Energy Policy 132 (2019) 122–131

Contents lists available at ScienceDirect

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

Will economic infrastructure development affect the energy intensity of T

China's manufacturing industry?
Boqiang Lin∗, Yu Chen
School of Management, China Institute for Studies in Energy Policy, Collaborative Innovation Center for Energy Economics and Energy Policy, Xiamen University, Fujian,
361005, China

ARTICLE INFO ABSTRACT

Keywords: China's manufacturing industry consumes more energy than the total amounts of Germany, Britain, France,
China's manufacturing industry Spain and Japan combined, and has a substantial impact on energy conservation and emissions reduction. This
Energy consumption paper investigates the influence mechanism of economic infrastructure on the energy consumption and energy
Energy intensity intensity of the sector. Using the China's provincial data during the period 2003–2016, the profit function is
Economic infrastructure
applied to incorporate infrastructure into the input-output system, and to avoid the endogenous problems caused
Energy price distortion
by the reverse causal relationship between energy consumption and infrastructure construction. The empirical
results indicate that economic infrastructure construction will increase energy consumption and reduce the
energy intensity in the long term. Although the energy consumption effect of infrastructure in the central and
western regions is less than that in the eastern region, the energy intensity of the western region declines the
most due to the infrastructure construction. Based on the results of this paper, some policy implications are
discussed. This paper also offers some targeted policy recommendations to improve policy design of the gov-
ernment.

1. Introduction On the other hand, in order to expand domestic demand and

maintain national macroeconomic stability, the Chinese government
As a traditional pillar industry in China, the manufacturing industry has implemented a proactive fiscal policy to promote economic devel-
has played a vital role in achieving rapid economic development, pro- opment through strengthening infrastructure construction. In the wake
moting employment, increasing people's income and accelerating the ur- of the global financial crisis of 2008, about 37.5% funds of China's “four
banization process (Lee and Zhang, 2012; Sun et al., 2018). However, the trillion” economic stimulus package was spent on infrastructure con-
rapid development of China's manufacturing industry (CMI) is at the ex- struction.2 In the work report of 2016, the Chinese government plans to
pense of huge energy consumption and CO2 emissions (Hasanbeigi et al., invest 800 billion yuan in railways and 1.65 trillion yuan in highways,
2013; Xu and Lin, 2017; Tian et al., 2018). The energy consumption of and focus on other infrastructure construction projects such as smart
CMI was 24.25 million tons of standard coal equivalent in 2016, exceeding power grids and urban rail transit.
the total amounts of Germany, Britain, France, Spain and Japan (19.96 According to the definition of the world bank, infrastructure refers
million tons of standard coal equivalent). The extensive developmental to permanent, complete sets of engineering construction, equipment,
pattern with high energy consumption has led to large-scale greenhouse facilities, which provide services for the common needs of all en-
gas emissions (Song et al., 2013; Li and Lin, 2016; Lin et al., 2018; Yu terprises' production and residents' life, including economic infra-
et al., 2018; Chen et al., 2019). The CO2 emissions of CMI were 5.19 structure and social infrastructure (World Bank, 1994). Public utilities
billion tons in 2016, accounting for 15.5% of the world.1 Therefore, it is of (electricity, communications, etc.), public works (water conservancy
great importance for the whole world to study the energy consumption projects, roads, etc.) and other transport sectors (railways, urban
and energy intensity issues of CMI (Li and Lin, 2015). transportation, etc.) are included in the former. The latter is comprised

∗
Corresponding author.
E-mail addresses: bqlin@xmu.edu.cn, bqlin2004@vip.sina.com (B. Lin).
1
The energy consumption data of CMI were obtained from China energy statistical yearbook, and the CO2 emissions of CMI are calculated by the authors. The
energy consumption and CO2 emissions data of Germany, Britain, France, Spain and Japan were available in BP statistical review of world energy, June 2017.
2
Source: Document entitled “Key investment and capital measurement of 4 trillion yuan investment” issued by the National Development and Reform Commission
of China. < http://www.ndrc.gov.cn/ > 2019 [accessed 27 January 2019].

https://doi.org/10.1016/j.enpol.2019.05.028
Received 20 February 2019; Received in revised form 16 May 2019; Accepted 17 May 2019
Available online 23 May 2019
0301-4215/ © 2019 Elsevier Ltd. All rights reserved.
B. Lin and Y. Chen Energy Policy 132 (2019) 122–131

of education and health care. The research object of this paper is eco- comprehension. We draw the conclusion that economic infrastructure
nomic infrastructure since it is closely related to the energy consump- construction will increase the energy consumption and reduce the en-
tion of CMI. The employment and income growth driven by infra- ergy intensity of CMI in the long term. The corresponding policy sug-
structural investment will lead to the upgrading of energy consumption. gestions are put forward in view of the empirical results. (2) Model
Furthermore, economic infrastructure is an important determinant of specification. The profit function is applied to incorporate infra-
energy efficiency in industrial enterprises (Moyo, 2013). For instance, structure into the input-output system of CMI, avoiding the endogenous
good electric power facility can improve the industrial productivity of problems caused by the reverse causal relationship between energy
unit energy consumption, while poor quality of power supply will result consumption and infrastructure construction. (3) Data processing.
in to frequent power cuts and difficult accessibility for industrial users, There is a problem of wide and inconsistent statistical caliber in the
which decreases the energy efficiency. estimation of China's infrastructure capital stock at present. By utilizing
Common proxy variables for infrastructure are in monetary form existing official statistics, we estimate the capital stock of economic
and physical form (Tan et al., 2018). However, the latter cannot build a infrastructure at the provincial level based on the perpetual inventory
comprehensive indicator, it is impossible to arithmetically sum up the method, providing relatively complete and reliable data for future re-
various physical values since they are of different measurement units. search. (4) We find the evidence that energy prices in China do not
In addition, physical indicators cannot accurately reflect the quality of reflect the environmental externalities and energy scarcity.
infrastructure. For instance, there are major discrepancies between the The remainder of this paper is structured as follows: Chapter 2
different levels of roads for the same 1 km. Agenor and Moreno-Dodson briefly reviews the existing relevant research. Methodology and data in
(2006) believe that infrastructure effects are derived from capital stocks the empirical analysis are described in chapter 3. In chapter 4, the re-
and should be reasonably estimated based on stock rather than flow sults are discussed and analyzed, and the robustness checks are con-
indicators. Hence, we use the monetary form of per capita capital stock ducted. Chapter 5 draws conclusions and proposes policy re-
to measure the development level of economic infrastructure. Two commendations. For the sake of fluency and readability of this paper,
physical forms, power infrastructure density and transportation infra- the abbreviations and symbols are depicted in Table 1.
structure density, are applied for robustness check.
Fig. 1 depicts the capital stock of economic infrastructure and en-
ergy consumption of CMI from 2003 to 2016. Both show a growth trend 2. Literature review
during the period. The average annual growth rate of capital stock and
energy consumption are 17.8% and 7.0%, respectively. Theoretically Energy consumption and energy intensity in manufacturing industry
speaking, economic infrastructure interacts with the energy consump- have always been hot topics of research. Park et al. (1993) decomposed
tion of manufacturing industry, causing the endogenous problems. On the changes of energy consumption in manufacturing industry into
the one hand, infrastructure investment will drive the development of energy intensity effect, output growth effect and industrial structure
steel, cement and other related manufacturing industries, increasing the effect by utilizing statistical data of manufacturing industry in 26
energy consumption. Sound infrastructure can improve the mobility of countries from 1973 to 1988. The study findings demonstrated positive
energy resources and promote the optimal allocation of resources correlation exists between energy consumption and output. Similar
among regions. On the other hand, the increase of energy consumption studies include Adenikinju (1998), Reddy and Ray (2011). Theriault
in the manufacturing industry can often promote the economic devel- and Sahi (1997) believe that the various stages and nature of manu-
opment of a region, which also means higher requirements for the in- facturing processes in OECD countries are the key influencing factors
frastructure level. for the differences in energy intensity. Miketa (2001) explored the
By calculating the short term and long-term elasticities of energy impact of total industrial output value, energy price and fixed capital
consumption, output, energy intensity with regard to economic infra- formation on energy intensity by employing panel data of manu-
structure, we investigate the influence mechanism of economic infra- facturing industry in 39 countries. Hang and Tu (2007) utilized time
structure on the energy consumption and intensity of the sector. The series data to calculate the elasticity of energy price to energy intensity,
main contributions of this paper are as follows. (1) Problem demonstrating that increasing energy price can decrease energy in-
tensity and improve energy efficiency. A similar conclusion was

Fig. 1. Capital stock of economic infrastructure and energy consumption of CMI.

123
B. Lin and Y. Chen Energy Policy 132 (2019) 122–131

Table 1 manufacturing industry (Achour and Belloumi, 2016; Xie et al., 2017;
The abbreviations and symbols in this paper. Chakamera and Alagidede, 2018). Seitz (2000) investigated the short-
Abbreviation and symbol Description term production effect and long-term regional capital distribution effect
of infrastructure by solving the structural equation of labor demand and
CMI China's manufacturing industry capital demand in the manufacturing industry. Paul et al. (2004) esti-
CO2 Carbon dioxide
mated the cost function of Canadian manufacturing public infra-
OECD Organization for Economic Cooperation and
Development
structure based on time series data, and proved the positive relationship
GDP Gross domestic product between the public infrastructure and the productivity of the manu-
PPI Producer price index facturing industry. Moyo (2013) used the survey data to explore the
SURE Seemingly unrelated regression estimation impact of power infrastructure quality on manufacturing productivity
OLS Ordinary least square
in Africa.
Y Industrial output
E Energy consumption Considering the huge energy consumption and the enormous in-
L Labor input dustrial output value of CMI, it is urgent and necessary to explore how
K Capital input they are affected by economic infrastructure. Most of the existing lit-
G Economic infrastructure
erature uses the production function as an analytical framework, con-
Py Output price
Ve Energy price
centrating on the influence of infrastructure on economic growth and
Vl Labor price employment. Discussions on the reciprocal effect between infra-
structure and manufacturing energy consumption are limited. This
paper incorporates infrastructure into the input-output system of CMI
obtained by Parker and Liddle (2016) by making use of the panel data. based on profit function, avoiding the endogeneity problem brought by
Kepplinger et al. (2013) adopted the mixed effect model to conduct a the production function method and the unbelievable flaw in the esti-
transnational analysis on the energy intensity of the manufacturing mation of the marginal output of private input. Based on the empirical
industry, and investigated the discrepancy in energy intensity between results, we investigate the influence mechanism of economic infra-
industrial sectors. The empirical results revealed that the energy in- structure on the energy consumption and intensity of CMI, hoping to fill
tensity of manufacturing decreases with time and economic develop- the gaps in existing research.
ment. Zhao et al. (2014) compared the energy consumption and energy
intensity of manufacturing industries in China and Japan, indicating
that the increase of energy intensity is brought about by the change of 3. Methodology and data
energy policy and industrial structure. By employing the divisia index
decomposition method to analyze the energy intensity of the South 3.1. Methodology
Korean manufacturing industry, Choi and Oh (2014) drew an analogous
conclusion. Lin and Chen (2018) adopted the co-integration model to To some extent, infrastructure is part of endogenous growth
examine the long-term equilibrium relationship between energy con- (Button, 1998). The analysis method based on the production function
sumption, industrial output value, energy price, and industrial structure implies the assumption of a causal relationship between infrastructure
of CMI from 1980 to 2013, and used the scenario analysis to predict the and output, thus generating endogenous problems. Dual theory is a way
energy consumption in the future. Chowdhury et al. (2018) evaluated to overcome endogenous problems. Starting from the profit function
the energy utilization technology and energy-saving potential of British and applying the dual theory to derive the total output and factor de-
manufacturing industry and proposed a business model to reduce the mand, the endogeneity problem of the explanatory variables in the
energy consumption of industry. production function model can be avoided by directly estimating the
Meanwhile, scholars are paying close attention to the research of simultaneous equations derived from the profit function.
infrastructure effects. Rosenstein-Rodan (1943) expounded the im- Referring to the study of Demetriades and Mamuneas (2000),;
portance of infrastructure construction for the country's comprehensive Kratena (2007) and Tan et al. (2018), We derive the theoretical model
development. Aschauer (1989) conducted a pioneering study on infra- of supply and demand based on the variable profit function, which can
structure by incorporating public capital into the production function avoid the endogeneity problem. In addition, China's economic infra-
for the first time and found that infrastructure contributed significantly structure construction is mainly funded by the government, thus the
to economic growth and productivity improvement. Since then, the cost and price do not fully reflect the nature of construction and op-
literature on the relationship between infrastructure and macro- eration activities. It is unreasonable to price the services provided by
economics has abounded. Calderón and Servén (2004) applied principal these facilities according to the marginal output (Straub and Stéphane,
component analysis to construct a comprehensive indicator including 2011). Contrary to the production function, the profit function does not
transportation, communication, electricity and other infrastructure to imply the assumption that the pricing of factors based on marginal
investigate the influence mechanism of infrastructure scale on GDP output. Assuming that producers use both variable inputs and fixed
growth rate. Pradhan and Bagchi (2013) adopted a vector error cor- inputs, only variable inputs can be flexibly adjusted in the short term.
rection model to reveal that the expansion of transportation infra- Firms choose optimal output and variable input to achieve short-term
structure will lead to the substantial economic growth in Indian. profit maximization with the constant public infrastructure capital and
Farhadi (2015) took 18 OECD countries as a case study to verify the private capital. In the long run, firms adjust their private capital stock
positive impact of transport infrastructure on total factor productivity through investment to maximize profits. In view of the hysteresis of
and labor productivity. Infrastructure construction can effectively im- productive services provided by fixed capital, we assume that fixed
prove energy efficiency (Garrone and Grilli, 2010). Wang et al. (2018) input plays a role with a lag of one period, then the profit function is
analyzed the impact of power grid infrastructure on China's regional formulated as follows:
total factor energy efficiency through empirical research. Tan et al. = F (Pt , Vt , It 1) (1)
t
(2018) used the seemingly unrelated regression estimation to find that
the road density exhibits a negative impact on energy intensity in the t is the profit in year t ; Pt represents the price of output in year t ; Vt
long term. represents the price of variable input in year t ；It 1 denotes the price of
Advanced infrastructure can stimulate innovation, bring economies fixed input in year t .
of scale and agglomeration effects, thereby reducing production costs In the short run, firms can only choose output y and variable input x
and improving the economic output and energy efficiency of the to maximize profits. The maximization problem can be formulated as:

124
B. Lin and Y. Chen Energy Policy 132 (2019) 122–131

Max { (Pt , Vt , It 1)} (2) Lt = py Y + ve E

= 0 + k Kt 1 + G Gt 1 ye py ve + kG K t 1 Gt 1
(Pt , Vt , It 1) = Pt y (Pt , Vt , It 1) Vt x (Pt , Vt , It 1) (3)
1 1 2 2 2 2
+2 kk ( Kt ) 2 + 2 ee ve yy py + kk K t 1 + GG Gt 1
According to Hotelling's Lemma (Hotelling, 1932) and envelope (10)
theorem (Kimball, 1952), the partial derivative of profit with respect to
Obviously, infrastructure stocks will affect the input-output decision
commodity price equals to the demand of the commodity. Therefore,
of the firms. The short-term and long-term effects of the infrastructure
output supply and variable input demand are respectively derived from
can also be further analyzed due to the dynamic nature of the model. In
the following equation:
equation (8) -(10), the coefficients eG and yG reflect the short-term
impact of infrastructure on energy consumption and output, separately.
y (Pt , Vt , It 1) = (Pt , Vt , It 1)/ Pt (4)
The corresponding elasticity in the short run can be expressed as:

x (Pt , Vt , It 1) = (Pt , Vt , It 1)/ Vt (5) G G

= , =
EG eG
E
YG yG
Y (11)
Suppose the profit function is a transcendental logarithmic function Production will reach equilibrium in the long run, and we can cal-
(Nadiri and Mamuneas, 1994; Bergman, 1997), the variable profit culate the optimal capital stock K by solving the long-term profit
function including adjustment cost can be expressed as: maximization problem:
N J 1 N H • •
t = 0 + q
n=1 n n
+ j=1 j I j, t 1 + 2 n=1
q q
h = 1 nh n h max{ (p , v , K , K , G ) qk K } = py (p, v, K , K , G )
•
(12)
1 M J N J vx (p , v, K , K , G ) qk K
+2 m=1 j=1 jm In, t 1 Im, t 1 + n=1
q I
j = 1 nj n j, t 1
+ j j
I j, t
qk stands for the price of investment, i.e. use-cost of capital. Take the
1
+2 j jj
( I j, t )2 + I
n j nj n, t 1
I j, t + n j nj qn I j, t
derivative on both sides of equation (12) with respect to K, we get the
(6)
first-order optimization conditions:
Where q = (P , V ) represents the price vector consisting of the
output price and variable input price. equation (6) satisfies the E |K = K = qk
K (13)
properties of symmetry and homogeneity: nh = hn , mj =
N H
= 0, n = 1 j = 1 nj = 0 . The items con-
N J This means the price of private capital equals the marginal profit of
jm, n = 1, n=1 h = 1 nh
taining I j, t constitute the adjustment cost. The marginal adjustment capital when K is at the long-term equilibrium level. By substituting
cost is 0 when the fixed input reaches the long-term equilibrium, i.e. equation (7) into equation (13), we obtain:
t/ I j, t | I j,t = 0, I j, t = I j,t = 0 . k + kG G + ek ve + yk py qk
Hence, j = nj = nj = 0, jj 0 . In this paper, variable inputs are K =
kk (14)
labor (L) and energy (E), fixed inputs are economic infrastructure (G)
and private capital (K). The rationality of considering infrastructure With the change of external factors in the economy, firms will adjust
capital as fixed input lies in the following aspects. (1) Infrastructure has private investment in the long term, ultimately affecting the level of
a long-life span with the characteristic of immovable and holistic. output and variable input demand. Thus external factors will influence
Although the demand for infrastructure services may increase rapidly in energy demand in two ways: one is the short-term direct impact on
the short term, it will take a long time for the inventory of infra- energy demand; the other is to change the long-term private capital
structure to adjust and increase supply. (2) The infrastructure in China equilibrium level by influencing private investment decisions and then
is largely provided by the government, and investment decisions mainly indirectly affect energy demand.
depend on public policies, which means that manufacturers cannot Hence, the long-term elasticities of energy consumption and output
flexibly adjust infrastructure stocks according to production require- of CMI with regard to economic infrastructure are:
ments. The investment and operation of infrastructure is weakly con- E E k G G
ek kG
nected with the market mechanism. EG = + = eG
G k G E kk E (15)
Besides, the labor price Vl is used to standardize the output price Py
and input price Ve , i.e. py = Py /Vl , ve = Ve /Vl , where py and ve refers to the Y Y k G yk kG G
output price and input price after standardization. Since infrastructure YG = + = yG
G k G Y kk Y (16)
construction is not in the investment program of producers, only the
adjustment cost of private net investment Kt is considered, and the Energy consumption per unit of output is used to denote energy
equation becomes: intensity, and the short-term elasticity of energy intensity of CMI with
regard to economic infrastructure is:
/ vl = 0 + y py + e ve + k Kt 1 + G Gt 1
G G
=
1 2 2 2 2
EG eG
E yG
Y (17)
+2 ee ve + yy py + kk K t 1 + GG Gt 1 + kk Kt2
The long-term elasticity is:
+ kG K t 1 Gt 1 + ye py ve + ek ve K t 1 + eG Ve Gt 1 + yk py K t 1 + yG py
ek kG G yk kG G
EG = eG yG
Gt 1 kk E kk Y (18)
(7)

On the basis of equations (4), (5) and (7), The total output supply 3.2. Data
equation and the variable input demand equation are:
The data in this research covers 30 provinces in China owing to the
Yt = y + yy py + ye ve + yk K t 1 + yG Gt 1 (8) data unavailability of Hong Kong, Macao, Taiwan and Tibet. Before
2003, the National Bureau of Statistics of China did not conduct sepa-
Et = + rate statistics on the environment and public facilities management
e ee ve + ye py + ek K t 1 + eG Gt 1 (9)
industry. In addition, the geological prospecting is also jointly counted

125
B. Lin and Y. Chen Energy Policy 132 (2019) 122–131

with the water conservancy, but generally speaking, the former does 4. Results and analysis
not fall within the scope of infrastructure defined by the world bank
report, which may cause our estimation to be biased. The statistical An error term was added to equation (8) -(10) for simultaneous
caliber has changed since 2003. Referring to the research of Jin (2012), estimation. The seemingly unrelated regression estimation (SURE)
the statistical scope of economic infrastructure investments is defined as method can take into account the simultaneous correlation between
the fixed asset investment for the four sectors in the economy in this heteroscedasticity and residuals of different equations. Hence, SURE
paper. They include (1) “transport, storage and post”, (2) “production method is applied to systematically estimate the parameters in this
and supply of electricity, heat, gas and water”, (3) “management of paper. Considering the regional discrepancy, we add the provincial
water conservancy, environment and public facilities”, (4) “information dummy variable in the regression. Pooled OLS estimation method is
transmission, software and information technology”. Therefore, the also used for comparison, which results are reported in Table 3. The
time interval of our provincial panel data is from 2003 to 2016. coefficients estimated by the two methods and their significance levels
Per capita economic infrastructure capital stock is used to measure are basically consistent, which proves that our estimation results are
the development level of regional infrastructure. The most commonly robust. In addition, the last line of Table 3 shows that the P value of “no
used method for estimating physical capital stock is the perpetual in- contemporaneous correlation” test between the disturbance terms of
ventory method proposed by Goldsmith (1951), which is formulated as each equation is 0.00, i.e., the null hypothesis that the disturbance
follows: terms of each equation are independent should be rejected at the 1%
significance level. Thus, the efficiency of system estimation can be
Kt = It + (1 t ) Kt 1 (19) improved by using SURE. The following analysis is on the basis of the
results of columns (1)–(3).
Kt and Kt 1 refer to the economic infrastructure capital stock of CMI in As we can see from column (2) of Table 3, an interesting result is
year t and year t 1,correspondingly; It refers to the economic infra- that the influence of energy price on the energy consumption is not
structure investment in year t ; t denotes the depreciation rate in year t . significant. Similar results include Hang and Tu (2007) and Tan et al.
The data of the economic infrastructure capital stock in the base (2018). Theoretically, the demand for a commodity should fall as its
year (2003) is derived from Jin (2012). The fixed asset investment data price rises. However, price distortions have occurred for energy pro-
of the four sub-subjects of economic infrastructure can be obtained ducts (Hang and Tu, 2007; Ju et al., 2017). Energy prices are controlled
from the Wind database. In terms of the calculation of depreciation rate by the Chinese government. For instance, electricity prices are tightly
of the capital stock, the existing research adopts the method of the regulated and natural gas prices are also strictly guided, making it is
constant depreciation rate (Young, 2003; Kamps, 2006), which does not difficult for energy prices to reflect environmental externalities and
strictly distinguish the diminishing efficiency of different fixed capital, energy scarcity. Moreover, there is a huge and rigid demand for energy
nor does it take into account the heterogeneity caused by the differ- products in the rapid development stage of CMI, thus the energy de-
ences in production technology and equipment between provinces and mand will not decrease when energy prices rise.
regions. Following the research of Zhang et al. (2004), this paper first
estimates the proportion of the three types of fixed assets, namely 4.1. Discussions of the impact of economic infrastructure on energy
construction, equipment and other costs, to the capital of the infra- consumption
structure. Then, the depreciation rates of these three types of fixed
capital are estimated based on the service life. The final weighted cal- Undoubtedly, the construction and development of economic in-
culation yields the variable depreciation rate of the capital of each frastructure has increased the energy consumption and improved the
province in each year. energy efficiency of the manufacturing industry. For instance, as a ty-
Relevant data for various sub-sectors of CMI in each province (gross pical province with large energy consumption in the central China,
industrial output, energy consumption, employment number, original Henan province has formed a safe, efficient, clean and low-carbon
and net value of fixed assets) are available in the provincial statistical modern energy security system supported by the road and railway
yearbook. We get the corresponding indicators of manufacturing in- network, power grid, oil and gas pipeline network, which accelerate the
dustry by summing up the data of each sub-sector. The producer price dissolve of overcapacity problem, eventually reducing the energy in-
index (PPI), the raw material and fuel price index, and the real wages tensity of the manufacturing industry.
index of manufacturing workers are used to measure the prices of The influence mechanism of economic infrastructure on the energy
output, energy and labor, respectively. The above indexes are counted consumption and energy intensity of CMI can be divided into the fol-
in the China statistical yearbook. The price of labor is applied to stan- lowing aspects. (1) In the process of infrastructure construction, the
dardize the price of output and the price of input. The population of cooperation of multiple manufacturing industries is needed, which is
each province is derived from China population and employment sta- easy to form the agglomeration effect. In addition, the construction of
tistics yearbook. The nominal variables are transformed into real vari- infrastructure is often planned and designed by the central government.
ables (2003 = 100) to avoid the effect of prices. The statistical de- The strong credibility will attract a lot of investments, which will
scriptions of all variables are reported in Table 2. shorten the construction period, form scale effect, and effectively boost
the energy consumption of the manufacturing industry. (2) The im-
provement of infrastructure is conducive to the realization of electrical
energy replacement. For example, the operation of power grids and
Table 2
railway facilities mainly uses electricity, which significantly reduces the
The statistical description of all the variables.
consumption of fossil energy, contributing to the adjustment of energy
Variable Obs Mean Std. Dev. Min Max Units structure. (3) A large amount of excess capacity of steel and cement is
consumed with the infrastructure construction, which helps reduce the
Y 420 172.9 239.1 1.6 1410.4 Ten billion yuan
E 420 66.52 51.84 3.4 267.6 Million tons of standard coal inventory of enterprises and eliminate backward production capacity,
L 420 3.597 4.694 0.1 22.5 Million persons ultimately improving the energy consumption and decreasing the en-
K 420 42.46 48.87 1.2 319.6 Ten billion yuan ergy intensity of the manufacturing industry.
G 420 18.97 13.59 2.2 82.8 Thousand yuan/person As reported in Table 4, the short-term average elasticity of energy
Py 420 1.268 0.236 0.9 2.36 \
consumption and output to economic infrastructure in CMI is 0.289 and
Ve 420 1.674 0.462 1 3.79 \
Vl 420 2.147 0.837 1 4.37 \ −0.489 separately. The long-term elasticity of the two is positive at
0.085 and 0.426 respectively. In other words, when the capital stock of

126
B. Lin and Y. Chen Energy Policy 132 (2019) 122–131

Table 3
Estimation results of parameters in equations 8–10
Variables (1) (2) (3) (4) (5) (6)

SURE estimation method Pooled OLS estimation method

Output equation Energy equation Labor equation Output equation Energy equation Labor equation

Kt 1 5.256*** (0.0654) 0.994*** (0.0367) 0.149*** (0.00994) 5.254*** (0.0660) 0.994*** (0.0370) 0.198*** (0.0130)
Gt 1 −1.062*** (0.336) −0.533*** (0.189) −0.136*** (0.0412) −1.088*** (0.341) −0.540*** (0.191) −0.165*** (0.0530)
py −69.67*** (25.43) 22.50 (14.46) −70.88*** (26.08) 22.16 (14.61)
ve 73.57*** (17.42) −15.62 (9.881) 72.45*** (17.82) −15.93 (9.983)
py ve 7.402 (5.976) 18.15** (8.080)
Kt 1 Gt 1 0.000233 (0.000378) −0.00189*** (0.000513)
( Kt ) 2 −0.00790*** (0.00117) −0.00545*** (0.00159)
ve2 −1.274 (2.234) −5.712* (2.933)
py2 −5.261 (4.039) −11.90** (5.449)

Kt2 1 −0.000308*** (3.36e-05) −0.000350*** (4.56e-05)

Gt2 1 0.000135 (0.000593) 0.00226*** (0.000805)
Constant −24.88 (17.83) 37.18*** (10.10) 0.523 (0.829) −22.66 (18.21) 37.78*** (10.20) 0.115 (0.942)
Observations 390 390 390 390 390 390
R-squared 0.953 0.676 0.585 0.953 0.676 0.633
Breusch-Pagan test of independence: chi 2 (3) = 185.98, pr = 0.00

Note: ***indicates p < 0.01, **indicates p < 0.05, *indicate p < 0.10. Standard errors are in parentheses.

Table 4 construction is a slow process. Infrastructure construction requires a

The elasticity of energy consumption and output in CMI to economic infra- large amount of capital which will lead to an increase in government
structure. expenditure and debt level. Under the circumstance of limited social
Province Region Short-term elasticity Long-term elasticity resources, the inclination of social resources to infrastructure con-
struction will restrict the output of manufacturing industry to some
EI to EI to output EI to EI to extent in the short term. That is, transport infrastructure investment
energy energy output
cannot promote economic growth, but may engender the opposite re-
Beijing E 0.888 −0.288 0.262 0.251 sult. However, the increase of economic infrastructure will promote the
Tianjin E 0.869 −0.259 0.256 0.226 flow of energy resources between regions in the long run, which not
Hebei E 0.056 −0.081 0.016 0.071 only improve the utilization of resources in regions with abundant
Liaoning E 0.123 −0.094 0.036 0.082
energy resources, but also guarantee the stable supply in regions with
Shanghai E 0.323 −0.124 0.095 0.108
Jiangsu E 0.056 −0.026 0.016 0.022
concentrated energy consumption, thereby promoting the output
Zhejiang E 0.087 −0.064 0.026 0.056 growth (Pradhan and Bagchi, 2013). Furthermore, large-scale infra-
Fujian E 0.335 −0.115 0.099 0.100 structure investment will drive the development of energy-intensive
Shandong E 0.033 −0.020 0.010 0.018 industries such as building materials, steel and cement industry, in-
Guangdong E 0.064 −0.026 0.019 0.022
directly stimulating the energy consumption of CMI. In the long run,
Hainan E 1.473 −1.491 0.434 1.300
Shanxi C 0.099 −0.412 0.029 0.359 economic infrastructure construction will improve the availability of
Jilin C 0.168 −0.174 0.050 0.152 energy resources in remote areas and enable more people to enjoy
Heilongjiang C 0.271 −0.373 0.080 0.325 energy services. For instance, the construction of power infrastructure
Anhui C 0.130 −0.087 0.038 0.076
will expand the coverage of power grid, provide more opportunities for
Jiangxi C 0.159 −0.171 0.047 0.149
Henan C 0.045 −0.048 0.013 0.042
primary energy transmission and more possibilities for the compatible
Hubei C 0.080 −0.109 0.024 0.095 utilization of renewable energy and distributed energy, which will re-
Hunan C 0.098 −0.103 0.029 0.090 sult in the increase of energy consumption (Wang et al., 2018).
Inner Mongolia W 0.227 −0.649 0.067 0.566 An interesting phenomenon is that the elasticity of energy con-
Guangxi W 0.143 −0.184 0.042 0.160
sumption with regard to infrastructure is higher than the long-term
Chongqing W 0.528 −0.295 0.155 0.257
Sichuan W 0.064 −0.101 0.019 0.088 elasticity in the short term. For one thing, due to the weak energy
Guizhou W 0.211 −0.564 0.062 0.492 transport system, the eastern energy consumption lacks sufficient
Yunnan W 0.164 −0.378 0.048 0.329 supply guarantee, and it is arduous for the central and western energy
Shaanxi W 0.214 −0.337 0.063 0.294
sources to carry out intensive development and external transportation
Gansu W 0.178 −0.483 0.052 0.421
Qinghai W 0.737 −4.187 0.217 3.651
(Xie et al., 2018). The improvement of infrastructure has released the
Ningxia W 0.628 −2.398 0.185 2.091 production and consumption capacity that was originally restricted by
Xinjiang W 0.219 −1.028 0.065 0.896 transportation bottleneck, leading to an increase in energy consump-
Eastern China 0.391 −0.235 0.115 0.205 tion. For another, infrastructure construction usually takes several years
Central China 0.131 −0.185 0.039 0.161
to complete. The stimulation of energy consumption and the direct
Western China 0.301 −0.964 0.089 0.841
Whole China 0.289 −0.489 0.085 0.426 promotion of related manufacturing industries such as transportation
equipment industry, steel industry and cement industry are mainly re-
Note:E, C, W in parentheses represents the east, central and west region, re- flected in the construction period. The direct pull effect basically dis-
spectively. appears after building, which role has shifted to energy transmission
and resource allocation, resulting in the lower long-term elasticity.
per capita economic infrastructure raises by one percent in the long Regional discrepancy cannot be ignored. When the economic in-
term, the energy consumption and output of CMI will increase by frastructure raises by 1% in the short run, the manufacturing energy
0.085% and 0.426%, correspondingly. The payoff for infrastructural consumption in the eastern, central and western regions will increase

127
B. Lin and Y. Chen Energy Policy 132 (2019) 122–131

by 0.391%, 0.131% and 0.301% respectively. The long-term elasticity manufacturing industry in the long term, which result is similar to Tan
in the three regions are 0.115, 0.039 and 0.089, correspondingly. This et al. (2018). There is a time lag in the decline of energy intensity due to
is not the case for output. In the short run, the elasticity of manu- the long construction cycle and large investment in infrastructure. In
facturing output with regard to economic infrastructure in eastern, the short term, manufacturing enterprises will take more profits into
central and western regions is −0.235, −0.185 and −0.964. On the account and ignore the issues of energy efficiency. Nevertheless, the
contrary, when the economic infrastructure raises by 1% in the long development of infrastructure is propitious to expanding the inter-re-
term, the output of the three regions will increase by 0.161%, 0.841% gional exchanges of people and commodities, thereby promoting the
and 0.426%, separately. The marginal effect of new infrastructure in- dissemination of knowledge and technology and improving energy ef-
vestment decreases after reaching the optimal value. Eastern China has ficiency in the long run (Chakamera and Alagidede, 2018). In addition,
developed economy and well-established infrastructure, hence the in- the abundance of infrastructure ultimately reflects changes in the cost
fluence of infrastructure investment on the output of the eastern region of energy consumption across sectors, regions and time periods. The
is smaller than that of the central and western regions. Moreover, in- construction of economic infrastructure can reduce the transportation
frastructure construction will strengthen the economic ties between cost brought by space distance, improve the regional accessibility, op-
districts, improve the interval trade and factor flow, affect the industrial timize the allocation of energy resources, leading to the decline of the
layout, and facilitate the migration and transportation of energy from energy intensity. As shown in Table 5, if the capital stock of per capita
the western and central regions to the economically developed eastern economic infrastructure raises by one percent, the energy intensity will
regions, resulting in the smaller energy consumption effect of infra- fall by 0.343% on average, with the largest decline in the western re-
structure in the western and central regions (Achour and Belloumi, gion reaching 0.752%. Therefore, more economic infrastructure in-
2016). vestment in the western regions is conducive to decrease the energy
intensity of CMI.

4.2. Discussions of the impact of economic infrastructure on energy

consumption 4.3. Discussions of the changes in energy consumption and intensity of CMI

According to the above analysis, the energy intensity of CMI will The capital stock of per capita economic infrastructure in the
decline in the long run since the economic growth effect of economic eastern, central and western regions increased by 14.7%, 18.0% and
infrastructure is stronger than the energy consumption effect. The re- 19.7% correspondingly during the period 2003–2016. Thanks to the
sults are shown in Table 5 which support our analytical conclusions. In implementation of the western development strategy, the infrastructure
the short term, the economic infrastructure of each province exhibits a construction in West China has been accelerated. Table 6 describes the
positive impact on energy intensity. However, the improvement of changes of CMI brought by economic infrastructure. Short-term energy
economic infrastructure can decrease the energy intensity of consumption in the three regions increased by 5.0%, 2.0% and 5.3%
respectively as a result of the infrastructure construction. In terms of
Table 5 long-term impact, the three regions increased by 1.6%, 0.7% and 1.7%
The elasticity of energy intensity in CMI to economic infrastructure. correspondingly. But this is not the case for energy intensity. On the
contrary, the improvement of infrastructure has promoted the struc-
Province Region EI to energy intensity
tural changes and upgrading of the manufacturing industry. The energy
Short-term elasticity Long-term elasticity intensity of manufacturing industries in the eastern, central and wes-
tern regions will decrease by 1.4%, 2.2% and 13.5% separately in the
Beijing E 1.175 −0.011 long run. The “13th Five-Year Plan for Energy Development” released
Tianjin E 1.127 −0.010
by the Chinese government puts forward a target of reducing energy
Hebei E 0.136 −0.054
Liaoning E 0.218 −0.046 consumption per unit of GDP by 15% by 2020 compared with 2015.3
Shanghai E 0.447 −0.013 Accelerating infrastructure construction in the western region will not
Jiangsu E 0.082 −0.006 only promote the development of CMI, but also help reduce the energy
Zhejiang E 0.151 −0.030 intensity and accomplish government planning (Sun et al., 2019).
Fujian E 0.451 −0.002
Shandong E 0.053 −0.008
Guangdong E 0.090 −0.003 4.4. Robustness test
Hainan E 2.963 −0.866
Shanxi C 0.511 −0.330
Jilin C 0.343 −0.103 For the sake of the robustness of our results, since the physical in-
Heilongjiang C 0.643 −0.245 dicators cannot be arithmetically summed up, we adopted the two phy-
Anhui C 0.217 −0.038 sical forms, transportation infrastructure density and power infrastructure
Jiangxi C 0.330 −0.102
density, as the proxy variables of economic infrastructure respectively for
Henan C 0.093 −0.028
Hubei C 0.190 −0.072
the regression.4 Furthermore, two other robustness checks are conducted
Hunan C 0.202 −0.061 to ensure the credibility of the test. The first is to recalculate the capital
Inner Mongolia W 0.876 −0.499 stock using the constant depreciation rate and perform the regression
Guangxi W 0.327 −0.118 again. The second method is to select the time interval of the two five-year
Chongqing W 0.823 −0.102
plans that China has passed (2006–2015), and to carry out sub-period
Sichuan W 0.165 −0.069
Guizhou W 0.775 −0.430 regression. The results of the robustness checks are reported in Table 7 and
Yunnan W 0.542 −0.281 Table 8. The coefficient and significance are basically the same without
Shaanxi W 0.551 −0.231 substantial change, which proves the robustness of our results.
Gansu W 0.661 −0.369
Qinghai W 4.925 −3.434
Ningxia W 3.025 −1.906 3
13th Five-Year Plan for Energy Development. < http://www.ndrc.gov.cn/
Xinjiang W 1.247 −0.832
> 2019 [accessed 27 January 2019].
Eastern China 0.627 −0.095 4
Central China 0.316 −0.122
Transportation infrastructure density is defined as mileage of road, rail and
Western China 1.265 −0.752 inland waterway per hundred square kilometers. Power infrastructure density is
Whole China 0.778 −0.343 defined as length of transmission and distribution electric lines per hundred
square kilometers.

128
B. Lin and Y. Chen Energy Policy 132 (2019) 122–131

Table 6
Changes in energy consumption and intensity of CMI brought by economic infrastructure.
Province Region Change in energy consumption Change in output Change in energy intensity

Short term Long term Short term Long term Short term Long term

Beijing E 0.071 0.024 −0.029 0.025 0.112 −0.007

Tianjin E 0.123 0.040 −0.042 0.037 0.178 −0.003
Hebei E 0.009 0.003 −0.014 0.012 0.024 −0.009
Liaoning E 0.017 0.006 −0.017 0.014 0.036 −0.009
Shanghai E 0.024 0.008 −0.011 0.009 0.037 −0.002
Jiangsu E 0.008 0.003 −0.004 0.003 0.013 −0.001
Zhejiang E 0.012 0.004 −0.009 0.008 0.022 −0.004
Fujian E 0.061 0.020 −0.022 0.019 0.090 −0.002
Shandong E 0.005 0.002 −0.003 0.003 0.009 −0.001
Guangdong E 0.007 0.002 −0.003 0.002 0.010 −0.001
Hainan E 0.213 0.069 −0.211 0.184 0.445 −0.115
Shanxi C 0.015 0.005 −0.073 0.064 0.091 −0.058
Jilin C 0.028 0.010 −0.037 0.032 0.069 −0.022
Heilongjiang C 0.036 0.012 −0.058 0.051 0.099 −0.039
Anhui C 0.023 0.008 −0.019 0.017 0.046 −0.009
Jiangxi C 0.025 0.008 −0.030 0.026 0.057 −0.018
Henan C 0.007 0.002 −0.008 0.007 0.016 −0.005
Hubei C 0.012 0.004 −0.018 0.015 0.031 −0.011
Hunan C 0.016 0.006 −0.019 0.017 0.038 −0.011
Inner Mongolia W 0.045 0.014 −0.145 0.126 0.194 −0.112
Guangxi W 0.026 0.009 −0.038 0.033 0.067 −0.025
Chongqing W 0.094 0.031 −0.061 0.053 0.165 −0.022
Sichuan W 0.011 0.004 −0.021 0.019 0.035 −0.015
Guizhou W 0.041 0.015 −0.130 0.114 0.182 −0.098
Yunnan W 0.027 0.009 −0.070 0.061 0.101 −0.052
Shaanxi W 0.042 0.013 −0.070 0.061 0.115 −0.048
Gansu W 0.032 0.011 −0.097 0.085 0.134 −0.074
Qinghai W 0.122 0.039 −0.695 0.606 0.826 −0.568
Ningxia W 0.106 0.034 −0.413 0.360 0.527 −0.326
Xinjiang W 0.036 0.011 −0.182 0.159 0.219 −0.148
Eastern China 0.050 0.016 −0.033 0.029 0.089 −0.014
Central China 0.020 0.007 −0.033 0.028 0.056 −0.022
Western China 0.053 0.017 −0.175 0.152 0.233 −0.135
Whole China 0.043 0.014 −0.085 0.074 0.133 −0.060

Table 7
Robustness check (a).
Variables Using transportation infrastructure density Using power infrastructure density

Output equation Energy equation Labor equation Output equation Energy equation Labor equation

Kt 1 5.296*** (0.0747) 1.037*** (0.0416) 0.165*** (0.0113) 5.412*** (0.0695) 1.084*** (0.0388) 0.160*** (0.00859)
Gt 1 −0.0737 (0.0719) −0.0784* (0.0401) −0.0222*** (0.00782) −0.717*** (0.131) −0.405*** (0.0733) −0.00646 (0.0143)
py −38.72* (23.05) 32.85 (20.14) −81.97*** (22.18) 13.62 (12.50)
ve 75.16*** (18.52) −11.62 (10.43) 111.6*** (18.82) 6.384 (10.61)
py ve 7.376 (6.219) 11.38* (6.288)
Kt 1 Gt 1 0.000468*** (0.000141) 0.000879*** (0.000228)
( Kt ) 2 −0.00661*** (0.00108) −0.00654*** (0.00108)
ve2 −1.774 (2.362) −2.276 (2.387)
py2 −0.536 (4.192) −5.364 (4.220)

Kt2 1 −7.02e-05 (5.99e-05) −9.11e-05* (5.33e-05)

Gt2 1 2.00e-05 (5.25e-05) 0.000111 (0.000140)
Constant −60.32*** (12.84) 22.42*** (7.197) −4.737*** (0.598) −51.90*** (11.24) 23.80*** (6.306) −3.139*** (0.491)
Observations 390 390 390 390 390 390
R-squared 0.951 0.672 0.573 0.955 0.693 0.602

Note: ***indicates p < 0.01, **indicates p < 0.05, *indicate p < 0.10. Standard errors are in parentheses.

5. Conclusions and policy implications economic infrastructure construction will increase the energy con-
sumption and reduce the energy intensity in the long term. The energy
5.1. Conclusions intensity of CMI will also decrease since the economic growth effect of
economic infrastructure is stronger than the energy consumption effect.
This paper estimates the capital stock of economic infrastructure at Furthermore, regional differences are significant. Although the energy
the provincial level based on the perpetual inventory method. By in- consumption effect of infrastructure in the western and central regions
corporating infrastructure into the input-output system, we investigate is less than that in the eastern regions, the energy intensity of the
the influence mechanism of economic infrastructure on the energy western regions declines the most due to the infrastructure construc-
consumption and intensity of CMI. Empirical results indicate that tion. Therefore, more economic infrastructure investment in the

129
B. Lin and Y. Chen Energy Policy 132 (2019) 122–131

Table 8
Robustness check (b).
Variables Constant depreciation rate Subperiod (2006–2015)

Output equation Energy equation Labor equation Output equation Energy equation Labor equation

Kt 1 5.257*** (0.0654) 0.995*** (0.0367) 0.150*** (0.00987) 5.279*** (0.0794) 1.045*** (0.0420) 0.137*** (0.0126)
Gt 1 −1.098*** (0.339) −0.557*** (0.191) −0.142*** (0.0416) −1.351*** (0.434) −0.493** (0.231) −0.146*** (0.0550)
py −71.47*** (25.58) 21.22 (14.54) −80.35** (35.00) 44.72 (28.81)
ve 74.40*** (17.45) −15.11 (9.897) 75.21*** (21.03) −20.27 (12.66)
py ve 7.114 (5.982) 5.815 (7.443)
Kt 1 Gt 1 0.000241 (0.000379) 0.00108** (0.000492)
( Kt ) 2 −0.00790*** (0.00117) −0.00662*** (0.00124)
ve2 −1.119 (2.237) −0.775 (2.652)
py2 −5.209 (4.041) −4.232 (5.406)

Kt2 1
−0.000309*** (3.40e-05) −0.000352*** (4.29e-05)
Gt2 1 0.000166 (0.000602) −0.000590 (0.000859)
Constant −23.90 (17.81) 37.95*** (10.09) 0.613 (0.831) −15.30 (22.28) 24.53** (11.93) 0.983 (1.050)
Observations 390 390 390 270 270 270
R-squared 0.953 0.677 0.587 0.950 0.707 0.577

Note: ***indicates p < 0.01, **indicates p < 0.05, *indicate p < 0.10. Standard errors are in parentheses.

western regions is conducive to decrease the energy intensity of CMI. Adenikinju, A.F., 1998. Productivity growth and energy consumption in the Nigerian
Finally, we find the evidence that it is difficult for energy price to reflect manufacturing sector: a panel data analysis. Energy Policy 26 (3), 199–205.
Agénor, P.R., Moreno-Dodson, B., 2006. Public Infrastructure and Growth: New Channels
environmental externalities and energy scarcity. and Policy Implications. The World Bank.
Aschauer, D.A., 1989. Is public expenditure productive? J. Monet. Econ. 23 (2), 177–200.
5.2. Policy implications Bergman, M.A., 1997. The restricted profit function and the application of the generalised
Leontief and the translog functional forms. Int. J. Prod. Econ. 49 (3), 249–254.
Button, K., 1998. Infrastructure investment, endogenous growth and economic con-
Based on the above conclusions of this paper, some targeted policy vergence. Ann. Reg. Sci. 32 (1), 145–162.
recommendations to improve policy design of the government are Calderón, C., Servén, L., 2004. The Effects of Infrastructure Development on Growth and
Income Distribution. The World Bank.
proposed. Chakamera, C., Alagidede, P., 2018. Electricity crisis and the effect of CO2 emissions on
Firstly, economic infrastructure such as electricity, transportation infrastructure-growth nexus in Sub Saharan Africa. Renew. Sustain. Energy Rev. 94,
and communications are the basic link to establish market relations. 945–958.
Chen, J., Xu, C., Cui, L., et al., 2019. Driving factors of CO2 emissions and inequality
China needs to build a massive economic infrastructure to boost its
characteristics in China: a combined decomposition approach. Energy Econ. 78,
manufacturing sector. In order to reduce energy intensity and improve 589–597.
energy efficiency while ensuring energy demand and economic growth, Choi, K.H., Oh, W., 2014. Extended Divisia index decomposition of changes in energy
infrastructure investment should be tilted towards the western region, intensity: a case of Korean manufacturing industry. Energy Policy 65, 275–283.
Chowdhury, J.I., Hu, Y., Haltas, I., et al., 2018. Reducing industrial energy demand in the
and the Chinese government should also give priority to infrastructure UK: a review of energy efficiency technologies and energy saving potential in selected
construction in the western region. sectors. Renew. Sustain. Energy Rev. 94, 1153–1178.
Secondly, compared with the eastern and central regions, the in- Demetriades, P.O., Mamuneas, T.P., 2000. Intertemporal output and employment effects
of public infrastructure capital: evidence from 12 OECD economies. Econ. J. 110
frastructure construction in western China is obviously falling behind, (465), 687–712.
which means that the infrastructure construction can still play an im- Farhadi, M., 2015. Transport infrastructure and long-run economic growth in OECD
portant role in the future. As public goods, infrastructure is mainly countries. Transport. Res. Pol. Pract. 74, 73–90.
Garrone, P., Grilli, L., 2010. Is there a relationship between public expenditures in energy
provided by the government. However, the western China is econom- R&D and carbon emissions per GDP? An empirical investigation. Energy Policy 38
ically backwards and the local governments have limited financial re- (10), 5600–5613.
sources. Therefore, the central government should take on the im- Goldsmith, R.W., 1951. A Perpetual Inventory of National Wealth, vol. 14. NBER, pp.
5–73 Studies in Income and Wealth.
portant task of allocating construction funds and put the infrastructure Hang, L., Tu, M., 2007. The impacts of energy prices on energy intensity: evidence from
construction in a more prominent position in the development strategy China. Energy Policy 35 (5), 2978–2988.
framework of the western region. Hasanbeigi, A., Price, L., Fino-Chen, C., et al., 2013. Retrospective and prospective de-
composition analysis of Chinese manufacturing energy use and policy implications.
Finally, in view of the energy price distortion in China, policy-
Energy Policy 63, 562–574.
makers should actively promote the market-oriented pricing me- Hotelling, H., 1932. Edgeworth's taxation paradox and the nature of demand and supply
chanism of raw coal and crude oil, insist on adjusting electricity prices functions. J. Political Econ. 40 (5), 577–616.
and resource tax reform. Market-oriented energy price is the funda- Jin, G., 2012. The estimation of China's infrastructure capital stock. Econ. Res. J. 4, 4–14
(In Chinese).
mental way and long-term mechanism to solving energy price distor- Ju, K., Su, B., Zhou, D., et al., 2017. Does energy-price regulation benefit China's economy
tions. and environment? Evidence from energy-price distortions. Energy Policy 105,
108–119.
Kamps, C., 2006. New estimates of government net capital stocks for 22 OECD countries,
Acknowledgements 1960–2001. IMF Staff Pap. 53 (1), 120–150.
Kepplinger, D., Templ, M., Upadhyaya, S., 2013. Analysis of energy intensity in manu-
The paper is supported by Report Series from Ministry of Education facturing industry using mixed-effects models. Energy 59, 754–763.
Kimball, W.S., 1952. Calculus of Variations: by Parallel Displacement. Butterworths
of China (No.10JBG013), and China National Social Science Fund Scientific Pubs.
(No.17AZD013). Kratena, K., 2007. Technical change, investment and energy intensity. Econ. Syst. Res. 19
(3), 295–314.
Lee, M., Zhang, N., 2012. Technical efficiency, shadow price of carbon dioxide emissions,
References
and substitutability for energy in the Chinese manufacturing industries. Energy Econ.
34 (5), 1492–1497.
Achour, H., Belloumi, M., 2016. Investigating the causal relationship between transport Li, J., Lin, B., 2016. Inter-factor/inter-fuel substitution, carbon intensity, and energy-re-
infrastructure, transport energy consumption and economic growth in Tunisia. lated CO2 reduction: empirical evidence from China. Energy Econ. 56, 483–494.
Renew. Sustain. Energy Rev. 56, 988–998. Li, K., Lin, B., 2015. The efficiency improvement potential for coal, oil and electricity in

130
B. Lin and Y. Chen Energy Policy 132 (2019) 122–131

China's manufacturing sectors. Energy 86, 403–413. manufacturing industries between India and China: a perspective of embodied energy
Lin, B., Chen, G., 2018. Energy efficiency and conservation in China's manufacturing in India-China trade. Energy Policy 113, 643–650.
industry. J. Clean. Prod. 174, 492–501. Sun, C., Zhang, W., Luo, Y., et al., 2019. The improvement and substitution effect of
Lin, B., Chen, Y., Zhang, G., 2018. Impact of technological progress on China's textile transportation infrastructure on air quality: an empirical evidence from China's rail
industry and future energy saving potential forecast. Energy 161, 859–869. transit construction. Energy Policy 129, 949–957.
Miketa, A., 2001. Analysis of energy intensity developments in manufacturing sectors in Tan, R., Liu, K., Lin, B., 2018. Transportation infrastructure development and China's
industrialized and developing countries. Energy Policy 29 (10), 769–775. energy intensive industries-A road development perspective. Energy 149, 587–596.
Moyo, B., 2013. Power infrastructure quality and manufacturing productivity in Africa: a Theriault, L., Sahi, R., 1997. Energy intensity in the manufacturing sector: Canadian and
firm level analysis. Energy Policy 61, 1063–1070. international perspective. Energy Policy 25 (7–9), 773–779.
Nadiri, M.I., Mamuneas, T.P., 1994. The effects of public infrastructure and R&D capital Tian, Y., Xiong, S., Ma, X., et al., 2018. Structural path decomposition of carbon emission:
on the cost structure and performance of U.S. Manufacturing industries. Rev. Econ. a study of China's manufacturing industry. J. Clean. Prod. 193, 563–574.
Stat. 76 (1), 22–37. Wang, Y., Yan, W., Zhuang, S., et al., 2018. Does grid-connected clean power promote
Park, S.H., Dissmann, B., Nam, K.Y., 1993. A cross-country decomposition analysis of regional energy efficiency? An empirical analysis based on the upgrading grid in-
manufacturing energy consumption. Energy 18 (8), 843–858. frastructure across China. J. Clean. Prod. 186, 736–747.
Parker, S., Liddle, B., 2016. Energy efficiency in the manufacturing sector of the OECD: World Development Report 1994, 1994. Infrastructure for Development. World Bank
analysis of price elasticities. Energy Econ. 58, 38–45. Publications.
Paul, S., Sahni, B.S., Biswal, B.P., 2004. Public infrastructure and the productive per- Xie, C., Bai, M., Wang, X., 2018. Accessing provincial energy efficiencies in China's
formance of Canadian manufacturing industries. South. Econ. J. 998–1011. transport sector. Energy Policy 123, 525–532.
Pradhan, R.P., Bagchi, T.P., 2013. Effect of transportation infrastructure on economic Xie, R., Fang, J., Liu, C., 2017. The effects of transportation infrastructure on urban
growth in India: the VECM approach. Res. Transport. Econ. 38 (1), 139–148. carbon emissions. Appl. Energy 196, 199–207.
Reddy, B.S., Ray, B.K., 2011. Understanding industrial energy use: physical energy in- Xu, R., Lin, B., 2017. Why are there large regional differences in CO2 emissions? Evidence
tensity changes in Indian manufacturing sector. Energy Policy 39 (11), 7234–7243. from China's manufacturing industry. J. Clean. Prod. 140, 1330–1343.
Rosenstein-Rodan, P.N., 1943. Problems of industrialisation of eastern and south-eastern Young, A., 2003. Gold into base metals: productivity growth in the People's Republic of
Europe. Econ. J. 53 (210/211), 202–211. China during the reform period. J. Political Econ. 111 (6), 1220–1261.
Seitz, H., 2000. Infrastructure, industrial development, and employment in cities: theo- Yu, S., Zheng, S., Li, X., et al., 2018. China can peak its energy-related carbon emissions
retical aspects and empirical evidence. Int. Reg. Sci. Rev. 23 (3), 259–280. before 2025: evidence from industry restructuring. Energy Econ. 73, 91–107.
Song, M.L., Yang, L., Wu, J., et al., 2013. Energy saving in China: analysis on the energy Zhang, J., Wu, G., Zhang, J., 2004. The Estimation of China's provincial capital stock:
efficiency via bootstrap-DEA approach. Energy Policy 57, 1–6. 1952—2000. Econ. Res. J. 10 (1), 35–44 (In Chinese).
Straub, Stéphane, 2011. Infrastructure and development: a critical appraisal of the macro- Zhao, Y., Ke, J., Ni, C.C., et al., 2014. A comparative study of energy consumption and
level literature. J. Dev. Stud. 47 (5), 683–708. efficiency of Japanese and Chinese manufacturing industry. Energy Policy 70, 45–56.
Sun, C., Ma, T., Xu, M., 2018. Exploring the prospects of cooperation in the

131