China’s regional rebound effect based on modelling multi-regional CGE

Improving energy efficiency has been regarded as an important measure to reduce energy consumption, yet the rebound effect has greatly shrunken the energy saving consequences of this measure. To investigate regional rebound effect in China, a multi-region computed general equilibrium (CGE) model is established in this paper. The results show that there are obvious regional differences in the rebound effect in China. The primary energy rebound effects are positive, whereas the production-side power rebounds are below zero in most regions. We also simulated the energy subsidy reform scenarios, which indicates that reducing or even eliminating coal and oil subsidies will increase the production-side rebounds. Finally, feasible policy recommendations are put forward based on the results.

Keywords: Rebound effect; energy efficiency; multi-region CGE; energy subsidy reform; energy tax structure

I. Introduction

Despite obvious achievements in energy conservation and emission reduction, the total energy consumption is still rising year after year in China. In 2017, China's total energy consumption had reached 4,490 Mtce, of which coal and oil consumption totalled 2,711.96 and 844.12 Mtce, respectively, whereas natural gas and secondary energy only account for a small proportion (7% and 13.8%, respectively).[1] In contrast, the share of natural gas consumption in OECD countries has exceeded 30%; The EU's renewable energy consumption has reached 15% and is expected to exceed 27% in 2030.[2] From this point, China's road to energy conservation and emission reduction still has a long way to go.

Forced by the pressure of domestic air pollution and global climate change, the 13th Five-Year Plan for Energy Development released at the end of 2016 proposed that energy consumption growth is expected to fall from an average of 9% to 2.5% since the 10th Five-Year Plan. Moreover, the Chinese government proposes that by 2030, the carbon dioxide emissions per unit of GDP will be 60–65% lower than the 2005 level (Yang and Li [40]). To achieve the proposed targets, China must accelerate the pace of promoting energy technology and improving energy efficiency to reduce energy consumption.

However, with the further development of energy consumption research, some studies have found that the energy efficiency increase has some discount on the reduction of energy consumption, and may even lead to an increase in it (Brookes [2]; Khazzoom [19]), which is defined as the 'energy rebound effect'(Greening 2000). Scholars have different perspectives on the magnitude of the rebound effect, who found that the rebound effect in various countries ranges from 0% to 100%, some are even negative or over 100%. Therefore, the rebound effect is indeed an obstacle to the implementation of energy policy, resulting in a certain degree of policy failure (Hymel and Small [16]; Schleich, Mills, and Dütschke [31]; Sorrell et al. [33]; Y.-J. Zhang et al. [42]).

Due to great differences in regional energy endowments and energy consumption structures, China's regional rebound effects will also be different (Jin, Duan, and Tang [18]; Wen et al. [38]). In 2016, the productions of raw coal in Tianjin, Shanghai, Zhejiang, Guangdong, and Hainan were all zero, while in Shanxi and Inner Mongolia, where coal reserves were abundant, raw coal productions were as high as 80.43 and 84.56 million tons, respectively. The energy demand also varies from region to region. Both of Guangdong and Jiangsu had total energy consumption of 300 million tons of standard coal in 2016, while Qinghai and Ningxia consumed only 41.11 and 55.92 million tons of standard coal, respectively. Furthermore, because of the large difference in energy demand between production and consumption in various regions of China, the resulting production- and consumption-side rebound will also differ (Lu, Liu, and Zhou [26]). Taking into account the above factors, this paper is aimed to study the production- and consumption-side regional rebound effects as well as dynamic rebound in China using a multi-region CGE model, so that precise regional energy policies can be developed. The CGE model, which can measure the regional rebound effects more comprehensively than other methods, consists of nine layers for the production structure, include four energies (coal, oil, natural gas, and electricity) and nine regions: Northeast, Northwest, Southwest, Central, South, (Bei)Jing–(Tian)Jin–(Hebei)Ji (JJJ), JJJ Rim, Yangtze River Delta, and Pearl River Delta.

The main contributions of this paper can be summarized as follows: Firstly, based on the existing studies (Lu, Liu, and Zhou [26]; Wen et al. [38]), we develop a multi-region CGE model to study China's static and dynamic rebound effect. By establishing the CGE model, we can not only comprehensively assess the rebound effects of various provincial regions in China based on regional differences and energy heterogeneity, but also predict the regional rebound changes in the next few decades, which are difficult to do with econometric models. Secondly, Wen et al. ([38]) studied the urban household rebound effect in various regions of China and divided it into direct and indirect rebound, but there is less research on subdividing the rebound effect into production- and consumption-side rebound at the regional level. We will extend Lu, Liu, and Zhou ([26])'s method of dividing China's overall rebound effect into production- and consumption-side rebounds. Thirdly, according to the decomposition results, we will further simulate the rebound effect after reducing even eliminating energy subsidies in various regions, then propose feasible policy recommendations.

The remainder of this paper is organized as follows. Section Ⅱ reviews previous studies. Section Ⅲ introduces the methodology, including the construction of multi–region CGE model and the calculation of rebound effect. Section Ⅳ introduces the data processing and scenario design. Section Ⅴ analyses the simulation results. Based on the previous results, Section Ⅵ implements regional energy subsidy policy simulation. Finally, Section Ⅶ provides the discussions and conclusions.

II. Literature review

The rebound effect is an important issue in energy economics. Khazzoom ([19]) first studied the rebound effect from a microscopic perspective. Then, Brookes ([2]) studied the macroscopic rebound, pointing out that technological advances do not necessarily lead to reduced energy consumption. Subsequently, Saunders ([29]) proposed the famous Khazzoom-Brookes hypothesis that improved energy efficiency would lead to a reduction in energy consumption. Since then, more and more literature have confirmed the existence of the rebound effect from the perspective of theoretical and empirical research (e.g. Brännlund, Ghalwash, and Nordström [1]; Herring [12]; Herring and Roy [13]; Lin and Liu [22]).

The empirical study for the rebound effect can be summarized from the research object and methodology: in terms of research objects, it includes the measurement of specific sectors' rebound and economy-wide rebound. Recent literature also began to investigate the rebound of different types of energy and decompose it to different dimensions. In terms of methodology, the model framework of empirical research is mainly divided into the econometric and the CGE methods. The econometric models are used to calculate the historical rebound, while the CGE method to simulate and predict the rebound effect.

Considerable literature have studied the rebound effects of various industries and household consumption (Druckman et al. [6]; Gonzalez [9]; Hymel, Small, and Van Dender [17]; Lin and Li [21]; Lin and Liu [23]; Wang, Zhou, and Zhou [37]; Y.-J. Zhang et al. [42]). The above studies found that the rebound rates of various industries vary widely, ranging from zero to 100%, and some even well below or exceed 100%.

Few studies have been done on the economy-wide rebound. Some scholars have measured the rebound effect of Kenya, Sweden, and other countries (Jin, Duan, and Tang [18]; Semboja [32]; Vikström [35]), from which it can be found that the rebounds of different countries are quite unequal. Recent studies have focused more on the rebound decomposition and rebound differences caused by energy heterogeneity. For example, Lu, Liu, and Zhou ([26]) analysed the rebound effects of various types of energy in China and decomposed them into the production–side and consumption-side rebound. They found that in terms of China's rebound effect, primary energy is greater than secondary energy, the macro level is greater than industry level, and power rebound is negative. Li et al. ([20]) further refined the energy types and simulated the impact of eliminating energy subsidies on the rebound effect.

The research methods of rebound effect are mainly based on econometric models (Freire-González [7]; Gonzalez [9]; Matos and Silva [28]; Wang et al. [36]; Wu et al. [39]). Hybrid macro-economic model and input-output analysis are also ways to study the rebound effect (Dimitropoulos [5]; Wen et al. [38]). In recent years, the CGE method is also used to measure the energy rebound. The CGE model takes the economic system as a whole and emphasizes the interaction between various sectors and variables of the economic system. This reflects the 'general equilibrium' characteristic of CGE models, which is beyond the reach of other econometric model. Moreover, the CGE model is compatible with the advantages of different models such as input-output analysis and linear programming. It also overcomes the shortcomings of these models that can not introduce the price mechanism into the model, and replaces the linear function with a nonlinear version in the traditional input-output model. Therefore, the simulation results are usually more detailed, comprehensive, and reasonable than those obtained by other models. That is the reason why we employ the CGE model to study the regional rebound effects in China. China's CGE-based rebound effect research started quite late. In the early studies, Glomsrød and Taoyuan ([8]) focused on the rebound of increased coal utilization efficiency by constructing a Chinese CGE model. They proposed that because China's coal consumption accounts for a large proportion of total energy, improving coal efficiency of using coal will lead to a significant expansion of production capacity, which will result in an increase in energy consumption. This study also shows that energy-saving expectations at the industrial level have not been reflected in the macro-economic level, and the energy savings and environmental benefits brought about by the development of clean coal technology are almost completely offset. Since then, the CGE method has been widely used in China's rebound effect research (Li et al. [20]; Wang et al. [36]). Some recent literature has also studied the difference in the rebound effect caused by energy heterogeneity (Lu, Liu, and Zhou [26]; Zhou et al. [43]).

In summary, the literature on rebound effect of China still has some limitations. Firstly, most of them are about the rebound effect of a particular sector, whereas little attention has been devoted on the overall rebound of China. Secondly, the application of CGE models in this area is still limited. Furthermore, there is less research on production – and consumption-side rebound at the regional level. Considering the above, we build a multi-regional CGE model to study the regional rebound effect in China. The main purpose is to study the overall, production–side and consumption-side regional rebound effect of various energies. To the best of our knowledge, this might be rare in existing literature. We also simulate the energy subsidy reform policy based on the rebound results of different regions. In this way, the impact of reducing or even eliminating energy subsidies on regional economies and rebounds can be observed, hence more scientific policy recommendations could be proposed. Additionally, to investigate dynamic trend of the rebound effect in each region, a dynamic CGE model is constructed as well.

III. Methodology

The multi–region CGE model

The multi-region CGE model established in this paper includes modules such as production, consumption, investment, government, and international trade. Only the production modules are described below, and the remaining are similar to the common CGE model.

Figure 1 presents the production structure, which is composed of nine layers. At the top layer, the total output consists of basic inputs and production taxes. The production taxes imposed by the government increase only a firm's production costs without affecting the output, so alternative possibilities between the taxes and costs are not available, and they are represented by the Leontief function. In the second tier, the basic inputs are divided into energy-capital-labour bundle and intermediate inputs, which also nested by the Leontief function. Compound intermediate inputs are composed of N commodities through the Leontief function in the third tier of intermediate inputs, Energy-capital compound products are a CES composite of energy and capital in the fourth tier. In the fifth layer, energies include fossil fuels and power, followed by fossil fuels divided into coal and non-coal in the sixth tier. In the seventh tier, non-coal fossil energy is composed of oil and gas. Notably, each type of fossil energy and intermediate inputs is composed of domestic and imported parts which nested by CES function (the eighth tier of oil and gas; the seventh tier of coal; the sixth tier of electricity; and the fourth tier of intermediate inputs). Domestic goods are nested by goods from multiple regions.

Graph: Figure 1. Production structure.

Here, only the coal nesting is taken as an example. Equation (1) indicates that the domestic coal used in industry i of period t in region r, is nested by the coal from all domestic regions.

(1)

Graph

(2)

Graph

(3)

Graph

(4)

Graph

where and are the quantity and price of total domestic coal demand of industry i in region r, respectively; and represent the quantity and price of coal purchased by industry i in region r from region s. is the scale parameter; represents the coal demand elasticity; and the parameter is an intermediate parameter related to substitution elasticity.

Equation (2) is the demand function that an enterprise makes a purchase decision based on the coal price of each region. The amount of coal purchased in region s is positively correlated with the total demand for domestic coal in the certain industry, and negatively correlated with the coal price in region, and is also affected by other parameters. Equation (3) illustrates the calculation process for the composite price of domestic coal used in industry, which is obtained by weighted average coal in each region. Equation (4) shows the relationship between the substitution elasticities and the intermediate parameters.

Regarding model closure, the exogenous variables selected are labour supply, energy efficiency, production tax rate, inventory change, international price and return on capital. The remaining variables are all endogenous.

Calculation of rebound effect

There are many economic mechanisms for the rebound effect, such as substitution and output effects (Saunders [30]; X. Zhang et al. [41]). The output effect reflects the increase in energy use from output expansion, whereas the substitution effect reflects the increase in energy use from producers substituting toward cheaper energy as its effective price is lowered.

The paper illustrates the impact of the rebound effect on the macro economy by analysing the changes in macroeconomic variables such as GDP and energy efficiency changes. In the CGE model, the increasing energy efficiency first leads to the improvement of the production efficiency of energy products, that is, when the energy input is constant, the output will be higher, which will undoubtedly reduce the cost of using energy elements. According to the zero profit conditions of production, the consumer price of energy products will also decrease. The same changes are made in government purchases and investments. In addition, the increase in energy demand will have a substitution effect on other commodities, causing changes in the prices and demands of these commodities, which will eventually lead to changes in CPI.

In terms of the measurement of rebound effect, the most widely used one is first proposed by Saunders ([29]). Then, according to the work of Hanley et al. ([11]) and Turner ([34]), the rebound effect can be expressed by Equation (5):

(5)

Graph

where is the ratio of energy efficiency improvement (such as 5%), and is the rate of change of energy consumption after energy efficiency is improved. since regional rebound effect is studied in this paper, when s represents different regions, the regional rebound can be expressed by Equation (6):

(6)

Graph

Following Lu, Liu, and Zhou ([26]), the rebound effect is decomposed into production-side and consumption-side rebound. The latter includes the rebounds of household, investment, inventory, government and export. The specific expression (Equations (7)–(8)) is as follows:

(7)

Graph

(8)

Graph

where is the total rebound; and are the rebound coefficients of production sectors and energy i, respectively. denotes energy demand changes. The subscripts HC, IN, GC, EX, IV and OP represent household consumption, investment, government consumption, exports, inventory and other production sectors, respectively.[3] Based on this calculation method, this paper focus on the production – and consumption-side rebound in various regions of China.

IV. Data and scenario design

Data

The basic data of the CGE model are the input-output table of 30 provincial regions (excluding Tibet) and 30 industries compiled by Liu, Zhipeng, and Jie ([25]). Because the purpose of this paper is to calculate the rebound effects in different regions of China, firstly we divide China into the East, West, Central and Northeast regions according to the regional division method published by the National Bureau of Statistics of China.[4] Secondly, considering that China's economic development and energy consumption are closely related to different economic circles, we will separately list the economic circles such as the JJJ (Beijing-Tianjin-Hebei), Yangtze River Delta, and Pearl River Delta regions. Thirdly, China's geography and climate conditions are quite different in its northern and southern regions, which may cause differences in energy consumption structure. We divide the West into the Southwest and the Northwest. Finally, since the rest regions are located in the south, we classify them as the South. The final regional division is shown in Figure 2.

PHOTO (COLOR): Figure 2. Zoning map.

The 30 industries of the original IO table are also split into eight industries: agriculture, coal, oil, gas, manufacturing, electricity, construction, and services. For a long time, oil and gas have been attributed to the same industry catalogue' oil and gas mining' in China IO tables, here we split the oil and gas industry based on the ratio of these two industries in the GTAP 9.0 database. Furthermore, to study the consumption situation of different sectors, terminal consumption has also been split into household and government consumption. At this point, the basic data required for the model have been prepared, and then the RAS method is applied for data balancing.

Scenario settings

We set four static simulation scenarios to investigate the rebound effect of each region. In the energy efficiency setting, considering that China's energy efficiency has been significantly improved over the past few decades, and most of the literature has set the energy efficiency to 5%, so this paper will also set the energy efficiency to 5% in the simulation. The specific settings are shown in Table 1.

Table 1. Static scenarios setting.

Scenario	Description
BAU-S	Energy efficiency in all regions remains unchanged
S01	Coal use efficiency increased by 5% in all regions
S02	Oil use efficiency increased by 5% in all regions
S03	Gas use efficiency increased by 5% in all regions
S04	Electricity use efficiency increased by 5% in all regions

Based on the static model, the paper will also simulate the dynamic rebound changes of various regions in China until 2030. The GDP growth rate is set by reference to the IEA World Energy Outlook 2015. We assume that the labour changes are consistent with the changes in the total population, which is derived from the World Population Prospects 2017.[5] Considering that China's energy efficiency has been significantly improved in the past few decades, and improving energy efficiency is also the focus of the Energy 13th Five-Year Plan, it is believed that China's energy efficiency will continue to increase in the future. Then, following Dai et al. ([4]), the energy efficiency of BAU-D is set as shown in Table 2. We will examine the changes in rebound effects when the energy efficiency of using coal, oil, gas, and electricity increases by 0.5%, 0.5%, 0.5%, and 0.3% per year, respectively (Chen et al. [3]; Ma, Wang, and Wei [27]).

Table 2. Dynamic scenarios setting.

Scenario	Description(2018–2030)
BAU-D	The energy efficiency improvements of using coal, oil, gas, and electricity are 3%, 3%, 2%, and 1%, respectively
D01	The energy efficiency improvement of using coal is 3.5%
D02	The energy efficiency improvement of using oil is 3.5%
D03	The energy efficiency improvement of using gas is 2.5%
D04	The energy efficiency improvement of using electricity is 1.3%

A 2014 report by the European Health and Environment Alliance (HEAL) claimed that China's subsidies to fossil energy consumers and producers amounted to $96 billion.[6] Globally, China's financial subsidies for fossil energy and renewable energy rank 8th and 5th, respectively. In contrast, almost all other countries (such as the Gulf States and Germany) have a clear bias towards their support.[7] According to previous studies, China's subsidies for coal, oil and natural gas are 6.46%, 19.52% and 35.46%, respectively (Hong, Liang, and Di [14]; Liu and Li [24]). In order to study the impact of fossil energy subsidy reform on energy consumption, we adjust the fossil energy subsidy rate in each region to set policy scenarios mainly according to static simulation results. Specific policy scenarios and the results can be found in Section Ⅵ.

V. Results and analysis

Regional rebound effect

The regional rebound rates under static scenarios with an energy efficiency increase of 5% are summarized in Figure 3. The overall rebound consists of the production–side (RP) and consumption-side rebounds (RC).

PHOTO (COLOR): Figure 3. Regional energy rebound effect under static scenarios (%).

As shown in Figure 3, energy efficiency improvement can indeed reduce energy demand, and there are differences in China's regional rebound effects. The increase in efficiency of primary energy (coal, oil, and gas) will result in positive regional rebounds, that is, the reduction in energy demand caused by advances in primary energy technology will not reach the theoretically reduced amount. In contrast, although the overall and consumer-side power rebounds are both positive, the production-side power rebounds in most regions are less than zero. This shows that the production-side energy consumption of these regions are lower than theoretically after the power efficiency is improved. Besides, the regional rebound effects of primary energy are generally greater than those of secondary energy (i.e. power).

In scenario S01, it can be seen that the rebound effect in every region is greater than 70%. Among the regions, the Central and the JJJ Rim region have relatively smaller overall rebounds (72.64% and 72.73%), while the Northwest, the Pearl River Delta, and the Northeast have the larger ones (86.98%, 85.29% and 85.02%). From the decomposition of the rebound effect, the production-side rebound in the Southwest is smaller than those in other regions, and the consumption-side rebounds in that region are significantly larger than those in others. Because the heavy industry accounts for a small proportion in this region, the coal demand for production purposes is naturally small. In contrast, in the South, the coal use in the production sector rebound the most. The reason is that the South is dominated by the service industry, and the coal used for production is quite rare. When coal utilization efficiency increases, compared to other regions, a slighter rise in coal use will lead to a larger rebound effect in the South.

Compared with the rebound effect of coal, the oil rebounds in various regions are less different. The overall and production-side rebounds in the Northeast, the Northwest and Yangtze River Delta are relatively small, and these regions happen to be large oil consumers. This may be because these regions are less sensitive to the oil cost change and they are very dependent on oil, and then their oil consumption will not change much. Additionally, the production-side oil rebound in each region is much larger than the consumption-side rebound. There are two reasons for this: First, the oil demand for production is larger than that for the consumption. The use of oil on the consumption side mainly comes from various types of transportation, and the number of vehicles is roughly at a steady growth level, which will not make oil rebound so big. Second, even if oil utilization efficiency is improved, China's consumption oil price changes will not be as large as the international oil price fluctuations, and the stimulus effect on oil consumption would not be too great. In scenario S03, as gas demand accounts for a very small proportion of total energy demand, the increase in gas use efficiency has little impact on the energy needs of each region. In other words, the theoretical and actual changes in energy demand are extremely small, and it is not surprising that the calculated rebound rates are around 100%. Moreover, compared with changes in terminal energy consumption, the increase in gas efficiency leads to much more changes in production-side energy demand. The reason is that the gas penetration rate of Chinese households is low, and historical data show that it is unlikely that domestic gas prices will be reduced.

When the power utilization efficiency is increased by 5%, the overall and the consumption-side rebounds are positive, the production-side rebounds in most regions are negative. The reason for this is that the proportion of thermal power in total power generation has exceeded 70% in China, and the proportion of coal-fired power generation in thermal power generation is as high as 90%. Therefore, the increase in power utilization efficiency will greatly reduce the amount of coal for power generation, and the rebound in power consumption will be offset by this effect. The South has the largest negative rebound, which exceeds −50%. This is mainly because improving power efficiency will increase demand for thermal power, but this will be affected by clean electricity, such as nuclear power, which accounts for a large proportion in the South. In the Northeast and the Northwest, because of the less developed manufacturing industry, their electricity uses are also lower among nine regions. Therefore, a slightly increased power usage will cause a large rebound, resulting in positive rebounds in these regions.

Figure 4 shows the dynamics of the regional rebound effect. Since the energy efficiency improvement in each dynamic scenario is extremely small (only 0.5% for fossil energy and 0.3% for electricity), the dynamic rebound in each region is generally higher. Overall, the gas rebounds are the same as the static results, around 100%. Except for the PRD and the South, energy rebound effects in other regions are greatest when oil efficiency increases, second in coal efficiency, and lowest in power efficiency, which is also consistent with static results. We have noticed that the dynamic rebounds of coal, oil, and electricity in the PRD and the South show a significant downward trend. The reasons are as follows: The industrial structures of the PRD and the South are dominated by the tertiary industry, and the proportions of fossil energy consumptions are relatively small. From the perspective of industrial energy use, in the case of improving coal, oil and electricity efficiency, respectively, the energy consumptions of the power sector in the PRD and the South will be significantly reduced compared to other regions in 2030. Considering China's thermal power-based power generation structure, the reduction in power demand indicates a significant reduction in fossil energy consumption. Furthermore, the Chinese government strongly supports the clean energy industry in the Pearl River Delta and the South. The progress of clean energy power generation technologies in these two regions will further curb their demand for fossil energy.

PHOTO (COLOR): Figure 4. Regional rebound effects under dynamic scenarios (%).

Impact of energy efficiency improvement on regional energy demand

The energy demand changes for each region under static scenarios are shown in Figure 5. Since the proportions of coal and electricity consumption in regional total energy consumption are not much different, there is no significant difference in the degree of decline in coal demand caused by advances in coal technology and the decline in power demand caused by increased power efficiency. When oil use efficiency increases, the energy demand of each region's energy sources will decrease, and the demand for gas will change most. The improvement efficiency of using gas will lead to a decline in gas consumption in the JJJ and other regions, indicating a rebound in gas consumption in these regions. This is because gas use efficiency has increased the cost of gas in these regions. In the same way, gas consumptions in the JJJ, Northwest, and Northeast will also increase due to differences in electricity use costs among regions caused by electricity efficiency improvement.

PHOTO (COLOR): Figure 5. Regional energy demand changes under static scenarios (%).

Impact on national and regional economy

Technological advances will increase equipment use efficiency, and the production costs and prices of corresponding energy products will decline, thus promoting faster economic growth. We found that in the macroeconomic aspect, in addition to the gas efficiency increase scenario (S03), all kinds of energy efficiency improvement will have a significant role in promoting macroeconomic growth. Specifically, the CPI is lower than that in the BAU under each scenario, while real GDPs will increase.

Table 3 shows the percentage of changes in regional real GDP, CPI, and household consumption relative to BAU under the static scenarios. The trend of regional economic variables is consistent with macroeconomic ones. That is, compared with BAU, the CPI of each region decreases under all scenarios, while the real GDPs will increase. At the same time, the impact of energy efficiency improvements on household consumption is also positive. These variables vary most significantly only when power efficiency is increased. Relatively speaking, increasing efficiency of using gas has little impact on the regional economy. As far as regional differences are concerned, the impact of energy efficiency improvements on the regional economy is mainly related to the regional energy consumption structure. For example, coal consumption in the JJJ region accounts for the largest proportion of total energy consumption. Therefore, the efficiency improvement of using coal has the greatest impact on the JJJ region, which has increased the GDP of the JJJ region by 0.54%. In scenario S01, the South's GDP increased by only 0.09% due to less coal use.

Table 3. Changes in regional economic variables relative to BAU (%).

	Northeast	JJJ	South	Northwest	Southwest	JJJ Rim	YRD	Central	PRD
Real GDP
S01	0.25	0.54	0.09	0.45	0.39	0.39	0.19	0.30	0.19
S02	0.44	0.15	0.06	0.64	0.12	0.14	0.18	0.10	0.15
S03	0.00	0.00	0.00	0.01	0.00	0.00	0.00	0.00	0.00
S04	0.53	0.83	0.51	0.75	0.56	0.58	0.59	0.56	0.66
CPI
S01	−0.57	−0.68	−0.54	−0.67	−0.60	−0.59	−0.54	−0.58	−0.54
S02	−0.32	−0.31	−0.28	−0.35	−0.30	−0.31	−0.29	−0.30	−0.28
S03	0.00	0.00	0.00	−0.01	0.00	0.00	0.00	0.00	0.00
S04	−1.12	−1.31	−1.30	−1.27	−1.16	−1.13	−1.10	−1.15	−1.12
Household consumption
S01	0.02	0.09	0.01	0.05	0.03	−0.01	0.02	0.02	0.02
S02	0.06	0.01	0.00	0.06	0.01	0.01	0.01	0.01	0.01
S03	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00
S04	0.05	0.16	0.13	0.09	0.05	−0.02	0.07	0.07	0.07

Notably, when the power efficiency is improved, the JJJ region has the fastest economic growth, but the highest proportion of power consumption is in the South. This indicates that the impact of energy efficiency on the economy does not depend entirely on the regional energy consumption structure, but also on the degree of energy marketization and industrial structure. In the JJJ region, especially in Tianjin and Hebei, there are many heavy industries and manufacturing industries. Therefore, as power efficiency increases, the demand for electricity and other energy sources in the JJJ will further increase. The dynamic trends of national GDP and CPI are consistent with static scenarios (see Figure 6). Improving gas efficiency has little impact on the macroeconomy.

PHOTO (COLOR): Figure 6. Changes in macro variables relative to BAU under dynamic scenarios (%).

VI. Simulating energy subsidy policies

In general, energy subsidies have an inhibitory effect on economic growth. First, energy subsidies squeeze out public spending. Second, energy subsidies lead to excessive resource mismatches in capital and energy-intensive industries. Third, energy subsidies lead to increased energy consumption and rebound effects, as subsidy policies can encourage energy use by lowering end-use prices (Khazzoom [19]). Therefore, energy subsidy policy reforms may lead to more efficient resource allocation. However, reducing or even completely eliminating energy subsidies would push up energy production costs, and the producers will pass on the higher costs to consumers, thereby raising the general price level, leading to a decline in consumer purchasing power, thereby reducing economic growth. Therefore, reducing subsidies for fossil energy does not necessarily bring benefits to economic growth.

China, as a developing country, has implemented a large subsidy for fossil energy, which has led to economic development becoming too dependent on fossil fuels, especially coal (Hong, Liang, and Minghua [15]; Liu and Li [24]). Our simulation results show that the impact of various energy efficiency improvements on different regions varies, so the energy policy should be adjusted according to the characteristics of each region and the type of energy.

We calculate changes in economic growth and regional rebound effects under the energy subsidy scenario. The rebound effect brought about by the improvement of primary energy and secondary energy efficiency is relatively large, and the production side rebound accounts for a large proportion of the primary energy rebound effect. Therefore, we focus on reducing the impact of energy subsidies on the production side. Although the natural gas subsidy rate is higher, compared with other fossil energy sources, natural gas consumption produce less pollution and improve natural gas utilization efficiency does not produce a big rebound. Temporarily, natural gas subsidies should not be reduced.

The baseline scenarios here are S01 and S02, where coal and oil utilization efficiency increased by 5%, respectively. We have set the following policy scenarios: Under scenario C01 (reducing S01–based coal production subsidies) and P01 (reducing – based oil production subsidies), regional subsidy rates are determined based on static rebound effects. For example, the coal production rebound in the northeast region is about 59.3%, so the coal subsidy in the region under the C01 scenario is 0.53%. Since the coal subsidy rate is very small, the setting of scenario C01 is almost equivalent to the cancellation of coal subsidies, so we do not analyse the scenario of cancelling coal subsidies. Scenario P02 is set to cancel the oil subsidy.

The rebound rates and real GDPs changes under different scenarios are given in Table 4. In each policy scenario, real GDP has fallen. In other words, the negative impact of reducing or eliminating energy subsidies exceeds its positive contribution to economic growth. In addition, the elimination or reduction of coal and oil subsidies would result in a reduction in overall and regional rebound effects, while the production-side rebound will increase. In particular, when the oil subsidy is abolished, the production-side rebound would increase sharply in all regions, and the consumption-side rebound would be negative. This is because reducing subsidies would lead manufacturers to pass on their production costs to consumers, which result in increased users' energy consumption costs, resulting in an involuntary reduction in their energy needs.

Table 4. Real GDP changes and rebound effects in policy scenarios (%).

	Northeast	JJJ	South	Northwest	Southwest	JJJ Rim	YRD	Central	PRD
Real GDP changes
C01	−0.089	−0.115	−0.044	−0.157	−0.087	−0.12	−0.029	−0.092	−0.028
P01	−0.114	−0.108	−0.007	−0.32	−0.076	−0.068	−0.019	−0.021	−0.015
P02	−0.401	0.033	−0.094	−0.538	−0.049	−0.143	−0.082	−0.069	−0.041
Overall rebound effect
v C01	71.28	68.66	77.67	75.93	71.7	72.66	81.29	68.12	80.82
P01	74.06	91.15	88.65	75.34	90.19	89.88	82.89	92.65	85.78
P02	55.81	86.45	81.01	49.95	83.04	84.59	72.11	88.55	77.64
Production–side rebound effect
C01	75.05	65.52	82.35	75.66	59.34	61.91	74.87	68.2	72.83
P01	78.59	88.08	88.16	79.19	90.19	88.19	85.45	89.45	85.55
P02	110.31	101.42	106.3	100.4	109.47	105.34	111.16	104.44	105.1

The results of scenarios P1 and P2 indicate that the more subsidies are reduced, the greater the negative impact on the economy. In terms of regional differences, areas with more heavy industries such as the Northeast and Northwest regions are more affected by subsidies, and the rebound effect in these areas has also declined more. From the above results, it can be seen that reducing or completely eliminating fossil energy subsidies not only negatively affects the economic growth in all regions, but also increases production-side energy rebound, resulting in loss of consumer welfare. Therefore, while implementing energy subsidy reform, other corresponding policies need to be adopted to prevent the transfer of energy production costs.

VII. Discussions and conclusions

A multi-regional CGE model was constructed to estimate the impact of energy efficiency improvement on China's regional economy, and to measure the magnitudes of the overall, production–side and consumption-side regional rebound effects. According to the rebound rates calculated, we set up different energy subsidy reform scenarios. Besides, we studied the long-term trends in the regional rebound effects. The main findings are as follows:

Energy efficiency improvement leads to an increase in the total output, and its impact on the regional economy is mainly related to the energy consumption structure of each region, which is consistent with the findings in Lu, Liu, and Zhou ([26]). Additionally, energy efficiency improvement can reduce energy consumption to a certain extent.

Wen et al. ([38]) proved that the direct and indirect household rebound effects are quite different among various regions in China, significant differences in the regional rebound effects of different energy sources were also found in this paper. For example, the production-side oil rebound is 82.9% in the Central region, whereas the production-side power rebound is below −50% in the South. Furthermore, the regional rebound effects of fossil energy are positive, whereas the production-side rebounds of power in most regions are below zero, which is similar to the results of Lu, Liu, and Zhou ([26]). The results of energy subsidy reform scenarios show that the reduction of coal and oil subsidies can effectively decrease the overall rebound effect, but would increase production-side rebounds and slow down the regional economic growth.

Since clean energies account for relatively large proportions of energy consumption structures in the Pearl River Delta and the South, the dynamic rebound effects of coal, oil and electricity in these regions show significant downward trends. According to the above results, the following policy recommendations are proposed:

Due to the obvious differences in the rebound effects of various regions, Chinas energy policy should vary from region to region. For example, the differences in regional oil rebounds are not large, so it is feasible to implement a unified oil subsidy policy in all regions, whereas the subsidy policies for coal in different regions should be different. Additionally, the consumption-side coal rebound effect of each region is relatively large, which can be partly reduced by the current coal to gas measure. Moreover, the long-term rebound trend in the region where the proportion of clean energy is large and the tertiary industry is dominant would decline, and then the region dominated by the secondary industry should accelerate the transformation of energy consumption structure and industrial structure.

The regional gas rebound effects are around 100% due to the low consumptions of natural gas. Accordingly, it is an urgent task to increase the use of gas. The government should vigorously promote the advancement of natural gas technology and provide more supportive policies besides subsidies.

The power rebound effect is mainly from the consumption side. Future power policies should aim to accelerate the development of renewable energy power generation technologies and minimize the use of fossil energy. In addition, electricity price reform is also an important policy tool to reduce the consumption-side rebound.

At present, it is difficult to change the huge demand for coal and oil in the industrial sectors simply by the policy of reducing energy production-side subsidies. This requires support for energy subsidy reform through overall macroeconomic policies. This requires coordination and support for energy subsidy reform through macroeconomic policies that are linked to the overall situation. In particular, the government should redistribute the benefits that energy producers receive from raising energy prices to compensate consumers who suffer welfare losses. In addition, in order to curb price volatility and related market distortions, transparent and coordinated fiscal, monetary and exchange rate policies need to be implemented as well.

Disclosure statement

No potential conflict of interest was reported by the authors.

1 Brännlund, R., T. Ghalwash, and J. Nordström. 2007. " Increased Energy Efficiency and the Rebound Effect: Effects on Consumption and Emissions." Energy Economics 29 (1): 1 – 17. 2 Brookes, L. 1990. " The Greenhouse Effect: The Fallacies in the Energy Efficiency Solution." Energy Policy 18 (2): 199 – 201. 3 Chen, Y.-H., S. Paltsev, J. M. Reilly, J. F. Morris, and M. H. Babiker (2015). " The MIT EPPA6 Model: Economic Growth, Energy Use, and Food Consumption (Tech. Rep.)." Massachusetts, America: MIT Joint Program on the Science and Policy of Global Change. 4 Dai, H., T. Masui, Y. Matsuoka, and S. Fujimori. 2012. " The Impacts of Chinas Household Consumption Expenditure Patterns on Energy Demand and Carbon Emissions Towards 2050." Energy Policy 50 : 736 – 750. 5 Dimitropoulos, J. 2007. " Energy Productivity Improvements and the Rebound Effect: An Overview of the State of Knowledge." Energy Policy 35 (12): 6354 – 6363. 6 Druckman, A., M. Chitnis, S. Sorrell, and T. Jackson. 2011. " Missing Carbon Reductions? Exploring Rebound and Backfire Effects in UK Households." Energy Policy 39 (6): 3572 – 3581. 7 Freire-González, J. 2011. " Methods to Empirically Estimate Direct and Indirect Rebound Effect of Energy-Saving Technological Changes in Households." Ecological Modelling 223 (1): 32 – 40. 8 Glomsrød, S., and W. Taoyuan. 2005. " Coal Cleaning: A Viable Strategy for Reduced Carbon Emissions and Improved Environment in China? " Energy Policy 33 (4): 525 – 542. 9 Gonzalez, J. F. 2010. " Empirical Evidence of Direct Rebound Effect in Catalonia." Energy Policy 38 (5): 2309 – 2314. Greening, L. A., D. L. Greene, and C. Difiglio. 2000. " Energy Efficiency and Consumptionthe Rebound Effecta Survey." Energy Policy 28 (6–7): 389 – 401. Hanley, N., P. G. McGregor, J. K. Swales, and K. Turner. 2009. " Do Increases in Energy Efficiency Improve Environmental Quality and Sustainability? " Ecological Economics 68 (3): 692 – 709. Herring, H. 2006. " Energy Efficiency a Critical View." Energy 31 (1): 10 – 20. Herring, H., and R. Roy. 2007. " Technological Innovation, Energy Efficient Design and the Rebound Effect." Technovation 27 (4): 194 – 203. Hong, L., D. Liang, and W. Di. 2013. " Economic and Environmental Gains of China's Fossil Energy Subsidies Reform: A Rebound Effect Case Study with EIMO Model." Energy Policy 54 : 335 – 342. Hong, L., D. Liang, and X. Minghua. 2011. " A Study on the Comprehensive Evaluation and Optimization of How Removing Gas and Electricity Subsidies Would Affect Households' Living." Economic Research Journal 2 (9): 100 – 112. Hymel, K. M., and K. A. Small. 2015. " The Rebound Effect for Automobile Travel: Asymmetric Response to Price Changes and Novel Features of the 2000s." Energy Economics 49 : 93 – 103. Hymel, K. M., K. A. Small, and K. Van Dender. 2010. " Induced Demand and Rebound Effects in Road Transport." Transportation Research Part B: Methodological 44 (10): 1220 – 1241. Jin, L., K. Duan, and X. Tang. 2018. " What Is the Relationship between Technological Innovation and Energy Consumption? Empirical Analysis Based on Provincial Panel Data from China." Sustainability 10 (1): 145. Khazzoom, J. D. 1980. " Economic Implications of Mandated Efficiency in Standards for Household Appliances." The Energy Journal 1 (4): 21 – 40. Li, H., Q. Bao, X. Ren, Y. Xie, J. Ren, and Y. Yang. 2017. " Reducing Rebound Effect through Fossil Subsidies Reform: A Comprehensive Evaluation in China." Journal of Cleaner Production 141 : 305 – 314. Lin, B., and J. Li. 2014. " The Rebound Effect for Heavy Industry: Empirical Evidence from China." Energy Policy 74 : 589 – 599. Lin, B., and H. Liu. 2015. " A Study on the Energy Rebound Effect of China's Residential Building Energy Efficiency." Energy and Buildings 86 : 608 – 618. Lin, B., and X. Liu. 2013. " Electricity Tariff Reform and Rebound Effect of Residential Electricity Consumption in China." Energy 59 : 240 – 247. Liu, W., and H. Li. 2011. " Improving Energy Consumption Structure: A Comprehensive Assessment of Fossil Energy Subsidies Reform in China." Energy Policy 39 (7): 4134 – 4143. Liu, W., T. Zhipeng, and C. Jie. 2014. Interregional Input-Output Table of China's 30 Provinces, Autonomous Regions and Municipalities in 2010. Beijing: China Statistics Press. Lu, Y., Y. Liu, and M. Zhou. 2017. " Rebound Effect of Improved Energy Efficiency for Different Energy Types: A General Equilibrium Analysis for China." Energy Economics 62 : 248 – 256. Ma, X., H. Wang, and W. Wei. 2019. " The Role of Emissions Trading Mechanisms and Technological Progress in Achieving Chinas Regional Clean Air Target: A CGE Analysis." Applied Economics 51 (2): 155 – 169. Matos, F. J., and F. J. Silva. 2011. " The Rebound Effect on Road Freight Transport: Empirical Evidence from Portugal." Energy Policy 39 (5): 2833 – 2841. Saunders, H. D. 1992. " The Khazzoom-Brookes Postulate and Neoclassical Growth." The Energy Journal 13 (4): 131 – 148. Saunders, H. D. 2013. " Historical Evidence for Energy Efficiency Rebound in 30 US Sectors and a Toolkit for Rebound Analysts." Technological Forecasting and Social Change 80 (7): 1317 – 1330. Schleich, J., B. Mills, and E. Dütschke. 2014. " A Brighter Future? Quantifying the Rebound Effect in Energy Efficient Lighting." Energy Policy 72 : 35 – 42. Semboja, H. H. H. 1994. " The Effects of an Increase in Energy Efficiency on the Kenya Economy." Energy Policy 22 (3): 217 – 225. Sorrell, S. 2007. The Rebound Effect: An Assessment of the Evidence for Economy-Wide Energy Savings from Improved Energy Efficiency. London : UK Energy Research Centre. Turner, K. 2013. " Rebound Effects from Increased Energy Efficiency: A Time to Pause and Reflect." The Energy Journal 34 (4): 25 – 42. Vikström, P. 2008. Energy Efficiency and Energy Demand: A Historical Cge Investigation on the Rebound Effect in the Swedish Economy 1957. Working paper, Umeå University, Sweden. Ekonomisk historia. Wang, H., D. Zhou, P. Zhou, and D. Zha. 2012. " Direct Rebound Effect for Passenger Transport: Empirical Evidence from Hong Kong." Applied Energy 92 : 162 – 167. Wang, H., P. Zhou, and D. Zhou. 2012. " An Empirical Study of Direct Rebound Effect for Passenger Transport in Urban China." Energy Economics 34 (2): 452 – 460. Wen, F., Z. Ye, H. Yang, and K. Li. 2018. " Exploring the Rebound Effect from the Perspective of Household: An Analysis of China's Provincial Level." Energy Economics 75 : 345 – 356. Wu, L., Y. Chen, M. R. Feylizadeh, and W. Liu. 2018. " Estimation of China's Macro-Carbon Rebound Effect: Method of Integrating Data Envelopment Analysis Production Model and Sequential Malmquist-Luenberger Index." Journal of Cleaner Production 198 : 1431 – 1442. Yang, L., and J. Li. 2017. " Rebound Effect in China: Evidence from the Power Generation Sector." Renewable and Sustainable Energy Reviews 71 : 53 – 62. Zhang, X., X. Zhao, Z. Jiang, and S. Shao. 2017. " How to Achieve the 2030 CO 2 Emission-Reduction Targets for China's Industrial Sector: Retrospective Decomposition and Prospective Trajectories." Global Environmental Change 44 : 83 – 97. Zhang, Y.-J., H.-R. Peng, Z. Liu, and W. Tan. 2015. " Direct Energy Rebound Effect for Road Passenger Transport in China: A Dynamic Panel Quantile Regression Approach." Energy Policy 87 : 303 – 313. Zhou, M., Y. Liu, S. Feng, Y. Liu, and Y. Lu. 2018. " Decomposition of Rebound Effect: An Energy-Specific, General Equilibrium Analysis in the Context of China." Applied Energy 221 : 280 – 298. Source: China Energy Statistics Yearbook 2017,http://www.stats.gov.cn/tjsj/tjcbw/201806/t20180612%5f1604117.html. Source: the 13th Five-Year Plan for Energy Development,http://www.ndrc.gov.cn/gzdt/201701/W020170117348373245462.pdf. For detailed derivation, see Lu, Liu, and Zhou ([26]). Source: National Bureau of statistics of China, http://www.stats.gov.cn//ztjc/zthd/sjtjr/dejtjkfr/tjkp//201106/t20110613_71947.htm. Source: United Nations, https://population.un.org/wpp/. Source: Catherine Early, Health cost of China's fossil fuel subsidies highlighted, Chinadialogue, 8 February 2017, https://www.chinadialogue.net/article/show/single/en/9956-Health-cost-of-China-s-fossil-fuel-subsidies-highlighted. Source: Financial Times, http://blogs.ft.com/the-world/files/2016/07/GR262Xcarbon%5ftax%5fmodern%5fenergy%5fSR%5fCHART.png.

By Ningjing Wang and Weixian Wei

Reported by Author; Author