario and the superscript 0 the reference scenario.

4. Scenarios There are five different scenarios defined for this analysis. One scenario does not involve climate change policies while all others are subject to climate change mitigation policy (constraining regional annual emissions). Three of the low-carbon scenarios are run with the elastic demand version of the TIAM–UCL model. In this context, the reference scenario is used as a benchmark for the lowcarbon scenarios. Since the elasticities for energy-service demands in the different end-use sectors are not well understood, we developed three scenarios with varied assumptions on the demand elasticity. All low carbon scenarios are based on the assumption that a cap-and-trade policy is in effect, i.e. any world region can trade emissions with other regions in order to meet its specific target. The scenarios are defined as follows:

 Reference scenario (REF): no climate change policy is applied. The standard version of TIAM–UCL is used, i.e. with no pricedriven demand reduction.

7228

F. Kesicki, G. Anandarajah / Energy Policy 39 (2011) 7224–7233

 Low-carbon scenario (LC-STD): constraining individual regions

  

to reduce global CO2 emissions to meet 450 ppm CO2 concentrations. In 2050 and post-2050, –80% target for Annex I countries, –30% target for China and India, and þ30% target for all other regions compared to the 2005 CO2 emission levels. The standard version of TIAM–UCL is used, i.e. with no pricedriven demand reduction. Low-carbon scenario-elastic demand (LC-MED): elastic demand version of the TIAM–UCL is used. Elasticities presented in Table 1 are used. Otherwise same as the LC-STD. Low-carbon scenario-higher elasticity (LC-HED): price elasticity of demand is 50% higher than that in LCS-MED. Low-carbon scenario-low elasticity (LC-LED): price elasticity of demand is 50% lower than that in LCS-MED.

Different targets have been chosen for different regions in order to take account of different responsibilities in tackling climate change. India and China have relatively strict targets compared to all other developing regions as they are emerging economies and account for the biggest share of emissions in 2050.

5. Results 5.1. CO2 emissions in REF scenario Regional CO2 emissions are presented in Fig. 2 in the REF scenario during 2005–2100. Global energy related CO2 emissions increase fourfold during this century in the absence of climate change mitigation policies. The current biggest emitter, China, will increase its contribution to global CO2 emissions from 18% in 2005

Fig. 2. Regional CO2 emissions in the reference scenario.

to 29% in 2100, while the share of the USA decreases from 22% to 11% during the same period. Overall, developing countries emissions increase more rapidly than developed countries, which currently contribute over half of the emissions, and are responsible for more than two third of the global emissions in 2100. The reason for this is the assumed high economic growth rate that affects energy-service demands. There is also a shift towards energy intensive industry in developing countries leading to a rapid increase in energy demand and consequently CO2 emissions. 5.2. Demand reduction The demand reduction level is influenced by the demand function that is constructed based on the price elasticity and implicitly constructed reference prices in the REF scenario. The level of demand reduction then depends on both the price elasticity of demand and the prices of alternative technologies and fuels available to meet the particular energy-service demand. For a particular energy-service demand, the demand reduction level will be high if alternative technologies possess a relatively high incremental cost (or vice versa). Demand reduction under different scenarios (high and low values for elasticity) are presented in Figs. 3–5 for residential, non-road transport (aviation, shipping, and rail transport) and road transport sectors respectively for selected regions and at the global level (GBL). Results show that demand reduction is sensitive to the respective elasticity in all sectors. Industry sector demand reduction is as high as 4.7% in the low elasticity scenario (LC-LED) and up to 12% in the high elasticity scenario (LC-HED). Since the growth rate of the underlying drivers are high in developing countries like China and India, the residential demand reductions are relatively high for those countries (Fig. 3). Elasticity plays a key role in the residential sector demand reduction. A higher elasticity for the residential sector energy-service demand in developing countries results in a higher demand reduction from developing countries. In the high elasticity scenario (LC-HED), non-road transport demand reduction goes up to 24%, which is relatively high as compared to that in road transport (maximum 6%) in 2100. This is due to the fact that the latter (for example cars) has low carbon alternative options at lower incremental cost while the former (for example air travel) has very limited low carbon alternative options with very high incremental cost. Demand reduction for road transport is relatively high for developing countries while the demand reduction for non-road transport such as aviation and shipping is relatively high for developed countries and reaches the ultimate limit (reduction floor) for WEU. This is due to the high elasticity values for the respective regions.

Fig. 3. Demand reduction level in residential sector under low and high elasticity scenario.

F. Kesicki, G. Anandarajah / Energy Policy 39 (2011) 7224–7233

7229

Fig. 4. Demand reduction level in road transport sector under low and high scenario.

Fig. 5. Demand reduction level in non-road transport sector under low and high scenario.

Fig. 6. Contribution of global demand reduction to overall CO2 emission reduction for different elasticities.

5.3. Contribution of demand reduction to CO2 mitigation 5.3.1. Overall contribution Results of the decomposition analysis indicate that the contribution of demand reduction to global CO2 reduction, when comparing the elastic demand low carbon scenarios with the REF scenario, is in all scenarios around 5% ( 72%) as shown in Fig. 6. This means that around 5% of the total CO2 reduction is achieved by reducing energy service demands in different end-use sectors, while the larger portion (95%) of the reduction is achieved by improving CO2 intensity (efficiency improvement and shifting to low/zero carbon technologies) and other measures, such as

sequestration. Demand changes have a very limited effect on emission reduction prior to 2020 since demand elasticities are assumed to be lower for this period. While the contribution of demand reduction tends to be highest in early periods (2020– 2030) with up to 9% due to the lack of cost-efficient low-carbon technologies, the share of demand reduction to CO2 emission reduction decreases slightly towards the end of the 21st century. This does not mean that the absolute demand reduction (expressed as a change in energy-service demand compared to the REF scenario) decreases over time towards the end of the century. On the contrary, the energy-service demand reduction increases over this period (Figs. 3–5). However, the contribution of technological options to meet the increasing CO2 reduction targets increases more strongly over the period due to greater availability of cheaper low/zero carbon technologies. Consequently, the CO2 emission reduction from structural changes and efficiency gains outweigh the demand contribution in the later part of the 21st century and decrease the demand contribution’s share of the total abated amount. This trend of a decreasing importance of demand reduction is counteracted by a shift towards more price sensitive energy service demand categories, such as residential heating or electric appliances. Consequently demand contribution stays relatively constant around 5% of overall CO2 reduction. The variation of the demand elasticity reveals that the contribution of demand reduction towards a mitigation of CO2 emissions is sensitive to the assumed elasticity level. Consequently, overall contribution of demand reduction to CO2 mitigation varies between 3% and 7%. This corresponds roughly to the 50% decrease and increase of demand elasticity in the different scenarios. Due to the fact that demand reduction contribution in the medium case is only 5%, even a high sensitivity to the elasticity level does not to lift the contribution of demand reduction above 7%.

7230

F. Kesicki, G. Anandarajah / Energy Policy 39 (2011) 7224–7233

5.3.2. Regional contribution In order to gain more insights into the global structure of the contribution of demand reduction to emission mitigation, regional results are depicted in Fig. 7 (USA and Western Europe) and Fig. 8 (India and China). Common features are that there is virtually no contribution from demand reduction before 2020 due to the fact that the emission trajectories between the reference and the mitigation scenarios only start to diverge from 2020 onwards. Furthermore, demand reduction contribution to CO2 mitigation tends to be highest in 2020–2030, while this contribution is diminished with time though the level of absolute demand reduction increases over time (Figs. 3–5) as described above. A major difference between the USA and Western Europe in the model is that the contribution of demand reduction is significantly higher in Western Europe (Fig. 7). While it is 2.8% in the medium scenario for the USA, it amounts to 4% for Western Europe. This can be explained with cheaper technological mitigation options due to a higher carbon and energy intensity that exist in the USA compared to Western Europe. This again has an influence on the cost for energy-service demands, which are higher in Western Europe and therefore trigger higher demand reductions. Compared with the developed countries, Fig. 8 highlights that demand reduction plays a more important role in India and China. Here, the cumulative contribution of demand reduction in the medium scenario is 6.6% for India and 6.2% for China. This is the result of a higher price sensitivity of demand in developing regions compared with industrialised regions. The demand reduction is 25% for China in 2020 in the medium scenario and 32% in the high elasticity scenario, compared with roughly 10% in both scenarios for India in 2020. As the overall CO2 reduction in China in 2030 is rather low however, this is not reflected in the

cumulative figure. The results for both countries in 2020 can be explained with a higher carbon intensity of energy service demand in China, which stands at 120 g CO2/MJ compared to 80 g CO2/MJ in India. This leads to higher energy-service prices in China as the availability of low-carbon technologies is limited and is reflected in an important demand reduction. Interesting to note is that the role of demand reduction in India tends to increase over time and is more sensitive to the assumed demand elasticity compared to China. The residential energy consumption is relatively important in India with a share of 40% in the total mix of energy service demands in 2020. This compares to 24% for China. Since residential demand becomes more elastic in India over time caused by an increased importance in more price-elastic demands such as electric appliances, space cooling and refrigeration, overall demand contribution increases as well. 5.3.3. Sectoral detail A look on the sectoral level can reveal insights into the importance of demand reduction at a more detailed level. For considerations at the sectoral level, emissions from secondary energy carriers, such as electricity, are accounted for in the enduse sectors. Demand reduction plays the most important role in the decarbonisation of the transport sector (Fig. 9). In the medium elasticity scenario 15.8% of the total mitigated CO2 emissions can be traced back to demand reduction in the transport sector, the respective figure is only 2.4% in the residential sector (Fig. 9). An explanation for this general trend is the much higher demand elasticity in the transport sector and the relatively high cost mitigation opportunities in the transport sector, as e.g. electric vehicles, increased share of biodiesel, compared to the residential sector, where relatively cheap abatement options exist

Fig. 7. Contribution of demand reduction to overall CO2 emission reduction in the USA (left) and Western Europe (right).

Fig. 8. Contribution of demand reduction to overall CO2 emission reduction in India (left) and China (right).

F. Kesicki, G. Anandarajah / Energy Policy 39 (2011) 7224–7233

7231

Fig. 9. Contribution of demand reduction to overall CO2 emission reduction in the transport sector (left) and residential sector (right).

600 500

US$/t-CO2

in the form of house insulation or increases in the efficiency of boilers. However, while in the years 2030 and 2040 the share of demand reduction in the transport emission reduction is very high (much higher than 30%), the emission reduction is very low during that period in the transport sector. In early periods, the transport sector has limited low carbon options as low carbon electricity based on nuclear power and renewables is not yet available. The CO2 price increases significantly from 2020 to 2040 and thereby increases the price for energy service demands, which again reduces the demand level particularly for shipping, aviation and railway transport. Structural changes in the transport sector start to be of importance from 2040 onwards explaining the decreasing share of demand contribution. Furthermore, fossil fuel prices increase with time due to the depletion of cheap resources so that biofuels become relatively cheaper. At the end of the model period a significant share of car travel is realised via methanol cars, where total fuel costs are about 80% more expensive compared to a gasoline internal combustion engine used in the reference scenario. This can be mainly explained with a better efficiency of a gasoline internal combustion engine car compared to a methanol car. However, if one takes into account the investment costs and operating and maintenance costs, the methanol alternative is only 14% more expensive. Those technologies also reduce the potential for further demand reduction, because there is only an incentive to reduce demand if CO2 is emitted, but no benefit if technologies do not emit any carbon. For decision-makers it is important to consider that demand reduction is a flexible means to reduce emissions in early periods at low to moderate carbon tax levels. In later periods, i.e. at higher carbon prices, when travel is almost completely decarbonised, there would be no incentive to further reduce demand as it would not result in carbon emissions savings. Concerning policies that address behavioural change, the transport sector could potentially be targeted first over the next few years before technological alternatives become available on a large scale since demand reduction can be relatively high compared with other sectors. In the residential sector, the total contribution of demand changes in terms of CO2 emission reduction remains very limited due to the fact that the demand elasticities are very low for energy services. These elasticities are close to zero for space heating, hot water and cooking that dominate energy consumption in the residential sector. Similar to the transport sector, the demand contribution is higher in 2030 and 2040 because demand responds to an increasing energy service price caused by a higher CO2 price. At this time the decarbonisation of electricity is still rather limited. Decarbonisation in comparison to the reference scenario in the residential sector is mainly characterised by the decarbonisation of electricity consumption, which increases rapidly over the course of the 21st century, while structural changes and efficiency gains play only a minor role. Nevertheless,

400 300 200

LC-STD LC-LED LC-MED LC-HED

100 0

2010 2020 2030 2040 2050 2060 2070 2080 2090 2100

Fig. 10. Marginal CO2 abatement cost under different scenarios.

Table 3 Share of selected regions in total CO2 emission in 2050 and 2100. 2050 (%)

CHI IND USA WEU

2100 (%)

STD

LED

MED

HED

STD

LED

MED

HED

50.8 7.0 4.7 3.9

50.3 6.4 5.3 4.0

48.5 6.0 6.9 5.0

47.5 5.7 7.5 5.6

56.8 0.7 4.8 0.9

54.1 5.4 4.4 0.6

52.1 8.6 3.6 0.8

46.5 9.6 3.4 1.4

the share of demand changes increases towards the end of the 21st century, which can be explained with an increased share of space cooling, electric appliances and refrigeration in the household energy mix. Those energy-services are assumed to be more price sensitive than cooking for example. 5.4. Implications of demand reduction cost and regional CO2 mix Fig. 10 presents the marginal CO2 abatement cost under different scenarios generated by TIAM–UCL. The elastic demand version of TIAM–UCL decreases the marginal CO2 abatement costs compared to the standard version of the model. As expected, the marginal abatement cost decreases when the elasticity is increased. Marginal abatement cost in 2050 is 337 US$/t-CO2 in LCD-STD and 219 US$/t-CO2 in LCD-HED. Respective figures for 2100 are 512 and 418 US$/t-CO2. The share of selected regions on the residual global emissions in 2050 and 2100 under different low carbon scenarios are presented in Table 3. It is based on the actual domestic emissions emitted by each region. Modelling results show that demand reduction, as a response to price changes, influences the regional CO2 mix post 2050. For example, the share of China in the residual

7232

F. Kesicki, G. Anandarajah / Energy Policy 39 (2011) 7224–7233

Fig. 11. Energy system cost (left) and welfare losses (right) under different scenarios.

CO2 emissions decreases under elastic demand scenarios (LC-LED, LC-MED and LC-HED) in 2050 and 2100 as compared to the standard version, where the energy-service demand is fixed, while the share of India decreases in 2050 and increases in 2100. The share of different regions in residual CO2 emissions varies under different scenarios. In 2100, China and the USA have their lowest share (i.e. the highest level of domestic mitigation) in the residual emissions when the demand elasticity is high, while India has its highest share in that scenario. Contribution to residual emissions under different elasticity scenarios are affected by levels of emission trading as well as technical factors. At the lowest level of elasticity, the contribution of demand reduction to meet the CO2 mitigation target is limited, domestic mitigation hence becomes expensive in China leading it to buy more emission credits in the international market. It is then economic for sellers like India and other developing countries to mitigate more than the target as they benefit by selling credits to China, especially in 2100. In contrast, at the highest level of elasticity, India emits more than its target as it is economic to buy cheaper credits available in the market to meet its target as China demands less credits in the market. This means that TIAM–UCL is sensitive to input parameters (price elasticities), i.e., if a region (like China) values its energy-services highly, i.e. a low price elasticity, in a competitive market, it will affect the consumption pattern of other regions. Fig. 11 shows the global discounted energy system cost under different scenarios and welfare losses in the mitigation scenarios compared with the reference scenario. Total welfare losses consists of costs associated with foregone demand and an increase in supply costs compared to the reference scenario. When demand reduction reduces energy system costs it needs a behavioural change, which comes as a cost to society in terms of welfare losses due to the un-served energy-service demand. Total welfare losses are about US$ 3 trillion in 2050 and US$ 7–8 trillion in 2100. The present analysis shows however that welfare losses are lower when the price elasticity of demand is high. Lower energy system cost at high-elasticity scenarios, i.e. savings due to lower energy production, outweigh the losses associated with foregone demand as the demand curve is flatter.

6. Conclusion The paper presented an analysis undertaken with the TIAM–UCL in combination with decomposition analysis in order to quantify the contribution of energy-service demand (useful energy) reduction to overall CO2 emission reduction. From the results, it can be concluded that price-induced demand reduction can play a role within a limited scope next to more important measures, in particular,

structural shifts towards carbon-free energy technologies. According to the model results, demand reduction contributes between 3 and 7% to overall CO2 emission reduction on a global level throughout the 21st century. At the sectoral level, it plays a significant role for selected energy-service demands especially in the transport sector contributing around 16% to emission reductions, while it is insignificant in the residential and commercial sector due to the assumption of low elasticity levels. Contribution of demand reduction is higher in early periods as the cheaper low/zero carbon technologies are not cost-efficient by then and an increasing CO2 price, which translates into higher energy-service demand prices, triggers the demand response. Further, demand reduction can affect the regional emission mix. In 2100, China and the USA have their lowest share in residual emissions when demand elasticity is high, while India and Africa have their highest share. Regional shares in residual emissions under different elasticity scenarios are affected by the level of emission trading next to technical factors. Demand reduction needs behavioural changes, which comes as an expense to society in terms of welfare losses due to un-served energy-service demand. Analysis shows that welfare losses decreases slightly as the price elasticity of energy-service demands increases. Welfare losses are about US$ 3 trillion in 2050 and US$ 7–8 trillion in 2100. Policy implications include that demand reduction is a flexible measure to reduce CO2 emissions early in the 21st century. Due to a higher elasticity, demand reduction is expected to play a comparably bigger role in developing countries and in the transport sector. When considering policies for behavioural change, the transport sector could be targeted first as the response is relatively high compared to other sectors. Nevertheless, one important result is that demand reduction can always only make up a limited amount of CO2 emissions reduction. Technological change is essential for a transition to a low-carbon society, so that fostering development and research in this area is key. As with all research that analyses the long-term development of the energy system and carbon emissions, it is important to bear in mind that unforeseeable changes over the century can significantly alter the results of this study. A major technological breakthrough concerning fuel cell or hydrogen vehicles, for example, could lead to a significantly smaller contribution from demand reduction. These conclusions are based on assumed values for the price elasticity of energy-service demands, which vary across sectors and regions, are uncertain and are poorly understood. Further, this study is merely based on own-price elasticity and does not include price-independent demand adaptations. It does not consider economy-wide interactions as it is focused on the energy system. In future research, it would be interesting to quantify the influence of cross-price elasticities that allow for substitution between transport modes for example.

F. Kesicki, G. Anandarajah / Energy Policy 39 (2011) 7224–7233

Acknowledgments This research formed part of the programme of the UK Energy Research Centre (UKERC). In addition, the first author gratefully acknowledges the support of a German Academic Exchange Service (DAAD) scholarship. References Agnolucci, P., Ekins, P., Iacopini, G., Anderson, K., Bows, A., Mander, S., Shackley, S., 2009. Different scenarios for achieving radical reduction in carbon emissions: a decomposition analysis. Ecological Economics 68, 1652–1666. Anandarajah, G., Kesicki, F., Pye, S., 2010a. Carbon tax vs. cap-and-trade: implications on developing countries emissions. In: Proceedings of the International Association for Energy Economics International Conference. Rio de Janeiro, Brazil. Anandarajah, G., Pye, S., Kesicki, F., Usher, W., McGlade, C., 2010b. TIAM–UCL Model Documentation. UKERC, London. Anandarajah, G., Strachan, N., Ekins, P., Kannan, R., Hughes, N., 2008. Pathways to a Low Carbon Economy: Energy Systems and Modelling. UKERC, London. Ang, B.W., Zhang, F.Q., Choi, K.-H., 1998. Factorizing changes in energy and environmental indicators through decomposition. Energy 23, 489–495. Chen, W., Wu, Z., He, J., Gao, P., Xu, S., 2007. Carbon emission control strategies for China: a comparative study with partial and general equilibrium versions of the China MARKAL model. Energy 32, 59–72. den Elzen, M., Lucas, P., Vuuren, D., 2008. Regional abatement action and costs under allocation schemes for emission allowances for achieving low CO2equivalent concentrations. Climatic Change 90, 243–268. Diakoulaki, D., Mavrotas, G., Orkopoulos, D., Papayannakis, L., 2006. A bottom–up decomposition analysis of energy-related CO2 emissions in Greece. Energy 31, 2638–2651. Dietz, T., Gardner, G.T., Gilligan, J., Stern, P.C., Vandenbergh, M.P., 2009. Household actions can provide a behavioral wedge to rapidly reduce US carbon emissions. Proceedings of the National Academy of Sciences 106, 18452–18456. ¨ Dreher, M., Wietschel, M., Gobelt, M., Rentz, O., 1999. Energy price elasticities of energy-service demand for passenger traffic in the Federal Republic of Germany. Energy 24, 133–140. ¨ Ekholm, T., Soimakallio, S., Moltmann, S., Hohne, N., Syri, S., Savolainen, I., 2010. Effort sharing in ambitious, global climate change mitigation scenarios. Energy Policy 38, 1797–1810. Fisher, B., Nakicenovic, N., Alfsen, K., Morlot, J.C., Chesnaye, F.d.l., Hourcade, J.-C., Jiang, K., Kainuma, M., Rovere, E.L., Matysek, A., Rana, A., Riahi, K., Richels, R., Rose, S., Vuuren, D.V., Warren, R., 2007. Issues related to mitigation in thel long-term context. In: Metz, B., Davidson, O.R., Bosch, P.R., Dave, R., Meyer, L.A. (Eds.), Climate Change 2007: Mitigation. Contribution of Working Group III to

7233

the Fourth Assessment Report of the Inter-governmental Panel on Climate ChangeCambridge University Press, Cambridge, pp. 169–250. Gerlagh, R., 2006. ITC in a global growth-climate model with CCS: the value of induced technical change for climate stabilization. Energy Journal 27, 223–240. Guertin, C., Kumbhakar, S.C., Duraiappah, A.K., 2003. Determining Demand for Energy Services: Investigating income-driven behaviours. International Institute for Sustainable Development, Winnipeg. Hanaoka, T., Kainuma, M., Matsuoka, Y., 2009. The role of energy intensity improvement in the AR4 GHG stabilization scenarios. Energy Efficiency 2, 95–108. Hinnells, M., 2008. Technologies to achieve demand reduction and microgeneration in buildings. Energy Policy 36, 4427–4433. Kanudia, A., Shukla, P.R., 1998. Modelling of uncertainties and price elastic demands in energy-environment planning for India. Omega 26, 409–423. Kawase, R., Matsuoka, Y., Fujino, J., 2006. Decomposition analysis of CO2 emission in long-term climate stabilization scenarios. Energy Policy 34, 2113–2122. Loulou, R., Labriet, M., 2008. ETSAP-TIAM: the TIMES integrated assessment model, part I: model structure. Computational Management Science 5, 7–40. Loulou, R., Labriet, M., Kanudia, A., 2009. Deterministic and stochastic analysis of alternative climate targets under differentiated cooperation regimes. Energy Economics 31, S131–S143. Pacala, S., Socolow, R., 2004. Stabilization wedges: solving the climate problem for the next 50 years with current technologies. Science 305, 968–972. Pye, S., Strachan, N., Anandarajah, G., Usher, W., 2010. The UK energy system in an uncertain world: insights from different modelling scales. In: Proceedings of the International Energy Workshop 2010. Stockholm, Sweden. Sartori, I., Wachenfeldt, B.J., Hestnes, A.G., 2009. Energy demand in the Norwegian building stock: scenarios on potential reduction. Energy Policy 37, 1614–1627. Sun, J.W., Malaska, P., 1998. CO2 emission intensities in developed countries 1980– 1994. Energy 23, 105–112. ¨ A., Ekholm, T., Savolainen, I., Holttinen, H., Peltola, E., 2008. Global Syri, S., Lehtila, energy and emissions scenarios for effective climate change mitigation— deterministic and stochastic scenarios with the TIAM model. International Journal of Greenhouse Gas Control 2, 274–285. Toke, D., Taylor, S., 2007. Demand reduction in the UK—with a focus on the nondomestic sector. Energy Policy 35, 2131–2140. ¨ Urge-Vorsatz, D., Metz, B., 2009. Energy efficiency: how far does it get us in controlling climate change? Energy Efficiency 2, 87–94. Vaillancourt, K., Labriet, M., Loulou, R., Waaub, J.-P., 2008. The role of nuclear energy in long-term climate scenarios: an analysis with the World-TIMES model. Energy Policy 36, 2296–2307. Webler, T., Tuler, S.P., 2010. Getting the engineering right is not always enough: researching the human dimensions of the new energy technologies. Energy Policy 38, 2690–2691. Zhang, F.Q., Ang, B.W., 2001. Methodological issues in cross-country/region decomposition of energy and environment indicators. Energy Economics 23, 179–190.