The price and income elasticity of China's natural gas demand: A multi-sectoral perspective

Energy Policy 113 (2018) 332–341

Contents lists available at ScienceDirect

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

The price and income elasticity of China's natural gas demand: A multi-sectoral perspective

MARK

⁎

Yi Zhanga,b, Qiang Jia,b, , Ying Fanc a

Center for Energy and Environmental Policy Research, Institutes of Science and Development, Chinese Academy of Sciences, Beijing 100190, China School of Public Policy and Management, University of Chinese Academy of Sciences, Beijing 100049, China c School of Economics & Management, Beihang University, Beijing 100191, China b

A R T I C L E I N F O

A B S T R A C T

Keywords: Natural gas Price elasticity Income elasticity Energy demand

Natural gas as a clean, low-carbon energy will play an important role in the world's low-carbon energy transformation. In this paper, research on the elasticity of natural gas demand is surveyed, and it is found that the price and income elasticities of natural gas demand in different sectors are distinctive. In particular, this paper constructs an autoregressive distribution lag model to study the elasticity of natural gas demand in various sectors of China. The results show that, except for the residents sector, the long-run price elasticity of natural gas demand in other sectors is greater than 0, which is contrary to the estimates of developed countries. The demand for natural gas is complementary to coal in industrial and power generation sectors, which is also different from developed countries. The elasticity of natural gas demand in residents sector is lack in price elasticity but abundant in income elasticity, which is similar to the developed countries. The results also shows that natural gas and oil are substitutes for each other in the transportation sector, and natural gas and coal are substitutes for each other in service sector.

1. Introduction Natural gas is a clean, high-efficient, low-carbon energy and constitutes one of the three pillars of the world energy mix along with oil and coal. In 2016 natural gas accounted for 24.13% of the world's primary energy consumption mix (BP, 2017b), and this proportion will rise to 25.17% in 2035 (BP, 2017a). In the background of energy security and climate change, the competitive advantage of natural gas relative to other energy sources is gradually emerging. As a clean fossil energy, natural gas is more economical than coal and oil. Natural gas prices in the North American fell sharply relative to coal and oil due to the North American shale gas revolution. The natural gas industry has become an important part of energy development strategy in many countries. China's natural gas demand is expected to account for a growing share of the world total demand (Shi et al., 2017). But, at present, China's energy consumption mix is still coal-based, and there is pressure to optimize and upgrade the energy consumption mix. At the end of 2015, China committed to lowering carbon dioxide emissions per unit of gross domestic product (GDP) by 60–65% compare to that of 2005 by 2030, posing a great challenge for China's transition to a low-carbon economy. Natural gas was expected to be an important way to achieve

⁎

the goal of carbon dioxide emissions reduction and became the focus of China's 13th Five-Year Plan. The plan proposed that by 2020 the proportion of China's natural gas consumption would rise from 5.9% in 2015 to about 10%, and reach about 360 billion cubic meters. Per estimates, the average annual growth of natural gas consumption in China will exceed 30 billion cubic meters from 2016 to 2020, with an average annual growth rate of about 17%. However, China's economy has entered a new normal, with slowed economic growth and weak energy demand, and natural gas consumption in 2016 increased by only 8%, well below the average of nearly 10 years. China's future demand for natural gas is not only affected by its own supply and demand, price and industrial policies, but also by coal, oil, electricity and other related energy prices and supply capacities. China's current coal production is surplus, and its price is relatively low. International oil prices are also at low levels. However, natural gas prices continue to rise with the market-oriented reforms, so natural gas is at a disadvantage in competition with coal and oil. There are many factors including economic growth, structural change in markets, environmental regulations, price and institutional changes contribute to the uncertainties in Chinese gas market (Shi et al., 2017). Thus, promoting the growth of natural gas consumption demand and achieving consumption targets is a challenge. In order to encourage, guide and standardize the use of natural gas,

Corresponding author at: Center for Energy and Environmental Policy Research, Institutes of Science and Development, Chinese Academy of Sciences, Beijing 100190, China. E-mail addresses: [email protected], [email protected] (Q. Ji).

https://doi.org/10.1016/j.enpol.2017.11.014 Received 25 July 2017; Received in revised form 24 October 2017; Accepted 6 November 2017 0301-4215/ © 2017 Elsevier Ltd. All rights reserved.

Energy Policy 113 (2018) 332–341

Y. Zhang et al.

input changes. It is symmetry, as shown in formula 2. Uzawa (1962) deduced that Allen's elasticities of substitution can be calculated from the cost function. C represents the cost function; Cij represents the mixed derivative of the cost function with respect to the prices of element i and the element j; Ci and Cj represent the partial derivative of the cost function with respect to the price of element i and j, respectively; and sj is the proportion of the input element j to the total cost, namely the cost share of input element j. Obviously, when the price elasticity of demand is the same, the element whose cost share is smaller the Allen's elasticities of substitution is larger, and the Allen's elasticities of substitution do not have an intuitive explanation.

optimize the energy mix, develop a low-carbon economy, promote energy conservation and improve the proportion of natural gas in the primary energy consumption mix, in 2012, the National Development and Reform Commission promulgated the “Natural Gas Utilization Policy”. The document classifies natural gas users into four categories: priority, allowing, restricted and prohibited, which covers different sectors, including industry, power generation, transportation, public services and residents. For different sectors, natural gas and other energy sources may be substitutional or complementary, and the different relationship of substitution or complement dictates the need for different policies to promote natural gas use. Natural gas demand price elasticity can accurately reveal the relationship between natural gas demand and coal, oil and electricity, providing a theoretical basis for policy measures to promote natural gas consumption. At present, there is little research on the elasticity of natural gas demand in China, especially about different sectors. Based on the literature review of natural gas demand elasticity, this paper investigates the price and income elasticities of natural gas demand for China's subsectors and reveals the substitutional or complementary relationship between natural gas and other major energy sources. It provides the basis for carrying out scientific and rational measures to promote natural gas consumption. In the second part of this paper, the research on natural gas elasticity is reviewed, and the existing conclusions about natural gas elasticity are compared with each other. The third part establishes the autoregressive distribution lag model of natural gas demand. The fourth part systematically studies the elasticity of natural gas demand in China's industrial sector, the power generation and supply sector, the transportation sector, service sector and the residents sector, and reveals the influence of price and income on natural gas demand. The fifth part elaborates on the main research conclusions and their policy implications.

x

Allen′s elasticities of substitution:σija =

∂ ln xi j

pj

CCij

=

Ci Cj

∂ ln p

=

ηij Sj

(2)

i

(σijm )

The Morishima elasticities of substitution are calculated by taking the percent change in the ratio of two demand factors ( xi ) and xj

dividing it by a percent change in price ( pj ), reflecting the impact of a factor price change on the relative input of the two elements, as shown in formula 3. Frondel (2011) pointed out that when the element's ownprice elasticity is less than 0 and small enough, the Morishima elasticities of substitution make it easier to conclude substitution compared to the cross-price elasticity and Allen's elasticities of substitution. x

Morishima elasticities of substitution:σijm =

∂ln xi j

∂lnpj

= sj (σija − σ jja) = ηij − ηjj (3)

2. Literature review

In practice, energy price policy, energy control measures and energy demand analyses are more concerned with the direct impact of energy price changes on related demand, and demand elasticity is the basis of Allen's elasticities of substitution and the Morishima elasticities of substitution. So this paper mainly focuses on the price elasticity of natural gas demand.

2.1. Elastic definition and comparison

2.2. Research methods on demand elasticity

Elasticity as commonly used in research has included the price elasticity of demand, Allen's elasticities of substitution and the Morishima elasticities of substitution (Griffin, 1977; Uri, 1978; Pindyck, 1979; Renou-Maissant, 1999; Urga and Walters, 2003; Serletis et al., 2010, 2011). There are differences as well as relations among them. Price elasticity of demand measures how sensitive the demand for a good is to changes in related prices. Allen's elasticities of substitution reflect how sensitive the ratio of two input factors is to changes in their price ratio. The Morishima elasticities of substitution reflect how sensitive the ratio of two input factors is to changes in one of their prices. Frondel (2004) deduces the quantitative relationship between the price elasticity of demand, Allen's elasticities of substitution and the Morishima elasticities of substitution, as shown in formulas 1, 2 and 3. The price elasticity of demand (ηij) is calculated by taking the percent change in quantity of a commodity demanded (xi) and dividing it by a percent change in price (pj), which reflects the response of factor demands to changes in the price of the relevant element. When i=j, it is called own-price elasticity, and when i‡j, it is called cross-price elasticity. As shown in formula 1, it is a measure of absolute substitution (Frondel, 2011). In formula 1, if the denominator is replaced by an income factor, the income elasticity of demand can be obtained.

The studies on demand elasticity can be divided into three categories based on the research method. The first kind of method establish a cost function or a cost share function deduced from a production function, then uses the translog cost function or the linear logit cost share function to study the price elasticity of demand. The second kind of method studies the short-run demand elasticity and long-run demand elasticity through the error correction model and cointegration analysis, respectively. The third kind of method examines the energy price elasticity and income elasticity of demand by establishing the energy demand function. The translog cost function is deduced from the production function according to dual theory based on a series of assumptions such as weak separability, homothetic, optimization and so on (Griffin, 1977; Uri, 1978; Pindyck, 1979; Renou-Maissant, 1999; Serletis et al., 2010). The cost function should satisfy the regular condition for the homogeneous, nondecreasing, continuous concave function of the factor price. Diewert and Wales (1987) point out that the regular conditions of the translog cost function impose an unacceptable premise constraint on price elasticity and cross-price elasticity. Another flaw in the translog cost function is that the impact of income on demand cannot be considered. The linear logit function is used to express the cost share derived from the translog cost function, ensuring the cost share is non-negative (Urga and Walters, 2003; Considine, 1989; Jones, 1995). Jones (1995) compared the logarithmic cost function and the linear logit cost share function based on US industrial sector data. When considering the time trend term, the linear logit cost share function performed better. There is no significant difference in the price elasticity of the two functions. These two models require an introduction of time factors into the functions to measure the short-run and long-run price elasticities of demand.

Price elasticity of demand:ηij =

∂ ln x i ∂ ln pj

(1) (σija )

are calculated by taking the Allen's elasticities of substitution percent change in the ratio of two demand factors ( xi ) and dividing it by xj

p

a percent change in their price ratio ( j ), reflecting that the impact of pi

the relative price changes of the two related elements on their relative 333

Energy Policy 113 (2018) 332–341

Y. Zhang et al.

Short-run price elasticity of demand measures demand response to a price change in the period which the price change occurs. Long-run price elasticity of demand measures the total demand response to a price change when the system returns to a new equilibrium after the change occurs (Donnelly, 1987). Hunt and Manning (1989), Bentzen and Engsted (1993) studied the energy demand elasticity of the British and Danish, respectively. An error correction model was applied to estimate short-run energy demand elasticity, and cointegration analysis was applied to estimate long-run energy demand elasticity. In the latest studies on the elasticity of natural gas demand, they usually used demand function (Asche et al., 2008; Erdogdu, 2010; Dagher, 2012; Yu et al., 2014). The demand function typically considered natural gas prices, the relevant substitution energy prices, climate factors, economic development levels or income levels, etc. Demand functions frequently used can be divided into two categories. One is the partial adjustment model in which the explanatory variables contain the first-order leg of the dependent variable in addition to the individual variables, and the long-run elasticity obtained is equal to the short-run elasticity divided by the adjustment rate. That is, model imposes the assumption that the response speed is the same, and this is where there is controversy. The other is an autoregressive distributed lag model, which contains multiple lags of the dependent variable and the independent variable, and short-run elasticity and long-run elasticity can be obtained directly, overcoming the defect of the partial adjustment model.

Fig. 1. Price elasticity of natural gas demand in power generation sector.

2.3. Existing research conclusions about the elasticity of natural gas demand The research literature on the elasticity of natural gas demand can be divided into two categories according to the calibre of the study data statistics. One focuses on the gross demand elasticity, and the other focuses on the elasticity for different sectors, which include power generation, industry, manufacturing, business and residents. There are some differences in the prices and substitutability of natural gas demand for different sectors, so research on the elasticities of natural gas demand in different sectors is more reasonable and more scientific. Al-Sahlawi (1989) reviewed a number of studies on the elasticity of natural gas demand in different sectors, noting that the reason for the different conclusions may be different sample periods, different data sources, geographical location differences, structural changes and different types of demand. Table 1 presents the abbreviation of all the mentioned countries in Fig. 1 to Fig. 5. Griffin (1977), Uri (1978) and Serletis et al. (2010) used static models to study the price elasticity of natural gas demand in the power generation industry. The results, shown in Fig. 1, reveal that the differences between different countries and the same country at different time periods are relatively large, such as the own-price elasticity of US

Fig. 2. Price elasticity of natural gas demand in industrial sector (static model).

Table 1 Abbreviation of countries for Fig. 1 to Fig. 5. Serial number

Abbreviation name

Full name

1 2 3 4 5 6 7 8 9 10 11 12

US FR WDE GB JP CA IT NL DK DE AT BE

The United States France West Germany the United Kingdom Japanese Canada Italy Netherlands Denmark Germany Austria Belgium

Fig. 3. Long-run price elasticity of natural gas demand in industrial sector (dynamic model).

334

Energy Policy 113 (2018) 332–341

Y. Zhang et al.

Europe. The conclusions, similar to that of Renou-Maissant (1999) show that both the short-run and long-run are inelastic. Regarding the elasticity of natural gas demand in the residents sector, Asche et al. (2008) used a dynamic model to study several European countries. Short-run and long-run price elasticity and income elasticity are shown in Figs. 4 and 5, respectively. Both short-run and long-run price elasticities of natural gas demand are small, but income elasticity is significantly greater than price elasticity, especially longrun income elasticity. This shows that residents’ natural gas demand is insensitive to the price, but the increase in income will significantly promote natural gas consumption. Alberini et al. (2011) examined the elasticity of residents’ natural gas demand for 50 metropolises in the United States. With an estimated value between −0.693 and −0.566, it is also lack in elasticity. With regard to the elasticity of gross natural gas demand, Serletis et al. (2011) studied that of 15 countries based on the standard quadratic cost function. They included six high-income countries (Canada, France, Japan, Italy, the United Kingdom and the United States); five upper-middle to high-income economies (Poland, Hungary, Mexico, Turkey and Venezuela); and four lower-middle to low-income economies (China, India, South Africa and Thailand). The conclusions indicate that long-run inter-fuel elasticities of substitution are generally significantly higher than their short-run counterparts. Natural gas and coal are mutually substitutable in some countries and complementary in others; there aren’t any significant differences among the three groups of countries in terms of inter-fuel substitution either in short-run or long-run. That is, inter-fuel substitution seems to be independent of the level of economic development. Bilgili (2014) investigated the elasticity of natural gas demand of eight OECD countries, the estimated elasticity is slightly greater than 1. Burke and Yang (2016) used national-level data of 44 countries to estimate the price and income elasticities of natural gas demand. The average long-run price elasticity of natural gas demand is around −1.25, and the average long-run income elasticity of natural gas demand is greater than 1. As can be seen from Fig. 1 to Fig. 5, the price elasticity of natural gas demand in the power generation sector is greater than that in the industrial sector in general, and the price elasticity of natural gas demand in the industrial sector is greater than that in the residents sector. The long-run income elasticity of natural gas demand in the residents sector is generally greater than 1. There are significant differences in the elasticity of natural gas demand among different sectors, so research on sector levels can better reflect the actual situation. To date, the research on China's natural gas industry has mainly focused on supply and demand forecasts, price reforms, industrial reforms, supply safety, liquefied natural gas (LNG), and natural gas power generation and so on. Therefore, research on the elasticity of natural gas demand in China is still relatively sparse. Xu and Wang (2010), Li et al. (2011), Zhang and Yang (2015) and Shaikh and Ji (2016) predicted the short-run, medium-term and longrun natural gas demand in China. Lin and Wang (2012), and Wang et al. (2016) studied the peak production of natural gas in China. Huo (2011), Hu and Dong (2015), Paltsev and Zhang (2015) and Zhang et al. (2017) studied the problems related to natural gas price reform in China. Shaikh et al. (2016b, 2017a) and Shaikh et al. (2017b) studied the safety of the natural gas supply in China. Shi et al. (2010), Lin et al. (2010) and Shaikh et al. (2016a) studied the China's LNG market. Skeer and Wang (2006), Kahrl et al. (2013), Arora et al. (2016) and Zhu et al. (2016) studied the effect of relevant factors on China's natural gas power generation costs. Shi et al. (2017) analysed the uncertainties of China's gas market. Yu et al. (2014) studied the price elasticity and income elasticity of natural gas demand in China's urban residents, finding a price elasticity of −1.431 and an income elasticity of 0.207. Sun and Ouyang (2016) studied the price and expenditure elasticities of Chinese residents energy demand, with natural gas demand at an own-price elasticity of −0.7794 and an expenditure elasticity of 0.7963.

Fig. 4. Short-run elasticity of natural gas demand in residents sector.

Fig. 5. Long-run elasticity of natural gas demand in residents sector.

natural gas demand; the maximum value of it is −0.136, and the minimum value of it is −1.4639. Natural gas and fuel oil is mostly mutually substitutable, so is natural gas and coal. Pindyck (1979) and Renou-Maissant (1999) studied a number of countries on the price elasticity of natural gas demand for the industrial sector. Pindyck (1979) used a static model to study the price elasticity of natural gas demand in industry based on the data of 1959–1973. As shown in Fig. 2, the conclusion indicates that the differences for the own-price elasticity of natural gas demand and the cross-price elasticity between natural gas and coal of different countries are relatively large. European countries’ and Japan's natural gas own-price elasticities are from −1.3 to −2.31, while those of the United States and Canada, which are −0.52 and −0.33, are relatively low. Cross-price elasticities between natural gas and fuel oil is relatively small, and so is between natural gas and electricity. Natural gas and coal are mutually substitutable; natural gas and fuel oil are complementary, as are natural gas and electricity. Renou-Maissant (1999) used a dynamic model to study the long-run price elasticity of industrial natural gas demand in seven OECD countries based on the data of 1960–1993, as shown in Fig. 3. The conclusion shows that natural gas demand is inelastic in the long run, though there are some small differences among different countries. Natural gas and oil are mutually substitutable, as are natural gas and electricity, which differs from what Pindyck (1979) found. Andersen et al. (2011) and Steinbuks (2012) used different methods to study the price elasticity of natural gas demand in manufacturing in 335

Energy Policy 113 (2018) 332–341

Y. Zhang et al.

Based on the previous literature, this paper studies the price and income elasticity of natural gas demand in China, analyses the short-run and long-run elasticity characteristics of natural gas demand in different sectors, and provides a theoretical basis and reference for the reform of China's natural gas market.

S ηinco = θ3,0

3. Models and data

Based on the above model, this paper measures the elasticities of natural gas demand in China's industry, electric power generation and supply, transportation, services and residents.

Long-run income elasticity of natural gas demand: v

L ηinco =

3.1. Model method In this paper, the natural gas demand elasticity of China's each sector is studied by establishing the natural gas demand function. The research of Pesaran and Shin (1999) showed that the autoregressive distributed lag (ARDL) approach yield consistent estimates of the coefficients irrespective of whether the regressors are I(1) or I(0). Bentzen and Engsted (2001) used the ARDL approach to estimate the energy demand function of Danish resident, and compared the estimates of ARDL approach with that of cointegration technique and error correction model (ECM). It turns out that the results of ARDL approach and the cointegration/ECM approach are very similar. Cuddington and Dagher (2015) pointed out that the ARDL model is the general form of the ECM and the partial adjustment model. The ECM and the partial adjustment model impose unreasonable constraints on short-run elasticity and long-run elasticity. One can avoid such restraints by using ARDL approach, whose lag terms were determined by the Akaike information criterion (AIC) or Schwartz information criteria (SIC). The research showes that the ARDL approach is an effective method to construct the energy demand function. This paper constructs an ARDL model to reveal the relationship between natural gas demand and natural gas prices, substitution energy prices and income levels for each sector. Climatic factors are not included in the demand functions because their coefficients are statistically not significant. Referring to Bentzen and Engsted (2001), we determine the lag number for variables according to F-test, t-test and finally by AIC. Dagher (2012) also finds that the long run equilibrium is reached around 18 months after a change in price or income has occurred. The general model form is as shown in Eq. (4): m

lnQt = θ +

n

d

j=0

q

s=1 k=0

∑ θ3,llnVt−l + ε l=0

(4)

m Where 1− ∑i = 1 θ0, i > 0 , Qt in year t ; Pt − j represents

represents the amount of natural gas demand the real price of natural gas in year t-j; Ps, t − k represents the real price of the substitution energy s for natural gas in year t-k; Vt − l represents the real income level in year t-l; and ε is a random error term. Short-run own-price elasticity of natural gas demand:

S ηself = θ1,0

(5)

Long-run own-price elasticity of natural gas demand: n

L ηself =

∑ j = 0 θ1, j m

1− ∑i = 1 θ0, i

(10)

(6) 4. Specific models and results of every industry

Short-run cross-price elasticity of natural gas demand with regard to substitution energy s:

4.1. Natural gas demand elasticity of industrial sector

S ηacro = θ2, s,0

(7) Natural gas consumption in the industrial sector is substitutional or complementary with coal, fuel oil and electricity. When constructing an ARDL model of the industrial sector natural gas demand, we mainly consider the real prices of industrial natural gas consumption, coal,

Long-run cross-price elasticity of natural gas demand with regard to substitution energy s: q

L ηacro =

m

1− ∑i = 1 θ0, i

The final energy consumption data of sub-sectors for natural gas comes from the “China Energy Statistics Yearbook” (except for the electric power generation and supply sector, which uses the data of transformation input). Annual natural gas price data come from the natural gas pricing document of National Development and Reform Commission.1 Prices of fuel oil, coal, industrial electricity, liquefied petroleum gas and urban residents’ disposable income data come from the “CEIC Data”. Crude oil prices come from the “International Statistical Yearbook”. The GDP per capita come from the “China Statistical Yearbook”. Since 1992, China has implemented classified natural gas prices for ex-factory price, which classifies users into urban residents living gas consumption, urban business gas consumption, fertilizer gas consumption and other industrial gas consumption. In December 2005, China simplified ex-factory price classification, combining urban residents living gas consumption, urban business gas consumption and small industrial user consumption which is supplied through the city natural gas pipeline network into city gas. After the reclassification of gas consumption, there were urban gas consumption, fertilizer gas consumption and industrial gas consumption which were supplied directly. In June 2013, the National Development and Reform Commission carried out a new natural gas price regulation policy, which supervised the city gate price of gas instead of ex-factory price. Therefore, the natural gas price used in this paper is limited to the exfactory price of natural gas from 1992 to 2012. Relevant monthly price data is converted to annual data by simple average conversion. Nominal price or value data are converted to real price or value data based on 1992. All data are logarithmically I(0) or I (1). The evolution of China's natural gas consumption mix is shown in Fig. 6. The total consumption of natural gas is 15.88 billion cubic meters in 1992 and 193.18 billion cubic meters in 2015. The industrial sector consumed the most part of the total natural gas, whose proportion is more than 60% all the time. In the industrial sector, natural gas consumption for electric power generation increases quickly, whose share grows from 4.56% to 23.77% in the total industrial sector natural gas consumption. The residents sector ranked second and its share in the total natural gas consumption reaches about 20% in the past ten years. Transportation sector's natural gas consumption grows fast and its ratio grows from 0.28% in 1992 to 12.30% in 2015. Service sector consumed 9.67 billion cubic meters of natural gas in 2015, which is only 0.34 billion cubic meters in 1992.

v

+

∑l = 0 θ3, l

3.2. Data sources

∑ θ0,ilnQt−i + ∑ θ1,jlnPt−j + ∑ ∑ θ2,s,klnPs,t−k i=1

(9)

∑k = 0 θ2, s, k m

1− ∑i = 1 θ0, i

(8) 1

Short-run income elasticity of natural gas demand: 336

Data for missing years are supplemented by linear interpolation.

Energy Policy 113 (2018) 332–341

Natural gas consumpon (billion cubic meters)

Y. Zhang et al. 200

oil processing, chemical manufacturing, metal smelting and calendaring, manufacturing of non-metallic mineral products, and electricity and heat production and supply are the industries that consume more natural gas. Both the short-run and long-run own-price elasticities of natural gas demand in the industrial sector are greater than 0. This result is different from that of Pindyck (1979) and Renou-Maissant (1999) who found that the own-price elasticities are negative. This is mainly because since the reform began in 1978 and with the national economy having sustained rapid development, energy supply is tense mostly of the time. Natural gas pricing is controlled by the government. Low prices lead to market distortions and resource mismatches, and the natural gas market does not reach equilibrium, so natural gas price increases do not lead to reductions in industrial natural gas final consumption. Natural gas and coal are complementary in the industrial sector. This is also inconsistent with the conclusion in Pindyck (1979) who found natural gas and cola are substitutes as shown in Fig. 2. This is related to China's long-run coal-based energy mix. Coal resources are relatively rich, and natural gas resources are scarce. Thus, industrial sector coal consumption is far more than that of natural gas. Natural gas consumption in 1992 was 5.15% of coal. Natural gas consumption has experienced rapid growth since 2000, but by 2012, natural gas consumption is only 17.55% of coal, with natural gas playing a supplemental role to coal. For industrial sector, natural gas consumption and electricity consumption are substitutes for each other, as are natural gas and fuel oil. This conclusion is consistent with Renou-Maissant (1999), who found that in most developed countries natural gas and electricity presents substitution. But the cross-price elasticity of natural gas demand with regard to electricity in this study is much bigger than that of RenouMaissant (1999). However, the results of Pindyck (1979) indicate that both natural gas and electricity and natural gas and fuel oil are complementary for each other. The difference maybe come from the different research methods, study object and sample period. Natural gas as a high-quality industrial fuel is a superior substitution to electricity and fuel oil. Natural gas heating has a higher energy efficiency than electric heating; natural gas combustion is cleaner than fuel oil, and less pollutants are emitted. Industrial natural gas consumption demand is rich in income elasticity. That is, the growth of the GDP per capita strongly stimulates the natural gas demand in industrial sector, as GDP per capita growth of 1% will lead to a natural gas consumption increase of 2.307%.

180 160 140 120 100 80 60 40 20 0 1992

1997

2002 Industry

Transportaon

2007 Service

2012

2015

Residents

Fig. 6. The natural gas consumption of each sector.

industrial electricity and that of fuel oil. The higher the level of economic development, the greater the energy consumption needs, so we also consider the impact of real GDP per capita on natural gas demand. We determine whether the ARDL model contains the autoregressive term and the lag number according to the procedure following Bentzen and Engsted (2001) by F-test, t-test and finally AIC. The following equation is obtained:

lnQin, t = Cin + α1lnPin, t + α2lnPin, t −1 + α3lnPc, t + α4lnPc, t −1 + α5lnPe, t + α6lnPfo, t + α 7lnPfo, t −1 + α8lnGDPt + εin

(11)

Where Qin, t : final energy consumption of natural gas in industry (excluding the raw materials) in year t; Cin : constant term; Pin, t : real natural gas price of industry consumption in year t; Pc, t : real coal price in year t; Pe, t : real electricity price of industry consumption in year t; Pfo, t : real fuel oil price in year t; GDPt : real GDP per capita in year t; and εin : the random error term. Based on formulas 5–10, for the industrial sector we can obtain from Table 2 that the short-run and long-run own-price elasticities of natural gas demand are 0.222 and 0.847 respectively. The short-run and longrun cross-price elasticities of natural gas demand with regard to coal are −0.694 and −1.093, respectively. The cross-price elasticity of natural gas demand with regard to electricity is 1.884 (short-run and long-run are the same). The short-run and long-run cross price elasticities of natural gas demand with regard to fuel oil are 0.323 and 0.519, respectively. The income elasticity of natural gas demand is 2.307. Natural gas final consumption in the industrial sector is mostly chemical raw materials and industrial fuels. This study eliminates the consumption of industrial feedstocks due to the relatively weak substitutability of feedstock use. Natural gas is a high-quality industrial fuel widely used in industrial boilers and kilns. Oil and gas extraction,

4.2. Natural gas demand elasticity of electric power generation and supply sector The electric power generation and supply sector is the energy transformation sector. The transformation input is much larger than its final energy consumption, and the demand elasticity of the transformation input of natural gas in this sector is more valuable than the demand elasticity of its final energy consumption. The natural gas demand of the electric power generation and supply sector has a substitutional or complementary relationship with respect to coal and fuel oil and is affected by the level of economic development. To construct an ARDL model of this sector, we consider factors including the real price of natural gas for electric power generation and supply sector use, the real price of coal, the real price of fuel oil and the real GDP per capita. According to the parameters and equation significant, the fuel oil real price does not enter the model, and we get the following equation:

Table 2 Coefficient estimation of natural gas demand equation in industrial sector. Coefficient

Estimated value

Standard deviation

t-Statistic

P-value

Cin α1 α2 α3 α4 α5 α6 α7 α8

−28.197 0.222 0.625 −0.694 −0.399 1.884 0.323 0.196 2.307

7.226 0.279 0.162 0.187 0.267 0.814 0.057 0.071 0.530

−3.902 0.796 3.850 −3.709 −1.497 2.314 5.697 2.756 4.351

0.003 0.443 0.003 0.003 0.163 0.041 0.000 0.019 0.001

lnQelec, t = Celec + β1lnQelec, t −1 + β2lnPelec, t + β3lnPc, t + β4lnPc, t −1 + β5lnGDPt + εelec

(12)

Where Qelec, t : natural gas transformation input quantity of electric power generation and supply sector in year t; Celec : constant term; Pelec, t :

Adjusted R2=0.99，F statistics=271.37，AIC=−2.19.

337

Energy Policy 113 (2018) 332–341

Y. Zhang et al.

Table 3 Coefficient estimation of natural gas demand equation in electric power generation and supply sector. Coefficient

Estimated value

Standard deviation

t-Statistic

P-value

Celtm β1 β2 β3

−36.625 0.162

6.240 0.139

−5.869 1.165

0.000 0.265

3.094

0.666

4.648

0.000

−1.084

0.458

−2.365

0.034

β4

−0.976

0.344

−2.840

0.014

β5

3.636

0.807

4.505

0.000

Table 4 Coefficient estimation of natural gas demand equation in transportation sector. Coefficient

Estimated value

Standard deviation

t-Statistic

P-value

Cts λ1 λ2 λ3 λ4 λ5

−10.447 0.650 −0.199 3.565 −2.013 1.058

8.140 0.150 1.100 1.183 1.051 0.453

−1.283 4.331 −0.181 3.012 −1.915 2.335

0.222 0.000 0.859 0.010 0.078 0.036

Adjusted R2=0.97，F statistics=133.21，AIC=0.62.

Adjusted R2=0.97，F statistics=165.41，AIC=−0.53.

constant term; Pts, t : real natural gas price of transportation in year t; Po, t : real price of crude oil in year t; and εts : the random error term. The results for the transportation sector shown in Table 4 show that the short-run and long-run own-price elasticities of natural gas demand are −0.199 and 3.866, respectively, and the short-run and long-run cross-price elasticities of natural gas demand with regard to crude oil are 1.058 and 3.023, respectively. A price increase will restrain natural gas demand in the transportation sector in the short run, but in the long run, it does not affect the growth of its consumption. This is mainly due to a surge in the number of natural gas cars boosting the demand for natural gas. Since their breakthrough invention in the late 1980s, natural gas vehicles have experienced explosive growth in China, climbing from 1.1 million vehicles in 2010 to about 5 million (of which about 200,000 are LNG vehicles) in 2015. The Energy Conservation Law of the People's Republic of China, which was implemented in 2008, stipulates that the state encourages the development and promotion of clean fuels and petroleum substitution fuels for the use of transportation. About 70% of the natural gas vehicles can be powered by both gasoline and natural gas. Wang et al. (2015) and Hao et al. (2016) show that the fuel cost of natural gas vehicles is much lower than that of gasoline and diesel vehicles. The promotion of dual fuel vehicles makes the price elasticity of natural gas demand with respect to oil flexible, whether in the short run or the long run.

real natural gas price of the electric power generation and supply sector in year t; Pc, t : real price of coal in year t; GDPt : real GDP per capita in year t; and εelec : the random error term. Table 3 presents the elasticity results for the electric power generation and supply sector based on formula 5–10. The results indicate that the short-run and long-run own-price elasticities of natural gas demand are 3.094 and 3.692, respectively. The short-run and long-run cross-price elasticities of natural gas demand with regard to coal are −1.084 and −2.458, respectively. The short-run and long-run elasticities of natural gas demand with regard to GDP per capita are 3.636 and 4.339, respectively. Both the short-run and long-run own-price elasticities of natural gas demand in the electric power generation and supply sector of China are positive and greater than 1. It is surprising results which is contrary to most previous research such as Griffin (1977), Uri (1978) and Serletis et al. (2010). The probably reason for our abnormal result is that the historical price of natural gas in China is low, and gas power generation has the advantages of flexible operation, high availability, quick startup and less construction space needed, allowing gas power generation to develop rapidly. The rise in natural gas prices did not affect the growth of its demand. Natural gas consumption and coal consumption in the electric power generation and supply sector are complementary. This conclusion is contrary to Uri (1978) and Serletis et al. (2010). However, the conclusion of Serletis et al. (2010) also shows that natural gas and oil in electric generation sector is mutually complementary in the United States. These unexpected results are generally based on specific industrial development stage and energy policies. During the sample period in this study, coal prices had undergone a descending trend which increase the demand of coal. However, in the meantime, natural gas demand also increased stimulated by government's industry policies and energy development strategies. This may contribute to their complementary relationship in industrial and electric sector. With the growth of GDP per capita, the power generation capacity has increased and the demand for electric power of various sectors has increased, leading to an increase in demand for natural gas in the electric power generation and supply sector.

4.4. Natural gas demand elasticity of service sector Natural gas consumption demand of services (excluding transportation) mainly relates to coal, kerosene, gasoline, fuel oil and electricity, and there may be a substitutional or complementary relationship between them. The level of economic development also may affect the consumption demand. Because there is not enough kerosene and gasoline price data, we use crude oil prices as the proxy variable. In addition, the price of electricity in service sector in different regions is quite diverse, and the relevant data are lacking. In the establishment of the natural gas demand equation for services, we take into account the real prices of natural gas, coal, crude oil and fuel oil in services, the real GDP per capita. The high linear relationship between the real price of crude oil and the real price of coal leads to serious collinearity. The model considers the real coal price. According to F-test and t-test, the real GDP per capita do not enter the model. We get the following equation:

4.3. Natural gas demand elasticity of transportation sector Transportation natural gas consumption is mainly related to gasoline and diesel and may be affected by the level of economic development. To construct an ARDL model of this industry, we consider the factors of the real price of natural gas in the transportation sector; the real price of crude oil — which is a proxy variable for gasoline and diesel — and take into account the real GDP per capita. According to Ftest and t-test, the real GDP per capita do not enter the model. We get the following equation:

lnQse, t = Cse + κ1lnQse, t −1 + κ2lnPse, t + κ3lnPse, t −1 + κ 4lnPc, t + κ5lnPc, t −1 + εse (14) Where Qse, t : natural gas consumption in services in year t; Cse : constant term; Pse, t : real natural gas price of services in year t; Pc, t : real price of coal in year t; and εse : the random error term. The results for service sector shown in Table 5 show that the shortrun and long-run own-price elasticities of natural gas demand are −1.004 and 5.730, respectively; the short-run and long-run cross-price elasticities of natural gas demand with regard to coal are 3.864 and 3.521, respectively. Due to the lack of relevant data, the demand equation does not directly consider the impact of gasoline, diesel and

lnQts, t = Cts + λ1lnQts, t −1 + λ2lnPts, t + λ3lnPts, t −1 + λ 4lnPts, t −2 + λ5lnPo, t + εts (13) Where Qts, t : natural gas consumption in transportation in year t; Cts : 338

Energy Policy 113 (2018) 332–341

Y. Zhang et al.

Table 5 Coefficient estimation of natural gas demand equation in service sector. Coefficient

Cse κ1 κ2 κ3 κ4 κ5

Estimated value −29.532 0.407 −1.004 4.402 3.864 −1.776

Standard deviation

t-Statistic

10.233 0.203 1.420 1.347 1.100 0.977

−2.886 2.004 −0.707 3.269 3.514 −1.818

Table 7 Summary of natural gas demand elasticity in five subsectors. P-value 0.012 0.065 0.491 0.006 0.003 0.090

I NG II

NG SR 0.222

NG

III

NG

IV

NG

Adjusted R2=0.91，F statistics=40.55，AIC=1.20.

electricity on natural gas demand, making the accuracy of elasticity results reduced. A rise in price will restrain the demand of natural gas in the short run but will not affect the demand increase in the long run. Natural gas and coal have a strong substitutional relationship in the short and long run. Natural gas consumption by service sector is found mainly in the wholesale, retail and accommodation catering markets for food production or cooking, where it enjoys the advantages of being cleaner, more efficient and safer compared to gasoline, diesel, fuel oil and coal.

V

NG

LR 0.847 NG SR 3.094 NG SR −0.199 NG SR −1.004 NG −0.233

C SR −0.694 LR 3.692 LR 3.866 LR 5.730 LPG 0.282

E LR −1.093 C SR −1.084 O SR 1.058 C SR 3.864 C SR 0.334

1.884 LR −2.458

FO SR 0.323 GDP SR 3.636

GDP LR 0.519

2.307

LR 4.339

LR 3.023 LR 3.521 DI LR 0.061

2.501

Note: I: industrial sector, II: electric power generation and supply sector, III: transportation sector, IV: service sector, V: residents sector. SR: short run, LR: long run. NG: natural gas, C: coal, E: electricity, FO: fuel oil, GDP: GDP per capita, O: crude oil, LPG: liquefied petroleum gas, DI: disposable income of urban residents.

relationship between natural gas and liquefied petroleum gas and between natural gas and coal. He et al. (2013) indicated that the affordability of Beijing's residents is higher than the actual price. Residents sector natural gas consumption is mainly used for cooking activities, and when the price of natural gas rises, residents will reduce the consumption of natural gas appropriately and increase the use of relatively cheap energy. Natural gas and liquefied petroleum gas (LPG) do not have a substitutional relationship for single families; families with natural gas pipelines are no longer using liquefied petroleum gas (LPG), because the two gas stoves are different. Otherwise, natural and liquefied petroleum gas are generally substitutable in the residents sector. Coal is mainly used by residents to heat and cook and is a substitute for natural gas. As income levels increase, the quality of life increases, and the corresponding energy consumption will increase, leading to the growth of natural gas consumption.

4.5. Natural gas demand elasticity of residents sector As there is no natural gas supply in rural areas, urban dwellers are the natural gas consumers for the residents sector. In addition to natural gas, residents’ use of coal, liquefied petroleum gas, electricity and their disposable income are important factors affecting energy consumption. The price difference of electricity consumption in each district is large, and the data that can be obtained is lacking, so the index cannot be included in the study. The real price of natural gas consumption, the real price of coal, the real price of civil liquefied petroleum gas and the real disposable income of urban residents are used to establish the natural gas demand ARDL model. We get the following equation:

lnQre, t = Cre + γ1lnPre, t + γ2lnPlpg, t + γ3lnPc, t + γ4lnPc, t −1 + γ5lnDIt + εre (15)

4.6. Comparison of natural gas demand elasticity in various sectors

Where Qre, t : natural gas consumption in the residents sector in year t; Cre : constant term; Pre, t : real natural gas price for the residents sector in year t; Plpg, t : real liquefied petroleum gas price in year t; Pc, t : real price of coal in year t; DIt : real disposable income of urban residents in year t; and εre : the random error term. For the residents sector, the results in Table 6 show that the ownprice elasticity of natural gas demand is −0.223 (short-run and longrun elasticities are the same); the cross-price elasticity of natural gas demand with regard to liquefied petroleum gas is 0.282 (short-run and long-run elasticities are the same); the short-run and long-run cross price elasticities of natural gas demand with regard to coal are 0.334 and 0.061, respectively; and the income elasticity of natural gas demand is 2.051. The results are relative greater than that of Asche et al. (2008) in absolute value. It indicates that the Chinese residents are more sensitive to the changes of price and income. Natural gas demand for the residents sector shows a lack of price elasticity but is rich in income elasticity. There is a substitutional

There are big differences in the price and income elasticities of natural gas demand in the different key sectors of China, as shown in Table 7. The short-run and long-run own-price elasticities of natural gas demand in the industrial and electric power generation and supply sectors are greater than 0, but that of industrial sector is less than 1, while that of electric power generation and supply sectors is greater than 3. The short-run and long-run cross-price elasticities of natural gas demand with regard to coal in the industrial and electric power generation and supply sectors are less than 0, but the absolute value of the latter is greater. Income elasticity of natural gas demand in the industrial and electric power generation and supply sectors is bigger than 2, but the elasticity of the latter is significantly greater than that of the former. The short-run price elasticity of natural gas demand in transportation and services is less than 0, but long-run elasticity is significantly greater than 1. The cross-price elasticity of natural gas demand with regard to oil in transportation is significantly greater than 1, which is rich in elasticity. The cross-price elasticity of natural gas demand with regard to coal in services is significantly greater than 1, also rich in elasticity. The natural gas demand of the residents sector lacks price elasticity but is rich in income elasticity.

Table 6 Coefficient estimation of natural gas demand equation in residents sector. Coefficient

Estimated value

Standard deviation

t-Statistic

P-value

Cre γ1 γ2 γ3 γ4 γ5

−14.289 −0.233 0.282 0.334 −0.273 2.051

1.453 0.223 0.121 0.148 0.135 0.113

−9.834 −1.048 2.334 2.262 −2.029 18.098

0.000 0.312 0.035 0.040 0.061 0.000

5. Conclusions and policy implications Based on the analysis of different elasticity concepts and demand elasticity methods, this paper constructs an autoregressive distribution lag model to study the elasticity of natural gas demand in China's various subsectors.

Adjusted R2=0.99, F statistics=1165.86，AIC=−2.69.

339

Energy Policy 113 (2018) 332–341

Y. Zhang et al.

References

Long-run price elasticities of natural gas demand in the industry, electric power generation and supply, transportation and service are greater than 0, and all are elastic except for industrial. This is contrary to the theory of price elasticity of demand but is in line with China's reality and fully explains that when price regulation makes natural gas prices relatively low, the natural gas market is in disequilibrium. The reform of natural gas pricing mechanism must be accelerated. Economic growth heavily drives the demand of natural gas in the industrial and electric power generation and supply sectors. There is a complementary relationship between natural gas consumption and coal consumption in the industrial and electric power generation and supply sectors, so it is necessary to change the status of natural gas as a supplement to coal and to take relevant measures to promote the substitution of natural gas for coal. Natural gas and petroleum products in the transportation sector are interchangeable. Natural gas consumption and coal in service sector are substitutable. The own-price elasticity of natural gas demand in the residents sector is less than 0 and greater than −1. The natural gas consumption of residents is substitutable by coal and liquefied petroleum gas consumption, and is rich in income elasticity. Based on the above conclusions, we believe the following measures should be taken to ensure the realization of the energy mix adjustment target and the natural gas consumption target during the 13th Five-Year Plan period:

Alberini, A., Gans, W., Velez-Lopez, D., 2011. Residential consumption of gas and electricity in the U.S.：the role of prices and income. Energy Econ. 33 (5), 870–881. Al-Sahlawi, M., 1989. The demand for natural gas: a survey of price and income elasticities. Energy J. 10 (1), 77–90. Andersen, T.B., Nilsen, O.B., Tveteras, R., 2011. How is demand for natural gas determined across European industrial sectors? Energy Policy 39 (9), 5499–5508. Arora, V., Cai, Y.Y., Jones, A., 2016. The national and international impacts of coal-to-gas switching in the Chinese power sector. Energy Econ. 60, 416–426. Asche, F., Nilsen, O.B., Tveteras, R., 2008. Natural gas demand in the European household sector. Energy J. 29, 27–46. Bentzen, J., Engsted, T., 1993. Short- and long-run elasticities in energy demand: a cointegration approach. Energy Econ. 15 (1), 9–16. Bentzen, J., Engsted, T., 2001. A revival of the autoregressive distributed lag model in estimating energy demand relationships. Energy 26 (1), 45–55. Bilgili, F., 2014. Long run elasticities of demand for natural gas: OECD panel data evidence. Energy Sources Part B Econ. Plan. Policy 9 (4), 334–341. BP, 2017a. BP Energy Outlook 2035 (2017 edition). BP, 2017b. BP Statistical Review of World Energy (66th edition). Burke, P.J., Yang, H., 2016. The price and income elasticities of natural gas demand: international evidence. Energy Econ. 59, 466–474. Considine, T.J., 1989. Separability, functional form and regulatory policy in models of interfuel substitution. Energy Econ. 11 (2), 82–94. Cuddington, J.T., Dagher, L., 2015. Estimating short and long-run demand elasticities: a primer with energy-sector applications. Energy J. 36 (1), 185–209. Dagher, L., 2012. Natural gas demand at the utility level: an application of dynamic elasticities. Energy Econ. 34 (4), 961–969. Diewert, W.E., Wales, T.J., 1987. Flexible functional forms and global curvature conditions. Econometrica 55 (1), 43–68. Dong, X.C., Pi, G.L., Ma, Z.W., Dong, C., 2017. The reform of the natural gas industry in the PR of China. Renew. Sustain. Energy Rev. 73, 582–593. Donnelly, W.A., 1987. The Econometrics of Energy Demand. Praeger Publishers, NewYork. Erdogdu, E., 2010. Natural gas demand in Turkey. Appl. Energy 87 (1), 211–219. Frondel, M., 2004. Empirical assessment of energy-price policies the case for cross-price elasticities. Energy Policy 32 (8), 989–1000. Frondel, M., 2011. Modelling energy and non-energy substitution: a brief survey of elasticities. Energy Policy 39 (8), 4601–4604. Griffin, J.M., 1977. Inter-fuel substitution possibilities: a translog application to intercountry data. Int. Econ. Rev. 18, 755–770. Hao, H., Liu, Z.W., Zhao, F.Q., Li, W.Q., 2016. Natural gas as vehicle fuel in China: a review. Renew. Sustain. Energy Rev. 62, 521–533. He, Y.X., Xia, T., Liu, Y.Y., Zhou, L.F., Zhou, B., 2013. Residential natural gas price affordability analysis–a case study of Beijing. Renew. Sustain. Energy Rev. 28, 392–399. Hu, A.L., Dong, Q., 2015. On natural gas pricing reform in China. Nat. Gas. Ind. B 2 (4), 374–382. Hunt, L., Manning, L., 1989. Energy price- and income-elasticities of demand: some estimates for the UK using the cointegration procedure. Scott. J. Political Econ. 36 (2), 183–193. Huo, J.L., 2011. Comparing the natural gas pipeline pricing between Europe and America and the revelation to China. Energy Procedia 5, 659–663. Jones, C.T., 1995. A dynamic analysis of interfuel substitution in U.S. industrial energy demand. J. Bus. Econ. Stat. 13, 459–465. Kahrl, F., Hu, J.F., Kwok, G., Williams, J.H., 2013. Strategies for expanding natural gasfired electricity generation in China: economics and policy. Energy Strategy Rev. 2 (2), 182–189. Li, J.C., Dong, X.C., Shangguan, J.X., Hook, M., 2011. Forecasting the growth of China's natural gas consumption. Energy 36 (3), 1380–1385. Lin, B.Q., Wang, T., 2012. Forecasting natural gas supply in China: production peak and import trends. Energy Policy 49, 225–233. Lin, W.S., Zhang, N., Gu, A.Z., 2010. LNG (liquefied natural gas): a necessary part in China's future energy infrastructure. Energy 35 (11), 4383–4391. Paltsev, S., Zhang, D.W., 2015. Natural gas pricing reform in China: getting closer to a market system? Energy Policy 86, 43–56. Pesaran, M.H., Shin, Y., 1999. An autoregressive distributed lag modelling approach to cointegration analysis. In: Strøm S, editor Proceedings of the econometrics and economic theory in the twentieth century: the Ragnar Frisch Centennial Symposium. Cambridge: Cambridge University Press. Pindyck, R.S., 1979. Interfuel substitution and the industrial demand for energy: an international comparison. Rev. Econ. Stat. 61, 169–179. Renou-Maissant, P., 1999. Interfuel competition in the industrial sector of seven OECD countries. Energy Policy 27 (2), 99–110. Serletis, A., Timilsina, G.R., Vasetsky, O., 2010. Interfuel substitution in the United States. Energy Econ. 32 (3), 737–745. Serletis, A., Timilsina, G., Vasetsky, O., 2011. International evidence on aggregate shortrun and long-run interfuel substitution. Energy Econ. 33 (2), 209–216. Shaikh, F., Ji, Q., 2016. Forecasting natural gas demand in China: logistic modelling analysis. Int. J. Electr. Power Energy Syst. 77, 25–32. Shaikh, F., Ji, Q., Fan, Y., 2016a. Assessing the stability of the LNG supply in the Asia Pacific region. J. Nat. Gas Sci. Eng. 34, 376–386. Shaikh, F., Ji, Q., Fan, Y., 2016b. Evaluating China's natural gas supply security based on ecological network analysis. J. Clean. Prod. 139, 1196–1206. Shaikh, F., Ji, Q., Fan, Y., 2017a. An ecological network analysis of the structure,

(1) Accelerate the reform of natural gas prices, straighten out the energy price system and change the status that the price of imported natural gas is more expensive than that of domestic. Since April 1, 2015, for non-resident gas users, the stock gas and incremental gas price has been unified; direct-supply gas price has been released (Dong et al., 2017). Due to the special nature of fertilizer, the reform of its gas prices was delayed. The reform if city gate gas prices (which include residential users, business users and small industrial users through the urban pipe network) should be based on different types of user classifications. It is reasonably carry out ladder price reform for residents and business users to avoid excessive consumption of natural gas. For different small industrial users, natural gas has different value. The government should expand the volatility of natural gas prices until the lifting of price regulation. (2) Speed up the establishment of a national carbon market, strengthen pollutant discharge standards and accurately reflect the environmental costs of energy use. It is conducive for gas power generation achieving cost advantage over coal-fired power generation to establish carbon market and charging carbon emissions (Skeer and Wang, 2006; Kahrl et al., 2013; Zhu et al., 2016). This would also help to improve the proportion of gas power generation in the power mix and promote the substitution of natural gas for coal. (3) Strengthen the subsidies, tax relief or incentives for substituting natural gas for coal in industrial sector to promote natural gas consumption. According to the natural gas supply situation, the government should gradually promote substituting natural gas for coal and change the situation that natural gas is a supplement of coal. (4) Optimize power grid scheduling and improve the proportion of gas power generation. In the premise of ensuring the supply of natural gas, the state grid should improve the proportion of gas power plants, change the load tracking, peak gas power plant into a base power plant, change the status that gas power plant is a back-up power for coal-fired power. Acknowledgement Supports from the National Natural Science Foundation of China under Grant No. 71774152, No. 91546109 and Youth Innovation Promotion Association of Chinese Academy of Sciences (Grant: Y7×0231505) are acknowledged. 340

Energy Policy 113 (2018) 332–341

Y. Zhang et al.

251–260. Uzawa, H., 1962. Production functions with constant elasticities of substitution. Rev. Econ. Stud. 29, 291–299. Wang, H.X., Fang, H., Yu, X.Y., Wang, K., 2015. Development of natural gas vehicles in China: an assessment of enabling factors and barriers. Energy Policy 85, 80–93. Wang, J.L., Mohr, S., Feng, L.Y., Liu, H.H., Tverberg, G.E., 2016. Analysis of resource potential for China's unconventional gas and forecast for its long-term production growth. Energy Policy 88, 389–401. Xu, G., Wang, W.G., 2010. Forecasting China's natural gas consumption based on a combination model. J. Nat. Gas Chem. 19 (5), 493–496. Yu, Y., Zheng, X., Han, Y., 2014. On the demand for natural gas in urban China. Energy Policy 70, 57–63. Zhang, W., Yang, J., 2015. Forecasting natural gas consumption in China by Bayesian Model Averaging. Energy Rep. 1, 216–220. Zhang, W., Yang, J., Zhang, Z.Y., Shackman, J.D., 2017. Natural gas price effects in China based on the CGE model. J. Clean. Prod. 147, 497–505. Zhu, N.P., Zhao, Q., Tian, L.X., Zhang, Q., 2016. Cost analysis and development strategies for China’ natural gas power generation industry under the situation of energy price's reformation. Energy Procedia 104, 203–208.

development and sustainability of China's natural gas supply system security. Ecol. Indic. 73, 235–246. Shaikh, F., Ji, Q., Fan, Y., Shaikh, P.H., Uqaili, M.A., 2017b. Modelling an optimal foreign natural gas import scheme for China. J. Nat. Gas Sci. Eng. 40, 267–276. Shi, G.H., Jing, Y.Y., Wang, S.L., Zhang, X.T., 2010. Development status of liquefied natural gas industry in China. Energy Policy 38 (11), 7457–7465. Shi, X.P., Variam, H.M.P., Tao, J., 2017. Global impact of uncertainties in China's gas market. Energy Policy 104, 382–394. Skeer, J., Wang, Y.J., 2006. Carbon charges and natural gas use in China. Energy Policy 34 (15), 2251–2262. Steinbuks, J., 2012. Interfuel substitution and energy use in the UK manufacturing sector. Energy J. 33 (1), 1–29. Sun, C.W., Ouyang, X.L., 2016. Price and expenditure elasticities of residential energy demand during urbanization: an empirical analysis based on the household-level survey data in China. Energy Policy 88, 56–63. Urga, G., Walters, C., 2003. Dynamic translog and linear logit models: a factor demand analysis of interfuel substitution in US industrial energy demand. Energy Econ. 25 (1), 1–21. Uri, N.D., 1978. Interfuel substitution possibilities: short-term prospects. Appl. Energy 4,

341