More use or cleaner use? Income growth and rural household energy-related carbon emissions in central China

HuayiChang^ab NicoHeerink^a WenWu^c JunbiaoZhang^bd

a Development Economics Group, Wageningen University & Research, 6706KN Wageningen, the Netherlands

b College of Economics & Management, Huazhong Agricultural University, Wuhan 430070, China

c Institute of Xi Jinping Thought on Socialism with Chinese Characteristics for a New Era, Peking University, Beijing 100871, China

d Hubei Rural Development Research Center, Wuhan 430070, China

Abstract

Rural household income growth may have two counteracting effects on energy-related carbon emissions, by stimulating total energy consumption and by changing the energy structure towards low-carbon energy sources. This study aims to provide more insight into this relationship for the case of rural central China, a region where coal-fired electricity dominates modern energy consumption. Based on survey data collected in 2019, energy-related carbon emissions (ECE) and energy-related carbon intensity (ECI) per unit of energy are calculated. A mediation effect model is employed to distinguish between energy structure changes and total electricity consumption as mediating factors in the income growth – carbon emission relationship. We argue that whether biomass fuels are treated as either carbon-intensive or as carbon neutral has major consequences for the analysis. When biomass fuel is treated as carbon-intensive, income-related changes in energy structure and electricity consumption have opposite effects on ECE, but both are conducive to the reduction of ECI. When biomass fuel is treated as carbon-neutral, the positive effect of growth in coal-fired electricity consumption on ECE is much larger than the negative effect of the change in energy structure. Based on these findings, we present several recommendations for balancing two main targets of China's rural revitalization vitalization policy, i.e. rural household income growth and the transition towards low-carbon energy sources in China.

Keywords

Energy use

Carbon emission

Income growth

Rural households

China

Introduction

With the continued growth of the economy and the rise in household welfare, residential demand for energy and the resulting carbon emission have grown rapidly in China in recent decades (Li et al., 2015; Liu & Diamond, 2005). As achieving energy conservation and sustainable development have become a global priority, the Chinese government promotes the shifting from solid fuel to gas or electricity by rural households to achieve the greenhouse gas mitigation targets in residential sectors (Chen & Chen, 2019; Xu & Ge, 2020). However, carbon emission reduction is rarely the main goal of rural households making energy decisions (Hahnel et al., 2020; Li et al., 2021). Thus, although income growth plays an important role in household energy-related emissions, the causal links between household incomes and carbon emissions remain unclear (Alola & Joshua, 2020; Goldstein et al., 2020; Li et al., 2019). Stimulating household income growth and protecting the living environment are two main targets of China's rural revitalization strategy adopted in recent years. Can the two targets be balanced, given that coal-fired power generation is the major source of modern energy in China? To the best of our knowledge, this question has not been thoroughly examined so far.

According to the energy consumption framework theory put forward by Barnes et al. (2011) and Khandker et al. (2012), households satisfy their basic energy demand and increase both the amount and types of energy they use as income goes further up. Carbon emissions, a type of unexpected output in this process, depend on the dual effects of changes in the total consumed amount of energy and changes in the structure of the consumed energy. Household incomes may affect both the total consumed amount and its structure (Zhang et al., 2009). Thus, the impact of household incomes on energy-related carbon emissions (ECE) needs to be explored by distinguishing two aspects, i.e. total energy quantity and its composition.

Numerous studies have found strong relations between income increases and household energy structure transition, and have come up with the ‘energy ladder’ hypothesis (Druckman & Jackson, 2008; Ekholm et al., 2010; Lenzen et al., 2006; Wang et al., 2017). Generally, wealthier households tend to substitute traditional biomass fuels or coal for electricity and gas with more modern energy sources. Other studies found evidence of the ‘energy stacking’ hypothesis, which states that income increases stimulate the uptake of clean fuel, but have no effect on the consumption of solid fuel (Baiyegunhi & Hassan, 2014; Carter et al., 2020; Cohen et al., 2005). Both the ‘energy ladder’ and ‘energy stacking’ theories see income growth as a major determining factor of household energy structure. Generally, the use of more modern energy sources and associated efficiency improving technologies is conducive to the mitigation of household energy-related carbon emissions (Kadian et al., 2007; Liu et al., 2012).

Income growth also tends to increase total household energy demand, especially the consumed amount of electricity (Gill & Moeller, 2018; Goldstein et al., 2020; Han & Wu, 2018; Li et al., 2019; Zou & Luo, 2019). For ‘energy stacking’ households, concurrent use of traditional sources of energy and increasing amounts of electricity may contribute to energy waste and endanger environmental sustainability by producing larger amounts of carbon emissions (Charlier et al., 2018; Chen et al., 2016; Liu et al., 2012; Liu et al., 2019). When electricity generation is mainly coal-based, growing use of electricity can lead to significantly higher CO₂ emissions even when households stop using traditional energy (Eludoyin & Lemaire, 2021; Liang et al., 2013; Ouyang et al., 2010; Sesan, 2021; Sukarno et al., 2017; Zhang et al., 2020). Macro-level studies on the relationship between economic growth and CO₂ emissions in developing countries generally find a positive relationship as well (Bekun et al., 2021; Bekun & Alola, 2022; Gyamfi et al., 2021; Zhang & Li, 2022).

It can be seen that the above two effects of income growth on energy-related carbon emissions may potentially counteract. However, few studies have explored their joint impact on energy-related carbon emissions. Especially for rural residents, the connotation of their energy transition is multi-level. From an environmental protection perspective, they are encouraged to substitute traditional biomass fuel and coal with gas or electricity. But to enhance welfare, more energy-intensive facilities (e.g. television, air conditioner, and refrigerator) are popularized to rural households. These policy targets or irreversible development trends may affect energy-related carbon emission in different directions.

Another important issue in assessing the impact of income on rural household energy-related carbon emissions is the lack of a suitable indicator that adequately reflects current policy goals in transition economies such as China. The reduction of total energy-related carbon emissions is in line with the long-term carbon neutrality target. However, the short-term energy policy objective in China is to reach the peak in carbon dioxide emissions before 2030. Taking into account that energy demand will continue to grow with continued economic growth and improvements in household welfare, reducing energy-related carbon intensity - defined as CO₂ emissions per unit of energy – becomes a major energy policy objective.

This study aims to provide more insight into the links between energy-related carbon emissions and rural household income growth in central China, a region where coal-fired electricity dominates modern energy consumption. Energy-related carbon emission (ECE) and energy-related carbon intensity (ECI) are calculated based on a household survey dataset collected in central China in 2019. We focus in particular on changes in the energy structure and on the growth in the total electricity consumption amount as main mechanisms through which household income is expected to affect ECE and ECI. The mediation effect model is employed to examine these two underlying mechanisms. By focusing on regions in central China having relatively similar resource and productivity characteristics, regional differences in available energy endowments on rural household energy choices play a minor role. Moreover, rural households in central China generally use a mix of traditional and modern fuels, which makes these households suitable objects for our study. The results obtained from our study may answer the question whether the two policy targets of energy-related carbon reduction and household income growth can be balanced under the technology constraints of power generation. Further, these findings can provide important insights for the design of improved policies aimed at net-zero emissions by 2050 in China's residential sector.

The remainder of the paper is structured as follows. We describe the data collection, the ECE and ECI calculation methods and the estimation methods in the second section. The results of the calculations and the estimation results are presented and discussed in the third section. The fourth section summarizes the main findings and uses them to present recommendations for future policy.

Data and methodology

Data collection

Our analysis is based on a unique household survey dataset of energy choices and consumption characteristics of rural households. The survey was conducted in July of 2019, and collected also information on the household composition and on economic aspects and agriculture production. >150 researchers were recruited and obtained interview training. They collected the data through face-to-face interviews. In total, 1080 households in Hubei, Hunan and Henan provinces in central China were interviewed.

Instead of the poor environmental endowment of western China and the prosperous industrialization of eastern China, central China is the main grain-producing area that supplies most rice, wheat and oil crops for the Chinese agricultural produce market (Lu et al., 2013; Zhu et al., 2021). As regards energy use, rural households in western China tend to use more bioenergy (Sichuan, Guizhou) or solar and wind energy (Inner Mongolia, Tibetan Plateau), given the available renewable energy endowments in the region (Liu et al., 2020; Zhang et al., 2009). Rural residents in eastern China are more used to consume modern energy because of their higher income levels (Zhang et al., 2009). Contrastingly, rural households in central China generally use a mix of traditional and modern fuels, and produce more energy-related rural carbon emissions than rural households in other regions (Zhang & Li, 2022). Thus, the region seems well-suited for examining the potentially offsetting effects of income-related growth in total energy consumption and change in energy structure on energy-related carbon emissions.

Central China is a geographical and a loosely defined cultural region that includes the provinces of Henan, Hubei and Hunan (see Fig. 1a). We selected the prefectures in these three provinces with roughly similar weather conditions and topography, located between 25°N and 36°N,108°E and 116°E (see Fig. 1b). We assume that these sample areas adequately reflect the variety in rural energy use characteristics in central China. A stepwise sample procedure was used for selecting the households (see Appendix C for details). In total 1080 households were selected. The main purpose of the survey was not to be representative of central China, but to obtain a sufficiently wide variation in household energy use characteristics and incomes that would allow us to obtain deeper insights into the relationship between income growth and energy-related carbon emissions.

Fig. 1. a Provinces in central China. b Sample cities in survey provinces.

Note: Names of the prefectures (cities) included in survey are in bold in panel b; Names of the surveyed counties are in parentheses.

ECI  accounting

Following Kadian et al. (2007) and Liu et al. (2012), the energy-related carbon emission ( ECE ) of a household is calculated by:(1)ECEi="∑j=1nEFj×Cijwhere  ECE  _i  is the CO₂ emission of the energy used by household  i  (kg).  C  _ij  is the consumption of energy type  j  by household  i  (energy unit, e.g. kg, cm³, and kWh).  EF  _j  is the CO₂ emission factor for energy type  j  (in kg CO₂ per energy unit). It can be further calculated as:(2)EFj="CEfj×NCVj×ORj×44/12where  CEf  _j  is the carbon content of energy type  j  (kgC/TJ).  NCV  _j  is the net calorific value of energy type  j  (TJ per energy unit).  OR  _j  is the oxidation rate of energy type  j  (percent). The  CEf  _j  of different commercial energy such as coal, LPG, and pipe gas are provided by Eggleston et al. (2006).

For the traditional biomass energy including straw and firewood, Liu et al. (2012) calculated the  EF  _j  by averaging the results of past studies. We adopt the  EF  _j  of straw and firewood of their study in this research. However, it is a controversial issue. The calculation of rural household CO₂ emission by Liu et al. (2013) did not include emissions from biomass. They argue that biomass fuels should be considered as carbon neutral, taking into consideration the function as carbon sinks during biomass production (Rabl et al., 2007). To meet the net-zero emission targets of the Paris Agreement, energy-related emissions should be considered in a wider perspective instead of in the residential sector merely (Pye et al., 2017; van Soest et al., 2021). If the removal of CO₂ during the creation of biomass is also taken into account, the links between income and rural household energy-related emissions should be re-examined (Liu et al., 2013; Rabl et al., 2007). Estimating the extent to which CO₂ emitted during the whole life cycle of biomass exceed the amount that it removes or may even be in balance is a complicated and unresolved issue. This paper therefore also presents additional results in which the  EF  _j  of biomass fuel is treated as equal to zero.

As electricity generation in China highly relies on coal, the  CEf  _j  of electricity depends to a large extent on the thermal coal consumption rate (TCCR) (Wang et al., 2020). According to ‘China Energy Statistical Yearbook 2019’, the power generations of Hubei, Hunan, and Henan in 2018 are 2817 × 10⁸kWh, 1540 × 10⁸kWh, and 3060 × 10⁸kWh, respectively. The raw coal inputs to power generation in Hubei, Hunan, and Henan are 4509.54 × 10⁷ kg, 3393.36 × 10⁷ kg, and 11,344.96 × 10⁷ kg, respectively. Thus, the TCCRs in Hubei, Hunan, and Henan are 0.1601 kg/kWh, 0.2204 kg/kWh, and 0.3708 kg/kWh, respectively. We calculate the  CEf  _j  of electricity based on the TCCRs and the emission factor of raw coal. Table 1 presents the CO₂ emission factors ( EF  _j ) that are used in this study.

Table 1. CO₂ emission factors used in this study.

The household energy-related carbon intensity ( ECI ) is calculated as the ratio of energy-related carbon emissions and household total energy consumption:(3)ECIi="ECEi∑j=1nCijwhere the units of  Cij  are converted to kgce (kg standard coal equivalent). Thus, the measurement unit of  ECI  is kgCO₂/kgce. The higher the  ECI , the more CO₂ emissions are produced per unit of energy used by the household.

Empirical method

The energy consumption decisions of rural households and the carbon emissions related to them can be expressed as follows:(4)Yi="α0+α1lnIncomei+∑iαiXi+fc+εiwhere  Y  _i  is the key outcome variable, which refers to ECE and ECI, respectively, while the natural logarithm¹ of household income is the key explanatory variable.  X  _i  is a vector of control variables.  f  _c  is the county-level fixed effect.  ε  _i  is the error term. The coefficient  α  ₁  reflects the direct effect of income on ECE and ECI.

We further investigate the mechanisms through which the household energy amount and types, and the related energy-related carbon emissions, change with income increases. The mediation effect model is used to examine whether traditional energy transition serves as a latent mediator in the relationships of income with ECE and ECI. It can be expressed as follows:(5)MVi="β0+β1lnIncomei+∑iβiXi+fc+μi(6)Yi="γ0+γ1MVi+γ2lnIncomei+∑iγiXi+fc+ϑiwhere  MV  _i  is the mediation variable, which reflects either the change in energy structure or the electricity consumption amount.  μ  _i  and  ϑ  _i  are the error terms, respectively. The indirect effect ( β  _i  ×  ɤ  _i ) is the amount of mediation, while the total effect is the sum of direct and indirect effects ( ɤ  _i+1 +  β  _i  ×  ɤ  _i ). To verify the effectiveness of the intermediary path, Sobel-test with bootstrap-based confidence intervals is employed.

Considering the absence or presence of markets for the different types of energy in a village, the energy use of residents in the same village may be homogeneous. Simultaneously, based on the prosperous social networks in rural China, the energy consumption decisions for different households are likely to be correlated because of the general communications and interactions between relatives, friends, or neighbours living in the same village (Cai et al., 2016). Village-level cluster standard errors are therefore employed to adjust the estimations for these correlations.

Measures

The key outcome variables in this study are ECE and ECI. Respondents were asked about their consumption of traditional fuels (firewood, straw), intermediate fossil fuels (raw coal, honeycomb coal) and modern energy (electricity, LPG, piped gas) (see Appendix A for the detailed energy module questionnaire). For these 7 types of energy, the study firstly asked the respondents whether they choose each type of energy in their daily life. If the households only use modern clean energy, it means they realize the suspension of traditional solid fuels. Otherwise, they either use both modern and traditional energy or they totally rely on traditional fuels. The usage rates and mean consumption levels of different energy sources are shown in Table B1, Appendix B. We further asked how much of each type of energy the household consumed last year, defined as  C  _ij  (in kg, kWh, m³). Based on the calculations presented in Eqs. (1), (2), (3), the values of ECE and ECI are presented in Table 2.

Table 2. Descriptive statistics for key outcome variables.

According to the study of Salari and Javid (2017), the factors affecting household energy consumption can be classified into five categories: social-economics and demographics, building characteristics, location, temperature, and energy price. Household income is the main explanatory factor that we examine in this study. The income distribution of the households in the sample is shown in Table B2, Appendix B. To address the potential endogeneity bias between household income and energy consumption, the household income in 2018 is used as a proxy of energy use in 2019.

Age and family size are widely shown to be important factors to influence household energy preference (Muller & Yan, 2018). Older people may perpetuate more traditional habits related to fuels (Baiyegunhi & Hassan, 2014; Rahut et al., 2014), but also can afford cleaner fuels more easily with fewer liquidity constraints (Gupta & Köhlin, 2006; Özcan et al., 2013). Although a larger household size may lead to scale economies on energy consumption that reduce energy-related carbon emissions (Wu et al., 2021), more studies support those larger households prefer traditional fuels because of the higher budget constraint or more time endowment (Chen et al., 2006; Ouedraogo, 2006; Özcan et al., 2013). Education is consistently regarded as a significant activator for household energy transition. Not only because of the growth of time opportunity costs, but also because better education can translate into greater awareness of body health, life efficiency, or environmental protection (Muller & Yan, 2018). Additionally, based on non-separable agricultural household models, household biomass use can be related to the farmland size, especially in most poor regions (Chen et al., 2006; Démurger & Fournier, 2011).

The other four categories of factors impact household energy use more directly. Detached house is more energy-intensive (Aydinalp et al., 2002). And the production systems of old or low-valued building are less efficient (Salari & Javid, 2017). Different location reflects different energy resource endowment and also different climate. High and low temperature both can stimulate energy demand (Moral-Carcedo & Vicéns-Otero, 2005; Salari & Javid, 2016). Finally, the negative own-price effects for energy demand are widely supported by a large number of studies (Muller & Yan, 2018; Wang & Lin, 2021).

We control the location and temperature factors with county-level fixed effect and consider the household social-economic and building characteristics and the energy price. If the reported price is higher than 0.8 CNY or lower than 0.4 CNY, we calculate the electricity price by the ratio of reported electricity expenditure and the consumed amount. If the calculated price is in the 0.4 to 0.8 interval, we substitute the reported price with the calculated price. Otherwise, the sample is regarded as an outlier. In total, 53 samples with anomalous electricity prices were dropped and 1027 samples are regarded to be valid. Further with 31 absence answers for the house value, 996 valid samples were used for the empirical analysis. The definitions of the control variables associated with household energy consumption are presented in Table 3. It can be found that most respondents are senior people, with an average education level of 6 years. >90 % of sample households are living in separated constructions and the average age of their houses is >15 years. The county-level fixed effect is controlled by setting dummy variables of 17 counties.

Table 3. Variable definition and descriptive statistics.

Results and discussion

Descriptive statistics

The average rural household consumption of each type of energy and ECE in Hubei, Hunan, and Henan in 2019 obtained from our survey data set are presented in Fig. 2. It can be found that rural households in central China on average consume 1700 kgce to 2200 kgce energy per year. Electricity and traditional biomass fuels are two main energy sources for rural household daily life in central China. The households in Hunan province consume more energy and the households in Henan province consume less. In other words, residents in the province with the highest heating demand use the least energy, while residents in the province with the lowest heating demand have the highest energy consumption. When we consider the residential energy policy in China, these results make sense. In 2017, the Chinese government started the ‘Coal to gas, coal to electricity’ policy, and many cities in Henan province were selected as pilots. Under this policy, subsidies and public services are provided to rural households to stimulate the use of gas or electricity as heating energy. As a result, the use of traditional fuels, like coal and firewood, reduced sharply in the pilot areas (Meng et al., 2019). Moreover, due to the higher efficiency of electricity and gas as compared to traditional fuels, households in the pilot areas were able to reduce energy consumption levels.

Fig. 2. Rural household energy consumption and ECE in Hubei, Hunan and Henan province, 2019. Source: Calculated by authors from survey data for 1027 households in central China.

Accordingly, when the emission from biomass fuels is included, the ECE in Hunan province is the highest. Consumption of traditional biomass fuels and intermediate fossil fuel is low in Henan, which may be related depend to the electricity heating policy in north China (Wang et al., 2020). The ECE in Henan is relatively lower than others. However, from the perspective that biomass fuel is carbon-neutral, the ECE in Hubei province becomes the lowest one. It reflects the high dependence on biomass fuels in rural Hubei.

Fig. 3 presents the consumption of five types of energy in different income quintiles of households for the whole sample. With the income level rising, the amount of electricity consumption increases visibly. While the consumption of traditional biomass fuels presents a converse trend. It means that the household energy consumption structure shows a transition trend from traditional fuels to modern energy with income increasing.

Fig. 3. Energy consumption for each quantile of household. Source: Calculated by authors from survey data for 1027 households in central China.

However, can the sharp growth of electricity consumption benefit CO₂ emission mitigation? Fig. 4 provides some descriptive evidence for this question. When the emissions from biomass fuels are included, the trend line reflects a slight reduction of ECE with income growth before household income reaches the highest quintile. The households with the highest income level produce most ECE of all income groups. When biomass fuel is regarded as carbon-neutral, the ECE shows a rising trend with the growth of income level, particularly for the two groups with the highest incomes. The amount of energy consumed by rural households (in kgce) shows nearly no increase before the income reaches the highest level. It may because electricity consumption increases especially for the highest two income groups, while biomass consumption declines especially between the middle- and high-income groups (shown in Fig. 3), which reflects the significant energy use structure change in richer households.

Fig. 4. Energy consumption and ECE for each quantile of household. Source: Calculated by authors from survey data for 1027 households in central China.

Generally, higher dependence on modern energy is regarded as a positive transition in energy use structure, but higher CO₂ emission are inconsistent with the clean transition goal. It seems that household income increases bring conflicting results with respect to energy structural transition and absolute CO₂ emissions. However, according to the policy aim to reach the peak in carbon dioxide emissions before 2030, reductions in the carbon emission intensity are also regarded as environmentally sustainable. Thus, ECI is further highlighted to assess the effects of China's rural energy transition.

The ECI in different income quintiles of households is shown in Fig. 5. When biomass fuels are considered to be carbon-intensive, ECI decreases significantly with the income growth. Considering the results in Fig. 4 together, although richer households tend to consume more energy than others, they produce less CO₂ per unit of energy consumption. It means that the household energy structure may be optimized with their income growth. When biomass fuels are assumed to be carbon-neutral, ECI increases with income growth before household income reach the highest level. It reflects the reduction of biomass fuel use with income growth. The lower ECI in the highest quintile further suggests the dramatic rise of electricity use in the households with the highest-level income.

Fig. 5. ECI for each quantile of household. Source: Calculated by authors from survey data for 1027 households in central China.

Biomass fuels as carbon-intensive residential energy

Table 4 presents the baseline estimation results of Eq. (4) from the (short-term) perspective that by burning straw or firewood households generate greenhouse gas emissions and biomass fuels are therefore considered to be carbon-intensive. The estimated coefficient of income is not significantly different from zero in the equation explaining ECE. The result is inconsistent with most studies that focus on electricity-using urban residences. For example, Goldstein et al. (2020) found that the per capita carbon footprints of wealthier Americans are about 25 % higher than those of low-income residents. Gill and Moeller (2018) and Li et al. (2019) also found that higher income was detrimental to household greenhouse gas emissions saving in Germany and China, respectively. The potential explanation is, with more diverse energy choices than urban residents, the nexus between income and ECE in rural households become more complex. In the equation for ECI, however, the coefficient of income is negative and statistically significant. A 1 % increase in income is estimated to reduce the energy-related carbon emission intensity by 0.041 kg/kgce. This finding is consistent with previous studies (Sanches-Pereira et al., 2016; Zhang et al., 2016). It suggests that the efficiency of energy consumption rises with economic growth or getting out of poverty.

Table 4. Baseline estimation results (OLS).

Among the household demographic factors, age and family size show positive and statistically significant impacts on both ECE and ECI. At a given income level, rural households with a larger size tend to spend a larger share of their income on energy to meet the demand of all its members. Moreover, larger households have more time endowment that can be spent on biomass fuels collection (Chen et al., 2006; Qiu et al., 2018). Furthermore, ECI is significantly lower for households with a higher education level. This finding is consistent with the expectation that more-educated people possess more knowledge about the efficiency and convenience of modern, more energy-efficient fuels and have a higher environmental awareness (Farsi et al., 2007; Muller & Yan, 2018).

Building characteristics do not significantly influence ECE, but do have some significant effects on the ECI of rural households. Specifically, newer and high-value houses have lower ECI than older and low-value houses. This results is consistent with the finding in previous research that more efficient energy facilities are generally installed in new and higher-value houses (Salari & Javid, 2017). Our results indicate that this lowers the energy intensity, but increases the total amount of energy that is consumed. As a result, the total amount of ECE does not significantly change. The electricity price has significantly negative effect on ECE and a significantly positive effect on ECI. This finding suggests that households use less energy in total and switch to energy sources with relatively high carbon emissions when the electricity price increases.

Regression results of the mediation analysis are presented in Table 5. Columns (1)–(3) present the results for the energy structure transition path. The estimated coefficient of income is positive and statistically significant in the equation for no traditional energy. It illustrates that wealthier rural households are more likely to stop using traditional fuels, which is in line with the “energy ladder” hypothesis. The estimated coefficients of no traditional energy in the equations for ECE and ECI are both negative and statistically significant. In other words, the substitution of traditional fuels by modern energy sources reduces ECE as well as ECI. These results are consistent with the findings of Liu et al. (2012) and Fowlie et al. (2018). Specifically, we find that the substitution of traditional fuels by modern energy types reduces ECE by 2240.3 kg and ECI by 0.576 kgCO₂/kgce on average.

Table 5. Estimation results of mediation analysis.

The estimation results of the mediation analysis for electricity consumption are shown in columns (4)–(6) of Table 5. The estimated coefficient of income is positive and statistically significant. This finding confirms that the demand for electricity-using appliances increases with income (Auffhammer & Wolfram, 2014). Moreover, wealthier households tend to be more involved in online activities and services that also increase household electricity use (Eludoyin & Lemaire, 2021; Yu et al., 2020). Interestingly, the estimated coefficient for electricity consumption is positive and significant in the equation for ECE and negative and significant in the equation for ECI. It reflects that, at a given income level, more electricity consumption is associated with higher total energy use but with lower CO₂ emissions per unit due to the lower-carbon trait of electricity as compared to traditional fuels.

The results of the Sobel tests (see Table 6) indicate that the mediation effects of no traditional energy use and electricity consumption are both statistically significant. The indirect effects on ECI are similar in magnitude and explain about 38 % of the effect of income on both indicators. But for ECE the story is quite different. The effect of income mediated through electricity consumption is positive and explains about 59 % of the total income effect, while the effect mediated through no traditional energy consumption is negative and explains about 32 % of the total income effect on ECE. This may at least partly explain why the total effect of income on ECE is insignificant in Table 4. These findings suggest that abandonment of traditional energy sources is of crucial importance for reducing carbon emissions generated by rural household energy use in central China. When rural households continue to use traditional energy sources, higher incomes will tend to increase total carbon emissions by the increased use of electricity.

Table 6. Results of Sobel test of mediation effects.

Carbon-neutral full life cycle of biomass fuel

Columns (1)–(2) in Table 7 show the baseline estimation results that use of biomass fuels is treated as carbon neutral. The total effect of income on ECE is found to be positive and statistically significant, while its impact on ECI is not statistically significant. These findings differ considerably from the non-significant effect on ECE and the significant, negative effect on ECI reported in Table 4. Hence, when the carbon sink function of biomass during its creation is also taken into account and assumed to be equal to emission during combustion, we find that income growth contributes to rising energy-related carbon emissions by rural households and does not significantly contribute to reductions in the carbon intensity of energy use.

Table 7. Estimation results if biomass fuels are assumed carbon-neutral^a.

The other four columns in Table 7 presents the mediation analysis results under the assumption that biomass fuels are carbon neutral. Results of the first step, explaining the mediation variable from the income level and other variables, are the same as reported in Table 5 and are not repeated here.

The estimated coefficient of no traditional energy use is negative and significant in the ECE equation but positive and significant in the ECI equation. The estimated reduction in ECE when households stop using traditional energy is 413.6 kg. Because use of biomass is considered as carbon neutral, this effect is much smaller than the estimated effect presented in Table 5. Because other traditional types of energy are no longer used as well, the negative effect is still significant (although only 18 % of the effect shown in Table 5). Interestingly, the carbon intensity of the energy used by rural households increases when they stop using traditional energy. Although traditional energy types with high emission factors like raw coal and honeycomb coal are no longer used, the replacement of biomass fuels by electricity and other types of modern energy more than offsets this negative effect on energy-related carbon intensity.

Electricity consumption has a significant positive effect on ECE growth, but no statistically significant effect on ECI. Not surprisingly, the estimated coefficient in the ECE equation is similar in magnitude to the one estimated under the assumption that biomass fuels do contribute to carbon emissions. But importantly, growing electricity use does not contribute to lower energy-related carbon emission intensity when the use of biomass fuels is considered as carbon neutral. On balance, the mix of (biomass fuel and other) traditional fuels that it partly replaces has a similar carbon intensity, resulting in a non-significant effect on ECI.

The Sobel test results presented in Table 8 indicate that the mediation effect of no traditional energy on ECE is significantly negative (at a 10 % testing level). It explains only a small proportion, ca. 7 %, of the income-induced change in total emissions when biomass fuels are considered as carbon-neutral. Its mediation effect on ECI is significantly positive, but is counterbalanced by other income-related effects on the emission intensity of energy use, resulting in a non-significant total effect of household income (see column (2) in Table 7). The mediation effect of electricity consumption on ECE is positive and highly significant. The proportion of the total effect mediated through this mechanism increases from 59.1 % to 73.5 % when biomass fuel is considered as carbon neutral instead of carbon-intensive. The mediation effect of electricity consumption on ECI is not statistically significant when biomass fuel is considered as carbon neutral. As discussed above, the replacement of (biomass fuel and other) traditional energy sources by electricity by rural households having relatively high incomes does not contribute to lower energy-related carbon emission intensities.

Table 8. Results of Sobel test of mediation effects, carbon-neutral assumption for biomass fuels.

Effects of regional differences in energy demand and policy implementation

Considering the temperature differences between different latitudes, we further examine how the variations in heating and cooling demands affect household energy use and its related carbon emissions. We assume that temperatures are lower at higher latitudes, and household heating demand increases and cooling demand goes down as a result.

Using the Yellow River and Yangzi River as boundaries, we grouped the sample counties into four temperature-based areas (Fig. 1b). Anyang, Xinxiang, Yanshi and Xinzheng are four north counties located around the Yellow River. The temperature in the Yellow River basin is the lowest among the selected counties. Households here demand more heating in winter and less cooling in summer. Nanzhao, Queshan, Zaoyang and Gucheng are four counties located between the Yellow River and the Yangzi River. These counties are regarded as a part of southern China geographically, but there are still many households that need heating in winter. Jianshi, Dangyang, Honghu, Huarong, and Luotian are five counties located in the Yangzi River basin. The winter here is warmer than in the first two groups of counties. Although some residents still use heating, the heating period in this region is shorter. The other five counties, Xinhuang, Liuyang, Longhui, Leiyang and Anren, are located south of the Yangzi River. The heating demand here is lowest, but cooling demand in summer is highest.

Taking the south of the Yangzi River area as the control group, we explore the effects of latitude-related variation in heating and cooling demand on household energy transition and its related carbon emissions. Location factors (dummy variables: Yellow River Basin, Between Yellow River & Yangzi River, and Yangzi River Basin) are added to Eqs. (4), (5), (6) to estimate the impact of location on household energy choices, electricity consumption and ECE and ECI, while controlling for income and other relevant factors.

The results for household energy choices and electricity consumption are shown in columns (1) and (2) in Table 9. Controlling for other explanatory factors, we find that rural households living in the Yellow River basin are significantly more likely to have stopped using traditional energy than rural households in the other three regions (see column (1)). Combined with the discussions based on the statistical results presented in Fig. 2, this result seems to reflect the fact that traditional biomass fuels are used more in cooking in the south of central China but are used less in heating. It may also be a result of a reduction in coal consumption caused by the ‘Coal to electricity, coal to gas’ policy, an ongoing project conducted in the northern rural China (Meng et al., 2019). This latter explanation is supported by the evidence presented in column (2) that, controlling for other explanatory factors, electricity consumption of the Yellow River basin households does not differ significantly from the control group; households living in the regions on the north of the Yangzi River and between the Yellow and Yangzi River consume significantly less electricity. These findings suggest that households at lower latitudes consume more electricity because of higher cooling demand, while households at relatively high latitudes use more electricity for heating to substitute for traditional fuels.

Table 9. Estimation results for regional effects.

The regression results for ECE and ECI are shown in columns (3)–(6) in Table 9. We find that carbon emissions (ECE) are significantly lower in the Yellow River basin when biomass fuels are considered as carbon-intensive (column (3)). On the other hand, when biomass fuels are considered as carbon-neutral, the relatively high heating demand in the northern part of central China contributes to significantly higher carbon emissions caused by electricity consumption (see column (6)).

There are 2 special administrative region and 26 prefectures in northern China that were selected as the pilots of the ongoing ‘coal to gas, coal to electricity’ project. All of them are located within or to the north of the Yellow River basin (Meng et al., 2019). In our survey regions, only the three northern prefectures in Henan province (Xinxiang, Anyang, and Zhengzhou) are located in the pilot area, no prefectures in Hubei and Hunan in the project. Considering the impacts of variations in temperature and energy use habits for the households located in the same basin will be relatively small. To obtain more insights into the effects of the energy transition policy, we compare the household energy use and carbon emissions characteristics in the Yellow River basin, where Xinxiang, Anyang and Xinzheng are pilot areas for the policy, while Yanshi is a non-pilot county.

We add three dummy variables, indicating whether a household is located in one of the three pilot prefectures, to Eqs. (4), (5) to explore the policy effects on household energy choices, electricity consumption, ECE and ECI, while controlling for income and other factors. The estimated policy effects are presented in Table 10. They suggest that there is no difference in the suspension of traditional energy between households in pilot and non-pilot areas in the Yellow River basin (see column (1)). But households in pilot areas consume significantly more electricity on average (see column (2)). These results supports the energy stacking model observed for rural China by Han et al. (2018) and Han and Wu (2018). Total carbon emissions (ECE) by rural households in the pilot prefectures are significantly higher as a result, when taking a short-term perspective (column (3)). But carbon intensity (ECI) is significantly lower in the pilot prefectures, both from a short-term and from a life-cycle perspective on biomass carbon emissions (see columns (4) and (6)).

Table 10. Estimation results for energy transition policy effects.

Conclusion and policy recommendations

Due to the serious challenges posed by climate change, the reduction of carbon emissions has become an international priority. With further improvements of living standards, household energy demand is likely to rise and carbon emissions are expected to increase if the carbon intensity of the consumed energy remains unchanged. A reduction of the total carbon emissions in the residential sector is therefore often difficult to achieve in the short term. As a consequence, most governments all over the world have set phased emission reduction targets.

In the case of China, the government has set ‘carbon peak 2030’ as a national strategy in the ‘14th Five Year Plan’ as a medium-term target. To achieve this goal, the expected growth in energy consumption needs to be accompanied by a significant reduction in energy-related carbon intensity (ECI) in order to achieve a reduction in carbon-related energy emissions (ECE). This study aims to provide more insight into the relationship between income growth and ECE and ECI for rural households in central China, a region where coal-fired electricity dominates modern energy consumption. Household energy consumption survey data collected in three provinces in central China are used to calculate total energy-related carbon emissions (ECE) and the energy-related carbon intensity (ECI) per unit of energy. A mediation effect model is employed to distinguish between energy structure changes and electricity consumption amount as mediating factors in the relationships.

We argue that whether biomass fuels are treated as either carbon-intensive or as carbon neutral has major consequences for the analysis. From a short-term perspective, the burning of biomass fuels is considered to be carbon-intensive. We find that household income growth in central China has no impact on ECE but can reduce ECI significantly. The change in energy structure and increase in electricity consumption that results from income growth have opposite effects on ECE, but both are conducive to the reduction of ECI. Taking a long-term perspective, including the carbon sink function of biomass during its creation into account, biomass fuel is considered to be carbon-neutral. This results in notably different findings. Household income growth is found to increase ECE but does not significantly affect ECI. The positive effect of growth in, coal-fired, electricity consumption on ECE is much larger than the negative effect of the change in energy structure.

Our findings imply that, under the present technological constraints (the high reliance on coal-burning for electricity generation), it is tough to balance two main targets of China's rural revitalization policy, i.e. rural household income growth and transition towards a low-carbon living environment. We provide evidence for the view that coal-based electricity is not the best alternative to traditional biomass energy for reducing carbon emissions. Vigorous promotion of electricity may in fact increase energy-related carbon emissions and thereby obstruct the achievement of carbon neutrality. We calculated electricity-related carbon emissions based on the conventionally used approach in this study. However, considering potential emissions during both upstream and downstream stages in the electricity production process, coal-based electricity may be even more carbon-intensive (Chen & Wemhoff, 2021). Thus, decarbonization in power generation should serve as a major target to achieve residential emission mitigation aims.

As the top emitter of CO₂, with power generation being the main culprit, the electricity sector plays a key role in meeting Chinese residential carbon neutrality goal. To reach the climate change target of the Paris Agreement, power generation sectors should focus more on the use of cleaner generation technologies, such as hydropower, instead of coal-fired generation. In addition, the use of renewable energy for power generation, such as solar, wind, and nuclear power, may be increased to stimulate carbon-neutrality of energy consumption. However, it will be not easy to realize the transition towards renewable energy. As the use of renewable energy resources mainly relies on the availability of local resource endowments, it highly depends on local geographical conditions. Moreover, large-scale investments are generally needed in the development of these resources.

For biomass fuels, it should be noted that the direct burning of biomass fuels causes serious indoor air pollution. Although biomass fuels can be regarded as a carbon-neutral source of energy in the whole ecological system, from a welfare-maximizing perspective rural households may be discouraged to use biomass fuels as their main energy source. In recent years, China's forest stock is increasing as a result of government policies aimed at boosting emission sinks. Forest protection combined with less use of firewood can be regarded as an effective measure to mitigate climate change.

Given the prominent role of bioenergy in limiting global greenhouse emissions in the long run, the direct-burning biomass fuels should preferably be replaced by biomass moulding fuel (BMF). This calls for the further development of the biomass industry. Bioenergy generation combined with carbon capture and storage may be cheaper decarbonization pathways than pathways based on other emission reduction technologies. Its key contribution is identified widely, and is not limited to the residential sector (Butnar et al., 2020; Demetriou & Hadjistassou, 2021). In rural residential sectors, some simple bioenergy generation practices such as biogas and bio-natural gas may be promoted. They may provide an important contribution to the clean energy transition of rural households.

There are some limitations for this study. The empirical results that we presented in this paper are based on cross-section data. This means that the results may be affected to some extent by omitted factors that are correlated with both the dependent and the independent variables. Future studies in this field may use panel data, that should preferably be collected in a larger area in (northern) China, to check the validity of the main findings of our research. In this study we focus on energy-related carbon emission and carbon intensity. We did not collect data on air pollution indicators, like PM2.5 and smoke, and therefore cannot assess the impacts of rural income growth on household indoor air pollution. When appropriate techniques for measuring indoor air pollution become available at a large scale, we recommend that research in this field will be extended in that direction.

Acknowledgments

Support for this research is provided by the National Social Science Foundation of China (20AZD091 & 19ZDA002) and Tsinghua Rural Studies PhD Scholarship (202106).

Tables, references and appendix omitted.

原文链接：“既要金山银山，又要绿水青山“的新时代探索

转载本网文章请注明出处