Abstract
Energy is becoming more and more important in the process of development. Through cluster analysis, 30 provinces chosen from the Chinese mainland are divided into high, medium, and low energy consumption regions, and the Theil index is used to analyze the characteristics of total energy consumption and other characteristics of regional differences. Based on this, an enhanced Panel-STIRPAT model is constructed. Through data and model inspection, panel models suitable for each region are selected for comprehensive analysis. The results show that: there are regional differences in energy consumption in China, and the total regional differences mainly come from within each region. The factors affecting energy consumption in different regions are the same, but due to differences in geographical environment and stress levels, the influence of these factors on energy consumption in different regions is different. Based on this, reasonable measurements to control energy consumption in different regions are presented.
Keywords: Energy consumption, regional differences, Theil index, Panel-STIRPAT model
Introduction
Since the steam age, human demand for energy has been expanding. Energy, as an indispensable factor for the progress and development of human society, is an important foundation for promoting rapid economic development and improving people's basic living standards. However, problems such as excessive demand, limited energy, and resource shortages have gradually become serious with China's rapid development. Only by rationally utilizing the limited resources and scientifically allocating the energy demand can the energy give full play to its maximum effectiveness.
Although China's total energy consumption is high, the per-capita consumption is not high due to the large population. In addition, due to differences in geographic environment, resource distribution, and political environment, the energy consumption and economic development of various provinces are also very different. Therefore, in order to better reflect the development law of energy consumption in different regions of China, it is necessary to cut in from the regional perspective and conduct in-depth analysis and research, to determine whether there are regional differences in China's energy consumption, and what factors are responsible for these differences, and finally propose policies and measures suitable for the development of energy consumption in various regions. This is of great significance for China to step out of the current energy situation that does not conform to the connotation of sustainable development.
Compared with the existing literature, two major contributions are made in this study.
First, this study not only classifies China's regional energy consumption by using cluster analysis but also analyzes the characteristics of total energy consumption and other characteristics of regional differences by the Theil index. The combination of the two methods is an innovation and makes the research conclusions more credible.
Second, this study constructs an enhanced Panel-STIRPAT model and conducts an empirical study on influencing factors of regional energy consumption. This is an enhanced STIRPAT model and empirical studies have shown better applicability.
The paper is organized as follows. Related works are reviewed in Section 2. Section 3 conducts China's energy consumption regional classifying. Section 4 presents empirical research on regional differences in energy consumption based on the Theil index. Section 5 conducts an analysis and model construction of influencing factors from China's energy consumption. An empirical study on influencing factors of regional energy consumption based on the Panel-STIRPAT model is shown in Section 6. Sections 7 and 8 present the research conclusions and policy recommendations.
Literature review
As global warming continues to intensify and the concept of sustainable development continues to spread, more and more scholars at home and abroad have begun to pay attention to the issue of energy consumption. Some experts, scholars and related organizations have conducted research and analysis on issues related to China's energy demand from different perspectives. Through sorting out the related literature on energy consumption, it is found that domestic and foreign scholars’ research on energy consumption influencing factors and regional differences mainly focuses on the following aspects.
Research on regional differences in energy consumption
Energy consumption in different regions
Blanco et al. evaluated the effects of the regional production of clean energy and identified the employment generated in the renewable sector. The adopted methodology was the shift-share analysis, frequently used by researchers to analyze territorial differences. 1 Yan and Su regarded that the spatial differences in energy consumption were distinct in China, with the greatest gap between Shanghai and Chongqing, which were mainly driven by the domestic Leontief structure effect and the total final demand effect. 2
Zhou et al. did an empirical study to confirm that not only how much is consumed and what is consumed, but also geographical origins (and, by implication, regionally specific production processes and methods) influence consumption-induced ecological impacts. 3 Eguchi et al. argued that the operational inefficiency is the main source of inefficiency in eastern and central China, however, the technology gap the differences in the quality of coal consumed for electricity production and in the equipment of the power plants among regions is the main source of inefficiency in western China. 4 Oswald et al. thought that energy-intensive goods tend to be more elastic, leading to higher energy footprints of high-income individuals. 5
Zhang et al. found that the place with the highest energy consumption in China is Guangdong, and Shanxi Province has the largest energy reserves in China. It is necessary to transport coal and other resources from the western region to the eastern region for energy consumption and economic development in the Chinese eastern region. 6 Renou-Maissant found some energy alternatives by analyzing the energy consumption structure of several major industrial countries in the world. 7
Chen et al. wanted to know what is the relationship between economic growth and energy consumption, and how the change in energy consumption affects economic growth. They used the MS-VAR model to study a lot of energy consumption data. The research results show that economic growth and energy consumption influence and restrict each other, and there is a strong asymmetry in the development relationship. 8 Cheong et al. demonstrated significant divergence presents across provinces, over time and within different regional groups. The results can pinpoint the transition mechanism within each region so that appropriate energy policy can be formulated to accommodate future demand for electricity for different regions in China. 9
Research on regional differences of influencing factors of energy consumption
Lutz et al. knew what drives the regional implementation of renewable energy is a prerequisite for energy transitions toward a post-fossil-based energy economy. They did an empirical analysis of driving factors for the regional implementation and use of renewable energy. Their findings confirm most of the driving factors identified in the literature, for example, the existence of key actors, knowledge exchange, or the use of goals and milestones, and they also observed differences in key driving factors between highly successful and less successful regions, especially regarding funding opportunities. 10 Park and Kwon argued that disutility from briquette consumption is considered an important factor. Using the census data on briquette-consuming households, it is found that the coupon program provides an adverse effect to switching fuels to clean energy while the disutility of briquettes is positively associated with the probability of fuel switching. 11
Adua and Clark revealed mixed relationships between investment in energy efficiency technologies and residential energy consumption, as some measures of efficiency technology are negatively related to residential energy consumption, while others are positively related to it. 12 Achour and Belloumi indicated that the overall effect of economic output, transportation intensity, population scale, and transportation structure on energy consumption is positive, whereas the overall effect of energy intensity is negative. It was shown that energy intensity played the dominant role in decreasing energy consumption during the study period. 13
Sun et al. analyzed and studied the energy consumption of Shanxi Province in recent years, and made a comprehensive analysis of the factors that will affect Shanxi. The results found that the factors that have less and less obvious impacts on energy consumption are mainly scaled effects and technological effects, while the more and more important factors are structural effects. 14 Through research, Zhang took out energy consumption and economic growth data from 1986 to 2005 to make a series of analyses on it and found that the impact of economic growth on energy consumption is different, which means the richer regions are more affected. 15
Nie also conducted a series of research and discussion on urban energy consumption data from 1996 to 2006. He concluded that the higher the level of urbanization, the higher the degree of impact on energy consumption, and the degree of impact is based on energy The economically rich regions, the rich regions in the energy economy, and the energy economically not rich regions show a sequential increase in the phenomenon. 16 Yu used spatial panel data models, and found that GDP per capita, transportation infrastructure, the level of marketization, and scientific and technological input significantly reduce the energy intensity; the ratio of heavy industries to total industries and the ratio of coal consumption to total energy consumption significantly expand the energy intensity; meanwhile, the coefficient of the ratio of export to GDP is not significant. 17
Research on the influencing factors of energy consumption
The energy consumption and its influencing factors are always a hot topic. Many scholars have used a variety of theoretical and mathematical modeling methods to analyze the influencing factors of energy consumption in different countries and regions. Each person's analysis method and the angle of analysis are different, so their conclusions are also very different.
Research on population size and energy consumption
Salim and Shafiei revealed that while total population and urbanization positively influence non-renewable energy consumption, population density has a negative impact on non-renewable energy consumption. 18 Zaharia et al. showed that factors such as greenhouse gas emissions, gross domestic product, population and labor growth have a positive relationship with both primary and final energy consumption, which means an increase in energy consumption. Meanwhile, factors such as feminine population increase, healthcare expenditures, or energy taxes have a negative relationship, which determines a reduction of energy consumption. 19
Abam et al. identified three decoupling states, weak negative decoupling, weak decoupling, and strong decoupling. The energy intensity, economic activity, population, and energy structure prevented decoupling during the study period, while the economic structure factor promoted decoupling. 20 Zarco-Soto et al. showed that cities with larger populations present higher consumptions per inhabitant and household. The smallest the population of a city is, the less energy the city consumes. 21 Zhang argued that social development is getting faster and faster, and people's living standards have gradually improved. When people start to pursue high-energy-consuming products, a connection will be created. When the population is increasing, the total energy consumption is also increasing. 22 After studying the energy consumption of developed countries, Wang believed that the greater the population, the greater the energy consumption. 23
Research on the level of urbanization and energy consumption
Yumashev et al. found that the size and rating of the human development index (HDI) are influenced by such factors as urbanization growth, gross domestic product (GDP), gross national income (GNI) per capita, the share of "clean" energy consumption by the population and business in total energy consumption, the level of socio-economic development, and R&D expenses. 24 Apergis and Li showed that the rural population migrating to cities consumes more energy services and produces larger emissions since urban lifestyles are generally more energetic and carbon-intensive. Migration and urbanization together drive China's energy consumption upwards and environmental quality downwards if the current trend continues over time. 25
Hu and Fan argued that the current expansion of China's city size tends to positively affect energy consumption; however, as city size continues to expand, energy consumption will exceed the critical value and change from increasing to decreasing. 26 Ahmed et al. found that financial development economic growth, industrialization, and urbanization have a positive impact on energy consumption in the short-run but industrialization has no impact in the long run. 27 Liddle thought that the higher the level of urbanization, the more cities, and the more convenient public transportation. With more popular resources, the per-capita level of people's energy consumption will decrease. 28
Mishra et al. also showed the same conclusions as Liddle. Of course, the process of urbanization requires construction, so in the initial stage of urbanization, energy consumption will increase significantly, but after urbanization, the long-term impact on energy consumption will gradually decrease. 29 Liu and Tang analyzed comprehensive data on urbanization and energy consumption from 1953 to 2011 since the founding of the People's Republic of China and found that there is a non-linear dynamic relationship between urbanization and energy consumption, rather than a positive linear correlation. 30 Lin and Jiang studied the panel data from 1997 to 2012, when the process of urbanization in China was relatively fast. He found that the impact of urbanization on energy consumption is positively correlated. 31 Yang constructed a STIRPAT model through modeling methods and used it to study the effect of urbanization in the three places of Beijing-Tianjin-Hebei on energy consumption. All of them are positive because of the large population differences in these places. The main factors affecting energy consumption are different in different places. 32
Research on economic development and energy consumption
Noorpoor and Kudahi did an analysis of socio-economic influencing parameters. It is performed by the stochastic impacts by regression on population, affluence and technology (STIRPAT) model using population size, gross domestic product (GDP) per capita, electricity intensity and consumption of energy resources for electricity generation. 33 Andreoni showed that the standard of living has been the main factor influencing the energy consumption increase. In addition, driven by structural changes and implementation of energy efficiency action plans, has contributed to reducing the energy used in relation to gross domestic product and human time. 34 ul Husnain et al. indicated short-run coherence among energy consumption and economic growth of all the top 10 energy-consuming countries. Long-run dependence between energy consumption and economic growth exists in the case of China with a mostly leading role of energy consumption over economic growth. 35
Mohsin et al. revealed that an increase in economic growth, urbanization, and energy consumption increased transport-based environmental degradation urbanization. Moreover, energy consumption has increased by 13.5%, and it shows a high dependence of economic growth on energy consumption. 36 Kraft J and Kraft A believed that economic growth needs energy supply support, and the effect on energy consumption is positive. 37 Yu and Jin adopted the co-integration technique to analyze the data of the United States from 1974 to 1990 in each quarter to see what kind of relationship exists between these data and energy consumption and income. They found that there is a relationship between these two things. There is no long-term balance. 38 Zhao and Fan also found that economic growth and energy consumption are not linearly correlated with positive proportions. The relationship between them is an asymmetry in the data model, and in different stages of economic development, the non-linear relationship between the more obvious. 39 Wang Peng adopted the panel co-integration theory to compare and analyze the energy consumption and economic growth of four major regions in China, and he found that the relationship between energy consumption and economic growth in these four regions is very different. 40
Research on industrial structure and energy consumption
Clark et al. suggested that export-oriented manufacturing increases coal consumption in developing nations but not in developed nations. 41 Yu et al. obtained the following results: (1) although sigma convergence and beta conditional convergence are observed, stochastic convergence is not observed for the 74 cities; (2) the per-capita GDP and energy structure are the key factors influencing the cities’ convergence, whereas the urbanization rate has a limited impact; and (3) improving the industrial structure and reducing the population density are effective methods of promoting the convergence of per capita carbon emissions among the capital and sub-provincial cities and municipalities, respectively. 42 Luan et al. indicated that a change in the industrial structure optimization leads to a change in the relationships among the industrial structure adjustment, technological progress, and energy intensity. This study also shows that the energy intensity in China's central and western regions has a huge potential downside, through the optimization of industrial structure. 43
Pan et al. showed that the tightening effect of energy intensity constraints on the industrial sector is most significant, followed by the tertiary industry, with the least impact on Agriculture and when there is no technological progress in the departments, the change of industrial structure is mainly reflected in the sharp decline in the proportion of Industry and the significant increase in the proportion of Tertiary Industry. In comparison of personnel consumption in these industries, researchers believe that the secondary industry's impact on energy consumption is too great. 44
Zhang pointed out that after adjusting the structure of the tertiary industry, the indispensability of energy consumption for economic development is decreasing. 14 Shang established a multiple linear regression model based on China's energy consumption data and economic development data from 1990 to 2009 and used this model to support the conclusion that adjusting the tertiary industry is useful for reducing energy consumption. 45
Literature summary
Based on the research on the existing literature on economic development and energy consumption, it can be seen the main research regions are population size, urbanization level, economic development status, and industrial structure, which mainly affect the economic situation. The regional research has also been carried out from multiple levels. However, there are some shortcomings in these studies. First, there are relatively few studies on the differences in energy consumption in China's different regions. This factor is the prerequisite and guarantee for the study of regional energy consumption. Therefore, this study will conduct certain research on this aspect. Second, when determining the influencing factors that can affect energy consumption, the research objects are relatively non-selective. The factors that affect energy consumption should be comprehensively analyzed, and some key factors should be selected for a systematic, comprehensive, multi-directional, and multi-angle in-depth analysis, then take huge data to find out the influence mode and degree of these factors. Third, the analysis of regional differences in energy consumption is relatively simple, and there is a lack of comprehensive research and analysis of regional differences in energy consumption under the influence of multiple factors.
Therefore, this study will analyze the differences and influencing factors of China's regional energy consumption from different perspectives, combine the actual situation, break the traditional regional administrative division restrictions, and use the systematic clustering method to regroup China's 30 provinces and municipalities into three types of energy consumption regions; and use the enhanced Panel-STIRPAT model, which embeds the social fixed asset investment, urbanization level and other influencing factors in it, to discover the degree and method of the influence of these influencing factors on the rich, low, and middle-rich regions of the energy economy.
China's energy consumption regional classifying
Since the reform and opening up, China has vigorously developed energy production and energy consumption for rapid economic development, and energy consumption has risen sharply, thus achieving rapid economic growth. In 2000, energy consumption was 1.45 billion tons of standard coal. In 2020, it consumed 4.98 billion tons of standard coal. The total energy consumption has tripled in 20 years. Economic construction is inseparable from energy. Energy is an indispensable material for life and industrial construction. Energy is of indispensable significance for economic development. However, China's energy distribution is extremely unbalanced. Energy distribution in China is affected by many factors, not only by human geography and economic levels, but also by energy consumption willingness. In order to study the differences in China's energy consumption in different regions, it is necessary to classify.
Determination of indicators
This study selects seven indicators as the standard indicators for cluster analysis. They are: , total energy consumption; , unit per capita GDP; , the proportion of tertiary industry in GDP; , the proportion of secondary industry in GDP; , population urbanization rate; , the population at the end of 2017; , fixed asset investment in the whole society.
Data source and processing
This article selects and analyzes some data from the China Energy Statistical Yearbook, as well as energy consumption data from other websites, such as the China Statistical Yearbook and the website of the National Bureau of Statistics. But the data contained in the "China Energy Statistical Yearbook" is still limited and cannot meet the research needs on energy consumption. To ensure that the data is complete and the empirical research can be carried out normally, this paper sets the data range of the research as the 18 years from 2000 to 2017. And in terms of regional selection, this study excludes Tibet and Taiwan because of the few energy consumption data relatively, remainders 30 provinces in the Chinese mainland.
Cluster analysis of energy consumption region
According to the established data indicators, 30 provinces in the Chinese mainland are regrouped into three types of energy consumption regions by using the hierarchical clustering method. The result is shown in Figure 1. The first category is energy-rich economic zones, including Beijing, Tianjin, and Shanghai. The second category is the rich regions in the energy economy, including 11 regions in Anhui, Jiangsu, Zhejiang, Henan, Guangdong, Shandong, Hebei, Sichuan, Hubei, Hunan, and Liaoning. The third category is economically under-energy regions, including 16 regions in Jiangxi, Hainan, Shaanxi, Shanxi, Heilongjiang, Jilin, Chongqing, Inner Mongolia, Fujian, Gansu, Ningxia, Qinghai, Guizhou, Guangxi, Xinjiang, and Yunnan.
Figure 1.
The energy consumption cluster tree of 30 provinces chosen from the Chinese mainland from 2000 to 2017.
Empirical research on regional differences in energy consumption based on the Theil index
As mentioned above, due to the different geographical environments, resources and economic levels of different regions in China, the energy consumption of different regions is also different. This study will use the Theil index to analyze whether there are different degrees of energy consumption differences among provinces, cities, and autonomous regions, as well as to find out the reasons and types of differences among them.
Theil index
Theil index is a kind of relative difference method. Theil coefficient is generally a positive number, and the larger the value, the greater the difference among different regions. The calculation formula is as follows.
(1) |
Where E represents the sum of energy consumption in each region, T represents the value of the calculated Theil index, represents the total energy consumption of the i province, n represents the number of provinces in each region; X represents the variable to be studied, for example, if represents the GDP, represents the province's GDP, in Equation (4-1), the numerical result indicates the degree of difference in energy consumption among regions.
The Theil index can also be disassembled, . The following are their respective expressions.
(4-2) |
(4-3) |
Where represents the regional Theil index, represents the inter-regional Theil index, represents the regional Theil index, represents the energy consumption of a certain province in a certain region, E represents the total energy consumption of China, represents the GDP or population of a regional province, X represents the GDP or the total population.
Further disassembling and analyzing the Theil index, the research status of the contribution rate in the region can be gotten. The related equations are shown as follows.
(4-4) |
(4-5) |
(4-6) |
Where Represents the contribution rate within each region, represents the inter-regional contribution rate, represents the intra-regional contribution rate.
Analysis of Theil index of energy consumption in three major regions
Theil index of energy consumption weighted by GDP
The Theil index of three major regions are represented by (rich region in energy), (medium-rich region in energy) and (energy-poor region). The calculation results are shown in Table 1 and Figure 2.
Table 1.
Theil index of energy consumption intensity of three major regions.
Year | GDP) | GDP) | GDP) | GDP) | (GDP) |
---|---|---|---|---|---|
2000 | 0.0090 | 0.0569 | 0.1621 | 0.0889 | 0.0137 |
2001 | 0.0071 | 0.0486 | 0.1695 | 0.0870 | 0.0161 |
2002 | 0.0067 | 0.0471 | 0.1719 | 0.0874 | 0.0178 |
2003 | 0.0056 | 0.0462 | 0.1638 | 0.0838 | 0.0186 |
2004 | 0.0075 | 0.0462 | 0.1547 | 0.0804 | 0.0182 |
2005 | 0.0050 | 0.0438 | 0.1334 | 0.0712 | 0.0183 |
2006 | 0.0070 | 0.0444 | 0.1329 | 0.0716 | 0.0182 |
2007 | 0.0100 | 0.0423 | 0.1290 | 0.0694 | 0.0179 |
2008 | 0.0088 | 0.0385 | 0.1137 | 0.0621 | 0.0162 |
2009 | 0.0100 | 0.0376 | 0.1103 | 0.0606 | 0.0167 |
2010 | 0.0125 | 0.0368 | 0.1064 | 0.0589 | 0.0146 |
2011 | 0.0152 | 0.0354 | 0.1093 | 0.0599 | 0.0148 |
2012 | 0.0169 | 0.0334 | 0.1136 | 0.0610 | 0.0156 |
2013 | 0.0206 | 0.0296 | 0.1279 | 0.0647 | 0.0173 |
2014 | 0.0193 | 0.0299 | 0.1289 | 0.0656 | 0.0194 |
2015 | 0.0225 | 0.0328 | 0.1348 | 0.0697 | 0.0210 |
2016 | 0.0235 | 0.0408 | 0.1428 | 0.0774 | 0.0225 |
2017 | 0.0232 | 0.0450 | 0.1495 | 0.0828 | 0.0259 |
Figure 2.
The evolution trend of the Theil index of energy consumption intensity in various regions.
From the analysis of the characteristics and differences among the curves in Figure 2, it can be concluded that the year-on-year trend of energy consumption intensity in regions with low energy economy is a slow decline and then an increase year by year, and a decline from 2000 to 2010, the difference between energy consumption and economic growth in the energy-poor region in the past decade has become smaller year by year, and the degree of matching between energy consumption and economic growth has increased year by year. After 2010, the energy intensity Theil index began to increase, the difference increased, and the degree of matching decreased. Moreover, the Theil coefficient of energy consumption intensity is the largest in the energy-non-rich regions. Compared with the rich regions and the medium-rich regions, there is no inevitable connection between the influencing factors of energy consumption and economic growth in the poor regions. The curves of Theil coefficients of energy intensity in the rich and medium-rich regions are relatively flat without major fluctuations, which indicates that the two factors of energy consumption and economic growth are closely linked between the two regions.
Theil index of energy consumption weighted by population
It can be seen from Figure 3 and Table 2 that the curve of the energy-poor economic zone is in a higher position, and the value of each year is ahead of other regions, indicating that the energy consumption of the energy-poor economic zone has the smallest matching degree with the population and the difference is the largest. The energy economy affluent region has the smallest per capita energy consumption Theil index (population), that is, the difference between energy consumption and population in this region is the smallest. The Theil index curve of per-capita energy consumption in rich regions in the energy economy is flat, indicating that there is a stable relationship between population and energy consumption.
Figure 3.
The evolution trend of the Theil index per-capita energy consumption in each region.
Table 2.
Theil index of per-capita energy consumption in the three major regions.
Year | ) | ) | ) | ) | (p) |
---|---|---|---|---|---|
2000 | 0.0033 | 0.0662 | 0.1418 | 0.0868 | 0.0333 |
2001 | 0.0028 | 0.0560 | 0.1537 | 0.0854 | 0.0312 |
2002 | 0.0027 | 0.0512 | 0.1606 | 0.0855 | 0.0275 |
2003 | 0.0044 | 0.0510 | 0.1583 | 0.0846 | 0.0201 |
2004 | 0.0025 | 0.0471 | 0.1556 | 0.0809 | 0.0164 |
2005 | 0.0035 | 0.0440 | 0.1542 | 0.0782 | 0.0129 |
2006 | 0.0039 | 0.0433 | 0.1601 | 0.0799 | 0.0111 |
2007 | 0.0047 | 0.0418 | 0.1631 | 0.0803 | 0.0093 |
2008 | 0.0081 | 0.0401 | 0.1572 | 0.0778 | 0.0077 |
2009 | 0.0090 | 0.0395 | 0.1481 | 0.0745 | 0.0063 |
2010 | 0.0130 | 0.0359 | 0.1413 | 0.0703 | 0.0054 |
2011 | 0.0176 | 0.0347 | 0.1453 | 0.0722 | 0.0037 |
2012 | 0.0198 | 0.0342 | 0.1454 | 0.0729 | 0.0032 |
2013 | 0.0215 | 0.0344 | 0.1486 | 0.0750 | 0.0034 |
2014 | 0.0213 | 0.0317 | 0.1452 | 0.0728 | 0.0031 |
2015 | 0.0223 | 0.0321 | 0.1443 | 0.0728 | 0.0032 |
2016 | 0.0212 | 0.0304 | 0.1428 | 0.0713 | 0.0033 |
2017 | 0.0184 | 0.0307 | 0.1467 | 0.0733 | 0.0035 |
Comparative analysis of two types of Theil index in each region
The evolution trends of (GDP) and (population) of energy consumption differences in the three major regions from 2000 to 2017 are shown in Figures 4 to 6.
Figure 4.
The evolution trend of energy consumption Theil index in wealthy energy regions.
Figure 6.
The evolution trend of energy consumption Theil index in energy-poor region.
The two types of Theil indices in the rich energy economy zone are represented by (GDP) and (population), respectively. This method is also applicable to the other two economic zones.
From Figure 4, it can be seen that in terms of energy economically rich regions, from 2000 to 2017, (GDP) and (population) are both in a steady upward trend, which shows that in energy economically rich regions, the population and the correlation between energy consumption are decreasing, and regional differences are becoming more and more obvious. Compared with the evolutionary trends of the two, before 2010, (GDP) was greater than (population), indicating that the similarity between energy consumption and population development was greater than that of GDP. One year later, the Theil index between energy consumption and the population is getting closer and closer, which shows that energy-saving measures are effective and per capita, energy consumption is gradually rationalized.
Figure 5 shows that the two types of Theil indices in the rich regions of the energy economy have shown a downward trend, indicating that population growth, economic development, and energy consumption in this region are gradually becoming more closely related over time. The overall energy consumption difference of the region is getting smaller and smaller. And the difference between the two variables is not very obvious. The Theil index of energy consumption under population weight decreased from 0.066 in 2000 to 0.03 in 2017. The Theil index of energy consumption under the weight of GDP began to rise in 2014, indicating that the difference between energy consumption and economic growth in the region has gradually increased since this year.
Figure 5.
The evolution trend of energy consumption Theil index in medium-rich regions in energy.
From Figure 6, it can be seen that there is a big difference in the correspondence between energy consumption, GDP, and population in regions with a low energy economy. From 2004 to 2016, (population) was always greater than (GDP), indicating that the similarity between energy consumption and population during this period was less than the similarity between energy consumption and economic growth, that is to say, there was a big difference in the region between energy consumption and population.
Contrastive analysis of Theil index contribution rates of three regions
Equations (4-1) to (4-6) can be used to analyze the contribution rate of the Theil index of energy consumption differences in different regions to the overall situation so that the reasons for the differences can be found more precisely. After analyzing and interpreting the Theil indices of the six formulas, a series of data arrangements are obtained as shown in Table 3.
Table 3.
Contribution rate of the Theil index of regional energy consumption under two weights.
Year | Wealthy energy region | Medium-rich region in energy | Energy-poor region | In the region | Between the region | |||||
---|---|---|---|---|---|---|---|---|---|---|
W(GDP) | W(P) | W(GDP) | W(P) | W(GDP) | W(P) | W(GDP) | W(P) | W(GDP) | W(P) | |
2000 | 0.0075 | 0.0024 | 0.3161 | 0.3146 | 0.5430 | 0.4059 | 0.8666 | 0.7228 | 0.1334 | 0.2772 |
2001 | 0.0059 | 0.0020 | 0.2679 | 0.2729 | 0.5705 | 0.4573 | 0.8443 | 0.7323 | 0.1557 | 0.2677 |
2002 | 0.0052 | 0.0020 | 0.2545 | 0.2576 | 0.5710 | 0.4968 | 0.8307 | 0.7564 | 0.1693 | 0.2436 |
2003 | 0.0041 | 0.0031 | 0.2617 | 0.2824 | 0.5525 | 0.5221 | 0.8182 | 0.8077 | 0.1818 | 0.1923 |
2004 | 0.0054 | 0.0019 | 0.2759 | 0.2850 | 0.5343 | 0.5444 | 0.8156 | 0.8313 | 0.1844 | 0.1687 |
2005 | 0.0038 | 0.0026 | 0.2920 | 0.2886 | 0.4996 | 0.5670 | 0.7954 | 0.8582 | 0.2046 | 0.1418 |
2006 | 0.0052 | 0.0028 | 0.2959 | 0.2848 | 0.4966 | 0.5906 | 0.7976 | 0.8782 | 0.2024 | 0.1218 |
2007 | 0.0075 | 0.0034 | 0.2897 | 0.2788 | 0.4981 | 0.6139 | 0.7953 | 0.8962 | 0.2047 | 0.1038 |
2008 | 0.0073 | 0.0061 | 0.2927 | 0.2795 | 0.4933 | 0.6241 | 0.7932 | 0.9097 | 0.2068 | 0.0903 |
2009 | 0.0083 | 0.0071 | 0.2899 | 0.2910 | 0.4855 | 0.6235 | 0.7837 | 0.9217 | 0.2163 | 0.0783 |
2010 | 0.0109 | 0.0110 | 0.2976 | 0.2824 | 0.4926 | 0.6353 | 0.8010 | 0.9287 | 0.1990 | 0.0713 |
2011 | 0.0125 | 0.0142 | 0.2799 | 0.2702 | 0.5097 | 0.6667 | 0.8021 | 0.9512 | 0.1979 | 0.0488 |
2012 | 0.0134 | 0.0157 | 0.2551 | 0.2620 | 0.5282 | 0.6798 | 0.7967 | 0.9575 | 0.2033 | 0.0425 |
2013 | 0.0152 | 0.0167 | 0.2082 | 0.2525 | 0.5652 | 0.6869 | 0.7886 | 0.9562 | 0.2114 | 0.0438 |
2014 | 0.0135 | 0.0166 | 0.2018 | 0.2392 | 0.5566 | 0.7027 | 0.7719 | 0.9585 | 0.2281 | 0.0415 |
2015 | 0.0147 | 0.0174 | 0.2072 | 0.2418 | 0.5465 | 0.6985 | 0.7685 | 0.9577 | 0.2315 | 0.0423 |
2016 | 0.0139 | 0.0168 | 0.2338 | 0.2334 | 0.5272 | 0.7053 | 0.7749 | 0.9554 | 0.2251 | 0.0446 |
2017 | 0.0123 | 0.0138 | 0.2357 | 0.2276 | 0.5139 | 0.7126 | 0.7619 | 0.9541 | 0.2381 | 0.0459 |
Note: In the table, W(GDP) represents the contribution rate of the Theil index of energy consumption under the weight of GDP, and W(p) represents the contribution rate of the Theil index of energy consumption under the weight of population
From Table 3, it can be seen that intra-regional differences are the main reason for the total differences in China's energy consumption. And under the weight of GDP, with the passage of time, the contribution rate of intra-regional differences shows a downward trend. The population weight is analyzed, and it is found that the different contribution rate within the region is higher and higher, while the changing trend of the different contribution rates between different regions is opposite to that of the population weight, showing a downward trend year by year.
The contribution rate of energy consumption Theil index under GDP weight
From Table 3, it can also accurately compare the respective difference contribution rate levels within and between regions, as well as a detailed data display. The following is the result of sorting out many analysis results of T(GDP), as shown in Figure 7.
Figure 7.
Contribution of regional differences under GDP weight.
According to Figure 7 and Table 3, the following analysis results can be gotten: the difference in energy consumption mainly comes from the internal differences in each region, and the contribution rate of intra-regional differences is much larger than that of inter-regions from 2000 to 2017, accounting for more than 70%, while the contribution rate of inter-regions only accounts for a small part. Analyzing the difference contribution rate between regions, it can be concluded that the difference contribution rate for the energy economy is not rich but very high, which is higher than the other two economic regions, and the smallest contribution rate is only 1% in the energy economy-rich regions. It shows that the energy utilization rates in the rich energy economy regions are still very high. However, for the rich regions of the energy economy, the differential contribution rate in the first ten years was relatively stable, and then there was a downward trend. This phenomenon shows that, with the development of the economy, the proportion of energy consumption and the proportion of GDP has begun to show a gap year by year, the energy efficiency of the energy economy is not rich regions, it can be seen from the chart that it has been at a low level, so it is necessary to maintain the energy consumption in the energy economy rich regions, increase the energy efficiency of the energy economy regions, so as to reduce differences in energy consumption among various regions.
Theil index contribution rate under population weight
The decomposition result of T (population) is shown in Figure 8.
Figure 8.
Contribution of regional differences under population weight.
In order to analyze the influence of population on the Theil index, combined with Figure 8 and Table 3, it can be seen from the figure that under the population weight, the contribution rate of intra-regional differences is higher and much higher than that of inter-regions. The 2017 data shows that the contribution rate of intra-regional differences accounted for more than 90%, which is much higher than the 4.5% difference contribution rate between regions. The contribution rate within the region also maintains an increase over time. The inter-regional contribution rate is decreasing year by year. In 2000, the inter-regional contribution rate was 27.72%, and by 2017 it was only 4.59%. Analyzing the contribution rate between regions from the data and curve display results in the figure, it is found that the contribution rate between regions is very small, and the contribution rate decreases as time goes by. The contribution rate of the energy economy is not rich in the three regions has been the largest among the three regions in the past 18 years, followed by the contribution rate of the rich region in the energy economy, and the contribution rate of the energy economy-rich region is the smallest. The contribution rate of the Theil index in the economically low energy economy has increased year by year, indicating that the energy consumption gap between the economically poor energy economy and the population matching is increasing.
Analysis and model construction of influencing factors of China's energy consumption
From the above empirical analysis of the Theil index, it can be found that there are regional differences in energy consumption, and the overall difference in energy consumption is mainly caused by regional differences. The following will further study and explore which factors have affected energy consumption and how.
Theoretical analysis of the influencing factors of China's energy consumption
There are many factors influencing China's energy consumption. The most important factors are economic factors, energy consumption structure factors, as well as different energy prices, regional populations, industrial nature, industrial structure, and so on. The following will focus on four aspects: the level of economic development, population size, industrial structure, and technological level, which will analyze the ways and extent of their influence on energy consumption.
Variable selection
Among the variables that reflect the level of economic development, economic growth, that is, GDP is the most direct and convenient. GDP can well represent the level of domestic economic development; in the industrial structure, the secondary and tertiary industries. The proportion of GDP is relatively important, so these two factors are selected to represent the industrial structure; the level of technological development can be expressed by the number of invention patent applications in a certain period of time. The larger the value, the more advanced the technology and the technological level of this place. Finally, this study selects the indicator of population growth to represent the demographic factor. In addition, compared with rural regions, China's urban development is more rapid and the energy consumption is also more. Coupled with the rapid growth of urbanization in recent years, people's lifestyles have undergone great changes. In order to reflect this situation, this study adds the indicator of population urbanization rate indicates the impact of population size on energy consumption.
The total energy consumption E of each region is regarded as the dependent variable, and the influencing factors (G, N, K, C, Z, R, S) are regarded as independent variables. The explanation of the independent variables is shown in Table 4.
Table 4.
Independent variable description table.
Influencing factors | Variable selection | Variable representation |
---|---|---|
The level of economic | •Economic growth G •Fixed asset investment in the whole society N |
•GDP •Fixed asset investment in the whole society |
Development population | •Population growth K •Population urbanization rate C |
•Permanent population at the end of the year •The proportion of urban population in total population |
Size technology level | •Acceptance of domestic invention patent applications Z | •Acceptance of domestic invention patent applications |
Industrial structure | •Secondary industry R •Tertiary Industry S |
•The output value of the secondary industry as a proportion of GDP •The output value of the tertiary industry as a proportion of GDP |
Data sources and descriptive statistics
The data mainly comes from the National Bureau of Statistics of China and the energy statistical yearbooks over the years, the same as in Section 3.2. Because the energy calculation methods of each province and municipality are different, the total energy consumption is different from the value obtained by the sum of each province. For the sake of unification, this study process the statistics of 30 provinces and municipalities, calculates their sum, and uses the sum to represent the energy consumption of the national population.
Panel-STIRPAT model construction
According to the research content of this article, an extended STIRPAT model based on the panel data model is constructed. The expression of the panel-STIRPAT model for each region is shown as follows:
(5-1) |
Where , respectively, represent the three types of regions with high, medium, and low energy consumption, namely, the above-mentioned cluster analysis of the energy economy-rich region, the energy economy-rich region, and the energy economy not rich region; i means Provinces, cities, and autonomous regions, represents the proportion of the second output value in GDP, represents time, represents the intercept term, represents the random error term, represents the third proportion of industrial output value in GDP, represents the gross domestic product, represents the total energy consumption, represents the fixed asset investment of the whole society, represents Population growth, represents the population urbanization rate, represents the number of domestic invention patent applications.
Empirical study on influencing factors of regional energy consumption based on the Panel-STIRPAT model
Based on the Panel-STIRPAT model, this section estimates the panel model for the rich regions of the energy economy, the rich regions of the energy economy, and the regions that are not rich in the energy economy.
This study selects the data of 30 provinces and municipalities across the country from 2000 to 2017, that is, .
For the rich energy economy region, ; for the energy economy-rich region, ; for the energy economy-not-rich region, . For the three regions, , so they are all long panel data.
Data validity check
Unit root test
In this study, the panel data of each region is imported into the software stata16.0, and the unit root test is performed by the LLC and IPS test methods.
The regression results are made into a graph form, as shown in Table 5, it can be seen that in terms of the energy economy-rich region, the five variables , , , , and reject the null hypothesis at a significant level of 5%, and only two variables are at 5%. Under the significance level, accept the null hypothesis. This shows that the , , , , and variables are stable, and only two variables are not stable. Therefore, this paper then conducts a unit root test on the first-order difference series of the data from the rich energy economy region. The results show that at the 5% significance level, only and without rejecting the null hypothesis, that is, only and are non-stationary.
Table 5.
Unit root test results of wealthy energy region.
Variables and test statistics | Original value | First-order difference value | |||
---|---|---|---|---|---|
Parameter | LLC | IPS | LLC | IPS | |
T | −2.535 | −1.312 | −2.57 | −1.409 | |
P | 0.0459 | 0.644 | 0.8829 | 0.577 | |
T | −1.999 | −1.33 | −4.206 | −2.220 | |
P | 0.1453 | 0.632 | 0.233 | 0.101 | |
T | −2.612 | −1.385 | −3.04 | −1.717 | |
P | 0.0315 | 0.594 | 0.7553 | 0.358 | |
T | −4.635 | −2.451 | −2.994 | −1.746 | |
P | 0.0138 | 0.045 | 0.2071 | 0.339 | |
T | −3.023 | −1.781 | −5.498 | −2.846 | |
P | 0.0165 | 0.316 | 0.0017 | 0.008 | |
T | −0.2046 | −1.548 | −1.0247 | −2.561 | |
P | 0.0869 | 0.477 | 0.068 | 0.029 | |
T | −3.308 | −1.722 | −4.291 | −2.261 | |
P | 0.0461 | 0.354 | 0.1228 | 0.089 | |
T | −2.449 | −1.236 | −3.267 | −1.867 | |
P | 0.7236 | 0.694 | 0.2316 | 0.263 |
Note: T means test value, P means probability value.
From Table 6, it can be seen that in terms of rich regions in the energy economy , , , , and significantly reject the null hypothesis. At the 5 percent significance level, only and accept the null hypothesis. This shows that the , , , , , and variables are stationary, while the other variables are not stationary. The results after the first-order difference show that at the 5% significance level, there is still only and no rejection of the null hypothesis, that is, only and are non-stationary.
Table 6.
Unit root test results of medium-rich region in energy.
Original value | First-order difference value | ||||
---|---|---|---|---|---|
Variables and test statistics | parameter | LLC | IPS | LLC | IPS |
T | −7.849 | −2.196 | −9.725 | −1.910 | |
P | 0.0001 | 0.009 | 0.0000 | 0.000 | |
T | −3.282 | −0.984 | −6.229 | −1.699 | |
P | 0.7425 | 0.967 | 0.0789 | 0.263 | |
T | −4.768 | −1.614 | −6.922 | −1.967 | |
P | 0.0226 | 0.366 | 0.0123 | 0.059 | |
T | −3.771 | −1.179 | −7.998 | −2.425 | |
P | 0.0178 | 0.879 | 0.0005 | 0.001 | |
T | −6.490 | −11.582 | −2.436 | −3.283 | |
P | 0.0000 | 0.0000 | 0.001 | 0.000 | |
T | −4.079 | −1.299 | −7.297 | −2.192 | |
P | 0.1779 | 0.774 | 0.0075 | 0.010 | |
T | −3.448 | −1.076 | −6.352 | −1.806 | |
P | 0.2079 | 0.935 | 0.3268 | 0.158 | |
T | −4.535 | −1.401 | −7.702 | −2.262 | |
P | 0.1167 | 0.655 | 0.0017 | 0.005 |
Note: T means test value, P means probability value.
From Table 7, it can be seen that as far as the energy economy is not rich, , , , and reject the null hypothesis at a significant level of 5%. The other two variables accept the null hypothesis at the 5% significance level. This shows that the , , , , , and variables are all stationary, while the other variables are non-stationary. After the first-order difference, at the 5% level of significance, there are still only and two variables that are non-stationary.
Table 7.
Unit root test results of energy-poor region.
Variables and test statistics | Original value | First-order difference value | |||
---|---|---|---|---|---|
parameter | LLC | IPS | LLC | IPS | |
T | −5.401 | −1.433 | −15.597 | −3.647 | |
P | 0.0185 | 0.635 | 0.0000 | 0.000 | |
T | −2.357 | −0.877 | −6.623 | −1.786 | |
P | 0.8341 | 0.996 | 0.3884 | 0.130 | |
T | −6.141 | −1.414 | −6.987 | −1.563 | |
P | 0.6028 | 0.666 | 0.8855 | 0.423 | |
T | −4.025 | −1.050 | −8.368 | −2.123 | |
P | 0.0063 | 0.974 | 0.0005 | 0.006 | |
T | −5.718 | −1.761 | −15.166 | −3.566 | |
P | 0.0124 | 0.1530 | 0.6702 | 0.000 | |
T | −4.727 | −1.438 | −11.523 | −2.746 | |
P | 0.1039 | 0.628 | 0.0000 | 0.000 | |
T | −1.442 | −1.239 | −9.310 | −2.251 | |
P | 0.9810 | 0.876 | 0.0955 | 0.001 | |
T | −6.696 | −1.734 | −9.194 | −2.295 | |
P | 0.0048 | 0.181 | 0.0092 | 0.001 |
Note: T means test value, P means probability value.
Cointegration test
There are generally three common cointegration tests: Pedroni test, Kao test, and Johansen panel cointegration test. Generally, when T < 10, the effect of the Kao test is better; with the increase of T, the effect of the Pedroni test will be better than that of the Kao test. Since in this article, this study will combine the Pedroni test and Kao test two inspection methods to test. The test results are shown in Table 8.
Table 8.
The results of the cointegration test in each region.
Region | Pedroni test | Kao test | |||
---|---|---|---|---|---|
Wealthy energy region | ADF | T | P | T | P |
−5.0283 | 0.0000 | −2.7192 | 0.0033 | ||
Medium-rich region in energy | ADF | T | P | T | P |
−3.6482 | 0.0001 | −4.4779 | 0.0000 | ||
Energy-poor region | ADF | T | P | T | P |
−5.2055 | 0.0000 | −2.8548 | 0.0022 |
Note: T means test value, P means probability value.
As shown in Table 8, the value of the statistical ADF of the energy economy-rich region is −5.0283, which rejects the null hypothesis. The value of the ADF statistic of the rich region in the energy economy is −3.6482, which rejects the null hypothesis. The value of the statistical ADF of the energy economy not rich region is −5.2055, which rejects the null hypothesis. This shows that in the three regions, there is a co-integration relationship between energy consumption and each variable, that is to say, the change of the dependent variable can be well explained by the change of the independent variable.
Model estimation
Model test and estimation of energy economy-rich regions
①F test
The inspection result shows that . The value is greater than 0.1, so the hybrid OLS estimation model is selected.
② Likelihood ratio test
The test result shows that . The P value is less than 0.1, so the random effects model is selected.
③Hausman test
The test result shows that , . According to the Hausman test result, the chi-square test value is greater than 0.1, so the random effects model is chosen.
According to the test results, the random effects model is selected to perform regression estimation on the factors affecting energy consumption in the energy economy-rich regions. The estimation results are shown in Table 9.
Table 9.
Estimated results of random effects model in energy economy-rich regions.
E1 | Coef. Std. Err. Z P > |z| [95% Conf. Interval] | |||||
---|---|---|---|---|---|---|
G1 | 0.5215857 | 0.0604362 | 8.63 | 0.000 | 0.4031329 | 0.640384 |
N1 | −0.1734693 | 0.0350253 | −1.13 | 0.260 | −0.1081245 | 0.0291722 |
K1 | 0.3894249 | 0.0797805 | 4.88 | 0.000 | 0.2330581 | 0.5457918 |
C1 | 0.4412713 | 0.0780657 | 5.65 | 0.000 | 0.2882654 | 0.5942773 |
Z1 | −0.1320541 | 0.0368982 | −3.58 | 0.000 | −0.2043732 | −0.059735 |
R1 | 1.085095 | 0.1060577 | 10.23 | 0.000 | 0.877226 | 1.292965 |
S1 | 1.161255 | 0.1826244 | 6.36 | 0.000 | 0.8033173 | 1.519192 |
_cons | 4.63302 | 0.6043062 | 7.67 | 0.000 | 3.448602 | 5.817439 |
From Table 9, it can be seen that the expressions among variables in wealthy energy regions are as shown in Equation (6-1):
(6-1) |
The inside R-square of the group is equal to 0.9861, which shows that the overall fit of the model is high, the independent variable can explain the dependent variable to the extent of 98.61%, and the regression effect is very good. Looking at the estimated P value of each parameter, except for the independent variable , the P values of the other independent variables are all less than 0.05, which has passed the significance test. Therefore, the panel-STIRPST measurement model for the rich energy economy region is valid. The coefficients of factors related to economic growth, population size, and industrial structure are all greater than zero, which shows that economic growth, population size, population urbanization rate, and industrial structure can all promote an increase in energy consumption. However, the elasticity coefficient of fixed asset investment and the technological level of the whole society is less than 0, which shows that these two factors will restrain the increase in energy consumption. The elasticity coefficient of the population urbanization rate is 0.4413, which shows that for every 1% increase in the population urbanization rate, energy consumption will increase by 0.4413% while keeping other factors unchanged. The GDP elasticity coefficient is 0.522, which means that for every 1 percentage point increase in GDP on the premise of keeping other factors unchanged, energy consumption will increase by 0.522 percentage points; the elasticity coefficient of technology level is −1.32, indicating that the increase in technology level will inhibit. With the increase in energy consumption, while keeping other factors unchanged, for every 1% increase in the level of science and technology, energy consumption will decrease by 1.32%.
According to the above analysis, in the energy economy-rich regions, the factor of industrial structure has the greatest impact on energy consumption, followed by economic growth, of which the impact of fixed asset investment in the whole society is the least.
Model test and estimation of energy economy medium-rich regions
①F test
The test result is , F-test selects the fixed-effects model.
② Likelihood ratio test
From the test result , it is found that the likelihood ratio test selects a random effect model.
③Hausman test
According to the test result of . Choose a random effects model.
According to the test results, a random effect model is selected to perform regression estimation on the factors affecting energy consumption in the energy economy medium-rich regions. The estimation results are shown in Table 10.
Table 10.
Estimated results of random effects model in the energy economy medium-rich regions.
E1 | Coef. Std. Err. Z P > |z| [95% Conf. Interval] | |||||
---|---|---|---|---|---|---|
G1 | 0.713738 | 0.0844926 | 8.45 | 0.000 | 0.5481356 | 0.8793405 |
N1 | −0.1734693 | 0.0490583 | −3.54 | 0.000 | −0.2696219 | −0.0773168 |
K1 | 0.5724383 | 0.2153895 | 2.66 | 0.008 | 0.1502827 | 0.9945939 |
C1 | 0.4736468 | 0.1112605 | 4.26 | 0.000 | 0.2555803 | 0.6917134 |
Z1 | −0.255411 | 0.0270521 | −0.94 | 0.345 | −0.0785622 | 0.0274801 |
R1 | 0.0721427 | 0.3004932 | 0.24 | 0.810 | −0.5168131 | 0.6610986 |
S1 | −0.5958835 | 0.2772188 | −2.15 | 0.034 | −1.139222 | −0.0525446 |
_cons | −0.6588745 | 1.683606 | −0.39 | 0.696 | −3.958681 | 2.640933 |
From Table 10, it can be seen that the expressions between variables in the energy economy medium-rich regions are as shown in Equation (6-2):
(6-2) |
The R-square within the group is equal to 0.9401, which indicates that the overall fit of the model is high, the independent variable can explain the dependent variable to the extent of 94.01%, and the regression effect is good. Look at the estimated P value of the parameters. Except for the independent variables , the P values of other independent variables are less than 0.05, which pass the significance test. From the economic significance of the parameter estimates, it can be seen that in the energy economy, the coefficients of economic growth, population size, and the proportion of the secondary industry's output value in GDP are all greater than 0. This shows that economic growth, population size, population urbanization rate, and secondary industry as the proportion of output value in GDP increases, energy consumption will increase accordingly. However, the elasticity coefficient of social fixed asset investment, technological level, and the output value of the tertiary industry as a proportion of GDP is less than 0, which shows that the increase of these factors will reduce energy consumption. The estimated coefficient of the population urbanization rate is 0.474, that is, under the premise of keeping other factors unchanged, for every 1% increase in population urbanization rate, energy consumption will increase by 0.474%. The elasticity coefficient of GDP is 0.714, that is, keeping other factors unchanged, every 1% increase in GDP will increase energy consumption by 0.714%. The coefficient of elasticity of science and technology levels is −0.255, that is, under the circumstance that other factors remain unchanged, every 1% increase in science and technology level will reduce energy consumption by 0.255%.
According to the above analysis, in the energy economy medium-rich regions, GDP has the greatest impact on energy consumption, followed by the tertiary industry's output value as a share of GDP, of which the level of science and technology has the least impact.
Model test and estimation of the energy economy-poor regions
①F test
According to the test result of , the fixed effects model is selected for this test.
② Likelihood ratio test
According to the test result of , the random effects model is selected under this test.
③ Hausman test According to the test result of , it can be known that the fixed-effect model is selected under this test.
Considering comprehensively, the fixed effect model is finally selected to estimate the factors affecting energy consumption in the energy economy-poor regions. The results are shown in Table 11.
Table 11.
Estimated results of random effects model for energy economy-poor regions.
E1 | Coef. Std. Err. Z P > |z| [95% Conf. Interval] | |||||
---|---|---|---|---|---|---|
G1 | 0.2864463 | 0.0632165 | 4.53 | 0.000 | 0.1619758 | 0.4109167 |
N1 | 0.1728833 | 0.0404579 | 4.27 | 0.000 | 0.0932235 | 0.2525432 |
K1 | 1.933154 | 0.1954567 | 9.89 | 0.000 | 1.548308 | 2.317999 |
C1 | 0.1551544 | 0.0749343 | 2.07 | 0.039 | 0.0076121 | 0.3026968 |
Z1 | −0.070009 | 0.0196189 | −3.57 | 0.000 | −0.1086378 | −0.0313803 |
R1 | 0.3914169 | 0.1319744 | 2.97 | 0.003 | 0.1315651 | 0.6512688 |
S1 | 0.0096488 | 0.1491456 | 0.06 | 0.948 | −0.2840123 | 0.3033099 |
_cons | −9.168153 | 1.500939 | −6.11 | 0.000 | −12.12344 | −6.212869 |
From Table 11 it can be seen that the expressions between variables in energy economy-poor regions are as shown in Equation (6-3):
(6-3) |
Among them, the square of R is equal to 0.9577, which means that the independent variable can explain the dependent variable at the level of 95.77%. In this test, except for the independent variables and , the P values of other independent variables are less than 0.05, which passed the model test.
The regression coefficients of economic growth, fixed asset investment in the whole society, population size and the output value of the secondary industry are all greater than 0, which shows that economic growth, fixed asset investment in the whole society, population size, population urbanization rate, and the proportion of the output value of the secondary industry. The increase in energy consumption can promote energy consumption. However, the level of technology and the elasticity coefficient of the output value of the tertiary industry is less than 0, which shows that these factors will inhibit the increase in energy consumption. The coefficient of the population urbanization rate is 0.474, which means that for every 1 percentage point increase in population urbanization rate under the premise of keeping the factors constant, energy consumption will increase by 0.474 percentage points. The GDP coefficient is 0.714, that is, when other factors remain unchanged, for every 1% increase in GDP, energy consumption increases by 0.714%; the regression coefficient for the level of science and technology is −0.255, that is, when other factors remain unchanged, for every 1 percentage point increase in the level of science and technology, energy consumption will decrease by 0.255 percentage points. The above analysis shows that in regions with a low energy economy, population growth has the greatest impact on energy consumption, followed by the proportion of the output value of the secondary industry in GDP, and the number of patent applications has the least impact.
Conclusions
This study has done a series of studies and empirical analyses on what factors are responsible for the differences in regional energy consumption in China, and summarized some of the more influential factors, and analyzed their impact.
Firstly, by calculating the Theil index under the weight of GDP, it is found that the development similarity between energy consumption and economic growth is the weakest in regions with a low energy economy. The evolution trend of the Theil coefficient of energy intensity in the energy economy-rich and medium-rich regions are relatively stable. It shows that the development similarity of energy consumption and economic growth in these two regions is also relatively stable.
Secondly, by calculating the Theil index under the population weight, the results show that the energy consumption in the energy economy-poor regions has the weakest similarity to the population development, while the energy economy in the rich regions has the better similarity to the population development, and the energy consumption and population in the medium-rich region have maintained a relatively stable development trend.
Thirdly, a comparative analysis of the Theil indices in two types of regions found that, for the energy economy-rich regions, energy consumption is becoming less and less similar to population and economic development, and the regional difference is becoming more and more obvious; for the energy economy medium-rich regions, the Theil indices in both two types of regions show the downward trend indicates that the economy of the energy economy-rich regions is directly proportional to the growth of population and energy consumption, and the overall difference in regional energy consumption is gradually shrinking; for the energy economy-poor regions, the regional difference between energy consumption and the population is relatively large.
Fourthly, by calculating the contribution rate of the Theil index, we found that the regional difference is the main reason for the total difference. And under the weight of GDP, the contribution rate of the Theil index in the region is declining year by year, while under the weight of population, it is rising. The contribution rate of the Theil index among regions is just the opposite. It shows an upward trend under the weight of GDP and a downward trend under the weight of the population.
Fifthly, the industrial structure has the greatest impact on energy consumption in the rich energy economy. The secondary industry's output value as a proportion of GDP has a positive effect in the three major regions; the increase in the tertiary industry's share of GDP will only promote the energy economy. The increase in energy consumption in affluent regions is negatively correlated with the other two regions. The two influencing factors of economic growth and population size can promote the increase of energy consumption in the three regions, and the population is the main influencing factor affecting the rich regions of the energy economy.
To sum up, there are regional differences in energy consumption in China, and the same factor has different effects on different regions, and different factors have different degrees and directions of influence on the same region. Different energy consumption should be adopted for each region. Energy-saving measures for coordinated development.
There are some limitations in this study. The selection of influencing factors index in this paper is not perfect, further research is needed in the future; limitation of the data collecting, some of the relevant research data are not published or incomplete, due to the coronavirus pandemic, so this study had to select the data of 2000 to 2017, which weakens a little bit of the research significance of this paper.
Policy recommendations
Through the sub-point analysis in the previous section, it can be seen that the same influencing factor has different degrees of influence on energy consumption in different regions. Only by formulating and implementing policies and measures that reflect the regional differences of influencing factors and achieve the goal of reasonably controlling total energy consumption. Based on the estimated results of the above model, this study proposes some policy recommendations:
Firstly, in the energy economy-rich regions, the economy is developing rapidly, and people's lifestyles are characterized by a high level and high-consumption. People's demand for high-energy-consuming products is gradually increasing, and more energy is consumed; it should reduce the appearance of private cars and reduce energy consumption in response to this situation by promoting public transportation such as buses and subways. It should also adopt feasible plans such as appropriate fuel tax and promotion of new energy vehicles to advocate civilized and frugal.
Secondly, in the rich regions of the energy economy, the secondary industry, namely the construction industry, has developed quite rapidly, and the urbanization process is at a rapid stage. Therefore, energy consumption in the rich regions of the energy economy is growing very fast. Meanwhile, it should control energy consumption and minimize harm to the environment in order not to affect the development of the rich regions of the energy economy.
Thirdly, for regions with low energy economy, they should popularize energy knowledge among local residents, raise their awareness of energy efficiency, pay attention to the development of local buildings and improve the progress of urbanization.
Author biographies
Da-Jin Yu is a Center Director and Professor in Management science and engineering. His area of research is regional development and statistical sciences.
Fen An holds a Master's Degree in Applied Statistics. Her area of research is energy economy.
Yuan-Fan Ye holds a Master's Degree in Quantitative economics. Her area of research is regional economic management.
Hui-Hong Zeng holds a Master's Degree in Quantitative economics. Her area of research is regional economic management.
Footnotes
Authors’ contributions: Study concept and design: Da-Jin Yu and Fen An. Acquisition of data: Da-Jin Yu and Fen An. Analysis and interpretation of data: Fen An and Da-Jin Yu. Drafting of the manuscript: Da-Jin Yu, Fen An, Yuan-Fan Ye, and Hui-Hong Zeng. Critical revision of the manuscript: Da-Jin Yu, Yuan-Fan Ye, and Hui-Hong Zeng.
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding: The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the the Humanities and Social Sciences Project for University of Jiangxi, Jiangxi Think Tank Research Project, the National Social Science Fund of China (grant number GL19108, 21zk19, 20BGL295).
ORCID iD: Da-Jin Yu https://orcid.org/0000-0002-5418-2116
References
- 1.Blanco M, Ferasso M, Bares L. Evaluation of the effects on regional production and employment in Spain of the renewable energy plan 2011–2020. Sustainability 2021; 13: 3587–3587. [Google Scholar]
- 2.Yan JN, Su B. Spatial differences in energy performance among four municipalities of China: From both the aggregate and final demand perspectives. Energy 2020; 204: 117915. DOI: 10.1016/j.energy.2020.117915 [DOI] [Google Scholar]
- 3.Zhou X, Imura H. How does consumer behavior influence regional ecological footprints? An empirical analysis for Chinese regions based on the multi-region input-output model. Ecol Econ 2011; 71: 171–179. [Google Scholar]
- 4.Eguchi S, Takayabu H, Lin C. Sources of inefficient power generation by coal-fired thermal power plants in China: a metafrontier DEA decomposition approach. Renew Sust Energy Rev 2021; 138: 110562. [Google Scholar]
- 5.Oswald Y, Owen A, Steinberger JK. Large inequality in international and intranational energy footprints between income groups and across consumption categories. Nat Energy 2020; 5: 231–239. [Google Scholar]
- 6.Zhang B, Chen ZM, Xia XHet al. et al. The impact of domestic trade on China's regional energy uses: a multi-regional input-output modeling. Energy Policy 2013; 63: 1169–1181. [Google Scholar]
- 7.Renou-Maissant P. Interfuel competition in the industrial sector of seven OECD countries. Energy Policy 1999; 27: 99–110. [Google Scholar]
- 8.Chen LM, Yan L, Chen FYet al. et al. Research on the dynamic relationship between economic growth and energy consumption. Techn Inn Manag 2020; 41: 75–82. [Google Scholar]
- 9.Cheong TS, Li VJ, Shi XP. Regional disparity and convergence of electricity consumption in China: a distribution dynamics approach. Chin Econ Rev 2019; 58: 101154. [Google Scholar]
- 10.Lutz LM, Fischer LB, Newig Jet al. et al. Driving factors for the regional implementation of renewable energy - A multiple case study on the German energy transition. Energy Policy 2017; 105: 136–147. [Google Scholar]
- 11.Park H, Kwon H. Effects of consumer subsidy on household fuel switching from coal to cleaner fuels: a case study for anthracites in Korea. Energy Policy 2011; 39: 1687–1693. [Google Scholar]
- 12.Adua L, Clark B. Even for the environment, context matters! States, households, and residential energy consumption. Environ Res Lett 2019; 14: 064008. [Google Scholar]
- 13.Achour H, Belloumi M. Decomposing the influencing factors of energy consumption in Tunisian transportation sector using the LMDI method. Transp Policy 2016; 52: 64–71. [Google Scholar]
- 14.Sun Q, Ji J, Xu C. Energy consumption driving factors and measuring models of regional integrated transport system. J Transp Systems Eng Inform Techn 2013; 13: 1–9. [Google Scholar]
- 15.Zhang XP. Analysis of regional differences and influencing factors of energy consumption intensity in China. Res Sci 2008; 6: 883–889. [Google Scholar]
- 16.Nie HG. Research on the problems and countermeasures of China’s energy consumption growth. Jilin, China: Jilin University, 2013. [Google Scholar]
- 17.Yu H. The influential factors of China's regional energy intensity and its spatial linkages: 1988–2007. Energy Policy 2012; 45: 583–593. [Google Scholar]
- 18.Salim RA, Shafiei S. Urbanization and renewable and non-renewable energy consumption in OECD countries: an empirical analysis. Econ Mod 2014; 38: 581–591. [Google Scholar]
- 19.Zaharia A, Diaconeasa MC, Brad Let al. et al. Factors influencing energy consumption in the context of sustainable development. Sustainability 2019; 11: 4147–4147. [Google Scholar]
- 20.Abam FI, Ekwe EB, Diemuodeke OEet al. Environmental sustainability of the Nigeria transport sector through decomposition and decoupling analysis with future framework for sustainable transport pathways. Energy Rep 2021; 7: 3238–3248. [Google Scholar]
- 21.Zarco-Soto IM, Zarco-Perinan PJ, Sanchez-Duran R. Influence of cities population size on their energy consumption and CO2 emissions: the case of Spain. Environ Sci Pollut R 2021; 28: 28146–28167. [DOI] [PubMed] [Google Scholar]
- 22.Zhang YD. Research on the evolutionary trends and influencing factors of the regional differences in energy consumption in China. Tianjin, China: Tianjin University, 2015. [Google Scholar]
- 23.Wang YC. Analysis of energy consumption differences and influencing factors in China. Sci Techn Plaz 2014; 3: 164–169. [Google Scholar]
- 24.Yumashev A, Slusarczyk B, Kondrashev Set al. et al. Global indicators of sustainable development: evaluation of the influence of the human development index on consumption and quality of energy. Energies 2020; 13: 2768. [Google Scholar]
- 25.Apergis N, Li J. Population and lifestyle trend changes in China: implications for environmental quality. AE 2016; 48: 5246–5256. [Google Scholar]
- 26.Hu W, Fan YM. City size and energy conservation: do large cities in China consume more energy? Energy Econ 2020; 92: 104943. [Google Scholar]
- 27.Ahmed MU, Hossain AKMN, Hasanuzzaman S. Exploring the depth of energy penetration in economic advancement: perspective of Bangladesh. Renew Sust Energy Rev 2015; 49: 1033–1047. [Google Scholar]
- 28.Liddle B. Demographic dynamics and per capita environmental impact: using Mishrapanel regressions and household decompositions to examine population and transport. Popul Environ 2004; 26: 23–39. [Google Scholar]
- 29.Mishrar V, Sharma S, Smythr R. Are fluctuations in energy consumption per capita transitory? Evidence from a panel of Pacific Island countries. Energy Policy 2009; 37: 2318–2326. [Google Scholar]
- 30.Liu HQ, Tang S. Analysis on the characteristics of energy consumption differences and research on influencing factors in various cities in Yunnan. Ecol Econ 2020; 3: 85–91. [Google Scholar]
- 31.Lin YF, Jiang Y. Economic structure, banking structure and economic development-an empirical analysis based on provincial panel data. Fin Res 2006; 1: 7–22. [Google Scholar]
- 32.Yang Y. Analysis of the impact of urbanization on energy consumption. Beijing, China: Beijing Jiaotong University, 2018. [Google Scholar]
- 33.Noorpoor AR, Kudahi SN. CO2 Emissions from Iran's power sector and analysis of the influencing factors using the stochastic impacts by regression on population, affluence and technology (STIRPAT) model. Carbon Manag 2015; 6: 101–116. [Google Scholar]
- 34.Andreoni V. From the Celtic tiger to the Celtic phoenix: the metabolic profile of Ireland and the main drivers of energy change. J Clean Prod 2020; 246: 118975–118975. [Google Scholar]
- 35.ul Husnain MIU, Nasrullah. Khan MA. Testing dependence patterns of energy consumption with economic expansion and trade openness through wavelet transformed coherence in top energy-consuming countries. Environ Sci Pollut R 2021; 28: 49788–49807. [DOI] [PubMed] [Google Scholar]
- 36.Mohsin M, Abbas Q, Zhang JJet al. et al. Integrated effect of energy consumption, economic development, and population growth on CO2 based environmental degradation: a case of transport sector. Environ Sci Pollut R 2019; 26: 32824–32835. [DOI] [PubMed] [Google Scholar]
- 37.Kraft J, Kraft A. On the relationship between energy and GNP. J Energy Dev 1978; 3: 401–403. [Google Scholar]
- 38.Yu ESH, Jin JC. Co-integration tests of energy consumption, income, and employment. Res Energy 1992; 14: 259–266. [Google Scholar]
- 39.Zhao JW, Fan JT. An empirical study on the internal compliance between economic growth and energy consumption. Econ Res 2007; 8: 31–42. [Google Scholar]
- 40.Wang P. Research on the regional differences of the relationship between energy consumption and economic growth in China. Zhengjiang, China: Jiangsu University, 2009. [Google Scholar]
- 41.Clark B, Jorgenson AK, Auerbach D. Up in smoke: the human ecology and political economy of coal consumption. Organ Environ 2012; 25: 452–469. [Google Scholar]
- 42.Yu SW, Hu X, Zhang XJet al. et al. Convergence of per capita carbon emissions in the Yangtze River Economic Belt, China. Energy Environ 2019; 30: 776–799. [Google Scholar]
- 43.Luan BJ, Zou H, Chen SXet al. et al. The effect of industrial structure adjustment on China's energy intensity: Evidence from linear and nonlinear analysis. Energy 2021; 218: 119517. DOI: 10.1016/j.energy.2020.119517 [DOI] [Google Scholar]
- 44.Pan XF, Li MN, Pu CXet al. et al. Study on the industrial structure optimization under constraint of energy intensity. Energy Environ 2021; 32: 134–151. [Google Scholar]
- 45.Shang M. Research on influencing factors and regional differences of energy consumption in China . Doctoral dissertation, Xi'an University Sci Techn 2011. [Google Scholar]