Energy Policy 39 (2011) 2944–2956

Contents lists available at ScienceDirect

Energy Policy

0301-42

doi:10.1

n Corr

Shangh

E-m

htyin@s1 Te

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

Industrial relocation and energy consumption: Evidence from China

Xiaoli Zhao a,c,1, Haitao Yin b,c,n

a School of Economics and Business, North China Electric Power University, Beijing 102206, Chinab Antai College of Economics and Management, Shanghai Jiao Tong University, Shanghai 200052, Chinac Erb Institute for Global Sustainable Enterprise, University of Michigan, Ann Arbor, MI 48109, USA

a r t i c l e i n f o

Article history:

Received 6 May 2010

Accepted 1 March 2011

Keywords:

Energy consumption

Industrial relocation

Externality

15/$ - see front matter & 2011 Elsevier Ltd. A

016/j.enpol.2011.03.002

esponding author at: Antai College of Ec

ai Jiao Tong University, Shanghai 200052, Chi

ail addresses: email99zxl@vip.sina.com (X. Zh

jtu.edu.cn (H. Yin).

l.: þ86 13910778294.

a b s t r a c t

With economic development and the change of industrial structure, industrial relocation is an

inevitable trend. In the process of industrial relocation, environmental externality and social cost

could occur due to market failure and government failure. Little attention has been paid to this issue. In

this paper, we address it with a theoretical analysis and an empirical investigation on the relationship

between China’s industrial relocation in the early 1990s and energy consumption which is the primary

source of CO2 emission, an environmental externality that causes increasing concerns. The macro-

policy analysis suggests that there would be a positive link between China’s industrial relocation in the

early 1990s and energy saving (and environmental externalities reduction). Using fixed-effect regres-

sion model and simulation method, we provide an empirical support to this argument. In order to

further reduce environmental externalities and social cost in the process of industrial relocation, we

provide policy suggestions as follows: First, strengthen the evaluation of environmental benefits/costs;

Second, pay more attention to the coordinated social-economic development; Third, avoid long-lived

investment in high-carbon infrastructure in areas with industries moved in; Fourth, address employ-

ment issue in the areas with industries moved out.

& 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Energy consumption in China has been growing very rapidlysince 1953, especially after 1978 when economic reform waslaunched (see Fig. 1). Over the period of 2000–2008, China’senergy consumption has increased at an average rate of 9.1% perannum. In 2009, China’s energy consumption accounted for one-quarter of the world’s total. The increase of energy consumptionposes two challenges: First, China heavily relies on fossil-basedenergy which takes up more than 90% of its total energyconsumption (DRCSC, 2009). As a result, China has become thelargest CO2 emitter, contributing one-third of the world’s totalCO2 emission. As climate change becomes an increasingly press-ing concern, China now is facing mounting international anddomestic pressures to cut back CO2 emission. Second, the pres-sure for China to reduce energy consumption also comes from theneed to secure energy independence. Fig. 2 displays China’s oilimport and export during the period from 1980 to 2007. It is clearthat China’s net oil import has increased quickly since the year

ll rights reserved.

onomics and Management,

na. Tel.: þ86 21 52301242.

ao),

1993. China has now become the world’s second largest con-sumer of oil, although its per capita oil endowment is only 6.1% ofthe world’s average (Jiang, 2008). According to a recent study,China’s dependence on imported oil has been close to 50% (Zhong,2009). International Energy Agency (IEA) predicts that it will beclose to 77% in 2020 (China’s Energy Outlook, 2004). Hence,energy saving has become an important and urgent issue thatChina must address in its future economic development.

Scholars have explored factors that drive energy consumptionand ways to reduce it. Economic growth is considered as one ofthe most important factors affecting energy consumption. A largebody of literature (Zhao and Yang, 2009; Akinlo, 2008; Shiu andLam, 2004; Ockwell, 2008; Karanfil, 2008; Narayan and Smyth,2008; Yuan et al., 2008; Lee and Chang, 2008) shows thatincreased energy consumption is a natural result of economicgrowth. As Stern (2006, 2010) argued, the ignorance of environ-mental externalities during the process of economic developmentis recognized as one of the significant factors to aggravateenvironmental degradation. The Chinese government has set itsGDP growth goal at a rate of 7.2% per annum between 2000 and2020 (Yang, 2008). The goal is largely based on the needs ofmaintaining employment and social stability in China. As adeveloping country, it is beyond doubt that China has an urgenttask to develop its economy at a relative higher rate. As Vennemoet al. (2009) argued, developing countries often find it difficult to

3000

2500

2000

1500

1000

500

0

mill

ion

tce

1952

1960

1965

1975

1980

1986

1988

1990

1992

1994

1996

1998

2000

2004

2006

2002

Fig. 1. Total energy consumption in China (1952–2007).

Source: China Statistical Yearbooks (2008).

Oil

Exp

ort a

nd Im

port

(Uni

tM

illio

n To

ns)

250

200

150

50

0

100

19801982

19841986

19881990

19921994

19961998

20002002

20042006

Year

Export Import

Fig. 2. Oil import and export in China.

Source: China Energy Statistical Yearbooks.

X. Zhao, H. Yin / Energy Policy 39 (2011) 2944–2956 2945

commit to an emission reduction since it might interfere with themuch-needed development of their economies. However, on theother hand, if China excessively strengthens its economic devel-opment and fails to pay enough attention to environmentalprotection and CO2 emission, the resultant economic and socialcost would be huge not only for China, but also for the entireworld in the future. As Stern (2006) noted, the global lossescaused by climate change have increased dramatically in therecent past, and the climate change is noted as the greatest failureever known. China could not continue the economic developmentmode that caused great adverse impact on the environment asthat in the United States and the United Kingdom. In the future,China needs to better integrate the consideration of environmen-tal externalities and social cost into its economic developmentstrategy.

The second factor that has been found to have an importantimpact on energy consumption is the adjustment of industrialstructure. The close relationship between the change of industrialstructure and energy consumption has been well recognized(Zhao and Yang, 2009; Liu et al., 1992; Ang, 1995; Vicent andRosa, 2004; Zhou and Li, 2006; Kambara, 1992). In the ‘‘Twelfth-Five Year Plan’’, Chinese government emphasized again to saveenergy consumption and reduce CO2 emission via further adjust-ing industrial structure. In general, industrial relocation wouldhappen along with the adjustment of industrial structure. Since1978, when the economic reform was launched, China hasexperienced two major industrial relocations. The first took placein the early 1990s, major industries were transferred to theeastern coastal areas from Northeastern areas, Western areas,and some Middle areas. Fujita et al. (2003) and Huang and Li(2006) illustrate that one characteristic of China’s industrialdevelopment in the last two decades was relocation and geogra-phical specialization. Jin et al. (2006) noted that the share ofeastern coastal provinces in the nation’s total industrial produc-tion increased quickly, while the share of western and

northeastern provinces decreased as a whole. Industrial produc-tion also retreats from Beijing, Shanghai, and Tianjin.

The second industrial relocation happened in the 20th century,and industries (mainly labor intensive and resource intensive ones)were transferred to Middle areas and Western areas from theeastern coastal areas (Jiang, 2009; Wu, 2009). Industrial relocationwould have great impact on the involved areas on a series ofeconomic and social issues, such as economic growth, employment,energy consumption, and environment. In this paper, we will focuson the impact of industrial relocation on energy consumption, andenvironmental externalities. We first provide a theoretical analysison the relationship between industrial relocation and energy con-sumption and then offer an empirical investigation of the relation-ship. To our best knowledge, little attention has been paid to therelationship between industrial relocation and energy consumption(environmental externalities). This study should prove useful forpolicy makers and scholars that concern about the externalities ofindustrial relocation.

The rest of the paper proceeds as follows. Section 2 analyzesthe externalities and social cost of industrial relocation. Section 3explores how public policies could have an impact on energysaving and externalities in the process of industrial relocation.Policy lessons regarding what policy factors would help Chinaachieve energy saving are presented. Section 4 introduces theregression model and simulation method we use to empiricallystudy the impact of China’s industrial relocation on energyconsumption in the early 1990s. The factors that contribute tothe positive correlation between China’s industrial relocation inthe early 1990s and energy consumption are discussed in Section5. Section 6 concludes the paper with a summary of main findingand a discussion of policy implications.

2. Externalities and social cost of industrial relocation

Industrial relocation is driven by both administrative powerand market power in China. The administration-oriented indus-trial relocation aims to improve the environment quality in keycities. For example, the Capital Iron and Steel Corporation , whichused to the worst source of pollution in Beijing, has moved itsprimary production capacity to Hebei before the 2008 OlympicGame, and the remaining manufacturing capacity will be trans-ferred to Hebei before the end of 2010.

The majority of industrial relocation in China is driven bymarket power. The market-oriented industrial relocation can bedivided further into two types: the first is to pursue higheroperation efficiency, and transferred to areas with cheaper pro-duction factors or lower transaction costs. For example, recently,some industries moved to the Western and Middle areas from theeastern coastal areas in order to get cheaper labor and energysources, while in the early 1990s, some industries moved to theeastern coastal areas which have more convenient transportation,broader market, and lower transaction cost. The second type ofmarket-oriented industrial relocation is to escape the strictenvironmental regulation in key and large cities. In general, asenvironmental regulations are more strictly enforced in wealthierurban areas, industries are moving to less affluent cities or ruralsites, taking their pollution with them (Holdaway, 2010). Thereare numerous reports about villages with a high prevalence ofcancer (Lora-Wainwright, 2010) and a variety of other environ-ment-related risks (Fang and Gerry, 2010).

Both administration-oriented and market-oriented industrialrelocation would cause environmental externalities due to ‘‘gov-ernment failure’’ and ‘‘market failure’’. An externality is anunpriced benefit or cost directly bestowed or imposed upon oneagent by the actions of another agent (Soderholm and Sundqvist,

Table 1The heavy industries relocation in early 1990s in China.

Data source: China Industry Economy Statistic Yearbooks (1985–2007); Statistical Data Collection of New China for 55 years (1949–2004).

Areas with industries moved out Areas with industries moved in

Raw chemical materials and chemical products Liaoning Shandong, Jiangsu

Smelting and pressing of ferrous metals (steel, etc.) Liaoning, Heilongjiang, Beijing, Shanghai, Xinjiang Jiangsu, Zhejiang, Fujian, Hebei

Nonmetal mineral products (cement, etc.) Heilongjiang, Guangong, Xinjiang Shandong, Fujiang

Coal mining Heilongjiang, Xinjiang Shandong, Guizhou

Electricity, gas and hot water Shanghai, Heilongjiang, Liaoning, Xinjiang Zhejiang, Shandong, Fujian, Guizhou, Yunnan

Table 2The light industries relocation in early 1990s in China.

Data source: China Industry Economy Statistic Yearbooks (1985–2007); Statistical Data Collection of New China for 55 years (1949–2004).

Areas with industries moved out Areas with industries moved in

Papermaking and paper products Liaoning, Beijing, Tianjin, Shanghai, Xinjiang Shandong, Jiangsu, Henan

Rubber products Liaoning, Heilongjiang, Tianjin Shandong, Anhui, Guizhou

Production and supply of tap water Heilongjiang Yunnan

Chemical fiber Heilongjiang, Beijing, Tianjin, Shanghai, Guangdong, Liaoning Zhejiang, Fujian, Shandong, Jiangsu

Pro

porti

on o

f add

ed v

alue

of s

ecto

r iin

eac

h pr

ovin

ce to

that

of t

he w

hole

coun

try (%

)

18

16

14

12

10

8

6

2

0

4

early1990s

2004 early1990s

2004 early1990s

2004 early1990s

2004 early1990s

2004

CementSteelElectricityCoalChemicalfibre

Shanghai

LiaoningXinjiang

GuangdongBejingTianjin

Heilongjiang

Fig. 3. Provinces with five most energy-intensive sectors moved out.

Source: Author’s calculation based on Statistical Data Collection of New China for 55 years (1949–2004), edited by State Statistical Bureau.

X. Zhao, H. Yin / Energy Policy 39 (2011) 2944–29562946

2003). The major characteristic of externality is the market pricedoes not reflect the true cost of some products since it fails tocapture the value of many social and environmental costsassociated with the transaction.

The externality of industrial relocation also has its owncharacteristics. First, industrial relocation will cause differentimpact on the environment of the areas with industries movedin and the areas with industries moved out. If industries that aremoved are energy intensive ones, such as what happened in Chinain the early 1990s (Tables 1 and 2, Figs. 3–6), the environmentalexternality to the areas with industries moved in would becomemore severe. Especially, when the industrial relocation belongs tothe replicative alternatives with energy-intensive industries mov-ing,2 the environmental externalities should be given moreattention. Second, industrial relocation induced by escape ofenvironmental regulation would yield worse environmentalexternalities since companies that move do not intend to improve

2 Biggiero (2006) argues that relocation strategies are divided into selective

and replicative alternatives, depending on the ability to preserve large kernels.

technologies and energy efficiency in the relocation, but aim totake advantage of the lax regulations in the areas that they moveinto. Moreover, the new plants built in the areas with industriesmoved in may have a long lifespan, during which it would cost alot more to improve energy efficiency of these built plants thanbuilding new plants from the scratch. As Stern (2006) noted, ‘‘it ismuch cheaper to build a new piece of capital equipment usinglow-emission technology than to retro-fit dirty capital stock.’’Hence, enterprises would have less enthusiasm to improve theexisting equipment in these areas later. In the meantime, Table 3shows that once investment is made, it can last for decades.A high-carbon or low-efficiency piece of capital stock will tend tolock the economy into a high emissions model (Stern, 2006).Hence, the areas that host industries that move in pursuit of laxenvironment regulations would face more severe environmentalexternalities in the future. Third, the environmental externalitiesin the areas with industries moved in and moved out should beconsidered as a whole. In general, the areas with industriesmoved in would face negative environmental externalities, andthe areas with industries moved out would face positive environ-mental externalities. Hence, we need to assess the combined

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0

9.0

%

LiaoningBeijingTianjinShanghaiXinjiang

Fig. 5. Moving out of Paper and Pulp Industry. The Y-axis refers to the proportion of added value of Paper and Pulp Industry in each province to that of the whole country.

The data in 1986, 1991, 1995, 1996, 1998, 2003, and 2004 is absent.

Data source: China Industry Economy Statistic Yearbooks (1985–2007).

0.00

5.00

10.00

15.00

20.00

25.00

%

ShandongJiangsuHenan

Fig. 6. Moving in of Paper and Pulp Industry. The Y-axis refers to the proportion of added value of Paper and Pulp Industry in each province to that of the whole country.

The data in 1986, 1991, 1995, 1996, 1998, 2003, and 2004 is absent.

Data source: China Industry Economy Statistic Yearbooks (1985–2007).

Pro

porti

on o

f add

ed v

alue

of s

ecto

r iin

eac

h pr

ovin

ce to

that

of t

he w

hole

coun

try (%

)

50454035302520

1050

15

early1990s

2004 early1990s

2004 early1990s

2004 early1990s

2004 early1990s

2004

CementSteelElectricityCoalChemicalfibre

ShandongJiangsuZhejiangFujianHebeiYunnanGuizhou

Fig. 4. Provinces with five energy intensive sectors moved in.

Source: Author’s calculation based on Statistical Data Collection of New China for 55 years (1949–2004), edited by State Statistical Bureau.

X. Zhao, H. Yin / Energy Policy 39 (2011) 2944–2956 2947

environmental externalities in assessment of the environmentalimpact of industrial relocation.

In addition to environmental externalities, other social costswould also deserve attention in the process of industrial reloca-tion. Social cost is the loss of social welfare. Social welfaredepends on the welfare of each individual in the community.When goods and services are defined in a broad way, they caninclude, for example, education, health and goods appearing atdifferent dates and in different circumstances (Stern, 2006). InChina, state-owned industries not only provide jobs but also theiremployees’ social welfare. It is very common for large, state-owned enterprises to have hospitals, schools, and residential

properties on site for their workers (Bai, 2002). Hence, the socialcosts in the areas with industries moved out would be reductionof education resources, increase of unemployment, and otherkinds of welfare losses as well as the negative impact on theeconomic development of the origin areas. And the social costs inthe areas with industries moved in would be the health impactdue to increase of pollution.

Stern (2010) illustrates further that the social marginal costwill depend on: (1) assumed future growth paths of the economyand emissions, both of which are highly endogenous in the sensethat they are strongly influenced by current and future decisionsand cannot be seen as an ‘‘external’’ input; (2) distributional

Table 3Lifespan for different kinds of investments (Years).

Data source: World Business Council for Sustainable Development (2004) and IPCC

(1999).

Building Coal

station

Nuclear

station

Gas

turbine

Aircraft Motor

vehicle

Lifespan 45 45 30–60 25 25–35 12–20

X. Zhao, H. Yin / Energy Policy 39 (2011) 2944–29562948

values both within and across generations; and (3) assumptionson the nature and magnitude of, and presumed attitudes towards,risk and uncertainty. According to these arguments, the relocationof energy intensive industries from China’s Eastern coastal areasto Western areas may face huge social marginal cost. China’sWestern areas are the ecological barriers of the Eastern, and thecradle of many major rivers, such as the Yangtze River and theYellow River. Most of China’s glaciers are in the Western areas.Hence, the undermining of the Western ecological environmentwould cause destructive impact both on the Western and Easternsustainable development (Wu, 2009), and it would have greatpotential social welfare loss to the future generations.

3. Energy saving and externalities in China’s industrialrelocation in the early 1990s: a policy perspective

In this section, we place China’s industrial relocation in itsmacro-policy changes in order to gain insights on what policyfactors would help China achieve energy saving and thereforereduce environmental externalities in the process of industrialrelocation.

3.1. Market-oriented economic reform and energy saving in the

process of industrial relocation

In the last 30 years, China has made a transition from plannedeconomy to market economy through a series of policy change.This process goes through three distinguishable yet intercon-nected stages. The first stage is from 1978 to 1992.3 The mostsignificant change in this stage is that collectively owned enter-prises and privately owned enterprises gain legal status and state-owned enterprises are granted more autonomy. A highly centra-lized planned economy was broken. In 1992, the second stage ofthe reform starts with the 14th National Congress of China’sCommunist Party (NCCCP) (Gao, 2008). During this stage, a freemarket for production factors was gradually established, as aresult of which resources are allocated following market signalsinstead of administrative plan. Since 2002, China has welcomedthe third stage of its economic reform.4 In this stage, theconstraint on private and foreign capital is further reduced. Insome traditionally publicly owned sectors, for example, watersupply, electricity, oil, and natural gas, market mechanism isgradually introduced. The representative policies include the‘‘Scheme of Electricity Price Reform (2003, [62])’’ and ‘‘Measuresfor Implementation of Electricity Price Reform (NDRC, 2005[514])’’.

In sum, China’s economic reform has been a market-orientedone. Relating to this study, this reform has created a free marketfor production factors (including energy) so that they can flowfrom areas with low efficiency to those with high efficiency in theprocess of industrial relocation. This is very different from a

3 The milestone is the Third Plenary Session of the China’s Communist Party

11th Central Committee in 1978.4 The milestones are (1) China became a member of WTO at the end of 2001,

and (2) the 16th NCCCP was held in 2002.

relocation that is strongly influenced by administrative plan.It would promote energy saving as manufacturing sectors transferfrom provinces with low energy efficiency to provinces with highenergy efficiency. Hence, China’s market-oriented economicreform since 1978 might lead to a positive linkage betweenenergy saving (and environmental externalities reduction) andChina’s industrial relocation in the early 1990s.

3.2. Gradient development strategy and energy saving in the process

of industrial relocation

Xiaoping Deng, the chief designer of China’s economic reform,set up a ‘‘gradient development strategy’’, in which some areasshould get developed first, and then these areas help othersachieve economic prosperity. Eastern coastal provinces werestrategically selected to be these that should be developed first.In 1980, four Special Economic Zones (SEZ), Shenzhen, Zhuhai,Shantou, and Xiamen, were set up in eastern coastal areas. In1984, 14 eastern coastal cities were further opened up to theoutside. Significant preferential policies for eastern coastal pro-vinces are put forward. First is preferential tax policy. Forexample, enterprises within ESZ enjoyed tax-exemption withinthe first 2 years and half tax-exemption in the following 3 years’’(Fang, 1994). Second, preferential investment policy. For example,the right to approve large-scale project was transferred fromcentral government to the local government of ESZ (Chen, 1998).As a result, transaction cost could be saved greatly and efficiencywas improved. Third, preferential policy for foreign trade. Forexample, the proportion of foreign exchange retention for ESZwas 100%, for Guangdong and Fujian it was 30%, much higherthan that of the interior provinces (Chen, 1998). With the abovepreferential policies, the eastern coastal provinces have attractedmuch more investment than other provinces. Over the period of1990–2004, the eastern coastal provinces had attracted 86.16% ofall FDI in China. Table 4 shows the difference in investmentbetween eastern coastal provinces and other provinces.

With FDI flowing into China, advanced machinery equipment,advanced technologies and management are also introduced intoeastern coastal areas which hosted most of the new addedindustrial production capacity (Xu and Tan, 2003; Chou and Wu,2009). What’s more, eastern coastal provinces are more populatedthan others. With new development of these areas, labor alsopoured into eastern coastal provinces. According to Wu and Ge(1996), 415 persons/km2 lived in eastern coastal provinces, com-paring with 123 persons/km2 in China.

In sum, the implementation of preferential policies in China’seastern coastal provinces promoted the rapid development ofthese areas. As a result, these provinces have attracted a wealth ofproduction factors: capital, technology and labor. Such concen-tration makes it possible to substitute energy with other produc-tion factors in eastern coastal provinces. The study on the inter-substitution between energy and other production factors inChina shows that there exists strong substitution between energyand capital (Lin, 2007). Hence, we argue that because of theconcentration of production factors caused by preferential poli-cies in eastern coastal provinces and because industries movedinto these areas, we might observe a positive linkage betweenenergy saving (and environmental externalities reduction) andChina’s industrial relocation in the early 1990s.

3.3. Policy that encourages energy efficiency and energy saving in

the process of industrial relocation

Since the 1990s China has put forward many market-orientedpolicies to encourage energy efficiency improvement. Major

Table 4Share of fixed asset investment in different areas of China.

Data Source: Li (1998)

Year Eastern coastal

areas

Middle

areas

Western

areas

Other

areas

1991 57.04 24.21 14.68 4.071992 59.72 22.91 13.83 3.541993 62.87 21.08 12.40 3.651994 63.35 20.85 12.04 3.761995 63.80 21.03 11.95 3.221996 62.21 22.16 12.54 3.09

X. Zhao, H. Yin / Energy Policy 39 (2011) 2944–2956 2949

measures can be categorized into four groups. First, reform ofenergy prices. Coal price was deregulated in 1993, followed by oilprice reform in 1998. Electricity price, which has the mostsignificant impact on industrial production, has experienced aseries of market-oriented reforms too, including the reform ofcapacity price, volume price and Peak-Valley Time of Use (TOU)prices.5 These reforms aim to promote energy saving throughmarket-based price mechanisms. Second, technological innova-tion for energy efficiency improvement is encouraged. Since the1990s, China has paid much attention to developing technologiesfor energy efficiency improvement and promoting the use ofenergy saving products. Third, legislation in energy saving. Energy

Saving Policy was published in 1998, and other complementarypolicies have been published from then on, including Energy

Saving Measures for Selected Organization (1999), Measures on

Electricity Saving (2000), and Measures on Certification of Energy

Saving Products (1999). Finally, finance and tax measures topromote energy saving. For example, according to the Notice ofTax (2000 [026]), the tax for purchasing energy saving products isreduced by half and exempted for fixed capital investment incombined heat and power generation projects and so on.

The aforementioned policies have effectively promoted thedevelopment and adoption of energy efficiency improvementtechnologies (Zhao et al., 2010). It is worth noting that thisprogress happened simultaneously with industrial relocation.When new plants are built up, it is natural for them, under thesepolicy incentives, to adopt more energy-efficient technologies andachieve energy saving. Because of this, we might observe apositive linkage between energy saving (and environmentalexternalities reduction) and China’s industrial relocation in theearly 1990s.

3.4. Environmental protection policy and externalities reduction in

the process of industrial relocation

Facing the mounting pressure of environmental protectionfrom both domestic and overseas, the Chinese government hasgiven more and more attention on the externalities reduction inthe process of industrial relocation. For example, in the documentof ‘‘the Circular of the State Council on Several Policy Measures

Concerning the Development of the Western Regions’’ (GuoFa [2000]No. 33), it is emphasized that the construction of ecologicalenvironment is one of the key tasks that will be given priority.At the same time, the financial transfer from central to westernlocal governments will focus on the protection of natural forest,and the pilot project of returning farmland to forest and grass-land, etc. In 2001, the State Environmental Protection Agencyissued ‘‘Several Opinions on the Management of Environmental

Impact of Construction Project in Western Development’’ (Huan Fa[2001] No. 4), and emphasized the prevention of pollutant

5 For a detailed discussion, please see Zhao et al. (2010).

intensive sectors and the outdated technologies and equipmentfrom transferring to the Western areas.

In order to reduce the negative impact on the environment, theChinese government required that new plants that are built up inthe process of industrial relocation must improve technology andenergy efficiency. For example, during the transfer of the CapitalIron and Steel Company (CISC), which used to be the worstpolluter in Beijing, from Beijing to Hebei, new equipments andadvanced technologies are adopted, and energy consumption isreduce by 30%.6 At the same time, the Chinese government putforward ‘‘Energy Saving Plans in the top-1000 largest industrial

enterprises’’ in April 2006. The ‘‘top-1000 largest industrial enter-prises’’ belong to energy intensive sectors such as steel, non-ferrous metal, coal, electric power, petroleum and petroleumindustry, chemistry, building materials, textile, and paper, andeach of them consumes at least 180,000 tons of coal equivalent(tce) in 2004. The energy consumption in the top-1000 largestindustrial enterprises accounted for 33% to the total energyconsumption, and 47% to the total industrial energy consumption(Price et al., 2010). These environment protection measuresundoubtedly would help reduce environmental externality.Because of it, we might observe a positive linkage between energysaving (and environmental externalities reduction) and China’sindustrial relocation in the early 1990s.

4. Empirical investigation of the industrial relocation impacton energy consumption

The previous policy discussions suggest that we might observe apositive linkage between energy saving (and environmental extern-alities reduction) and China’s industrial relocation. In this section,we will provide an empirical investigation of such impact during theprocess of China’s industrial relocation in the early 1990s.

4.1. The measurement of industrial relocation

To make ‘‘industrial relocation’’ more precise, it is useful tointroduce a few notations. We first define the years from 1985 to1989 as a baseline period since industrial location remainsrelatively stable during these years and industrial relocationbecomes intense after that. Let AVMi0 denote the added value ofmanufacturing sectors in province i in baseline period, and AVMi1

the added value of manufacturing sectors in province i for yearsafter the baseline period. We further define

Pi0 ¼AVMi0Pn

i ¼ 1 AVMi0

ð1Þ

Pi1 ¼AVMi1Pn

i ¼ 1 AVMi1

ð2Þ

Here, Pi0 is defined as the proportion of added value ofmanufacturing sectors in province i to that of the whole countryduring the baseline period and Pi1 for the years after the baselineperiod. A relationship of Pi1oPi0 suggests that the province i’smanufacturing sectors shrink or grow slower than its peers. Inthis sense, we say that manufacturing sectors are moving out ofprovince i. On the other hand, if Pi14Pi0, it suggests thatmanufacturing sectors have moved into province i.

Based on this definition, we found that after the early 1990s, theprovinces with manufacturing sectors moved out are as follows:Liaoning, Heilongjiang, Beijing, Tianjin, Shanghai, Guangdong, and

6 Data source: The relocation of CISC is not the transfer of pollution and the

previous melting will be stopped by the end of 2010. http://www.southcn.com/news/

china/zgkx/200506100468.htm, 2005, 6, 10.

Fig. 7. Change of economic structure in Beijing.

Source: China Statistic Yearbooks.

Fig. 8. Change of economic structure in Tianjin.

Source: China Statistic Yearbooks.

Fig. 9. Change of economic structure in Shanghai.

Source: China Statistic Yearbooks.

Fig. 10. Change of economic structure in Guangdong. Source: China Statistic

Yearbooks.

X. Zhao, H. Yin / Energy Policy 39 (2011) 2944–29562950

Xinjiang. The provinces with manufacturing sectors moved ininclude Shandong, Jiangsu, Zhejiang, Fujian, Hebei, Sanxi, Henan,Jiangxi, Anhui, Yunnan, Guizhou, and Guangxi.

Industrial relocation in China in the early 1990s takes on twomajor trends. First, manufacturing sectors moved from provincesthat historically hosted heavy industries, such as Heilongjiang andLiaoning, to eastern coastal, Southern and several Middle pro-vinces. Second, industries moved from the relatively developedprovinces or municipality, such as Beijing, Shanghai, Tianjin andGuangdong, to eastern coastal provinces, Southern and severalMiddle provinces.

4.2. The drivers behind China’s industrial relocation in the early

1990s

The trend of industrial relocation discussed above is mainlydriven by three forces. First, since the early 1990s, China’sindustrial strategy was shifted from heavy manufacturing indus-tries oriented to light manufacturing industries oriented. Duringthis period, light manufacturing sectors developed more quicklythan heavy manufacturing sectors. Eastern Coastal provinces havecomparative advantages in labor, technology, capital, transporta-tion, and market size that are critical for the development of lightmanufacturing sectors. Moreover, Huang and Huang (2008)argued that Eastern coastal provinces in China have more advan-tages and experiences in trading with foreign markets because oftheir geographic location and preferential policy. Wen (2004)presents further that the coastal regions have attracted foreigninvestment not only because of the preferential policies, but alsodue to their geographical proximity to big export ports. Thisadvantage becomes more significant since the early 1990s withthe accelerated steps for opening up to the outside world.

Second, southern and several Middle provinces that haveindustries moved in are relatively abundant in natural resources.Taking coal as an example, China’s coal reserve mainly lies alongthe west line of DaXingAnLing-TaiHang mountain-XueFengmountain. The coal reserve in eleven provinces locating in thewest part of this west line, such as Neimenggu, Shanxi, Sichuan,Guizhou, etc., account for almost 91.83% of the total reserve inChina (CES, 2007). The provinces that have energy intensivesectors moved in, such as Shanxi, Guizhou, Anhui, and Yunnan,are abundant in coal. Other provinces, such as Yunnan, Guizhou,and Guangxi are abundant in hydro-electric power. This type ofrelocation of manufacturing sectors is driven by the desire to cutdown energy transportation cost.

Third, the move-out of industrial sectors from Beijing, Tianjin,Shanghai, and Guangdong is mainly driven by the adjustment ofeconomic structure in relatively developed provinces. There existsclose relationship between economic growth and economic struc-ture change (Medlock and Soligo, 2001). When economic develop-ment reaches a certain stage, the economic structure tends to shifttowards tertiary sector. In the aforementioned provinces, where theeconomy developed more quickly than others, industrial sectorsretreats and tertiary sectors replace them as the major engine foreconomic prosperity. Figs. 7–10 show the adjustment of economicstructure in the aforementioned four provinces. It is evident that thetertiary sector in these provinces increased quickly, especially inBeijing, Tianjin and Shanghai, where the contribution of the tertiaryproduction to their GDP already outweighed the secondary (includ-ing industrial) sectors.

4.3. Method and data

Regressions and simulations are used to examine the impactof industrial relocation on total energy consumption. We usefixed-effect regressions to investigate the relationship between

11 A large-scale industrial structural adjustment was launched during the 9th

5-year period (1996–2000). The aim was to change the mode of economic growth

from ‘‘high energy consumption – high pollution discharge – high growth’’ to ‘‘low

X. Zhao, H. Yin / Energy Policy 39 (2011) 2944–2956 2951

the development of industrial sector and energy consumption,which is described by

Eit ¼ aþb0lagEitþb1 lnTþb2pitþb3zþb4pitzþeit ð3Þ

where Eit denotes the energy consumption in provincial entity7 i

and year t. In order to capture the general increase trend in energyconsumption, we include one-period lagged dependent variablesin our specification. The lags proxy for general increase rate ofenergy consumption by allowing for lagged energy consumptionto influence current energy consumption. Logarithmic time trend,ln T, is included to capture the impact of technological change onenergy consumption. Logarithmic time trend arises naturallyfrom a translog functional form and have frequently been usedto proxy for exogenous technological change (Calmfors andForslund, 1991; Auffhammer and Carson, 2008). A logarithmictime trend suggests that the impact of technological change onenergy consumption decreases over time. We include provincefixed effect in the model to control any factors that vary acrossprovince but stay constant over time.

The variables that are of our particular interest are added valuefrom manufacturing sector, which is pit in Eq. (3). We furtherallow the coefficient of added value from manufacturing sectorsto vary across provinces by including interactions betweenindustrial added value and area dummies (z in Eq. (3)), whichallows for differential impact of the growth of manufacturingsector. The differential impact might be due to differences inresource endowment and/or energy efficiency across provinces.The coefficient for manufacturing sector’ added value describesthe difference in energy consumption that a larger manufacturingsector can make in the ‘‘reference province’’8 (we refer it asreference difference hereafter). The coefficient for the intersectionterm between added value of manufacturing sector and areadummy measures how the difference in energy consumption thata larger manufacturing sector can make in other provincesdeviates from the reference difference (Irwin and McClelland,2001). A negative coefficient, for example on the interaction termbetween Guangdong province and added value, would indicatethat the difference in energy consumption that the same amountof production growth from manufacturing sector makes is smallerin Guangdong province than that in the reference province, whichis most likely caused by reasons such as regional difference inenergy efficiency or resource endowment. eit captures otherfactors that affect the total energy consumption but do notcorrelate with explanatory variables in the model.

The following strategy is used to identify the impact ofindustrial relocation on energy consumption. Using the resultsfrom the regression analysis, we construct a counterfactualestimation for what the energy consumption would be if indus-trial location had not been changed after 1990,9 that is, if variouseconomic sectors had been growing proportionately across dif-ferent provinces. Then the predicted energy consumption withreal industrial relocation and this counterfactual estimation ofenergy consumption are compared.10

7 Beijing, Shanghai, and Tianjin are provincial-level municipalities; Guangxi,

Inner Mongolia, Ningxia, Tibet, and Xinjiang are autonomous regions. Chonqing

was elevated to the level of a provincial-level municipality in 1997, but we still

count it as part of Sichuan. We refer to provinces and the entities mentioned in

this footnote as provinces.8 It’s Jiangxi province in the paper. The selection of reference province is

arbitrary and will not have an impact on the results.9 We use 1985–1989 as a benchmark. The average share of manufacturing

production to the GDP over these 5 years is calculated and assumed to stay

unchanged in the years after 1990 in the counterfactual scenario.10 Here we compared the counterfactual estimation of energy consumption

with the predicted energy consumption instead of the real energy consumption in

order to reduce the impact from the inaccuracy of the prediction model.

The data on the added value of manufacturing sector iscollected from the Database of China Center for EconomicResearch (CCER). The energy consumption data is collected fromChina Energy Statistic Yearbooks (CESYs): 1986–2007. The firstCESY was published in year 1986, which means that the data onenergy consumption at provincial level before 1985 is not pub-licly available. Hence, our data ranges from the year of 1985–2006. Manufacturing sector production is converted to the valueof 1985 based on the index of manufacturing sector production.This index is collected from China Statistics Yearbooks (CSYs):1986–2007.

There are 31 provincial entities in China. Chongqing munici-pality was separated from Sichuan in 1995, for the sake ofconsistency, we combine Chongqing with Sichuan. We droppedHainan and Tibet because the energy consumption data in thesetwo regions are not released officially for several years. Thisshould not have a significant impact on final results due tominimal energy consumption share of Hainan and Xizang in thewhole country. As a result, the number of provincial entities thatare included in the final analysis is 28.

4.4. The regression result

Table 5 reported the coefficient estimation and T-value. Thecoefficient for ln T is negative and statistically significant, reflect-ing the fact that China’s energy intensity has largely maintained adeclining trend in the last three decades, which, accordingto Garbaccio et al. (1999), is mainly due to technological changes.

Table 5 shows that the coefficient on added value is positiveand statistically significant, which suggests that in the referenceprovince, the growth of industrial sector leads to a significantincrease in energy consumption. The linear combination of thecoefficients for industrial added value and the interaction termbetween industrial added value and province dummy is some-times positive, suggesting that the growth of industrial produc-tion leads to significant increase of energy consumption in someprovinces; and sometimes not statistically different from zero,suggesting that the growth of industrial production does notresult in significant increase in energy consumption in others.There are two possible reasons for the latter phenomenon. First,as discussed in Section 4.2, since the early 1990s, China’sindustrial strategy had shifted from heavy manufacturing indus-try oriented to light manufacturing industry oriented. Lightmanufacturing industries do not require as much energy as heavyones. Second, the Chinese government encourages replacingenergy-intensive equipment and production process with moreefficient ones.11 Therefore, as time goes, the same amount ofmanufacturing output does not require as much energy as itused to.

energy consumption – low pollution discharge – high growth’’. The industrial

structural change policies continued in the 10th 5-year period (2001–2005). The

themes at this stage are to phase out obsolete technologies and associated

production capacity, and to optimize product structure to save energy and reduce

pollution. Furthermore, macro-policies of industrial structural change has been

supplemented by sector-specific environmental friendly development plans as

well as implementation measures and schemes in price, tax, investment, technol-

ogy, environmental protection, and government administration. Some measures

and schemes include: (1) Remove preferential power price and increase energy

prices for energy intensive industries; (2) Reduce or remove tax returns at the

export stage for some energy intensive and pollution intensive industries, and

strengthen export supervision; (3) Phase out obsolete technologies and encourage

investment in technologies with higher energy efficiency and lower pollution

discharge by publishing national catalogs of preferred technologies and invest-

ments and providing associated policy support.

Table 5Regression result.

Serial numberof provinces Coefficient T-value

lagEC 1.15 (47.72)nn

1 AVindAdj Anhui (reference

province)7.32 (2.86)nn

2 AVindAdjnBeijing �5.36 (1.04)3 AVindAdjnFujian �2.57 (0.69)4 AVindAdjnGansu 5.43 (0.79)5 AVindAdjnGuangdong �5.30 (1.24)6 AVindAdjnGuangxi 4.28 (0.72)7 AVindAdjnGuizhou �5.45 (0.16)8 AVindAdjnHebei �3.65 (0.39)9 AVindAdjnHelongjiang �3.71 (0.23)

10 AVindAdjnHenan 8.86 (1.24)11 AVindAdjnHubei �1.90 (0.20)12 AVindAdjnHunan 9.31 (1.07)13 AVindAdjnJiangsu �2.46 (0.44)14 AVindAdjnJiangxi �8.85 (2.19)n

15 AVindAdjnJilin 10.02 (1.22)16 AVindAdjnLiaoning 33.33 (1.86)17 AVindAdjnNeimenggu 39.92 (4.38)nn

18 AVindAdjnNingxia 105.41 (2.89)nn

19 AVindAdjnQinghai 39.32 (2.30)n

20 AVindAdjnShaanxi 3.53 (0.75)21 AVindAdjnShandong 6.16 (0.55)22 AVindAdjnShanghai �9.29 (2.15)n

23 AVindAdjnShanxi �0.42 (0.02)24 AVindAdjnSichuan �5.90 (1.07)25 AVindAdjnTianjin 12.28 (1.06)26 AVindAdjnXinjiang �2.18 (0.41)27 AVindAdjnYunnan �2.49 (0.36)28 AVindAdjnZhejiang �4.95 (1.14)

Ln(time) �104.54 (2.53)n

Province fixed effect Yes

Observations 588

R2 0.99

Robust t-statistics in parentheses.

n Significant at 5%.nn Significant at 1%.

Table 6The impact of industrial relocation on energy consumption (10,000 tce).

Year DCIa DCOb TDc

1990 �78.7 38.75 �39.95

1991 �3.41 �59.75 �63.16

1992 �68.35 �171.04 �239.39

1993 �103.77 �225.66 �329.43

1994 �37.78 �252.77 �290.55

1995 96.97 �230.99 �134.02

1996 112.79 �1085.64 �972.85

1997 329.81 �1083.95 �754.14

1998 356.54 �1284.12 �927.58

1999 406.67 �1321.7 �915.03

2000 445.1 �1365.99 �920.89

2001 466.5 �1440.71 �974.21

2002 542.49 �1558.12 �1015.63

2003 610.6 �1655.18 �1044.58

2004 734.6 �1659.73 �925.13

2005 771.3 �1350.87 �579.57

2006 763.1 �1255.45 �492.35

Sum 5344.46 �15962.90 �10618.50

a DCI: The difference in energy consumption between real scenario and

counterfactual scenario for provinces with industrial moved in. The positive

number indicates energy consumption increases in these provinces due to

industrial growth.b DCO: The difference in energy consumption between the two scenarios for

provinces with industrial moved out.c TD: The aggregate energy savings due to industrial relocation, which is equal

to column DCI plus column DCO.

X. Zhao, H. Yin / Energy Policy 39 (2011) 2944–29562952

For our research purpose, the most important finding is thatthe coefficients for the intersection terms between industrialproduction and area dummy vary a lot. This suggests that theenergy consumption increase due to the same amount of indus-trial growth is bigger (positive coefficient on the interactionterms) than the reference province in some provinces whilesmaller (negative coefficient on the interaction terms) in others.That is to say, the same amount of industrial growth maytranslate into different increase of energy consumption, all elseequal. This geographical difference is most likely due to factorsthat vary across provinces such as natural resources endowmentand energy efficiency. This is the very reason that industrialrelocation may lead to changes in energy consumption. Anotherfinding is that the R2 is 99%. This is important as we are going touse this model as the basis to predict energy consumption indifferent scenarios.

4.5. Comparing energy consumption under two scenarios

To examine the impact of industrial relocation on energyconsumption, we examine two groups of provinces: provinceswith industries moved in and those with industries moved out.We investigate two scenarios. One scenario is what actuallyhappened, which we refer as real scenario. Another scenario iswhat would have happened if industrial location had stayedunchanged after 1990, which we refer as counterfactual scenario.Table 6 reported the difference in energy consumption between

these two scenarios. The first column shows that provinces withindustries moved in use more energy compared to counterfactualscenario, while the second column suggests that provinces withindustries moved out use less energy than counterfactual sce-nario. Column 3 displays the aggregate result. It suggests that ifno industrial relocation had happened, the energy consumptionwould have increased about 106 mtce. This means that China’sindustrial relocation since the early 1990s has greatly promotedenergy consumption saving.

5. Further discussion

The empirical investigation shows that there is positive corre-lation between China’s industrial relocation in the early 1990sand energy saving. The first driver behind the positive correlationis the macro-policy changes, which have been discussedin Section 3. The second driver behind the positive correlationwould be the industrial relocation in the early 1990s, whichpromotes the substitution of energy with capital and labor. Griffinand Gregory (1976) find that energy is a substitute for capital andlabor; Lin (2007) illustrates that there exists strong substitutionbetween energy and capital in China. One of important character-istics of China’s industrial relocation in the early 1990s is theindustrial agglomeration in the eastern coastal areas, and theeastern coastal areas have advantages in attracting investments,especially foreign investment since their geographical position,convenient transportation condition, and more open and biggermarket. Table 7 shows that the capital availability and populationintensity in the eastern coastal areas is much more than that inthe western areas and northeastern areas. Moreover, with thequicker economic development in eastern coastal areas, laborforce and capital are observed to move from the interior to thecoastal provinces with industrial agglomeration (Jian et al., 1996).Hence, it is possible that capital and labor substitute for energy inthe process of industrial relocation to eastern coastal areas.

The third driver behind the positive correlation betweenindustrial relocation and energy saving is the gap in energy

Table 7Comparing of production factors between different areas in China.

Population intensity (person/km2) Available capital of fixed asset per province (billion RMB)

Eastern coastal areas Northeastern areas Western areas Eastern coastal areas Northeastern areas Western areas

1985 389.72 118.47 40.47 10.02 10.54 5.931990 425.95 125.56 47.12 22.88 17.31 6.771995 445.83 130.45 49.79 122.08 55.02 19.642000 467.58 134.05 52.36 194.35 85.53 41.532005 482.30 135.34 54.22 570.02 263.30 148.902007 475.79 136.27 55.06 852.22 485.22 246.11

Data source: Author’s calculation based on China Statistic Yearbooks (1986–2009).

1985 1992 1994 1999 2001 2007year

EI (

1000

0 to

ns o

f SC

E/1

00m

illio

n R

MB

yua

n)

30.00

25.00

20.00

15.00

10.00

5.00

0.00

Eastern coastal areas

Northern areas

Western areas

Central areas

Fig. 11. Energy intensity comparison of paper and pulp industry in different areas

in China. Eastern Coastal Areas include Shanghai, Zhejiang, Fujian, Guangdong,

Jiangsu, Shandong (The data of Jiangsu is absent, hence Jiangsu is omitted);

Northern Areas include: Beijing, Tianjin, Hebei, Liaoning, and Helongjiang; Central

Areas include: Anhui, Henan, Jiangxi; Western Areas include: Guangxi, Guizhou,

Sanxi, Yunnan, and Xinjiang.

Source: Authors’ calculation based on China Industry Economy Statistic Yearbooks

(1985–2007) and relevant provincial entities statistic year books (1986–2008).

1985 1992 1994 1999 2001 2007year

EI (

1000

0 to

ns o

f SC

E/1

00m

illio

n R

MB

yua

n) 60.0050.00

80.0070.00

40.0030.0020.0010.00

0.00

Eastern coastal areasNorthern areas

Western areasCentral areas

Fig. 12. Energy intensity comparison of raw chemical materials and chemical

products in different areas in China.

Source: Authors’ calculation based on China Industry Economy Statistic Yearbooks

(1985–2007) and relevant provincial entities statistic year books (1986–2008).

1987 1990 19941992 1997 1999 2002 2005 2007year

EI (

1000

0 to

ns o

f SC

E/1

00m

illio

n R

MB

yua

n)

50

40

30

20

10

0

Eastern coastal areasNorthern areas

Western areasCentral areas

Fig. 13. Energy intensity comparison of non-metallic mineral products in different

areas in China.

Source: Authors’ calculation based on China Industry Economy Statistic Yearbooks

(1985–2007) and relevant provincial entities statistic year books (1986–2008).

12 Data source: the website of State Development and Reform Committee.

X. Zhao, H. Yin / Energy Policy 39 (2011) 2944–2956 2953

efficiency across provinces. The energy efficiency gap betweendifferent areas in China is significant in magnitude, and theeastern coastal areas have higher energy efficiency than otherareas (He and Wang, 2009; Qu, 2009; Shi et al., 2008). The energyefficiency difference is caused by energy consumption structure,industrial structure, per capita GDP, enterprise size, geographicallocation, landform and physiognomy, temperature, duration ofsunshine, degree and duration of wind (Yang et al., 2009), and byendowment factors, including capital, labor, resource abundance,such as coal, natural gas, oil, and hydro-electricity (Wang andZhong, 2009). The difference in energy efficiency across differentprovinces in China (Figs. 11–13) makes it possible to achieveenergy saving through industrial relocation. As Shi (2006) argued,if we observed industrial production moved from areas with lowenergy efficiency to those with high energy efficiency and as aresult which the gap in energy efficiency narrowed down acrossareas, this would be an indication for overall energy saving.Fig. 14 demonstrates the change of energy efficiency in bothprovinces with manufacturing industries moved out and thesewith manufacturing industries moved in from 1985 to 2006. It isworthwhile to note two interesting phenomena. First, the energyefficiency in provinces with manufacturing industries moved in isoriginally higher than that in provinces with manufacturingindustries moved out. Second, the difference in energy efficiencybetween the two types of provinces was reduced. As Shi (2006)argued, the convergence of energy efficiency would suggestenergy saving as a result of industrial relocation.

In sum, China’s industrial relocation in the early 1990s wasdriven mainly by market-oriented reform, preferential policies ineastern coastal areas, and policies encouraging efficiencyimprovement. The in-flow of production factors to the areas with

higher operation efficiency improves energy efficiency, and as aresult, energy saving is realized.

Although there exists positive correlation between China’sindustrial relocation in the early 1990s and energy saving (envir-onmental externalities reduction), the issue of environmentalexternalities during the process of industrial relocation is stillun-ignorable. At the same time, the government failure speededup the environmental externalities. One of examples is that sinceJune 2, 2004, the Chinese government carried out discriminatehigher electricity price on six energy intensive sectors includingelectrolytic aluminum, iron alloy, carbide, caustic soda, cement,and steel.12 However, a few local governments in mid-westernareas still provided preferential electricity price to the energyintensive industries in order to get more economic benefit. Forexample, more than half of the 20 Industrial Parks in Neimenggu

1985

1986

1988

1989

1993

1994

1996

1998

2000

2002

2003

2001

2004

2005

2006

1987

1990

1991

1992

1995

1997

1999

year

1200

1000

800

600

400

200

0Ene

rgy

Inte

nsity

(tce

per

mill

ion

RM

B)

Moving-in Provinces Moving-out Provinces

Fig. 14. Difference in energy efficiency between provinces with industries moved in and moved out.

Source: Author’s calculation based on China Energy Statistic Yearbooks (CESYs): 1986–2007.

X. Zhao, H. Yin / Energy Policy 39 (2011) 2944–29562954

that are mainly supported by the local government producedenergy intensive products. Hence, many energy intensive indus-tries are transferred to these areas without technological progressand energy efficiency improvement.13 The failure of enforcingcentral government regulation led to the local environmentundermining and negative impact on the global climate change.

6. Conclusion and policy suggestions

This paper illustrates the characteristics and drivers of China’sindustrial relocation, and pays special attention to the impact ofindustrial relocation on energy consumption and environmentalexternalities. Based on relative theoretical analysis, we put for-ward the natures of environmental externalities during theindustrial relocation process: First, the environmental externalityto the areas with industries moved in would be a very significantissue deserved to be concerned, especially when the industrialrelocation belongs to the replicative alternatives of energy inten-sive industries moving; Second, environmentally induced indus-trial relocation would yield more badly environmentalexternalities; Third, the environmental externalities in the areaswith industries moved in and moved out should be considered asa whole.

Our macro-policy evolution analysis suggests a positive link-age between energy saving (and environmental externalitiesreduction) and China’s industrial relocation in the early 1990s.We argued that: (1) Market-oriented economic reform creates afree market for production factors to flow from areas with lowefficiency to ones with high efficiency; (2) Preferential policies foreastern coastal areas together with these areas’ endowment madethem rich in capital, technology and labor, which made it morelikely to substitute energy with other production factors; (3)Policies for energy efficiency improvement promotes the devel-opment of technological progress; (4) Strengthened environmen-tal protection policy gave special attention to the ecologicalenvironment protection in the Western areas. All of these policiescontribute to energy saving (and environmental externalitiesreduction) in the process of industrial relocation.

We further provide an empirical analysis on the relationshipbetween industrial relocation and energy consumption. We foundthat China’s industrial relocation in the early 1990s has resultedin significant energy saving. In addition to macro-policy changes,we present two other factors that may be responsible for the

13 Data source: Preferential electricity price is cancelled, and the energy intensive

industries in mid-western areas are facing a severe test. http://info.finance.hc360.

com/2006/07/27111851483.shtml.

observed positive relationship: substitution of energy with capitaland labor, and energy efficiency gap between different areas.

With economic development, industrial relocation is an inevi-table trend. For example, it is anticipated that by the end of 2010,the output value of industries transferred out of Guangdong,Shanghai, Zhejiang, and Fujian would reach RMBf1400 billion(Jiang, 2009). Hence, in order to reduce environmental external-ities and social cost further in the process of industrial relocation,we present the following policy suggestions:

First, strengthen the evaluation of environmental benefits/costs and reduce the market failure. Stern (2006) pointed outthat the ignorance of externalities is the greatest market failureever known. In the standard theory of externalities, there are fourways in which negative externalities can be approached: tax,carbon price, quantity restriction, and a full set of property rights(Stern, 2006). Soderholm and Sundqvist (2003) also argues thattaxes and subsidies are usually taken as measures that reflect theexternal costs or benefits, and ensure that profit-maximizingfirms select the mix of goods and production technologies thatbest satisfy environmental and economic goals. In order toascertain the appropriate price of carbon or other kinds ofemission (such as SO2), and determine the proper taxes andsubsidies, it is essential to evaluate environmental benefits/costs.

Second, pay more attention to the coordinated developmentbetween economy, environment, and society rather than empha-size only the GDP growth. Investment push and high growth rateis the most important characteristics of China’s economic devel-opment (Lin, 2009). If the economic development mode was notchanged, no matter where industries move to, the energy savingand CO2 emission reduction that may be achieved throughindustrial relocation would be very limited. Only when China’sgovernment pays more attention to the importance of coordinateddevelopment between economy, environment, and society, canthe environmental externalities in the process of industrial reloca-tion be reduced. In the ‘‘Twelfth-Five Year Plan’’, the Chinesegovernment has used the word of ‘‘economic development’’ inreplacement of ‘‘economic growth’’, and the index of energy savingand environmental pollution in each province will be publishedregularly. This suggests that the central government has paid moreattention to environmental and social issues, a broadening fromthe narrowest of economic principles. However, this is only thefirst step, as the shift of local government attention and enter-prises’ behavior in particular are the key factors in reducingenvironmental externalities of industrial relocation.

Third, avoid long-lived investment in high-carbon infrastruc-ture in the process of industrial relocation. In the process ofindustrial relocation, much investment will be made and manyinfrastructures will be constructed. Stern (2006) presents that

X. Zhao, H. Yin / Energy Policy 39 (2011) 2944–2956 2955

‘‘China is expected to increase their capital infrastructure sub-stantially over coming decades, with China alone accounting foraround 15% of total global energy investment. If they use low-emission technologies, emission savings can be ‘locked in’ for thelifetime of the asset.’’ Hence, to ensure that new investmentsare made in low-carbon infrastructure or low-carbon technolo-gies in the areas with industries moved in is vital for thereduction of environmental externalities and for the climatechange mitigation.

Last, but not the least, in the process of industrial relocation,addressing the employment issue is also important and it shouldbe given attention in the areas with industries moved out. For acountry like China with about 1.4 billion population, the employ-ment issue is always of the greatest concern. Hence, how toreduce the negative impact of industrial relocation on theemployment in the areas with industries moved out should beconsidered. CISC has set an example in this respect. When itmoved from Beijing to Hebei, it offered new jobs to its more than28,000 former employees in Beijing by developing non-steelsectors, such as electronics industry, electrical and machineryindustry, building industry, real estate, catering industry, andtravel industry in Beijing, and many of the former employees arehired by these new enterprises. This example reflects that in theareas with industries moved out, some other new sectors such astertiary industry are expected to develop so as to address theemployment issue, promote the industrial structure change, andalso reduce the possible social conflict caused by industrialrelocation.

Acknowledgment

Xiaoli Zhao wishes to thank the financial supports fromNational Natural Science Foundation of China (Project number70773040 and 71073053). Haitao Yin wishes to thank supportsfrom Shanghai Pujiang Program, Innovation Program of ShanghaiMunicipal Education Commission as well as Project on Huma-nities and Social Sciences Research of Ministry of Education of thePRC. The authors appreciate the anonymous reviewers for theirvaluable comments, highly responsible attitude and hard work.Their comments benefit the authors not only in the improvementof this paper, but also for rethinking about some economic issuesfrom deeper social-cognitive theoretical perspective. The authorsthank Nicholas Powers and participants at the University ofMichigan Erb Colloquium for very helpful comments. In addition,the authors thank Sufang Zhang for her help in the grammarcorrection of the paper, and thank Chao Ouyang, Cui Song, and NaLi for research assistance.

References

Akinlo, A.E., 2008. Energy consumption and economic growth: evidence from 11Sub-Sahara African countries. Energy Economics 30, 2391–2400.

Ang, B.W., 1995. Multilevel decomposition of industrial energy consumption.Energy Economics 17 (1), 39–51.

Auffhammer, M., Carson, R.T., 2008. Forecasting the path of China’s CO2 emissionsusing province-level information. Journal of Environmental Economics andManagement 55 (3), 229–247.

Bai, X., 2002. Industrial relocation in Asia. Environment 44 (5), 8–21.Biggiero, L., 2006. Industrial and knowledge relocation strategies under the

challenges of globalization and digitalization: the move of small and mediumenterprises among territorial systems. Entrepreneurship & Regional Develop-ment 18, 443–471.

Calmfors, L., Forslund, A., 1991. Real-wage determination and labour marketpolicies: the Swedish experience. The Economic Journal 101, 1130–1148.

Chen, W.H., 1998. The impact of policies on the development of eastern andwestern areas in China. Industrial Technology Economics 19 (6), 16–17.

China’s Energy Outlook, 2004. The compiling team of China’s Outlook Institute ofNuclear and New Energy Technology, Tsinghua University. Tsinghua Univer-sity Press, World Scientific.

CES: China Energy Strategy, 2007. /www.chinaembassy.org.in/chn/kj/P020080725509142506360.docS. Last accessed: 10 October 2009.

Chou, Y., Wu, J.J., 2009. The impacts of trade agglomeration in China’s easterncoastal areas on technological diffusion effect. System Engineering 27 (7),112–115.

DRCSC: Development Research Center of the State Council, 2009. Constraintfactors and pathway choice for electricity industry entering into Low CarbonAge. /http://www.drcnet.com.cn/DRCnet.common.web/DocS.

Fang, J., 1994. Constraint on Special Economic Zone from international legal normsand readjustment of preferential policies in China’s Special Economic Zone.International Business Study 5, 28–31.

Fang, J., Gerry, B., 2010. China’s rural health system and environment-relatedhealth risks. Journal of Contemporary China 19 (63), 23–35.

Fujita, M., Henderson, J.V., Kanemoto, Y., Mori, T., 2003. Spatial distribution ofeconomic activities in Japan and China. Handbook of Urban and RegionalEconomics, pp. 49–95.

Gao, S.Q., 2008. Review of China’s thirty years’ reform. China’s Worker Education10, 3–5.

Garbaccio, R., Ho, M., Jorgenson, D., 1999. Why has the energy-output ratio fallenin China? The Energy Journal 20, 63–91.

Griffin, J., Gregory, P., 1976. An intercountry translog model of energy substitutionresponses. American Economic Review 66, 845–857.

He, C.F., Wang, J.S., 2009. Economic transition and energy intensity in China.Scientia Geographica Sinica 29 (4), 461–469 (in Chinese).

Holdaway, J., 2010. Environment and Health in China: an introduction to anemerging research field. Journal of Contemporary China 19 (63), 1–22.

Huang, J., Li, K., 2006. Foreign trade, local protection and China’s industriallocation. China Economic Quarterly 5 (3), 733–760.

Huang, J., Huang, J., 2008. Market size and industrial growth in China’s Provinces.China Economic Quarterly 7 (4), 1317–1334.

Irwin, R.J., McClelland, G.H., 2001. Misleading heuristics and moderated multipleregression models. Journal of Marketing Research 38 (1), 100–109.

Jian, T., Sachs, J., Warner, A., 1996. Trends in regional inequality in China. ChinaEconomic Review 7 (1), 1–21.

Jiang, H., 2009. Current situation and issues of the industrial relocation in China’seastern coastal areas and policy suggestions. Journal of China Trade Guide 18,5–6 (in Chinese).

Jiang, Z., 2008. Reflections on energy issues in China. Journal of Shang Hai JiaoTong University 42 (3), 345–359.

Jin, Y., Chen, Z., Lu, M., 2006. China’s regional industrial accumulation: economicgeography, new economic geography, and economic policy. EconomicResearch Journal 4, 79–89.

Kambara, T., 1992. The energy situation in China. China Quarterly 131, 608–636.Karanfil, F., 2008. Energy consumption and economic growth revisited: does the

size of unrecorded economy matter? Energy Policy 36, 3029–3035.Lee, Chien-Chiang Lee, Chang, Chun-Ping, 2008. Energy consumption and eco-

nomic growth in Asian economies: a more comprehensive analysis using paneldata. Resource and Energy Economics 30, 50–65.

Li, W.W., 1998. Policy analysis on China’s regional economic development. SocialScience 3, 20–25 (in Chinese).

Lin, B.Q., 2007. The drivers and mechanism for energy saving and emissionreduction. China Construction 12, 16–17 (in Chinese).

Lin, B.Q., 2009. Energy saving and emission reduction is the guarantee of China’ssustainable economic development (in Chinese). /http://www.lrn.cn/stratage/ressecurity/200908/t20090824_404754.htmS.

Liu, X.O., Ang, B.W., Ong, H.L., 1992. The application of the divisia index to thedecomposition of changes in industrial energy consumption. Energy Journal13 (4), 161–177.

Lora-Wainwright, A., 2010. An anthropology of ‘‘cancer villages’’: villagers’ perspectivesand the politics of responsibility. Journal of Contemporary China 19 (63), 79–99.

Price, L., Wang, X., Yun, J., 2010. The challenge of reducing energy consumption ofthe top-1000 largest industrial enterprises in China. Energy Policy 38,6485–6498.

Medlock, K.B., Soligo, R., 2001. Economic development and end-use energydemand. The Energy Journal 2, 77–105.

Narayan, P.K., Smyth, R., 2008. Energy consumption and real GDP in G7 countries:new evidence from panel cointegration. Energy Economics 30, 2331–2341.

Ockwell, D.G., 2008. Energy and economic growth: grounding our understandingin physical reality. Energy Policy 36, 4600–4604.

Qu, X.E., 2009. China’s interprovincial total factor energy efficiency change decomposes.The Journal of Quantitative & Technical Economics 8, 29–43 (in Chinese).

Shi, D., 2006. Regional difference in China’s energy efficiency and conservationpotentials. China’s Industrial Industry 10, 49–58.

Shi, D., Wu, L.X., Fu, X.X., 2008. Study on regional difference of energy efficiency inChina and the drivers. Management World 2, 35–43.

Shiu, A., Lam, P.L., 2004. Electricity consumption and economic growth in China.Energy Policy 32, 47–54.

Soderholm, P., Sundqvist, T., 2003. Pricing environmental externalities in thepower sector: ethical limits and implications for social choice. EcologicalEconomics 46, 333–350.

Stern, N., 2006. Review on the economics of climate change. /http://webarchive.nationalarchives.gov.uk/þ/http://www.hm-treasury.gov.uk/independent_reviews/stern_review_economics_climate_change/stern_review_report.cfmS.

Stern, N., 2010. Presidential address imperfections in the economics of publicpolicy, imperfections in markets, and climate change. Journal of the EuropeanEconomic Association 8 (2–3), 253–288.

X. Zhao, H. Yin / Energy Policy 39 (2011) 2944–29562956

Vennemo, H., Kristin, A., He, J., Hu, T., Li, S., 2009. Benefits and costs to China ofthree different climate treaties. Resource and Energy Economics 31, 139–160.

Vicent, A., Rosa, D., 2004. Comparison of energy intensities in European union

countries: results of a structural decomposition analysis. Energy Policy 32 (2),177–189.

Wang, J., Zhong, W.Z., 2009. The research of regional energy intensity difference in

China—from the factor endowment perspective. Industrial EconomicsResearch 6, 44–51.

Wen, M., 2004. Relocation and agglomeration of Chinese industry. Journal of

Development Economics 73, 329–347.Wu, A., 2009. Industrial relocation is one kind of organic interactions. Reformation

2, 143–145 (in Chinese).Wu, D.T., Ge, Y.J., 1996. Changing orbit of regional economic increase, resources

development and environmental change in eastern coastal areas in China.

China Population, Resources and Environment 6 (3), 26–31.Xu, L.D., Tan, W.H., 2003. Empirical analysis on China’s FDI aggregation effect.

Management World 7, 38–44.

Yang, H.L., Shi, D., Xiao, J., 2009. Influence of environmental factors on energyefficiency—analysis on regional theoretic energy saving potential and realenergy saving potential. China Industrial Economics 4, 73–84.

Yang, M., 2008. China’s energy efficiency target 2010. Energy Policy 36, 561–570.Yuan, J.H., Kang, J.G., Zhao, C.H., Hu, Z.G., 2008. Energy consumption and economic

growth: evidence from China at both aggregated and disaggregated levels.Energy Economics 30, 3077–3094.

Zhao, X.L., Yang, J., 2009. Comparable analysis of key factors affecting industryenergy consumption between China and Beijing. China Energy 5, 19–25.

Zhao, X.L., Ma, C.B., Hong, D.Y., 2010. Why did China’s energy intensity increaseduring 1998-2006: decomposition and policy analysis. Energy Policy 38,1379–1388.

Zhong, J.J., 2009. The import oil dependence is nearly 50% last year. January 15,/http://funds.hexun.com/2009-01-15/113448232.htmlS. Last accessed:March 28, 2009.

Zhou, Y., Li, L.S., 2006. The action of structure and efficiency on China’s energyintensity—an empirical analysis based on AWD. Industrial EconomicsResearch 4, 68–74.