Input-Output Modeling for Urban Energy Consumption in Beijing: Dynamics and Comparison Lixiao Zhang1*, Qiuhong Hu1, Fan Zhang2 102461 1 State Key Joint Laboratory of Environmental Simulation and Pollution Control, School of Environment, Beijing Normal University, Beijing, China, 2 The World Bank, Washington DC, United States of America Abstract Input-output analysis has been proven to be a powerful instrument for estimating embodied (direct plus indirect) energy usage through economic sectors. Using 9 economic input-output tables of years 1987, 1990, 1992, 1995, 1997, 2000, 2002, 2005, and 2007, this paper analyzes energy flows for the entire city of Beijing and its 30 economic sectors, respectively. Results show that the embodied energy consumption of Beijing increased from 38.85 million tonnes of coal equivalent (Mtce) to 206.2 Mtce over the past twenty years of rapid urbanization; the share of indirect energy consumption in total energy consumption increased from 48% to 76%, suggesting the transition of Beijing from a production-based and manufacturing-dominated economy to a consumption-based and service-dominated economy. Real estate development has shown to be a major driving factor of the growth in indirect energy consumption. The boom and bust of construction activities have been strongly correlated with the increase and decrease of system-side indirect energy consumption. Traditional heavy industries remain the most energy-intensive sectors in the economy. However, the transportation and service sectors have contributed most to the rapid increase in overall energy consumption. The analyses in this paper demonstrate that a system-wide approach such as that based on input-output model can be a useful tool for robust energy policy making. Citation: Zhang L, Hu Q, Zhang F (2014) Input-Output Modeling for Urban Energy Consumption in Beijing: Dynamics and Comparison. PLoS ONE 9(3): e89850. doi:10.1371/journal.pone.0089850 Editor: Ben Bond-Lamberty, DOE Pacific Northwest National Laboratory, United States of America Received September 11, 2013; Accepted January 27, 2014; Published March 3, 2014 Copyright: ß 2014 Zhang et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: This work was supported by the National Natural Science Foundation of China (Grant No. 41371521) and National Science Foundation for Innovative Research Group (Grant No. 51121003). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. * E-mail: zhanglixiao@bnu.edu.cn Introduction A more holistic way of analyzing city energy consumption is to account for both direct and indirect energy consumption, the latter Cities account for the largest share of energy consumption and of which refers to energy usage that is embodied in the also provide concentrated opportunities for large energy savings intermediate products and material that are then passed on to [1]. This is particularly true in China, where cities accounted for other sectors until they reach the final consumer. By focusing on 84% of China’s total commercial energy consumption in 2006 [2]. embodied (direct and indirect) energy consumption, one can By the end of 2012, urban population in mainland China reached identify the drivers of the overall energy consumption and 712 million, or 53% of the total population, rising from 26% in quantifies the amount of energy usage for which each sector is 1990 [3]. At the same time, the government is projecting the ultimately responsible [7–9]. The ratio of direct and indirect urbanization of an additional 350 million people–greater than the energy use can also provide essential information concerning a population of the entire United States–over the next 15 years [1]. city’s economic structure [10–12]. Such large-scale urbanization would have huge implications for Input-output (I-O) technology-based embodiment analysis has future energy demand and related environmental impacts. recently become a popular method for benchmarking systems The traditional way of analyzing urban energy consumption energy accounting, which could facilitate a deeper appreciation of highlights only the direct energy consumption of end-user sectors, each sector’s total energy requirements, including both the direct, which normally consist of agricultural, industrial, transport, visible requirements and the indirect, hidden energy costs[13–27]. commercial, and residential sectors [2,4–6]. However, urban With the help of the Leontief inverse matrix (see more details in economic systems are highly complex and integrated. Each section 2.1), such an approach could account for the cumulative economic sector not only consumes energy in a direct way in energy requirements of a sector regardless of the complexity and forms such as electricity, oil, coal, and natural gas, but also in an length of the production process [28–31]. Chen and his research indirect way by consuming energy-intensive intermediate inputs group carried out an analysis of embodied resources and emissions produced by other sectors. Ignoring the linkage between sectors at China’s national level [32–34]. Liu et al [11] use 2007 input- and focusing only on direct and final energy consumption would output table to analyze embodied energy use in China’s industrial underestimate the amount of city-level energy consumption and sectors. At the city level, Zhou et al. carried out an embodied undermine the efforts of energy savings. resource accounting analysis of Beijing’s economy [35]; Liang et PLOS ONE | www.plosone.org 1 March 2014 | Volume 9 | Issue 3 | e89850 Urban Energy Consumption of Beijing al. conducted a case study of Suzhou, involving the identification final demands with elements Yi; Ed ~(Ed 1 ,Ed 2 ,:::,Edn )T and of key sectors [12]. Ef ~(Ef 1 ,Ef 2 ,:::,Efn )T denotes the outputs exported to the As the capital city of China, Beijing exemplifies the rapid domestic and foreign markets from sector i; Similarly, urbanization and economic growth that has occurred in China Id ~(Id 1 ,Id 2 ,:::,Idn )T and If ~(If 1 ,If 2 ,:::,Ifn )T denote the imported over the past two decades. Currently, Beijing is entering into a inputs from the domestic and foreign markets to sector i. A is the phase of extensive deindustrialization and economic restructuring, technology coefficient matrix (n6n) with elements aij, describing with the relocation of the Beijing Shougang Iron and Steel Plant the amount of intermediate demand of output from the domestic from Beijing to Caofeidian, Tangshan a vivid example of such a sector i used by domestic sector j; strategy [5]. If E (16n) denotes the direct energy inputs for each sector from In this paper, we use an environmental input-output model to the perspective of sectoral production, V (16n) denotes the factor analyze the embodied energy consumption of Beijing from 1987 to vector of the energy intensity for each sector, then 2007. The main aims of this paper are: (1) to find the variations of total energy consumption (directly and indirectly) for Beijing during the past 20 years alongside rapid urbanization, (2) to Ei Vi ~ ð2Þ compare the sectoral distribution of embodied energy consump- Xi tion between 1987 and 2007 with regard to economic structural changes, (3) to reveal the transmission mechanism of energy flow through the entire supply chain or among sectors. The rest of the If e (16n) denotes the embodied energy demand per unit of paper is structured as follows: Section 2 provides an introduction production of each sector within the city, then of the environmental input-output method and data source. Section 3 presents the results and discussions. Concluding remarks e~V(I {A){1 ð3Þ are provided in Section 4. where I represents the identity matrix, (I {A){1 is the Leontief Methodology inverse. We know E is the total direct energy consumption from the Environmental input-output model production perspective and Ei is the direct energy demand for The I-O method has been applied to embodied energy production activities by different sectors. In contrast to the direct accounting many times over the past decades, without major energy requirement E (Ei), the indirect items of energy inputs are changes in methodology [36]. In fact, some references from the relative concepts. The term ‘‘indirect’’ can be defined in two ways 1970s cited in this paper read as if they had been written in depending on the context. For the whole city, it refers to any twenty-first century [37]. Technically, an environmental input- primary, secondary, or final energy consumed outside the output model has to be built through integrating the economy with boundary of the metropolitan area in order to produce energy, its energy consumption by industrial sectors. Each sector in the goods or services consumed by any entity, public or private, inside economy uses primary energy as a direct energy input into their the metropolitan area [44]; For a certain sector i, however, production process, which eventually gravitates towards final indirect energy usage is associated with energy embodied in demand. Thus, every sector takes direct energy in the form of coal, intermediate use from other sectors. natural gas and so on, and indirect energy through the embodied Therefore, for the whole city, we have reformulated the energy in inputs from other sectors. It should be noted that this equation (1): method is based on some key assumptions and subject to certain constraints and limitations [9,32,38]. For instance, it is assumed that imported commodities (both domestic and foreign imports) V(AX zY zEd zEf )~V(X zId zIf ) ð4Þ have the same embodied energy intensity as local products. Certainly this is not accurate, but we need to work with such an assumption because we simply do not know the import structure After several steps of transformation, we get, for ports of origin outside of China [39]. Although the development of a multiregional input–output (MRIO) model can e(Y zEd zEf ){VX ~eA(AX zY zEd zEf {X {Id {If ) partially address this issue [40–42], it is nearly an impossible task ð5Þ ze(Id zIf ) to build 9 MRIO models covering domestic provinces and various countries in the world with regard to the data intensive feature of the MRIO model and the limited data availability. Since AX zY zEd zEf {Id {If ~0 and VX ~E , it can be With reference to [34] and [40], an environmental input-output proved that the indirect energy requirement for the entire urban table for the urban economy was built to integrate the economic system equals the embodied energy imported from outsides. activities and energy consumption or emissions (see table 1). The embodied energy analysis framework using the I-O table has been described extensively in the literature [8,9,17,23,32,43]. In the e(Y zEd zEf ){E ~e(Id zIf ) ð6Þ following, only the most important aspects of this methodology are summarized. With regard to the inputs and outputs, the central balance equation is: For a specific sector, the indirect energy necessary for output production can be described as: AX zY zEd zEf ~X zId zIf ð1Þ ei (AXi zYi zEdi zEfi )~ei (Xi zIdi zIfi ) ð7Þ where X ~(X1 ,X2 ,:::,Xn )T is the (n61) vector of gross outputs for each economic sector i; Y ~(Y1 ,Y2 ,:::,Yn )T is the (n61) vector of PLOS ONE | www.plosone.org 2 March 2014 | Volume 9 | Issue 3 | e89850 Urban Energy Consumption of Beijing which can be further denoted as Total output (X) ei Xi {Ei ~ei AXi ð8Þ … Xn X1 X2 Indicators of influence coefficient and response coefficient As mentioned above, every sector takes direct energy from Import from foreign (If) primary energy resources, and indirectly through the embodied energy in inputs from other sectors within the urban economic system. From an energy management perspective, it is important … Ifn If1 If2 to understand the links between sectors in energy consumption. Therefore, the influence coefficient (IC) and response coefficient (RC) were used as indicators to identify the key sectors in energy transmission mechanism along the production chain. The IC refers to the effect induced on all sectors due to an increase of one domestic (Id) Import from unit of final demand in a particular sector. It reflects the backward linkage effect, i.e., the severity of impact in a sector that diffuses to all other sectors. The RC refers to the effect induced in one Idn Id1 Id2 … particular sector due to an increase in one unit of final demand Table 1. Scheme of the environmental input-output table as an integration of urban economy and energy[35,40]. Export to foreign (Ef) from all sectors. It reflects the forward linkage effect, i.e., the amount a particular sector receives from all other sectors [31]. These two indicators have often been used to analyze industrial linkages. Based on traditional correlation theory, Rasmussen extended these indicators to the energy analysis of industrial sectors [45]. Efn Ef1 Ef2 … The IC and RC of energy consumption by industry sector can Export to domestic (Ed) be derived as follows. If the demand increases by k percent, then the change in the embodied energy of the total final demand can be expressed as: E DEE ~V(I {A){1 kY ~ (I {A){1 kY X ð9Þ Edn Ed1 Ed2 E … EE ( ) ~ EE (I {A){1 kY X Final demand (Y) Where EE is the embodied energy consumption of total final Final use demand, and DEE is the change in the embodied energy of the total final demand due to an increase in final demand. Let B = E/ … Yn Y1 Y2 EE = (B1, B2 …Bi), where Bi refers to the ratio of direct energy to embodied energy in the total final demand of sector i: x1n xn1 x21 … Xn EEB En vn n DEE ~ (I {A){1 kY ð10Þ X … … … … DEE B ~ (I {A){1 kY ð11Þ EE X doi:10.1371/journal.pone.0089850.t001 Intermediate use xn2 x12 x21 … X2 E2 v1 2 Let k = 1; then DEE B ~ (I {A){1 Y xn1 x11 x21 ð12Þ … X1 E1 v1 1 EE X Total input (X) Direct energy Value added Sector inputs … n 1 2 PLOS ONE | www.plosone.org 3 March 2014 | Volume 9 | Issue 3 | e89850 Urban Energy Consumption of Beijing By diagonalizing the B and Y vectors and to let energy consumption, there are two rapid growth periods: one from 1997 to 2002 and another from 2005 to 2007. The construction B activities as a result of real estate development were, to a large h~ (I {A){1 Y ð13Þ extent, responsible for the rapid growth in indirect energy X consumption, further illustrated by an embodied energy analysis where h is the industrial linkage matrix of energy consumption by of the construction sector in sections 3.2 and 3.3. To cope with the sector, of which the ij-th element is BimijYj/Xi, denoting the change 1997 Asian financial crisis and to fuel domestic demand, real estate of energy consumption in sector i when the demand of sector j development had been encouraged by the government during increases by one percent, and mij is the coefficient of the Leontief 1997–2002. The preparation of 2008 Olympics Game contributed inverse matrix, i.e., the total demand required to produce one unit to the large wave of urban construction during 2005–2007. The of product in sector j. boom and bust of real estate development over time in Beijing and The RC is the sum of the row elements and can be expressed as: in China in general is also worth mentioning. Specifically, the declining trend of indirect energy consumption during 2002–05 X n can be attributed to the real estate development suppression L~ hij , ð14Þ policies. j ~1 Overall, the total embodied energy consumption of Beijing increased from 38.85 Mtce in 1987 to 206.2 Mtce in 2007 Mtce, which represents the percentage change in energy consumption by representing a growth rate of 9% per year. Although total energy sector i when the demand of other sectors increases by one consumption has significantly increased over time, the energy percent. intensity has, on the contrary, declined, with direct energy The IC is the sum of the column elements and can be expressed intensity decreasing from 2.66 tce/ten thousand yuan to as: 0.70 tce/ten thousand yuan, and embodied energy intensity decreasing by 52%, from 5.16 to 2.71 tce/ten thousand yuan, X n during the past 20 years (Figure 2). b~ hij , ð15Þ In addition, it is interesting to note that the proportion of direct i ~1 energy in embodied energy has declined from 52% to 26%, while that of indirect energy has increased from 48% to 76%. Since which represents the percentage change in energy consumption by indirect energy demand for the whole city reflects how a city relies other sectors when the demand of sector j increases by one on outside sources, the increase of indirect energy use reveals the percent. evolution of Beijing’s economic structure, which will be discussed intensively in the later section of this paper. Imagine an urban Data source economy in which direct energy makes up only about one quarter The analysis in the paper is based on the input-output tables of of consumption, where technology has been efficient for some the following years: 1987, 1990, 1992, 1995, 1997, 2000, 2002, time, and further reductions are likely to be less efficient. 2005, and 2007, which are compiled by the Beijing Statistical Bureau [46]. The tables for 1987, 1990, 1992, and 1995 cover 33 Comparison of sectoral energy consumption between economic sectors, and 40 sectors for 1997 and 2000, while the 1987 and 2007 tables for 2005 and 2007 include 42 sectors. We aggregate the Figure 3 compares sectoral embodied energy consumption in different sectors in different years into 30 sectors to keep these 1987 and that in 2007. Figure 4 illustrates the comparison results datasets consistent. The 30 sectors are then further divided into 5 of sectoral energy intensity of these two years. Overall, the sectoral more aggregated sector groups, i.e. agriculture (A), mining (B), embodied energy consumptions were substantially higher while manufacturing (C), construction (D) and services (E). The entries intensities were relatively lower in 2007 than their counterparts in in the tables valued in the producers’ prices are converted into 1987; Sectors in Group C, D, E account for most of the energy 1987 constant prices using the GDP price index. The sectors’ consumption in 1987 and 2007, while sectors in Group A and B classification and sector codes are shown in Table 2. The direct consumed much less. Beijing lacks natural resources, demonstrated energy consumption data for each sector are derived from the by the fact that all of the natural gas and crude oil consumed by energy balance table (EBT) of the Beijing Statistics Yearbook for the city, as well as 95% of the coal, 64% of the electricity, and 60% the corresponding years [47] (see Table S1 of appendix). For the of the refined oil consumed are imported from outside [49], which use of energy data in the EIO framework, it is necessary to allocate explains why Group B is not a large energy user. all energy back to the primary carrier [48]. With regard to sector-by-sector comparison, substantial changes can be found in the distribution of direct and indirect energy Results and Discussion consumption during this time period. For instance, both the largest direct and indirect energy consumers are in Group C in 1987, Dynamic changes of energy consumption for whole such as sector chemicals (No. 13), while in 2007 the largest direct urban economy of Beijing energy consumer is the transport sector in Group E and the largest Figure 1 shows the direct and indirect energy demand of Beijing indirect energy consumer is in Group D, i.e., the construction in the 9 years analyzed. With regard to the aggregate city-level sector; Some traditional manufacturing sectors in Group C energy consumption in Beijing, the total direct energy consump- decreased their direct energy consumption due to the downsizing tion increased by 2.64 times, rising from 20.31 Mtce in 1987 to of the sectors (e.g., textiles sector) or technology upgrade (e.g. 52.79 Mtce in 2007. This represents a 5% annual growth rate. chemical sector). There is no doubt that the current extensive Prior to 1997, direct energy consumption was always greater than deindustrialization and economic restructuring in Beijing have indirect energy consumption. This trend, however, reversed after affected energy consumption in many ways. For example, the 2000, suggesting that Beijing has become more reliant on outside relocation of Beijing Shougang Iron and Steel Plant from Beijing sources for energy demand. Regarding the increase of indirect to Caofeidian, Tangshan, Hebei province may have contributed to PLOS ONE | www.plosone.org 4 March 2014 | Volume 9 | Issue 3 | e89850 Urban Energy Consumption of Beijing Table 2. Aggregated sectors and groups for input-output analysis [35,40]. Group code Group name Sector code Sector name A Agriculture 1 Agriculture B Mining 2 Mining and washing of coal 3 Extraction of petroleum and natural gas 4 Metal ore mining 5 Nonmetal mineral mining C Manufacturing 6 Food industries 7 Textiles 8 Wearing apparel, leather, furs, feather and related production 9 Sawmill products and furniture 10 Paper products, printing, and recording media reproductions 11 Electricity, steam, and hot water production 12 Petroleum processing and coking 13 Chemicals 14 Nonmetallic mineral products 15 Metal smelting and pressing 16 Metal products 17 General- and special-purpose machinery 18 Transportation equipment 19 Electric equipment and machinery 20 Electronic and telecommunication equipment 21 Instruments and other measuring devices 22 Other manufactured products D Construction 23 Construction E Service 24 Transportation, storage, and postal services 25 Business services 26 Accommodation and food services 27 Public services 28 Professional, scientific, and technical services 29 Finance and insurance 30 Public administration and other sectors doi:10.1371/journal.pone.0089850.t002 the changes of industrial business of sector No. 2 and No. 17 in direct energy consumption with an increase of embodied energy Group C. Of all the sectors, the direct energy consumption of consumption can be attributed to the ongoing industrial transition sector 24 (transportation, storage, and postal services) increased and updates in these sectors. the most, by 7.61 Mtce, followed by sector 12 (petroleum Another important indicator to analyze is the ratio of direct and processing and coking), with an increase of 6.48 Mtce, which indirect energy consumption by sectors, or the proportion of direct are highly interrelated. The rapid growth of direct energy and indirect energy in embodied energy. We found that in almost consumption in transportation and petroleum processing can be all cases, the indirect energy consumption in a production process attributed to the rapid development of the transportation system in was higher than direct energy. The average ratio of direct energy Beijing: the number of motor vehicles in Beijing increased 11.28 consumption to total energy consumption for the 30 sectors under times, from 0.27 million vehicles to 3.1 million during this period. study was 33% in 1987 but 22% in 2007(see Table S1 and S2 in As to the embodied energy consumption, the most dramatic the appendix). The change of this ratio further illustrates the changes occurred in the downstream sectors, such as the transition of Beijing’s economy from a production-based and construction sector (No. 23) in Group D, the transportation, manufacturing dominated system to a consumption-based and storage, and postal services sector (No. 24), the public services service-dominated system. In other words, urban economic sector (No. 27), and the professional, scientific, and technical activities moved along the production chain and closer to the services sector (No. 28) in Group E. Although the direct energy end-users. China emphasized the development of heavy industry consumption of sector No. 8 increased, its embodied energy from 1949 until the late 1970s. In response to the state consumption declined. The reason for this change remains a topic industrialization strategy, Beijing built the first side-blown for future study. In contrast, the embodied energy consumption of converter for the Shougang Iron and Steel Plant for steel sectors 13, 17, 19 increased largely from 1987 to 2007; however, production in 1958 and launched the Sinopec Beijing Yanshan their direct energy consumption showed a decline. A decrease in Group in 1970. However, more recently the industrial structure PLOS ONE | www.plosone.org 5 March 2014 | Volume 9 | Issue 3 | e89850 Urban Energy Consumption of Beijing Figure 1. Embodied energy consumption of Beijing city from 1987 to 2007. doi:10.1371/journal.pone.0089850.g001 has been undergoing a transformation from traditional heavy heavy industries in Group C, such as nonmetallic production and industries to modern high technology and high-value-added metal production still remain as the most energy-intensive sectors industries, a change which has been actively promoted by Beijing’s in terms of direct energy intensity, although the energy efficiency municipal government. This transformation of the urban economy of these sector has largely improved. For instance, the direct undoubtedly has had a great influence on the level of each sector’s energy required for unit output of metal smelting and pressing has direct and indirect energy consumption and the distribution of substantially decreased from 1.15E-03 tce/yuan in 1987 to 4.62E- energy consumption among sectors. 04 tce/yuan tce in 2007. Despite of the decline, the energy From the perspective of energy efficiency as indicated by energy demand per unit of GDP in petroleum processing and coking intensity or energy consumption per unit of GDP, traditional increased greatly, which can be ultimately attributed to the Figure 2. The changes of energy consumption intensity of Beijing city from 1987 to 2007. doi:10.1371/journal.pone.0089850.g002 PLOS ONE | www.plosone.org 6 March 2014 | Volume 9 | Issue 3 | e89850 Urban Energy Consumption of Beijing Figure 3. Sectoral distribution of embodied energy consumption between 1987 and 2007 in Beijing (see Table 2 for sector code definitions). doi:10.1371/journal.pone.0089850.g003 increasingly stringent environmental regulation (e.g., extra energy has become more energy efficient than it was 20 years ago. In is needed for auxiliary equipment associated with pollutant general, there was not much change in the proportion of direct treatment). Another interesting finding is that the service sector and indirect energy consumption of most of the energy-intensive Figure 4. Embodied energy intensity by sector between 1987 and 2007 in Beijing (see Table 2 for sector code definitions). doi:10.1371/journal.pone.0089850.g004 PLOS ONE | www.plosone.org 7 March 2014 | Volume 9 | Issue 3 | e89850 Urban Energy Consumption of Beijing sectors. This is because most of these sectors consumed energy in a affected by other sectors. Sectors located in quadrant IV have direct form while the amount of indirect energy consumed during relatively higher IC values and lower than average RC values. A production was typically low. In contrast, there are great change in their demand would therefore have a great effect on differences between direct and embodied energy intensity in some other sectors’ energy consumption. These sectors are usually other sectors. For example, the construction sectors and profes- located in the downstream segment of the production chain and sional, scientific, and technical services sectors, all of which have a have backward linkages with other sectors. relatively low direct energy intensity but required much higher As shown in Figure 5, in 1987 the three most important sectors indirect energy consumption. were metal smelting and pressing (No. 15), chemicals (No. 13), and construction (No. 23). Sectors 15 and 13 in Group C had a Transmission mechanism among economic sectors much higher RC than other sectors, which means these two The IC and RC for different economic sectors in 1987 and 2007 sectors had a much stronger forward linkage with other sectors, were calculated and plotted in Figure 5 and Figure 6 in the form of while sector 23 in Group D had a much higher IC, indicating that a planar graph. The average values of the two indicators in 1987 it was a key sector in backward linkages with other sectors in the and 2007 were 0.033 and 0.034, respectively. These average energy system. Figure 6 shows that great changes have occurred in values were then used to define the starting point of the coordinate the interrelationships among sectors over the last two decades, a systems shown in Figure 5 and Figure 6. Sectors located in period corresponding to Beijing’s rapid urbanization. In 2007, the sectors located in quadrant I with the closest correlations with quadrant I are characterized by higher values of IC and RC, other sectors were associated with not only sectors belonging to which implies that an increase in their final energy demand would Group C, but also sectors in Group E. Not surprisingly, metal have a great effect on other sectors’ direct energy consumption. smelting and pressing sector (No. 15) and petroleum processing Conversely, these sectors themselves would also have been greatly and coking sectors (No. 12) had a relatively higher RC and affected by energy demand increases in other sectors. In other therefore a stronger forward linkage with other sectors, while words, these sectors demonstrated a powerful ability to diffuse and sectors construction (No. 23), public service (No. 27) and profes- to receive impacts. Because these sectors have both backward and sional, scientific, and technical services (No. 28) had a much forward linkages, they are usually regarded as key sectors in the higher IC, indicating a stronger driving force for energy energy system. Sectors located in quadrant II are characterized by consumption of other sectors. In addition, the transportation, higher RC and lower IC, which implies that their energy storage, and postal services sector (No. 24) has strong backward consumption is influenced by the demand of other sectors and and forward linkages with its upstream and downstream sectors. thus have a strong forward linkage effect. In general, these sectors Nevertheless, sector 15 remains the sector with the largest forward are usually in an upstream position in the production chain. They linkages, while sector 23 has the largest backward linkages with are responsible for providing embodied energy for the downstream another sector in both 1987 and 2007. It can be easily concluded sectors. Sectors located in quadrant III have lower-than-average that energy saving strategies should be taken from a supply chain RC and IC values, indicating that their energy consumption is less perspective, not ender user perspective. Sectors like construction Figure 5. Planar graph of RC and IC of the 30 sectors in 1987 (see Table 2 for sector code definitions). doi:10.1371/journal.pone.0089850.g005 PLOS ONE | www.plosone.org 8 March 2014 | Volume 9 | Issue 3 | e89850 Urban Energy Consumption of Beijing Figure 6. Planar graph of RC and IC of the 30 sectors in 2007 (see Table 2 for sector code definitions). doi:10.1371/journal.pone.0089850.g006 should also be considered as ‘‘energy intensive’’, although their replaced with urban commercial energy services. To achieve the consumption is concentrated in the supply chain activity rather ambitious urbanization strategy, there is a need to adjust energy than their direct energy consumption [11]. consumption behavior in order to meet the great challenges for Given this, transmission mechanism such as forward and energy demand and environmental protection. This would also backward linkages among sectors, are especially important for have strong implications for the world energy market and climate the development of energy policy. Specifically, the upstream and change mitigation activities. downstream sectoral collaboration along the whole supply chain is Beijing provides an example of the kind of change in energy essential to establish policy mix for effective and efficient urban consumption pattern associated with rapid urbanization. In this energy management. For those traditional heavy industries with paper, we analyze the direct and indirect energy consumption for high RC (for instance, sectors 15 and 12 in 2007), efforts should be the whole city and by 30 economic sectors in Beijing during 1987– focused on the improvement of efficiency through cleaner 2007. Results show that total energy consumption has steadily production, energy audits, technology updates, compulsory increased despite a decline in overall energy intensity. At the sector phase-out/shutdown of inefficient manufacturing facilities, capa- level, we find that most of the sectors experienced a decrease in bility-building programs on energy saving awareness, which would direct energy consumption, but an increase in indirect energy contribute to the improvement of embodied energy intensity of consumption, especially the construction, transportation and sectors in downstream; for those sectors with higher indirect service sectors. Changes in the pattern of energy consumption energy consumptions, efforts should focus on addressing their have shown that Beijing has shifted from a production-based and supply chain energy consumption, such as greening their supply manufacturing-dominated system to a consumption-based and chain and controlling the irrational final demand. service-dominated system. Analysis in this paper demonstrates that the traditional way of Concluding remarks analyzing urban energy consumption, which focuses solely on Over the last 20 years, China has undergone unprecedented direct energy consumption, ignores the complex energy flows rapid urbanization and profound structural changes in the urban among sectors and is insufficient for making robust energy policies. economy. Recently the Chinese government has reemphasized Energy saving measures and efficiency improvement policies urbanization as an important development strategy, and projects should not only consider those traditional heavy industries, but another 350 million more people to join the current Chinese should also pay attention to other sectors along the supply chain. urban population over the next decade. Such a large-scale For instance, greater attention should be paid to the construction urbanization will most likely spur energy demands for the sectors and the professional, scientific, and technical services construction of new buildings and infrastructure; additional sectors, which demonstrated relatively low direct energy intensity, residential energy uses will increase as well, as rural biomass is but demanded much higher indirect energy consumption. It is well PLOS ONE | www.plosone.org 9 March 2014 | Volume 9 | Issue 3 | e89850 Urban Energy Consumption of Beijing known that China’s urban economy, such as Beijing, has been Supporting Information driven primarily by infrastructure construction and capital investment. Shifting investment to areas such as education and Table S1 Direct sectoral energy consumption associat- technological innovation would be helpful not only for long-term ed with the concerned 9 years (unit: Mtce). economic growth but also for achieving energy security policy (DOCX) objectives. Table S2 Embodied energy consumption by sectors In summary, an economy-wide system accounting analysis associated with the concerned 9 years (unit: Mtce). allows us to trace the direct, as well as indirect energy consumption (DOCX) along the supply chains of an economy. It is therefore a useful tool for systematic policy making. Our analysis highlights that it is Acknowledgments embodied energy as opposed to direct energy that provides a We would like to thank the anonymous reviewer and PLoS ONE Section holistic picture on urban economy’s energy consumption. Corre- Editor Dr. Ben Bond-Lamberty for very helpful comments and suggestions. spondingly, energy saving and efficiency improvement policy should be based on such a system accounting approach, e.g., Author Contributions enhancing end treatment for those sectors with high RC, while addressing source control and supply chain management for those Conceived and designed the experiments: LZ QH FZ. Performed the sectors with higher IC. experiments: LZ QH. Analyzed the data: LZ QH. Wrote the paper: LZ FZ. References 1. Fridley D (2010) Embodied energy: an alternative approach to understanding 25. Peters GP, Hertwich EG (2008) CO2 embodied in international trade with urban energy use. Available: http://www.theoildrum.com/node/6842 Accessed implications for global climate policy. Environ. Sci. & Technol. 42(5): 1401– 20 August 2013. 1407. 2. Dhakal S. (2009) Urban energy use ad carbon emissions from cities in China and 26. Phdungsilp A. (2003) A thermodynamic framework for urban energetic policy implications. Energ. Policy 37(11): 4208–4219. metabolism. ENETT2550-003. Available: http://www.aseanenergy.info/ 3. Fung P, Liu K (2013) New Push for Reform: Signals from China’s Two Abstract/33006312.pdf Accessed 20 August 2013. Parliamentary Conferences. Available: http://www.kpmg.com/CN/en/ 27. Tiwari P (2000). An analysis of sectoral energy intensity in India. Energ. Policy IssuesAndInsights/ArticlesPublications/Documents/new-push-for-reform- 28(11): 771–778. China-Lianghui-201303-v1.pdf Accessed 20 August 2013. 28. Han XL, Lakshmanan TK. (1994) Structural changes and energy consumption 4. Minadali O, Raveh A, Salomon I. (2004) Urban density and energy in the Japanese economy 1975–85: an input–output analysis. Energ. J. 15(3): consumption: a new look at old statistics. Transportation Research Part A: 165–188. Policy and Practice 38(2): 143–162. 29. Beule F D (2000) Backward linkages of foreign subsidiaries in Guangdong, 5. Zhang LX, Feng YY, Chen B (2011) Alternative scenarios for the development China: A country-of-origin analysis. Available: https://lirias.kuleuven.be/bitstream/ of a low-carbon city: a case study of Beijing, China. Energies 4(12): 2295–2310. 123456789/386124/1/BackwardlinkagesofforeignsubsidiariesinGuangdongChina.pdf 6. Zhang LX, Yang ZF, Liang J, Cai YP (2011) Spatial variation and distribution of Accessed 20 August 2013. urban energy consumptions from cities in China. Energies 4(1): 26–38. 30. Liu QY, Peng CY (2002) Analysis of changes in industry linkages during the 7. Brown MT, Herendeen RA (1996) Embodied energy analysis and EMERGY process of transformation of the economic structure since the 1990s in China analysis: a comparative view. Ecol. Econ. 19: 219–235 from the perspective of energy consumption. Economy and Management of 8. Bullard CW, Herendeen RA (1975) The energy cost of goods and services. Metallurgy 6: 12–14 (in Chinese). Energy Policy 3: 268–278. 31. Yabe N. (2004) An analysis of CO2 emissions of Japanese industries during the 9. Costanza R (1980) Embodied energy and economic valuation. Science 210: period between 1985 and 1995. Energ. Policy. 32(5): 595–610. 1219–1224 32. Chen GQ, Chen ZM. (2010) Carbon emissions and resources use by the Chinese 10. Bin S, Dowlatabadi H (2005) Consumer lifestyle approach to US energy use and economy in 2007: a 135-sector inventory and input-output embodiment. the related CO2 emissions. Energ. Policy 33: 197–208 Commun. Nonlinear Sci. and Numer. Simulat 15(11): 3647–732. 11. Liu Z, Geng Y, Lindner S, Zhao HY, FujitaT, Guan DB (2012) Embodied 33. Chen ZM, Chen GQ, Zhou JB, Jiang MM, Chen B (2010). Ecological input- energy use in China’s industrial sectors. Energ. Policy 49: 751–758 output modeling for embodied resources and emissions in the Chinese economy 12. Liang S, Wang C, Zhang TZ (2010) An improved input-output model for energy in 2005. Commun. Nonlinear Sci. and Numer. Simulat 15(7): 1942–65. analysis: a case study of Suzhou. Ecol. Econ. 69: 1805–1813 34. Zhang B, Chen GQ (2010) Methane emissions by the Chinese economy: 13. Alcantara V, Padilla E (2003) ‘‘Key’’ sectors in final energy consumption: an inventory and embodiment analysis. Energ. Policy 38(8): 4304–4316. input-output application to the Spanish case. Energ. Policy 31: 1673–1678. 35. Zhou SY, Chen H, Li SC (2010) Resource use and greenhouse gas emissions in 14. Chung WS, Tohno S, Shim S Y (2009) An estimation of energy and GHG the urban economy: ecological input-output modeling for Beijing, 2002. emission intensity caused by energy consumption in Korea: an energy IO Commun. Nonlinear Sci. and Numer. Simulat. 15(10): 3201–3231. approach. Appl. Energ. 86(10): 1902–1914. 36. Baynes T, Lenzen M, Steinberger J, Bai XM (2011) Comparison of household 15. Fan Y, Liang QM, Wei YM, Okada N (2007) A model for China’s energy consumption and regional production approaches to assess urban energy use and requirements and CO2 emissions analysis. Environ. Model & Software 22(3): implication for policy. Energ. Policy 39: 7298–7309. 378–93. 37. Lenzen MR, Wood R, Foran B. (2008) Direct versus Embodied Energy – The 16. Labandeira X, Labeaga JM. (2002) Estimation and control of Spanish energy- Need for Urban Lifestyle Transitions. Urban Energy Transition. In: Droege P related CO2 emissions: an input–output approach. Energ. Policy 30(7): 597–611. editor. Urban Energy Transition. Netherlands: Elsevier. pp. 91–120, 2008. 17. Lenzen M. (1998) Primary energy and greenhouse gases embodied in Australian 38. Rosado L, Ferra˜ o P (2009). Measuring the embodied energy in household goods: final consumption: an input-output analysis. Energ. Policy 26(6): 495–506. application to the Lisbon City. In: Havra ´ nek M. editor. ConAccount 2008: 18. Lenzen M, Dey C, Foran B (2004) Energy requirements of Sydney households. Urban Metabolism, Measuring the Ecological City. Prague: Charles University Ecol. Econ. 49(3): 375–399. Environment Center. pp. 159–181. 19. Leontief W. (1970) Environmental repercussions and the economic structure: an 39. Lindner S, Legault J, Guan D. (2012) Disaggregating input-output models with input-output approach. Rev. Econ. Stat. 52(3): 262–71. incomplete information. Economic Systems Research 24: 329–347. 20. Liu HT, Guo JE, Qian D, Xi YM (2009) Comprehensive evaluation of 40. Chen GQ, Guo S, Shao L, Li JS, Chen ZM (2013) Three-scale input–output household indirect energy consumption and impacts of alternative energy modeling for urban economy: Carbon emission by Beijing 2007. Commun policies in China by input-output analysis. Energ. Policy 37(8): 3194–3204. Nonlinear Sci Numer Simulat. 18: 2493–2506. 21. Liu HT, Xi YM, Guo JE, Li X (2010) Energy embodied in the international 41. Feng KS, Davis SJ, Sun LX, Li X, Guan DB, Liu WD, Liu Z, Hubacek K (2013) trade of China: an energy input-output analysis. Energ. Policy 38(8): 3957–3964. Outsourcing CO2 within China. PNAS, doi/10.1073/pnas.1219918110. 22. Machado G, Schaeffer R, Worrell E (2001) Energy and carbon embodied in the 42. Liang QM, Fan Y, Wei YM (2007) Multi-regional input–output model for international trade of Brazil: an input-output approach. Ecol. Econ. 39(3): 409– regional energy requirements and CO2 emissions in China. Energ. Policy 35: 424. 1685–1700. 23. Miller R, Blair P (1985) Input-Output Analysis: Foundations and Extensions. 43. Bullard CW, Penner PS, Pilati DA (1978) Net energy analysis-handbook for Englewood Cliffs: Prentice-Hall. combining process and input-output analysis. Resources and Energy 1: 267–313. 24. Park HC, Heo E (2007) The direct and indirect household energy requirements 44. Kaneko S, Nakayama H, Wu LB (2003) Comparative study on indirect energy in the Republic of Korea from 1980 to 2000: an input-output analysis. Energ. demand, supply and corresponding CO2 emissions of Asian Mega-cities. Policy 35(5): 2839–2851. Proceedings of International Workshop on Policy Integration Towards PLOS ONE | www.plosone.org 10 March 2014 | Volume 9 | Issue 3 | e89850 Urban Energy Consumption of Beijing Sustainable Urban Energy Use for Cities in Asia East West Center, Honolulu, 47. Beijing Statistics Bureau (1988, 1991, 1993, 1996, 1998, 2003, 2008) Beijing Hawaii. Statistical Yearbook. Beijing: China Statistics Press. 45. Rasmussen P N. (1956) Studies in Inter-Sectoral Relations. Amsterdam: North- 48. Peters GP, Weber CL, Guan DB, Hubacek K (2007). China’s growing CO2 Holland. pp. 58–72. emissions-a race between increasing consumption and efficiency gains. Environ. 46. Beijing Statistics Bureau (1990, 1991, 1995, 1996, 1999, 2001, 2004, 2006, 2009) Sci. & Technol. 41: 5939–5944. Beijing Input–Output Table 1987, 1990, 1992, 1995, 1997, 2000, 2002, 2005, 49. Feng YY, Chen SQ, Zhang LX (2013) System dynamics modeling for urban 2007. Beijing: China Statistics Press. energy consumption and CO2 emissions: A case study of Beijing, China. Ecol. Mode. 252, 44–52. PLOS ONE | www.plosone.org 11 March 2014 | Volume 9 | Issue 3 | e89850