How Does Income Inequality Affect Rural Households’ Transition to Clean Energy? A Study Based on the Internal Perspective of the Village

Abstract

Promoting clean energy transition in rural areas is a key path to achieving global sustainable development, protecting public health, and promoting ecological livability. Based on data from the China Family Panel Studies (CFPS), this paper employs a multi-dimensional fixed effects model to evaluate the impact of income inequality on rural households’ clean energy transition (CET) and examines its underlying mechanisms. Research findings indicate that income inequality significantly suppresses rural households’ CET, primarily by reducing basic energy consumption and hindering the upgrading of basic energy consumption structures. Government governance quality exerts a significant negative moderating effect on the relationship between income inequality and rural households’ CET. Further analysis shows that the inhibitory effect of income inequality on CET is more significant in the regions with a low economic development level and low coal resource endowment, and in the western and northeastern regions of China. Therefore, while continuously promoting rural income growth, the government should prioritize equitable distribution, strengthen institutional capacity-building, improve the social service and security system, and facilitate rural households’ CET.

1. Introduction

With the intensification of global climate change, the clean energy transition (CET) has become a central issue for countries in achieving sustainable development. However, social issues, such as energy justice and access and the use of clean fuels, which are facing the current energy transition process, still need to be solved urgently [1]. In the clean energy transition, developed countries and elites tend to adopt clean energy faster, while low-income groups are marginalized due to high cost thresholds and the lack of infrastructure [2]. International organizations have keenly recognized the multifaceted challenges in the CET process and are urging nations to address its inherent issues. The United Nations Sustainable Development Goal (SDG 7) emphasizes “ensuring access to affordable, reliable and sustainable modern energy for all”; it emplaces CET into the sustainable development framework, requiring cross-goal coordination to build a more resilient and inclusive future and promote the sustainable development of the global economy, society, and environment. As the world’s largest investor in clean energy and the world’s largest carbon emitter, achieving “carbon peaking” and “carbon neutrality” represents China’s solemn commitment to the global community. The Chinese government has also actively promoted CET through continuous policy innovation to facilitate the realization of the Sustainable Development Goals. However, in line with the global trend of clean energy transformation and development, China also faces deep-rooted structural challenges such as social income inequality and regional development imbalance, which are quietly affecting the process and benefits of clean energy transformation.

In some areas of northeast and northwest China, the average proportion of coal and fuelwood mixed burning reaches 75% (<Research Report on Comprehensive Treatment of Bulk Coal in China 2023>: https://www.ccetp.cn/newsinfo/6390799.html accessed on 8 July 2025) and the process of rural CET is still slow [3]. The prolonged use of non-clean fuels reduces rural labor supply [4], jeopardizes residents’ health, and traps socioeconomically disadvantaged households in an environment–health–poverty nexus [5], while generating negative regional environmental externalities [6], which is not conducive to regional sustainable development. Compared to non-clean alternatives, clean energy adoption significantly improves user health outcomes [7] and mitigates air pollution and GHG emissions, thereby achieving low-carbon co-benefits and promoting regional green, high-quality [8], and sustainable development in the region. Regarding household head characteristics, existing studies have demonstrated the impacts of age, education attainment [9], and gender [10] on CET. Specifically, younger age and higher education levels of household heads positively facilitate household CET, while traditional gender roles may constrain economic opportunities for female-headed households, leading to greater reliance on polluting energy sources. Household characteristics, household non-agricultural employment [11], household income [12], household wealth and household population size [13], and family social networks all have significant impacts on CET [14]. Specifically, urban location, higher non-agricultural employment share, greater household income and wealth, and more expansive social networks positively correlate with clean energy adoption, while a larger household size exerts a suppressive effect on household CET. At the macro-regional level, digital economy enhancement [15] and other factors can significantly advance the CET in rural households. In addition, other scholars have found that improving redistribution policies [16] and Low-Carbon Pilot Programmes [17] have emerged as key catalysts for rural households to transition to clean energy, proving the critical role of national policies in driving sustainable energy transitions. Further research points out that the quality of government governance can promote CET [18], especially in rural areas, and that a higher government governance quality can significantly improve rural clean energy fuel accessibility [19] and promote rural household CET.

Scholars have conducted extensive research on how income inequality and inequality affect clean energy consumption. First, income inequality exacerbates urban–rural disparities in clean cooking fuel accessibility by simultaneously increasing urban access while reducing rural availability [19]. Secondly, at the national level, studies have found that income inequality significantly exacerbates energy poverty [20] and suppresses CET in rural households [21]. Empirical evidence indicates that income inequality reduces aggregate energy demand while altering consumption patterns—decreasing renewable energy use and increasing reliance on non-renewables [22]. Furthermore, widening income disparities constrain energy access for low-income populations, forcing a shift toward cheaper, less sustainable energy alternatives. This dynamic subsequently suppresses clean energy technology innovation and delays its adoption [23], which is not conducive to the sustainable development of the economy, society, and environment. However, some studies reveal that income inequality can stimulate energy consumption [24], with further evidence identifying an inverted U-curve relationship [25]—initial inequality expansion exacerbates energy use, but beyond a threshold, it may suppress demand. Current scholarship regrettably lacks consensus on income inequality’s impact on clean energy adoption, with most analyses relying on national panel data that examine urban–rural disparities while overlooking intra-regional inequality’s influence on local energy consumption decisions. In particular, there is no clear answer to the question of whether intra-village income inequality will affect rural household decision-making and what effect it will have on household CET.

Other reasons cannot be ignored for studying village income inequality and the CET of rural households within rural regions. On the one hand, compared to cities, rural infrastructure, education, and medical resources are relatively backward; the people are relatively closed; the distribution of resources is uneven; and the mobility is poor, which will make it easier for the able minority in rural areas to access most resources [26,27] thus exacerbating the income gap within rural areas. On the other hand, traditional biomass and non-market fuels remain the primary energy source for essential household requirements among 3 billion individuals globally, accounting for 30–40% of the population, with this demographic concentrated predominantly in less-developed economies and rural regions [28]. The CET of rural households is a key path to comprehensively promoting ecological livability and sustainable rural development. Thus, there is an urgent need to clarify the factors influencing rural households’ CET amidst intra-rural income disparities to formulate countermeasures to foster CET and realize rural inclusive green development. Based on the above analysis, this article will first examine whether income inequality within villages in rural areas has a significant impact on the CET of rural families. If such an impact exists, it will further explore what role energy consumption plays in it and whether the quality of government governance can have a significant impact on its correlation.

This study builds upon existing research by constructing a multi-period unbalanced panel using CFPS data to examine income inequality’s impact on rural households’ CET. Here, rural household CET specifically refers to the energy type used for daily cooking—a fundamental survival need where energy choices substantially reflect household energy patterns. Existing research shows cooking energy accounts for 44% of total residential energy consumption in rural China (<Research on Household Energy Consumption in China (2015)>: https://casisd.cas.cn/zkdt/gn/201610/W020161017502592202510.pdf accessed on 8 July 2025), justifying the use of basic cooking fuel choices as a proxy for rural households’ CET status. The marginal contributions of this study may manifest in three key dimensions: First, it breaks through the boundary of previous research in the research scope of regional inequality and further refines the measurement scope of regional inequality by focusing on the income gap within rural areas. Considering the impact of physical distance and the living environment on rural household decision-making, this paper accurately captures the relative income status of rural households within the village by calculating the relative deprivation index of individual households and provides a village-level reexamination of income inequality’s effects on rural households’ CET. Secondly, considering the limitations of the questionnaire, it innovatively employs the proportion of household energy expenditure to the total household expenditure to measure rural households’ energy affordability. From the perspective of energy consumption in rural households, this paper further clarifies the operational mechanisms through which total basic energy consumption and basic energy consumption structure mediate the impact of income inequality on rural domestic CET. Finally, this paper attempts to construct a comprehensive index system of regional government governance quality to break through the limitation of the single perspective of previous research. From the perspective of local government governance quality, this paper incorporates government governance quality into the model to test what role the government governance quality plays in the relationship between income inequality and the CET of rural households, and to further enrich relevant research.

2. Theoretical Analysis

2.1. Income Inequality and CET of Rural Households

As posited by the energy ladder framework, the choice of household energy use type is limited by the household income level. Combined with the investment decision theory, low-income rural households demonstrate a higher propensity for non-clean energy utilization for the following reasons: On the one hand, rural households with relatively low income cannot afford the early investment in CET due to economic constraints, and household expenditure will give priority to meeting the short-term needs of family life, which imposes barriers to clean energy diffusion [29]. On the other hand, rural households with low relative income may be unwilling or unable to afford the high use and maintenance costs of clean energy [30]. The relative income hypothesis holds that income inequality will amplify the consumption demonstration effect of high-income households on low-income households, and that households with higher relative incomes prefer to choose clean energy [31]. However, with the expansion of income inequality, it is easier for high-income groups with a relatively low share to obtain most social wealth through resource integration, increasing the proportion of low-income groups in society [32]. Income and wealth are the decisive factors of household CET [33], and households with low relative income cannot replicate the use pattern of clean energy due to their limited budget [34]. More importantly, households with low relative incomes, constrained by low social networks, may lack knowledge of the key information about CET [35]. Consequently, higher intra-village income inequality significantly impedes rural households’ CET.

Hypothesis 1.

Income inequality will inhibit the CET of rural households.

2.2. Energy Consumption Effect of Income Inequality

Empirical studies consistently demonstrate that widening income inequality significantly constrains household consumption capacity. From a psychological perspective, income inequality inhibits rural households’ CET by reinforcing cognitive biases. Low-income rural households tend to be loss-averse, perceiving the high upfront costs of adopting clean energy as a “loss,” which makes them more prone to exclusion [36]. According to prospect theory, rural households with low-income status prioritize immediate gains, and traditional non-clean energy sources—due to their affordability and accessibility—better align with their psychological preferences [37]. This present bias leads rural households to systematically overestimate the costs of clean energy adoption and maintenance while underestimating its long-term benefits, thereby hindering optimal CET decision-making [38] and resulting in low household energy consumption. From an economic perspective, income inequality reduces rural households’ energy consumption by increasing precautionary savings due to heightened financial risk. Income inequality reduces the total basic energy consumption of rural households. Higher income inequality encourages households to save more [39], especially for low-income groups. Reducing current consumption and increasing savings can help to cope with uncertain risk shocks [40], relieve household liquidity constraints [41], and improve household social status [42] to achieve the long-term healthy development of the family. Based on budget constraint theory, rural households’ subsistence energy consumption declines correspondingly when their overall consumption expenditure contracts.

Further, income inequality makes the affordability of household energy expenditure worse, worsens rural households’ basic energy consumption structure, and blocks the upgrading of the basic energy consumption structure. When households reduce consumption, the consumption expenditure (hedonic and developmental consumption) that is limited by the budget constraint/high-household-demand elasticity will decline more [43], while the decline of subsistence consumption will be lower due to its rigid demand. In addition, non-farm employment can effectively increase household income, reduce household income inequality [44], and promote the CET of rural households [45]. However, when household members shift to non-agricultural employment, labor availability for collecting non-clean energy (e.g., biomass) decreases [46]. This reduces households’ ability to obtain near-zero-cost traditional fuels, forcing them to increase market-based energy expenditures. Consequently, the transient increase occurs in the proportion of household budgets allocated to essential energy needs, worsening energy affordability in the short term. Under income inequality effects, clean energy’s high costs not only constrain adoption in energy-budget-constrained households but also worsen energy affordability for low-income families [1]. This deteriorates basic energy consumption structures, impedes their upgrading, and reinforces reliance on non-clean energy.

Hypothesis 2.

Income inequality inhibits rural households’ CET by reducing their basic energy consumption.

Hypothesis 3.

Income inequality inhibits rural households’ CET by blocking the upgrading of the basic energy consumption structure.

2.3. Effect of Government Governance

The market failure theory holds that there may be various types of failures and imperfections in the free market mechanism, leading to inefficient resource allocation and unreasonable social welfare, and emphasizes the effectiveness of government intervention to safeguard overall social and public interests. From the perspective of distributive justice within energy justice theory, effective government intervention can promote redistribution fairness by reducing corruption [47] and enhancing regional governance, thereby mitigating income inequality [48], with low-income groups benefiting the most. This directly alleviates the issue of “energy poverty” caused by income inequality, providing institutional support for the economic foundation of rural households’ CET. From the standpoint of procedural justice, high-quality government governance ensures the effectiveness of information transparency mechanisms, broadens public participation channels, and improves feedback pathways [49]. This maximizes the engagement of all stakeholders, including rural households, in energy decision-making, fostering trust in government policies and institutions [50], thus reducing resistance to rural households’ CET. From the perspective of recognition justice, effective government intervention achieves this by formulating differential policies tailored to the real-world contexts across diverse regions and population segments [51]. This safeguards equal rights and opportunities for low-income rural households to access modern, clean, and affordable energy services, thereby facilitating their CET. Social capital theory holds that the stability and effectiveness of government and social systems will affect individuals’ social capital accumulation and trust, and then affect their decision-making behavior. From the perspective of CET, household energy use and consumption types are closely related to the governance quality of regional governments; therefore confidence in government policies and their implementation has become a critical factor significantly influencing household cooking energy choices [52]. Effective government intervention can reduce government corruption [22], strengthen public trust [53], and promote energy transition to increase renewable energy consumption [54]. Many studies have shown that the quality of government governance has an impact on income inequality and household CET from all aspects.

Hypothesis 4.

Government governance quality exerts a negative moderating effect on the relationship between income inequality and rural households’ CET.

As illustrated in Figure 1, within the conceptual framework of this study, intra-village income inequality suppresses household CET by reducing basic energy consumption and hindering the upgrading of energy consumption structures among rural households. Moreover, higher-quality government governance can mitigate the negative impact of rural income inequality on household CET.

3. Variable Selection and Model Setting

3.1. Source of Data

After matching the household database with the individual database and retaining the rural samples, the variables are processed as follows: (1) Eliminating all the samples with abnormal and missing key information; according to the official advice, the financial respondents in each household are treated as the household head, and only samples with household heads aged 16 or older are included. (2) All income, consumption, and asset variables are winsorized bilaterally at 1% to reduce the interference of extreme outliers. (3) The samples are selected and the survey samples that have been interviewed at least twice during the four phases of the survey are kept. (4) Villages with only one survey sample in the same survey year are excluded, and the data are further processed as the unbalanced panel data of 2014, 2016, 2018, and 2020. Since the last publication of the village-level data of CFPS was in 2014, the village-level variable data used in this paper are all obtained by matching in 2014. The provincial coal-power price indices were obtained from the National Development and Reform Commission’s Price Monitoring Center portal, with missing values supplemented using interpolation methods.

3.2. Description of Variables

3.2.1. Explained Variable: CET of Rural Households

Referring to the existing research [3], and according to the household questionnaire item in the CFPS database, “Which fuel do you mainly use for cooking?”, if rural households primarily use electricity, natural gas, biogas, solar power, piped gas, and liquefied petroleum gas—all classified as clean energy sources—they are considered to be utilizing clean energy, and a value of 1 is assigned. A value of 0 is assigned to households whose primary energy sources consist of non-clean options including firewood and coal.

3.2.2. Explanatory Variable: Income Inequality

Social capital, which is rooted in the geographical network of rural households, has a direct impact on rural household decision-making. Referring to the calculation method of the relative deprivation index [55], this study takes the village as the basic unit of analysis and measures the income deprivation of each household by comparing its income with that of other households in the same village.

3.2.3. Mediating Variable: Rural Households’ Basic Energy Consumption

Households’ basic energy consumption, as a mediating variable, consists of two dimensions: total consumption and consumption structure. The total consumption metric measures the scale of household basic energy use, reflecting both energy consumption capacity and consumption patterns. Empirically, this study calculates it by taking the logarithm of the sum of annual household fuel expenses and electricity costs plus one, where a higher value indicates greater clean energy adoption among rural households.

Research based in the UK proposed that when a family’s energy consumption is higher than 10% of its household income, the family’s energy burden is considered to be relatively high and in an energy-poverty state [56], but the single experience derived from developed countries may not be universally applicable to all situations [57]. China, being the globe’s biggest developing nation, has a rural population accounting for 36.11% (<Communiqué of the Seventh National Population Census>: https://www.gov.cn/guoqing/2021-05/13/content_5606149.htm accessed on 8 July 2025)—significantly higher than developed economies such as the US and UK. Thus, the 10% threshold may not be applicable in this context. Furthermore, due to the privacy of income issues, most respondents tend to conceal their income and exaggerate their expenditures when being surveyed. This may lead to consumption not being completely comparable to income, thereby affecting the estimation results. Therefore, this study measures the affordability of rural households’ basic energy consumption by the ratio of total annual fuel and electricity expenditures to the total household expenditures. A higher value of this ratio indicates a more inefficient energy consumption structure, implying a greater probability of falling into energy poverty. The advantage of doing so lies in that, first of all, the consumption indicators are comparable in the same dimension; secondly, compared to the possible large fluctuations in rural household income, the existence of consumption inertia enables the article to estimate the energy affordability of rural households more accurately and measure the current energy consumption structure of rural households.

3.2.4. Moderating Variable: Quality of Government Governance

Government governance quality is a composite variable measuring the level of regional governance performance by local authorities. The household head, as the primary decision-maker, significantly influences family decision-making. Accordingly, this study constructs a comprehensive government governance quality index using the entropy method, based on survey indicators including household heads’ attitudes toward local officials, evaluations of government performance, and the perceived severity of government corruption. A higher index value indicates superior regional governance performance. The specific indicator construction is detailed in

Table 1. Index system of government governance quality.

Indicator	Indicator Direction
Trust in cadres	+
Evaluation of this county and municipal government	+
The severity of corruption in the Chinese government	−

.

3.2.5. Control Variable

Following prior research [9], this study incorporates control variables including household head traits and family characteristics to improve model precision. Household head characteristics include age, gender, marital status, health status, and education level. Rural household characteristics comprise dependency ratio, labor force size, land assets, financial assets, and net worth.

Table 2. Descriptive statistics: Income inequality’s effect on rural households’ CET.

Name of Variables	Definition of Variables	Mean	SD
CET	Electricity, natural gas, gas, liquefied gas, solar energy and other clean energy = 1, firewood and coal = 0	0.562	0.496
Income inequality	Relative deprivation of household income at the village scale	0.357	0.285
Household head’s age	Actual age of respondents	51.778	12.967
Household head’s gender	Male = 1; female = 0	0.590	0.492
Household head’s marital status	In marriage, cohabitation = 1; unmarried, divorced, and widowed = 0	0.880	0.325
Household head’s health status	Unhealthy = 1; general = 2; relatively healthy = 3; very healthy = 4; extremely healthy = 5	2.872	1.268
Household head’s education level	Need not read/illiterate/semi-literate = 0; primary school = 1; junior high school = 2; high school = 3; junior college and above = 4	1.231	1.063
Household burden to population ratio	The proportion of people under 16 and over 65 in the household population	0.326	0.296
Number of people in the household labor force	Number of working-age persons in the household	2.697	1.437
Family land property	Take the log of household ownership of land assets plus one	8.177	4.036
Household financial assets	Take the log of total household financial assets plus one	6.385	4.751
Household net worth	Take the log of household net worth plus one	11.925	1.182

presents the descriptive statistics, which introduces the definition of the key variables in detail and shows the corresponding mean and standard deviation of each key variable.

Table 3. Correlation analysis results and statistics.

	CET	Income Inequality	Age	Gender	Spouse	Health	Surplus	Population Burden Rate	Labor Count	Land Asset	Finance Asset	Total Asset
CET	1
Income inequality	−0.124 ***	1
Gge	−0.064 ***	0.333 ***	1
Gender	−0.052 ***	−0.014 *	0.112 ***	1
Spouse	0.009	−0.209 ***	−0.072 ***	0.001	1
Health	0.062 ***	−0.158 ***	−0.218 ***	0.073 ***	0.027 ***	1
Surplus	0.192 ***	−0.260 ***	−0.253 ***	0.152 ***	0.081 ***	0.118 ***	1
Population burden rate	−0.033 ***	0.231 ***	0.364 ***	0.022 ***	−0.100 ***	−0.052 ***	−0.143 ***	1
Labor count	0.000	−0.388 ***	−0.378 ***	0.002	0.265 ***	0.090 ***	0.123 ***	−0.526 ***	1
Land asset	−0.121 ***	−0.132 ***	−0.092 ***	0.069 ***	0.161 ***	0.043 ***	−0.007	−0.147 ***	0.232 ***	1
Finance asset	0.151 ***	−0.226 ***	−0.123 ***	0.047 ***	0.055 ***	0.098 ***	0.199 ***	−0.044 ***	0.031 ***	0.019 **	1
Total asset	0.225 ***	−0.409 ***	−0.199 ***	0.017 **	0.188 ***	0.119 ***	0.252 ***	−0.157 ***	0.280 ***	0.170 ***	0.344 ***	1

Note: ***, **, and * represent significance at 1%, 5%, and 10%.

displays the covariance matrix, presenting Pearson correlation coefficients for pivotal variables under investigation alongside variance inflation factor diagnostics.

Table 3 demonstrates a substantial correlation among the key variables in this study, and that the absolute values of the correlation coefficients among all variables are less than 0.05, which basically rules out the existence of serious multicollinearity problems among the variables in the article.

3.3. Model Setting

To more accurately measure the impact of rural household income inequality on rural household CET, drawing on relevant studies, this article uses the Hausman test to determine the model selection type. The Hausman test yields a p-value of 0.000 (p < 0.05), providing strong evidence supporting the selection of the fixed effects model. The high-dimensional fixed effects model can effectively reduce the estimation bias caused by omitted variables and significantly improve the estimation efficiency. Therefore, the article further selects the high-dimensional fixed effects model. At the same time, to control for potential heteroscedasticity in the model, referring to existing studies [58,59], this study further adopts cluster-robust standard errors at the village regional level to alleviate the interference of heteroscedasticity on statistical inference and thereby enhance the accuracy of parameter estimation. The specific model setting is as follows:

$${CET}_{itj}={\alpha }_0+{{\alpha }_{1}II}_{itj}+{\alpha }_2{C}_{itj}+{\mu }_i+{\phi }_t+{\pi }_j+{\epsilon }_{itj}$$

(1)

where $\mathrm{i}$, t, and $\mathrm{j}$ are individual, time, and region, respectively; ${II}_{itj}$ represents rural household income inequality; ${CET}_{itj}$ represents rural household CET; $C_{itj}$ is a control variable; and individual fixed effects ${\mu }_i$ are included to control for time-invariant unobservables at the unit level. To control for time-invariant unobserved factors, time fixed effects ${\phi }_t$ are introduced; meanwhile, village-level fixed effects ${\pi }_j$ are incorporated to mitigate potential regional confounding. Here, ${\epsilon }_{itj}$ denotes the random disturbance term.

To verify whether energy consumption mediates the relationship between rural household income inequality and rural household CET, following existing research [60], this study constructs the following mediation effect model for mechanism testing:

$${TEC}_{itj}={\beta }_0+{{\beta }_{1}II}_{itj}+{\beta }_2{C}_{itj}+{\mu }_i+{\phi }_t+{\pi }_j+{\epsilon }_{itj}$$

(2)

$${ECS}_{\begin{array}{c}itj\\ \end{array}}={\gamma }_0+{{\gamma }_{1}II}_{itj}+{\gamma }_2{C}_{itj}+{\mu }_i+{\phi }_t+{\pi }_j+{\epsilon }_{itj}$$

(3)

To examine whether government governance quality moderates the linkage of rural household income inequality to rural household CET, the following model is constructed:

$${CET}_{itj}={\delta }_0+{{\delta }_{1}II}_{itj}+{\delta }_2{C}_{itj}+{\delta }_3{II}_{itj}\times {GOV}_{itj}+{\delta }_4{GOV}_{itj}+{\mu }_i+{\phi }_t+{\pi }_j+{\epsilon }_{itj}$$

(4)

where ${GOV}_{itj}$ denotes governance quality.

4. Analysis of Empirical Results

4.1. Benchmark Regression Analysis Results

Table 4. Benchmark regression results of rural household income inequality on CET.

Variables	CET
	(1)	(2)	(3)	(4)
Income inequality	−0.062 **	−0.150 ***	−0.157 ***	−0.057 **
	(0.027)	(0.018)	(0.017)	(0.025)
Age of the head of household	0.001	−0.001 ***	−0.002 ***	−0.000
	(0.001)	(0.000)	(0.000)	(0.001)
Gender of the head of household	−0.073 ***	−0.022 ***	−0.020 **	−0.015
	(0.014)	(0.008)	(0.008)	(0.013)
Marital status of the head of household	−0.020	−0.053 ***	−0.050 ***	−0.027
	(0.019)	(0.014)	(0.014)	(0.024)
Health status of the head of household	0.012 ***	0.005	0.004	0.005
	(0.004)	(0.003)	(0.003)	(0.005)
Educational attainment of the head of household	0.064 ***	0.031 ***	0.026 ***	0.006
	(0.007)	(0.004)	(0.004)	(0.009)
Household burden to population ratio	−0.050 **	−0.095 ***	−0.100 ***	−0.028
	(0.025)	(0.016)	(0.016)	(0.035)
Number of people in the household labor force	−0.022 ***	−0.021 ***	−0.019 ***	−0.007
	(0.006)	(0.004)	(0.004)	(0.007)
Family land property	−0.017 ***	−0.012 ***	−0.011 ***	−0.004 ***
	(0.002)	(0.001)	(0.001)	(0.002)
Household financial assets	0.006 ***	0.003 ***	0.001 *	0.001
	(0.002)	(0.001)	(0.001)	(0.001)
Household net worth	0.083 ***	0.037 ***	0.033 ***	0.015 **
	(0.007)	(0.004)	(0.004)	(0.006)
Constant	−0.309 ***	0.418 ***	0.504 ***	0.481 ***
	(0.101)	(0.058)	(0.058)	(0.092)
Household FE	NO	NO	NO	YES
Year FE	NO	NO	YES	YES
Village FE	NO	YES	YES	YES
Observations	14,354	14,354	14,354	14,354
R2	0.106	0.422	0.431	0.721

Note: ***, **, and * represent significance at 1%, 5%, and 10%, respectively; village-clustered robust standard errors appear in parentheses.

presents the effects of rural household income inequality on CET across columns. To accurately assess the impact of rural households’ income inequality on CET status, various control variables and fixed effects are gradually added to the model estimation. Specifically, column (1) reports the estimation results with control variables but without fixed effects. Columns (2), (3), and (4) successively add village-level regional fixed effects, time fixed effects, and individual fixed effects. It can be observed that as fixed effects are added one by one, rural households’ income inequality exerts a statistically significant negative impact on their CET at the 5% significance level. The estimation results in column (4) indicate that for every 1 percentage point increase in rural household income inequality, the probability of rural household CET decreases by 0.057 percentage points. This finding is generally consistent with existing research conclusions [22], and Hypothesis H1 is thus verified. The relative income hypothesis posits that farmers’ consumption decisions are not only determined by their household income levels but are also influenced by the income levels of others. The expansion of regional income disparities amplifies perceptions of relative deprivation within low-income populations, making them more cautious about investing in clean energy, which is detrimental to both rural household CET and regional sustainable development. Additionally, the budget constraint theory suggests that household consumption is restricted by household income. When household income is relatively low, the consumption budget is limited, which is not conducive to the shift towards higher-cost clean energy, resulting in potential social benefit losses. Moreover, compared to high-income groups, low-income groups have relatively lower cognitive literacy and are unable to accurately recognize the negative impacts of non-CET on family health and the surrounding environment. As a result, they are more likely to fall into an energy poverty trap and continue to use non-clean energy for a long time.

4.2. Robustness Test

4.2.1. Replace the Key Explanatory Variable

When focusing on the community level, the CFPS sample size may be insufficient, potentially introducing estimation bias. Therefore, this study shifts the regional focus to the district/county level and recalculates the rural household income inequality index. Meanwhile, to avoid the limitation of potential selection bias caused by using a single variable for measurement, the article further employs the Yitzhaki index to remeasure the rural household income inequality index at the village level [61]. Since the Yitzhaki index itself is non-restricted data, the article scales it down by 10,000 to unify the dimension. The test results of replacing the core explanatory variable are shown in columns (1) and (2) of

Table 5. Robustness checks for income inequality effects on rural household CET.

Variables	CET
	(1)	(2)	(3)	(4)	(5)
Income inequality	−0.059 **	−0.010 **	−0.071 **	−0.096 **	−0.052 **
	(0.027)	(0.005)	(0.031)	(0.041)	(0.023)
Controls	YES	YES	YES	YES	YES
Constant	0.493 ***	0.464 ***	0.419 ***	0.522 ***	0.354 ***
	(0.089)	(0.091)	(0.108)	(0.154)	(0.100)
Household FE	YES	YES	YES	YES	YES
Year FE	YES	YES	YES	YES	YES
Village/county FE	YES	YES	YES	YES	YES
Village × year FE	NO	NO	NO	NO	YES
Standard variables × year FE	NO	NO	NO	NO	YES
Observations	14,354	14,354	11,200	7066	10,655
R2	0.719	0.721	0.726	0.732	0.754

Note: *** and ** represent significance at 1% and 5%. The regional fixed effect in column (1) is the district and county fixed effect, and the figures in brackets are the robust standard errors clustered to district and county. The specifications for other equations maintain identical control variables and fixed effects as the baseline regression.

, both of which are significant at the 5% confidence level, indicating that the previous estimation results are robust.

4.2.2. Change the Range of Sample Selection

To avoid the error of the estimation results caused by the insufficient sample size at the village level, the sample size selection threshold is adjusted, and the villages with less than 10 and 15 households interviewed in the same year are excluded, respectively. Column (3) of Table 5 presents the estimation results using village samples with ≥10 surveyed households, while column (4) reports the results for villages with ≥15 surveyed households. The results demonstrate consistent statistical significance at the 5% level across all alternative sample specifications, confirming the robustness of our findings.

4.2.3. Increase the Interactive Fixed Effect

Since the survey at the community level in the CFPS database is not conducted continuously, only controlling the regional fixed effects at the village level may still cause the interference of the missing variables in the village and of the unobservable factors on the model, resulting in false and biased estimation results. Building upon the baseline regression, we incorporate village–year interaction fixed effects. Using the 2014 village database, we construct standardized variable–year interaction fixed effects by multiplying three regional characteristic variables (village elevation, village topography, and village-to-county distance) with year dummies. Table 5’s column (5) reports estimates incorporating interaction fixed effects, maintaining significance at p < 0.05 and reinforcing our core findings’ robustness.

4.2.4. Instrumental Variable Method

Potential endogeneity issues may exist between income inequality and rural households’ CET due to reverse causality. Studies show that households transitioning from polluting to clean energy experience significant developmental capacity improvements [62]. Moreover, clean energy adopters demonstrate a higher social participation likelihood [63], which enhances social capital and reduces rural income inequality [64]. Given the disruptive impact of emerging digital social capital on traditional rural networks [65], this study measures household-level social capital using transportation/communication expenditures, and constructs regional social capital as the mean value of other households’ social capital within the same year and area. The theoretical logic of selecting regional social capital as the instrumental variable of rural household income inequality lies in the following: when the research perspective is a region, the growth of regional social capital may exacerbate the income inequality within the region. The reason is that the high-income group wins the first-step advantage of obtaining information due to its high social status, which is more conducive to integrating resources and obtaining benefits, while the low-income group has difficulty breaking through the class barrier due to the network closure [66], thus further aggravating the intra-regional income inequality and meeting the correlation requirements of instrumental variables. However, the mean social capital of other rural households, excluding the household in question, does not directly affect the household’s CET status, satisfying the homogeneity requirement of the instrumental variable. Therefore, using regional social capital as an instrumental variable (IV) for rural household income inequality is empirically justified. The estimation results of instrumental variables are shown in the first two columns of

Table 6. IV estimation results of income inequality on rural households’ CET.

Variables	Income Inequality	CET	Household Income
	(1)	(2)	(3)
Regional Social Capital	0.116 ***		0.025
	(0.021)		(0.071)
Income Inequality		−0.472 *
		(0.255)
Controls	YES	YES	YES
Constant			7.741 ***
			(0.189)
Household FE	YES	YES	YES
Year FE	YES	YES	YES
Village FE	YES	YES	YES
Observations	14,354	14,354	14,354
Kleibergen–Paap rk LM Statistic	41.586 [0.000]
Weak Identification F-test	30.879 {16.38}
R2			0.732

Note: *** and * represent significance at 1% and 10%, number in [] is the p-value, and in {} denotes the weak identification test’s threshold value at the 10% significance level.

. The estimation results of the second stage show that the income inequality of rural households has a significantly negative impact on the CET of rural households, which is consistent with the conclusions of the benchmark regression.

Moreover, household income level is a crucial factor in determining whether a household uses clean energy. Referring to previous studies [61], the article conducts an exclusion test on the instrumental variable after incorporating household income as the outcome variable and the instrumental variable as a predictor in the regression model. Table 6 column (3) documents the estimation output. The instrumental variable has no significant impact on household income level, indicating that the instrumental variable does not indirectly affect the rural household CET transformation by influencing household income. This verifies the IV’s adherence to the exogeneity condition and corroborates the reliability of prior estimates.

4.2.5. Exclude Other Factors

In 2020, China made historic progress in solving the problem of absolute poverty and entered a new stage of development to promote common prosperity. There are still a small number of poor people who have not been lifted out of poverty before this time, which may have an impact on the accurate estimation of the model in the article. According to the national poverty classification standard (Source of data: https://www.stats.gov.cn/zs/tjws/tjzb/202301/t20230101_1903716.html and https://news.cctv.com/2021/04/06/ARTIKemhGKDmE36ukw0ypKPO210406.shtml accessed on 8 July 2025), this study accurately identifies rural households in poverty during the survey period and incorporates an absolute poverty dummy variable into the model to test result stability.

Table 7. Robustness tests on income inequality’s effects on farm households’ CET after controlling for confounding factors.

Variables	CET
	(1)	(2)	(3)	(4)
Income inequality	−0.058 **	−0.057 **	−0.061 **	−0.536 ***
	(0.026)	(0.025)	(0.024)	(0.164)
Absolute poverty	0.003
	(0.019)
Energy price		−0.034
		(0.046)
Controls	YES	YES	YES	YES
Constant	0.480 ***	0.688 **	0.488 ***
	(0.092)	(0.296)	(0.095)
Household FE	YES	YES	YES	YES
Year FE	YES	YES	YES	YES
Village FE	YES	YES	YES	YES
County × year FE	NO	NO	YES	NO
Observations	14,354	14,354	14,354	4692
R2/Nagelkerke R2	0.721	0.721	0.746	0.266

Note: *** and ** represent significance at 1% and 5%. The errors in the parentheses in column (4) are standard errors, and R² is reported as Nagelkerke R².

column (1) demonstrates the persistently significant negative association at the 5% level between income inequality and CET, supporting our main conclusions’ robustness. In addition, energy prices significantly influence rural households’ CET, and this paper adds energy prices as a control variable to the benchmark regression. Since the energy price index is monthly data and is currently only published until January 2020, this paper takes the annual mean value of the coal and electricity price index of each province as the proxy variable of the energy price. The mean value of the coal and electricity price index in each province is taken as the natural logarithm of the unity dimension. The estimates in Table 7 column (2) maintain consistent significance levels, verifying result robustness.

Although the baseline regression controls for household fixed effects, time fixed effects, and village-level regional fixed effects, certain competing national policies—such as the Winter Clean Heating Plan in northern regions and coal power phase-out initiatives—may simultaneously influence regional income inequality and rural households’ CET. This could potentially confound the regression results. The policies issued by the state play a guiding role, and the specific actions are carried out by local government units. To mitigate potential confounding effects from competing policies, this study incorporates county-by-time fixed effects into the baseline regression model. Demonstrated in column (3) of Table 7, after including these interaction fixed effects, the income inequality coefficient exhibits increased magnitude while maintaining 5% significance and a negative sign, reinforcing our results’ robustness.

Furthermore, since whether rural households carry out CET is set as a 0–1 variable in this article, this article further changes the model to an xtlogit model. The corresponding outcomes are displayed in Table 7’s column (4). Subsequent to model respecification, the negative impact of regional income inequality on rural household CET remains significant, verifying the correctness of the previous conclusion. At the same time, since the article uses unbalanced panel data, the xtlogit regression automatically eliminates samples where the rural household CET does not change within the observation period. Therefore, the sample size decreases and there is no intercept term.

4.3. Analysis of Heterogeneity

First, China’s significant regional economic disparities directly influence household energy consumption patterns [67]. This study employs the natural logarithm of provincial GDP per capita as a measure of territorial economic advancement. Regions exceeding the annual average economic threshold are categorized as being in the high-development tier, whereas those below constitute the less-developed tier. The empirical results are shown in columns (1) and (2) of

Table 8. Heterogeneity analysis of income inequality’s impact on rural households’ CET.

Variables	High Economic	Low Economic	High Coal Resources	Low Coal Resources	Eastern and Central Regions	Northwest and Northeast Regions
	(1)	(2)	(3)	(4)	(5)	(6)
Income Inequality	−0.012	−0.085 **	−0.027	−0.098 **	−0.024	−0.086 **
	(0.033)	(0.037)	(0.035)	(0.038)	(0.031)	(0.037)
Controls	YES	YES	YES	YES	YES	YES
Constant	0.582 ***	0.405 ***	0.527 ***	0.380 **	0.632 ***	0.365 ***
	(0.150)	(0.117)	(0.130)	(0.171)	(0.129)	(0.129)
Household FE	YES	YES	YES	YES	YES	YES
Year FE	YES	YES	YES	YES	YES	YES
Village FE	YES	YES	YES	YES	YES	YES
Observations	6357	7512	6840	5418	6777	7576
R2	0.734	0.706	0.732	0.714	0.726	0.683
p	0.047		0.079		0.048

Note: *** and ** represent significance at 1% and 5%. The inter-group coefficient difference is tested using Fisher’s permutation test with 1000 bootstrap replications.

. Income inequality’s detrimental effect on rural families’ CET manifests more prominently in economically disadvantaged areas. At the same time, the coefficient disparity across groups is significant at the 5% level (p < 0.05), indicating substantial heterogeneity between groups. The possible reason is that among the regions with a better overall economic situation, on the one hand, rural households may have more abundant resources and easier access to higher income. On the other hand, with better regional economic development and more tax revenue, the government can invest more resources in infrastructure construction, so that rural households can obtain and use clean energy more affordably. On the contrary, in the low-economic-level group, rural households may be more vulnerable to income inequality, leading to a negative impact on their CET.

Secondly, existing studies have demonstrated that regional resource endowments exert a nuanced influence on rural households’ energy selection [68]. It is the basic national condition of China that coal is rich, oil is poor, and gas is low. Since the open data of the CFPS questionnaire can only be traced back to the provincial level, provincial raw coal output is used to measure the regional coal resource endowment. Specifically, after eliminating the provinces with missing data, it is determined that the provinces with raw coal output in the top 50% of the year have a higher coal resource endowment, while the other way around, they have a lower coal resource endowment. Table 8 columns (3) and (4) report the empirical findings, revealing a more pronounced detrimental effect of income inequality on rural households’ CET in coal-resource-deficient areas. Tests reveal a statistically discernible cross-group coefficient discrepancy (p < 0.10), indicating substantial heterogeneity between groups. The possible reasons are that, on the one hand, regions with high coal resource endowment can rely on local coal mines to develop the economy, improve household income, and mitigate the detrimental consequences of financial constraints on CET; on the other hand, regions with abundant coal resources have developed stronger environmental awareness during early resource extraction, gaining a deeper understanding of non-clean energy hazards. Consequently, residents in these areas exhibit a greater willingness to adopt clean energy.

Finally, China is vast and has significant regional development disparities. The social and economic conditions and infrastructure construction levels vary greatly among different regions. Therefore, this article divides the central and eastern regions into one group and the western and northeastern regions into another based on the actual development differences in the regions. The reason for doing so is that the western and northeastern regions, as underdeveloped areas, have lagging economic growth, and their economic development is largely dependent on resource-based industries and central transfer payments. Moreover, there is a significant outflow of population and a low urbanization rate, which objectively differ significantly from the eastern and central regions. The empirical results are shown in columns (5) and (6) of Table 8, and the negative impact of income inequality on the CET of rural households is more significant in the western and northeastern regions. At the same time, the coefficient difference between the two groups is significant at the 5% confidence level, proving that there are significant differences between the two groups. The possible reason for this is that, on the one hand, the urbanization rate in the western and northeastern regions is relatively low, and their economic development is relatively lagging, which may lead to higher income gaps in these regions. On the other hand, rural households generally have lower incomes, and the infrastructure construction in these regions is relatively backward, making it more difficult to obtain clean energy. This makes it difficult for households to afford the initial investment and maintenance costs of clean energy equipment, thereby hindering the CET of rural households.

4.4. Test of Mechanism

It has been verified that income inequality can significantly inhibit the CET of rural households, but its influencing mode and channel need to be further clarified. On the one hand, income inequality enables lower-income groups to better cope with risk shocks and pursue higher social status, thus increasing savings and reducing household basic energy consumption, inhibiting household CET. On the other hand, income inequality constrains the upgrading of rural households’ consumption structure, widens consumption disparities, and consequently deteriorates energy consumption patterns. This elevates energy poverty risks and adversely affects CET in rural households. Therefore, this study examines how income inequality affects both the total volume and structural composition of rural households’ basic energy consumption to verify the underlying mechanisms. As evidenced in

Table 9. Mechanism test of the impact of income inequality on farm households’ CET.

Variables	TEC	ECS	CET
	(1)	(2)	(3)
Income Inequality	−0.149 ***	0.024 ***	−0.058 **
	(0.056)	(0.004)	(0.025)
Gov			−0.032
			(0.027)
Income Inequality × Gov			0.193 **
			(0.096)
Controls	YES	YES	YES
Constant	6.124 ***	0.072 ***	0.494 ***
	(0.244)	(0.015)	(0.093)
Household FE	YES	YES	YES
Year FE	YES	YES	YES
Village FE	YES	YES	YES
Observations	14,354	14,354	14,354
R2	0.637	0.547	0.721

Note: *** and ** represent significance at 1% and 5%.

column (1), income inequality reduces total household energy consumption, thereby suppressing rural households’ CET—confirming Hypothesis H2. Column (2)’s results indicate that income inequality significantly increases the share of basic energy expenditure in rural household budgets. Furthermore, it worsens consumption structures, raising energy poverty risks and adversely affecting CET, thus validating Hypothesis H3.

One possible explanation is that, against the backdrop of increasing income inequality, the risk aversion theory suggests that individuals tend to increase savings to avoid risks when facing uncertainty and risks. When affected by income inequality, lower-income groups face greater economic risks and instability. Therefore, low-income groups may be more inclined to cut unnecessary household expenditures and increase savings to prepare for potential risk shocks, thereby reducing household basic energy consumption. However, at the same time, low-income families have poorer resource integration capabilities and often face higher energy acquisition costs. Under the dual impact of overall consumption reduction and increased energy acquisition costs, these families are more likely to fall into energy poverty, hindering the upgrading of the household basic energy consumption structure and suppressing the use of clean energy. Maslow’s hierarchy of needs theory posits that when household income is limited, basic survival needs will be prioritized. Therefore, when households reduce consumption, the reduction in development-oriented and enjoyment-oriented consumption is greater than that in survival-oriented consumption, leading to an increase in the proportion of household basic energy consumption in the total household expenditure.

Clean energy transition serves as a crucial lever for implementing sustainable development principles and driving green, high-quality economic growth. Eliminating income inequality and establishing a more equitable, inclusive developmental paradigm constitute the cornerstone for fostering a collective human future, both of which rely on effective governance by the government. This study further examines how the governance quality mode rates income inequality’s impact on rural households’ CET. Column (3) of Table 9 reveals that governance quality significantly attenuates the adverse effect of income inequality on rural households’ CET. The estimator for the income inequality–governance interaction stands at 0.193 and is significant (p < 0.05), revealing that enhanced governance efficacy dampens the inhibition of household CET by income inequality. Hypothesis H4 is proved. The possible reasons are as follows: On the one hand, regions with good government governance quality usually provide better resource allocation, policy support, and service guarantees. Through measures such as subsidies for clean energy equipment, infrastructure optimization, and financial support, they directly reduce the upfront costs of CET for rural households, thereby mitigating income inequality’s adverse consequences for rural households’ CET. On the other hand, good government governance quality can build trust and cooperative relationships with the public, enhancing the trust of rural households in the government. This makes rural households more willing to respond to government policy calls, including accepting clean energy technologies, thereby changing the negative impact of income inequality on rural households’ CET.

5. Discussion and Policy Recommendations

The inhibitory effect of income inequality on CET has been widely documented, yet existing studies predominantly focus on the national level [21] or urban–rural disparities [19], overlooking how intra-rural income inequality affects household CET—particularly in regions where non-clean energy use remains most prevalent. Moreover, few studies integrate basic energy consumption, regional income inequality, and household CET within a unified framework, nor do they extend the analysis to examine how total energy consumption and consumption structure further influence this relationship. Additionally, while government governance exerts a subtle yet significant influence on rural household decision-making, related research has only explored isolated aspects such as corruption [22] or institutional trust [52] to assess how governance quality affects CET. Few studies construct comprehensive regional-level indicators to measure government governance quality, leaving its precise capacity as an interacting factor between income inequality and CET unresolved.

This study utilizes four-wave unbalanced panel data from the CFPS database and deploys a multi-dimensional fixed effects model to investigate how income inequality influences rural households’ CET and its determinants. The findings reveal the following: First, income inequality significantly suppresses rural households’ CET. Second, income inequality’s inhibitory effect on farmers’ CET is more pronounced in areas with lower economic development and low coal resource endowment, and in the western and northeastern regions of China. Third, income inequality curbs rural households’ CET by reducing their basic energy consumption and hindering the upgrading of their basic energy consumption structure. Lastly, the quality of public administration significantly and negatively moderates the association between income inequality and rural households’ CET. As governance quality improves, income inequality’s adverse influence regarding clean energy consumption progressively attenuates over time. This study empirically examines the impact of income inequality on rural households’ clean energy adoption from a micro-level perspective and investigates the underlying mechanisms involving energy consumption and government governance, thereby providing valuable insights into the existing literature.

This investigation’s outcomes generate substantive implications for policymaking. First, income inequality is a key constraint on clean energy consumption. While vigorously developing the economy and increasing the income of rural households, the Chinese government should design effective redistribution policies and provide financial support to low-income rural households through reasonable tax systems and subsidy mechanisms to promote equitable distribution and enhance their clean energy consumption capacity. Secondly, we should play the leading role in technological innovation. The government should strengthen policy guidance to incentivize enterprises to accelerate R&D and the adoption of new clean energy technologies, further improve the clean energy market, and reduce clean energy prices through policy leadership, technological innovation, and market mechanisms—thereby lowering household adoption costs and facilitating rural households’ CET. Thirdly, we must leverage the facilitative role of both social finance markets and service support systems through launching financial products and “green credit” programs for rural households to ease liquidity constraints; and establishing and improving the social service security system to ensure that rural households can receive timely support when they encounter uncertain risks, reducing precautionary savings and promoting the clean energy consumption of rural households. At the same time, this will improve the awareness of rural households regarding the use of clean energy. Grassroots governments should actively conduct publicity campaigns to highlight the environmental and health hazards of non-clean energy use, raise farmers’ awareness of clean energy benefits, and foster household adoption of clean energy sources. Finally, the quality of government governance is a key element in promoting the CET of rural households. The government should strengthen self-improvement, improve administrative efficiency, ensure effective policy implementation, and avoid local protectionism to enhance the authority of government leadership, improve people’s sense of identity, and enhance the confidence of rural households in CET. This study also provides an important reference for other countries, especially developing countries, to promote clean energy transition.

6. Research Deficiencies and Prospects

There are still some limitations in this paper. First of all, although this paper follows academic conventions by using a village-level relative deprivation index calculated from rural household income to represent income inequality among rural households, this single-dimensional metric based solely on income may overlook the impact of factors such as wealth, land access, or credit availability on rural income inequality. Future research should comprehensively incorporate relevant factors that could significantly influence rural household income inequality levels, aiming to more accurately measure the extent of such inequality.

Secondly, since this paper quantifies rural households’ income inequality by benchmarking against the economic status of fellow villagers, it has a strong correlation with the same village. At the same time, owing to the unavailability of village-tier data in public datasets, it is impossible to control the relevant confounding factors at the village level, so the village-level regional fixed effects are used to alleviate their possible impact on the estimation results. The explained variables of this paper are categorical variables of 0 and 1, and theoretically, the xtlogit model should be used for estimation. However, because the xtlogit model has many limitations in adding fixed effects and clustering robust standard errors, the high-dimensional fixed effects model is adopted for regression tests following the existing research. Even though the xtlogit model is deployed in subsequent robustness checks to reconfirm the validity of the paper’s core findings, there is still a certain gap in the estimation results of different models. Therefore, in the subsequent research, the current model should be further optimized and improved according to the research objectives to safeguard the scholarly validity and robustness of this paper’s investigation.

Furthermore, the current study only examines how income inequality affects rural households’ CET from two perspectives: energy consumption and government governance quality. In future research, it would be valuable to adopt a more systematic and comprehensive approach by incorporating additional factors at both the household and regional levels that may influence rural households’ CET. This would help further enrich the relevant literature.