Dynamic disparities in clean energy use across rural–urban, regional, and ethnic boundaries in China

Abstract

The consequences of relying on solid fuels are disproportionately borne by minorities, the marginalized, and rural communities. However, the social disparities in transitioning from polluting energy to clean energy are not well understood. We track changes in the main energy source used for cooking among Chinese households between 2010 and 2018. We find that the proportion of households who rely on clean energy increased from 53.7% in 2010 to 80.1% in 2018. We detect substantial disparities in clean energy use between rural and urban areas, across regions, and between ethnic minorities and the Han majority. Urban status, regional variations, and household characteristics entirely accounted for the observed ethnic differences in clean energy use. Over time, disparities across rural–urban, regional, and ethnic boundaries declined, and household characteristics became irrelevant to the ethnic differences. Therefore, China’s efforts to mitigate the imbalance in socioeconomic development also reduced ethnic inequalities in clean energy use.

Supplementary Information

The online version contains supplementary material available at 10.1007/s13280-023-01913-5.

Introduction

To promote global sustainability, the United Nations has established 17 Sustainable Development Goals (SDGs) to be achieved by 2030 (United Nations 2015). In SDG 7 (affordable and clean energy), target 7.1 is to “ensure universal access to affordable, reliable and modern energy services.” However, around 2.4 billion people in the world still rely on polluting energy sources such as wood and charcoal for cooking, and more than 700 million people do not have access to electricity (IEA et al. 2022). Promoting the universal use of clean energy such as electricity and liquefied petroleum gas (LPG) to replace reliance on coal and biomass fuels is critically important. For instance, it is estimated that about 3.3 million premature deaths per year worldwide can be attributed to outdoor air pollution, and emissions from residential energy use have the largest impact on premature mortality (Lelieveld et al. 2015). Transitioning from polluting energy to clean energy also has tremendous impacts on socioeconomic development via mechanisms such as poverty alleviation, improvement in educational opportunities, and environmental conservation (World Health Organization 2016; Bailis et al. 2017; Chen et al. 2017). In addition, progress in access to clean energy can also support the achievement of many other SDGs (Liu et al. 2018a; Nerini et al. 2018).

Disparities in energy use also reflect economic inequalities. Most of the people who have limited access to clean energy are in relatively poor countries, and the vast majority of them live in Asia and Sub-Saharan Africa (Gonzalez-Eguino 2015). Between 2010 and 2020, the number of people in the world who relied on polluting energy for cooking dropped by 20%, from around 3 billion to around 2.4 billion, which mainly reflects advancements in the transition toward clean energy in Asia. In Sub-Saharan Africa, in contrast, the number of people who relied on polluting energy for cooking nearly doubled between 1990 and 2020 (IEA et al. 2022). Substantial disparities in access to clean energy also exist across different regions within countries due to regional economic inequalities, particularly between rural and urban areas (Rao and Reddy 2007; Masera et al. 2015).

Universal access does not guarantee transition to clean energy because clean energy is often more expensive than traditional energy sources. In fact, the number of people in the world who rely on polluting energy for cooking is more than three times the number of people who do not have access to electricity (IEA et al. 2022). As described in the “energy ladder” model, as household income levels rise, biomass fuels tend to be replaced first by coal, kerosene, and oil, which are in turn to be replaced by clean energy sources (Hosier and Dowd 1987). In many circumstances, as the “energy stacking” model suggests, households may use multiple energy sources simultaneously rather than completely replacing solid fuels such as coal and wood with clean energy when their income improves (Masera et al. 2000). As a result, the consequences of relying on polluting energy sources, such as premature deaths and poverty, are disproportionally borne by ethnic minorities, the marginalized, and rural communities, and people in low- and middle-income countries (Landrigan et al. 2018; Zhao et al. 2019). However, our current understanding of the dynamic disparities in energy use as well as the underlying mechanisms that determine the disparities is still limited.

China has the highest rate of premature mortality attributable to outdoor air pollution in the world (Lelieveld et al. 2015). Therefore, the transition from polluting energy to clean energy in China deserves careful examination in order to ensure effective policy interventions. China’s economy has experienced rapid growth over the past four decades; it has been the world’s second-largest economy since 2010, and is now the largest CO₂ emitter. Thanks to the recent development of a national power grid infrastructure, almost all Chinese people are now connected to electricity. However, many people still rely on solid fuels for cooking and space heating, leading to ambient and indoor air pollution. It is estimated that solid fuels used for cooking contributed approximately 56% more to indoor air pollution than those used for heating on average among Chinese households in 2015 (Zhao et al. 2018). In fact, the contribution of residential use of solid fuels to the emissions of primary PM_2.5 (particulate matter with an aerodynamic diameter equal to or smaller than 2.5 µm) is far more than that of the transportation and power sectors combined in China (Liu et al. 2016a). As a result, air pollution from residential use of solid fuels is considered a leading risk factor for morbidity and mortality from air-pollution-related diseases such as ischemic heart disease, chronic obstructive pulmonary disease, and respiratory disease (Cohen et al. 2017). In 2015 alone, residential use of solid fuels was estimated to be responsible for more than 500 000 premature deaths in China (Zhao et al. 2018; Yun et al. 2020).

The rapid economic development in China has substantially advanced the progress toward pursuing sustainable development (Xu et al. 2020), including the promotion of the use of clean energy (Tao et al. 2018). However, China’s economic development has been uneven, with structural forces such as rural–urban disparity and regional variations being the most important determinants (Wu 2019). For instance, rural–urban disparity and regional variations accounted for about 10% and 12% of overall income inequality, respectively (Xie and Zhou 2014). Unbalanced development also creates disparities between the Han majority and the 55 ethnic minorities in China. Ethnic minorities, accounting for less than 10% of the nation’s population, lag behind those of the Han people in socioeconomic outcomes such as income and educational attainment (Wu and He 2016) and may also face disadvantages in using clean energy.

In this study, we analyze 2010–2018 data from a nationally representative survey dataset that covers 25 provinces representing about 95% of the population of mainland China (Xie and Hu 2014). (See Fig. 1 for a visual breakdown of these provinces into eastern, central, and western regions.) We refer to electricity, LPG, biogas, and solar energy as “clean energy,” which usually has a much lower impact on public health than coal and biomass fuels. We track the temporal trend in clean energy use and the share of different energy sources as the main energy source for cooking between 2010 and 2018. We then identify trends in the disparities in the use of clean energy across rural–urban, regional, and ethnic boundaries. The effects of urban status, regional variation, and ethnicity on clean energy use are evaluated with a random-effects logit model. Due to the unbalanced development between ethnic minorities and the Han majority and the uneven geographic distribution of ethnic minorities across China, we estimate the effects of urban status, regional variations, and household characteristics on the disparities in clean energy use between ethnic minorities and the Han people with nearest-neighbor matching. To the best of our knowledge, no study has identified the underlying mechanisms that drive the dynamic inequalities in transitioning from polluting energy to clean energy.

Fig. 1.

Distribution of western, central, and eastern regions, and two major rivers in China. Names of provinces, autonomous regions, and municipalities are shown

Materials and methods

Data and measures

We analyze data from a national survey, the China Family Panel Studies (CFPS), that has been administered biennially by the Institute of Social Science Survey at Peking University since 2010. The survey is a comprehensive longitudinal social survey that is nearly nationally representative (Fig. 1) because it covers 25 provinces (excluding Hainan, Inner Mongolia, Ningxia, Qinghai, Tibet, Xinjiang, Hong Kong, Macau, and Taiwan due to challenges in logistic arrangements), representing about 95% of the population of mainland China (Xie and Hu 2014). Through a three-stage probability-proportional-to-size sampling with implicit stratification, it drew a sample of 19 986 households, and successfully interviewed 14 960 households containing 57 155 family members in 2010. The survey aims to track these family members and interview their corresponding households throughout their lives. Overall, 13 315, 13 946, 14 019, and 14 218 households were successfully interviewed in 2012, 2014, 2016, and 2018, respectively (Xie 2020).

The survey solicited comprehensive information at both the household and individual levels. In particular, respondents were asked about the main energy source that they used for cooking, which included electricity, LPG, coal, wood and crop residue, biogas, and solar energy. For the regression analysis reported in Table 1, we create a dummy variable, energy (energy = 1 if the main energy source was electricity, LPG, biogas, or solar energy; energy = 0 if the main energy source was wood and crop residue or coal).

Table 1.

Estimated effects of ethnicity, urban status, regional variations, and household characteristics on clean energy use

Independent variables	Model 1		Model 2		Model 3
	Parameters	Standard error	Parameters	Standard error	Parameters	Standard error	Marginal effects
Ethnicity	− 0.60***	0.08	− 0.38***	0.08	0.04	0.08	0.01
Urban			2.46***	0.05	2.29***	0.05	0.32***
Central					0.84***	0.07	0.11***
Eastern					2.08***	0.07	0.27***
Log(expenditure)	0.71***	0.02	0.60***	0.02	0.58***	0.02	0.08***
Family size	− 0.32***	0.01	− 0.24***	0.01	− 0.21***	0.01	− 0.03***
Mean education	1.23***	0.03	0.97***	0.03	0.92***	0.02	0.13***
Senior	− 0.08*	0.04	− 0.16***	0.04	− 0.20***	0.04	− 0.03***
Year12	0.69***	0.04	0.76***	0.05	0.78***	0.05	0.09***
Year14	0.78***	0.04	0.81***	0.04	0.84***	0.05	0.10***
Year16	1.06***	0.05	1.10***	0.05	1.16***	0.05	0.13***
Year18	1.67***	0.05	1.73***	0.05	1.79***	0.05	0.18***
Constant	− 3.56***	0.09	− 4.02***	0.09	− 5.08***	0.11
${\sigma }_{\mu }$	2.75***	0.04	2.57***	0.04	2.50***	0.04
$\rho$	0.70***	0.01	0.67***	0.01	0.65***	0.01
χ2	12 931***		11 431***		10 692***

Significant parameters for ${\sigma }_{\mu }$ and $\rho$ suggest that the random-effects model is appropriate, and significant test statistic for a χ² test indicates that the random-effects model is preferred to the pooled model. Number of observations = 62 933; Number of households = 14 597

*p < 0.05, ***p < 0.001

We divide the respondents into social groups based on ethnicity, urban status, and region (Xie and Zhou 2014; Wu 2019). Urban status of households (urban = 1 for urban households; urban = 0 for rural households) was identified based on the standard of China’s National Bureau of Statistics (Xie et al. 2017). We use a three-category variable to capture significant and systematic differences in socioeconomic development across the eastern, central, and western regions of China (Fig. 1). Households are divided into two ethnic groups, “minority” (ethnicity = 1) if any one of the household members was an ethnic minority, and “Han” (ethnicity = 0) if all the household members were Han people. In addition, we also include the following covariates in our regression analysis reported in Table 1, as they are known to be determinants of household energy use (Chen et al. 2012; Jeuland et al. 2015; Muller and Yan 2018): total household expenditure in the past year (household expenditures from different years were converted into 2010 real values; in 1000 yuan with log transformation), household size, mean educational attainment of adult household members (with ages >  = 22), and the presence of senior household members (senior = 1 for households with members > 60 years old; senior = 0 for households without senior members). See Table S2 for descriptive statistics of household characteristics. Educational attainment is measured in educational levels in ascending order (1 = less than an elementary school education; 2 = elementary school; 3 = middle school; 4 = high school; 5 = technical school; 6 = 4-year college degree; 7 = graduate degree).

Analytical methods

We calculate the proportion of households who used specific energy sources as the main energy source for cooking in each of the surveyed years between 2010 and 2018. To separate out the net effects of ethnicity, urban status, and region on clean energy use, we pool data across different years in a regression analysis. Empirically, clean energy use is modeled with a random-effects logit model (Wooldridge 2010):

$$P({\text{energy}}_{\mathit{it}}=1|H_{\mathit{it}},Y_t,u_i)=\mathrm{\Lambda }\left(H_{\mathit{it}}\beta +Y_t\gamma +u_i\right)$$ 1

where $\mathrm{P}{\text{(energy}}_{\mathit{it}}=1\text{)}$ is the probability of the ith household using clean energy as the main energy source for cooking in year t; $\mathrm{\Lambda }(.)$ is the logistic transformation function; $H_{\mathit{it}}$ represents ethnicity, urban status, region, and household characteristics associated with the ith household in year t; $Y_t$ is a dummy variable representing year t; $\beta$ and $\gamma$ are parameter vectors associated with household characteristics and year dummy variables, respectively; and $u_i$ represents the unobserved random effects associated with the ith household. In the logit model, the marginal effects of continuous variables are obtained as

$$\frac{\partial P(\text{energy}=1)}{\partial X}=\mathrm{\Lambda }\left(X\beta \right)\left[1-\mathrm{\Lambda }\left(X\beta \right)\right]\beta ,$$ 2

where X represents all continuous explanatory variables, and the derivative is calculated at the mean of the explanatory variables (Greene 2012). The marginal effect for a dichotomous variable (d) is obtained as

$$P(\text{energy}=1|{\overline{X}}_{\left(d\right)},d=1)-P(\text{energy}=1|{\overline{X}}_{\left(d\right)},d=0),$$ 3

where ${\overline{X}}_{(d)}$ represents the means of all other variables in the model.

We estimated the effects of urban status, regional variations, and household characteristics on the disparities in clean energy use between ethnic minorities and the Han majority. In order to reduce the bias due to covariates, we used the nearest-neighbor matching method to estimate the Average Treatment Effect (ATE) where ethnic-minority households formed a treatment group while the Han households formed a control group (see SI Appendix for details). We first conducted exact matching on urban status and province to estimate the disparities that remain when households from the treatment and control groups were selected from the same provinces and urban status. We then conducted nearest-neighbor matching on urban status, province, and the four household characteristics (expenses were used without log transformation) that were used in the regression analyses presented above to test if any disparities remain when these characteristics were matched between treatment and control groups. The matching used the Mahalanobis distance metric with exact matching on urban status, provinces, and the presence of senior household members. In addition, we conducted exact matching on village (for rural households) and community (for urban households) to test if there were any disparities at the local level (i.e. within a village or community). All matching was one-to-one with replacement (i.e. each treated household was matched to one control household). We used a bias-adjustment procedure that asymptotically corrects the conditional bias in finite samples (Abadie and Imbens 2006, 2011). Matching was conducted in Stata 16. In addition, we also tested the difference in clean energy use between ethnic minorities and the Han people for the unmatched data with a chi-square test.

Results

Even though over 99% of the respondents reported that they had access to electricity in 2010, 46.3% of the households still relied on solid fuels (i.e. coal or wood and crop residue) for cooking. The proportion of clean energy as the main energy source for cooking dramatically increased by about 26.4% points in the following eight years, from 53.7% in 2010 to 80.1% in 2018 (Fig. 2a). The rapid increase in clean energy use was due to increases in the use of LPG and electricity (Fig. 2b), rising, respectively, from 39.4% of overall energy use for cooking in 2010 to 55.0% in 2018 and from 12.9% in 2010 to 24.8% in 2018. As a result, the share of solid fuels declined substantially. The share of wood and crop residue was cut by about half from 36.4% in 2010 to 17.4% in 2018, while the share of coal dramatically dropped from 9.9% in 2010 to only 2.4% in 2018. The share of solar and biogas was minimal and only 1.5% and 0.3% of households used these fuels as their main energy source in 2010 and 2018, respectively.

Fig. 2.

Change in the main energy used for cooking among Chinese households between 2010 and 2018. a Proportion of clean energy. b Share of different energy sources

We detect a substantial disparity in use of clean energy for cooking between rural and urban households, albeit a disparity that declined between 2010 and 2018 (Fig. 3a). In urban households, clean energy use increased by about 15.5% points, from 75.1% in 2010 to 90.6% in 2018, while in rural households it increased by 28.9% points, from 33.5% in 2010 to 62.4% in 2018. As a result, the disparity between rural and urban households in clean energy use declined from 41.6% points in 2010 to 28.2% points in 2018. In rural communities, wood and crop residue accounted for 56.7% of energy sources in 2010, while the combined share of LPG and electricity was only 30.9% (Fig. 3b). Between 2010 and 2018, the share of LPG and electricity increased by 16.3 and 14.7% points, respectively, while the proportion of wood and crop residue fell by 23.1% points and coal by 5.8% points. As a result, more rural households relied on LPG than on wood and crop residue in 2018. In urban communities, LPG was predominant among energy sources for cooking, while the share of wood and crop residue was slightly greater than that of electricity in 2010 (Fig. 3c). Between 2010 and 2018, the share of LPG and electricity in urban China increased by 6.3 and 9.4% points, respectively. As a result, the share of solid fuels dropped to only 9.4% by 2018.

Fig. 3.

Comparison of changes in the main energy used for cooking between rural and urban households. a Proportion of clean energy. b Share of different energy sources for rural households. c Share of different energy sources for urban households

Compared to the disparity between rural and urban households, the differences in clean energy use for cooking across different regions were much smaller (Fig. 4a). The regional gradient in use of clean energy mirrors the regional gradient in socioeconomic development: highest in the most-developed eastern region, and lowest in the least-developed western region. However, the regional disparity is reduced by regionally differential rates of increase in the use of clean energy in favor of less developed regions: 33.7, 26.2, and 20.3% point increases, respectively, in western, central, and eastern China between 2010 and 2018. As a result, the disparity in clean energy use between eastern and western China declined substantially from 25.9% points in 2010 to 12.5% points in 2018. In 2010, wood and crop residue dominated the share of energy sources in western China, LPG dominated the share of energy sources in eastern China, while LPG and wood and crop residue each accounted for about one-third of energy sources in central China (Fig. 4b–d). The increases in the share of LPG and electricity between 2010 and 2018 were 15.7 and 20.6% points in western China (Fig. 4b), 17.4 and 9.9% points in central China (Fig. 4c), and 11.3 and 9.3% points in eastern China (Fig. 4d), respectively. By 2018, the share of both LPG and electricity was greater than that of solid fuels among households in all three regions of China.

Fig. 4.

Comparison of changes in the main energy used for cooking among households in western, central, and eastern China. a Proportion of clean energy. b Share of different energy sources for households in western China. c Share of different energy sources for households in central China. d Share of different energy sources for households in eastern China

We also found a substantial disparity in clean energy use for cooking between ethnic minorities and the Han majority (Fig. 5a). Clean energy use was greater for Han people than for ethnic minorities, a disparity that also declined substantially, from 19.6% points in 2010 to 11.7% points in 2018, as the clean energy use of ethnic minorities increased faster than that of Han people. In 2010, for Han households the share of LPG as the primary energy source for cooking was greater than that of wood and crop residue, while only 12.2% of households used electricity as the primary energy source (Fig. 5b). In comparison, the share of wood and crop residue was greater than that of LPG and electricity combined for ethnic minorities in 2010 (Fig. 5c). The increases in the share of LPG and electricity in Han households between 2010 and 2018 were 15.4 and 10.9% points (Fig. 5b), and 13.9 and 21.7% points in ethnic-minority households (Fig. 5c). By 2018, ethnic minorities had joined Han households in relying more on LPG and electricity than wood and crop residue for cooking.

Fig. 5.

Comparison of changes in the main energy used for cooking between ethnic-minority and Han households. a Proportion of clean energy. b Share of different energy sources for Han households. c Share of different energy sources for ethnic-minority households

The three contributors of inequality used in this study—urban status, region, and ethnicity—are associated. To separate out their net effects, we conduct a regression analysis and present the results in Table 1. When urban status and regional variations are not controlled for, ethnicity had a significant negative effect on clean energy use, shown in Model 1, replicating our findings reported earlier in Fig. 5. Examining the three models in Table 1, we observe that ethnicity had a weaker relationship with clean energy use when urban status is controlled for, and was no longer associated with clean energy use when both urban status and regional variations are controlled for. In other words, urban status and regional variations accounted for the ethnic differences in clean energy use.

We further interpret the effects of independent variables on clean energy use in Model 3 (Table 1). Urban dwellers were more likely to use clean energy as the main energy source for cooking than were their rural counterparts. It is estimated that urban households were 32% more likely to rely on clean energy than rural households on average. This reflects the fact that infrastructures that provide clean energy, particularly LPG, in urban areas are often more developed than in rural areas, while wood and crop residue are often more available to rural households relative to their urban counterparts. We also detect significant regional variations in clean energy use. Households in central and eastern China were, respectively, 11% and 27% more likely to rely on clean energy than those in western China. Household socioeconomic characteristics also had significant effects on clean energy use. Households with higher expenditures were more likely to use clean energy. A 1% rise in household yearly expenditure increased the likelihood of relying on clean energy by 8%. Family size was negatively related to clean energy use. An additional family member reduced the likelihood of relying on clean energy by 3%. Household members with higher educational attainment were more likely to use clean energy. An additional level in mean educational attainment among adult family members increased the likelihood of clean energy use by 13%. Households with senior members were 3% less likely to use clean energy for cooking than households without senior members. In addition, the effects of year dummy variables are consistent with the results from descriptive statistics, reflecting steady increases between 2010 and 2018 in the proportion of Chinese households who mainly used clean energy for cooking.

The effects of urban status, regional variations, and household characteristics on the disparities in clean energy use between ethnic minorities and the Han people changed over time. When ethnic-minority households and Han households were matched on urban status and province, the ethnic difference in the means of clean energy use in 2010 and 2012 dropped from 19.6 to 5.7 and from 12.4 to 7.9% points, respectively (Fig. 6). These differences became statistically insignificant between 2014 and 2018. When matched on urban status, province, and the four household characteristics that were used in the regression analyses above, no significant difference between ethnic minorities and the Han people was found throughout 2010–2018. Therefore, while the ethnic differences in clean energy use can be entirely attributed to urban status, regional variations, and household characteristics, the underlying mechanisms that drive the ethnic disparities changed over time. In addition, we find no significant difference between ethnic minorities and the Han people when they were matched on villages (for rural people) and communities (for urban people), suggesting that there was no local-level ethnic difference in clean energy use.

Fig. 6.

Difference in means of clean energy use between ethnic minorities and the Han households before and after matching on urban status, regional factors, and household characteristics. Asterisks refer to the level of significance of the difference (*p < 0.05; ***p < 0.001)

Discussion

China made substantial progress toward affordable and clean energy between 2010 and 2018. The rapid transition toward clean energy reflects the rapid economic growth of the nation. The gross domestic product (GDP) of China doubled between 2010 and 2018 (National Bureau of Statistics of China 2018). We detect substantial disparities in clean energy use across rural–urban, regional, and ethnic boundaries, which mainly reflect differences in socioeconomic development and the development of the infrastructure for the supply of clean energy between rural and urban areas and across different regions. We also find declining disparities between rural and urban areas, across different regions, and between Han and ethnic minorities, at least partially due to China’s investment in energy supply infrastructure, nation-wide conservation and development policies, and rapid urbanization.

China has been rapidly developing infrastructure for the supply of clean energy. With the construction of an additional 34 000 km of pipeline for the transportation of petroleum gas, the total length of pipeline in the country increased by 75.6% between 2010 and 2018 (Sun et al. 2011; Gao et al. 2019). By 2018, LPG accounted for more than half of the share of energy sources for cooking (Fig. 2b). There are significant benefits of transitioning residential energy toward clean energy (Liu et al. 2016b). For instance, the reduced household use of solid fuels between 2005 and 2015 is estimated to have averted about 400 000 PM_2.5-related premature deaths annually (Zhao et al. 2018).

The disparity in clean energy use between rural and urban households, albeit still substantial, has been declining, with wood and crop residue being replaced by clean energy (Fig. 3). In addition to the increased access to and affordability of clean energy, the reduction of wood fuels among rural communities was also attributable to China’s great efforts in conserving forest ecosystems. For instance, since the beginning of the millennium, China has been implementing two “payments for ecosystem services” (PES) programs—the Natural Forest Conservation Program and the Grain-to-Green Program—which are among the largest PES programs in the world (Liu et al. 2008). In addition to restoring forests through logging bans and afforestation, these programs also aim at improving rural livelihoods and alleviating poverty by providing economic incentives and creating employment in forest conservation (Yang et al. 2013). As a result, a great amount of wood fuels has been replaced with clean energy in many rural communities across the nation (Chen et al. 2017, 2020). Residential energy use was also affected by China’s efforts to combat ambient air pollution through a coal use ban and increased supply of electricity and LPG (The State Council of China 2013). Between 2010 and 2018, the share of coal as the main energy for cooking declined by 85.6% and 59.0%, respectively, in urban and rural areas (Figs. 3b, 3c).

The disparity in clean energy use by region reflects unbalanced development in China. Compared to the rugged topography and long distance to the coast in western China, eastern China is relatively flat in topography with favorable climate conditions, resulting in more advanced socioeconomic development. The decline in clean energy disparity between eastern and western China has mainly resulted from China’s policies. In order to narrow regional disparities in development, the Chinese government has been implementing the Western Development Strategy since 2000 to improve environmental and socioeconomic conditions in western China (Liu et al. 2018b). The proportion of the Chinese government’s fiscal transfers allocated to western China increased from 25% in 2010 to about 40% in 2015, substantially reducing the gap in economic development between the western region and the rest of China (Zhang et al. 2019).

China also launched the Small Hydropower for Fuel project in 2003 that aims to build small hydropower plants (mostly in western China; installed capacity each of < 50 megawatts) to provide electricity in rural communities in order to replace solid fuels, especially wood fuels, promote rural economic development, and generate employment in the energy sector (Kong et al. 2016). As of 2015, a total of 993 small hydropower plants with an aggregate installed capacity of 1913.6 megawatts had been built, providing electricity to millions of rural households (Zhang et al. 2021). In addition, the above-mentioned two PES programs have mainly focused on protecting forest ecosystems in the upper and middle reaches of the Yangtze and Yellow river basins in central and western China (Liu et al. 2008), incentivizing households in these regions to replace wood fuels with clean energy (Chen et al. 2017). By 2018, electricity accounted for the largest share of energy use in western China while LPG accounted for the largest share in central and eastern China among different energy sources for cooking (Fig. 4b–d).

The proportion of Han households who relied on clean energy for cooking was substantially greater than that of ethnic minorities, a disparity that is now narrowing. Our finding that urban status and regional variations accounted for most of the ethnic differences in clean energy use is consistent with the fact that more ethnic minority people live in rural and western areas than in urban and eastern areas (Wu and He 2016). Therefore, the declining disparity in clean energy use between Han and ethnic minorities can largely be attributed to China’s efforts in environmental conservation and reducing gaps in socioeconomic development between rural and urban areas and across different regions. While household characteristics also contributed to the ethnic differences in 2010 and 2012, they became irrelevant between 2014 and 2018, suggesting that the underlying mechanisms that drive the ethnic disparities are dynamic. Unfortunately, the survey dataset that we use did not cover 6 provinces, or their administrative equivalents (i.e. autonomous regions), of mainland China (see Fig. 1) with high concentrations of ethnic minorities (Table S1). The ethnic-minority population in these 6 provinces accounted for about 24.3% of the total ethnic-minority population in mainland China (National Bureau of Statistics of China 2012). Although we cannot generalize our results to the provinces where the survey was not conducted, it is plausible that residential energy use in those provinces has also been affected by similar forces as in other provinces because the development and conservation policies discussed above are nation-wide efforts.

In addition to government policies, the progress of Chinese households toward using clean energy for cooking may also have been affected by rapid urbanization and the associated migration of rural people to urban areas (Shen et al. 2017). Between 2010 and 2020, the urban population in mainland China increased from about 666 million to about 902 million, while the rural population decreased from about 674 million to about 510 million (National Bureau of Statistics of China 2010, 2020). As a result, the number of people who have access to the infrastructures that provide clean energy has increased. Even if people have access to and can afford clean energy, certain households may be reluctant to change from using solid fuels due to traditional culture and habits (Ravindra et al. 2019). For instance, the cultural practice of cooking pig fodder and smoke-drying pork for future use were main determinants driving the demand for fuelwood in southwest China’s Wolong Nature Reserve (Chen et al. 2017). Campaigns that improve the public understanding of the consequences, especially health impacts, of the residential use of solid fuels can induce behavioral changes toward adopting clean energy (Chen et al. 2013).

As China’s economy continues to grow, the transition toward clean energy among Chinese households is likely to continue. In addition to the above-mentioned policies, which are expected to continue to play important roles, a few recent policies may also substantially promote clean energy use and reduce disparities across rural–urban, regional, and ethnic boundaries. In 2012, the Chinese government proposed a strategic policy known as Ecological Civilization Construction, which aims to reduce negative environmental impacts of economic development and strengthen the protection of natural ecosystems (Wu et al. 2021). In 2016, the government published 23 indicators, including those specifically for forest conservation and reduction in the emissions of primary PM_2.5, that were to be used for performance evaluation of government officials (National Development and Reform Commission of China 2016).

Another national policy is the Targeted Poverty Alleviation that was launched in 2013 in order to lift the remaining 70 million people out of poverty by 2020, through mechanisms such as economic development, vocational training, resettlement, PES, and social security systems (Zhou et al. 2018). In fact, under the auspices of Targeted Poverty Alleviation the development of photovoltaic power has been adopted as one of ways to create access to electricity and generate income for people in poverty (Li et al. 2018). At the local level, the most relevant policy is probably the coal-to-electricity program of Beijing Municipality that has been implemented since 2016, and which has effectively replaced substantial residential coal consumption with clean energy (Barrington-Leig et al. 2019). Given that most people who rely on solid fuels live in rural areas of central and western China, policies that target residential fuel needs of specific regions for exclusive transition toward clean energy can be critical for achieving target 7.1 of the SDGs.

China’s rapid transition toward clean energy has important global implications. Based on our estimation, the number of Chinese people who relied on polluting energy for cooking dropped by about 341 million between 2010 and 2018, which accounts for more than half of the progress of the transition toward clean energy for cooking worldwide during this period (IEA et al. 2022). China’s experience in developing infrastructure for the supply of clean energy and implementing development and conservation policies can provide important reference for the transition toward clean energy in many other countries, particularly in Asia and Sub-Saharan Africa. For instance, China is now Africa’s biggest trading partner, and is investing in numerous projects, including power generation, in Africa. In many Asian and African countries where people still largely rely on polluting energy sources it is important to track trends in energy use and their relation to socioeconomic development and foreign investment. Future studies may also explore the interactions between disparities in the transition toward clean energy and disparities in the progress toward achieving many other SDGs.

Conclusions

We find that ethnic minorities, rural communities, and residents in central and western China lagged far behind the Han majority, urban communities, and residents in eastern China in using clean energy for cooking. The ethnic differences can be attributed to rural–urban and regional differences, and household characteristics initially, while household characteristics became irrelevant to the ethnic differences over time. Following China’s environmental and developmental policies, we identify a steady increase in clean energy use between 2010 and 2018, accompanied by declining disparities across different social groups. The progress of the transition toward clean energy for cooking among Chinese households accounted for more than half of the total worldwide progress of transition toward clean energy for cooking between 2010 and 2018. China’s strategies of investing in the infrastructure for the supply of clean energy and implementing development and conservation policies may also be adopted to reduce disparities in clean energy use worldwide.

Supplementary Information

Below is the link to the electronic supplementary material.

Biographies

Xiaodong Chen

is a Professor at the Research Center for Eco-Environmental Sciences, Chinese Academy of Sciences. His research interests include human–environment interactions, ecosystem services and human well-being, and systems modeling.

Yu Xie

is Bert G. Kerstetter ‘66 University Professor at Princeton University. His research interests include social stratification, demography, statistical methods, Chinese studies, and sociology of science.

Qiong Wu

is a Research Associate Professor at Peking University. Her research interests include survey research methodology, cognitive function, and educational achievement.

Yan Sun

is a Research Associate Professor at Peking University. Her research interests include survey research methodology.

Jianguo Liu

is Rachel Carson Chair in Sustainability and University Distinguished Professor at Michigan State University. His research interests include coupled human and natural systems, telecoupling, and systems integration.

Funding

Funding was provided by National Natural Science Foundation of China (42071265), Second Tibetan Plateau Scientific Expedition Program (2019QZKK0308).