Perceived Status and Sustainable Actions: How Subjective Socioeconomic Status Drives Green Energy Consumption in Chinese Households

Abstract

Household energy consumption plays a critical role in the context of global climate change. Utilizing data from the 2018 China Social Survey (CGSS), this study empirically examined the impact of subjective socioeconomic status (SES) on household green energy consumption behaviors using probit and ordered probit models. The mechanism of influence was further analyzed through the mediated effect approach. The results found include the following: (1) Although the proportion of households participating in green energy policies is similar to those not participating, the proportion of households deeply participating in multiple policies is very low; (2) subjective SES significantly influences both the rate and depth of household participation in green energy policies; (3) internet usage and understanding of green energy policies serve as mediating mechanisms for the promotive effect of subjective SES; and (4) subjective SES showed significant heterogeneity in its effects on different gender and education level groups. These findings contribute to the understanding of the drivers of household green energy use decisions and provide an important reference for governmental policymaking to enhance participation rates and degrees in green energy participation. Implications of these findings highlight the potential for targeted policies that address internet accessibility and educational outreach, which could significantly enhance the effectiveness of green energy initiatives across diverse socioeconomic groups.

1. Introduction

As global climate change and environmental degradation intensify, the consumption and promotion of green energy have become focal concerns of the international community. The Sixth Assessment Report (AR6) of the Intergovernmental Panel on Climate Change (IPCC) has warned society of the inevitability of future temperatures breaching the 1.5 °C warming stipulated in the Paris Agreement, irrespective of whether the socio-economic development model is radical or negative. Carbon dioxide emissions resulting from human activities, especially energy consumption, are the main driver of global warming [1]. How to balance energy consumption and environmental protection has become a major issue for the government and the public [2]. Encouragingly, the cost of green energy use is declining substantially, making it a key option for combating climate change [3]. In China, family participation in green energy policies, as the basic unit of society, directly influences the optimization of the national energy structure and the improvement of environmental quality [4]. A decade ago, the energy consumption of Chinese households was mainly dominated by biomass (such as biogas, animal manure, firewood, and tangerine rings) and coal primarily dominated Chinese household energy consumption, contributing to a large amount of greenhouse gas emissions [5,6]. To improve the structure of household energy consumption, the Chinese government has implemented a series of initiatives aimed at promoting green energy consumption for sustainable development [7]. For example, the implementation of the “three reform linkage” policy (energy-saving and carbon-reducing transformation of coal power, flexibility transformation, and heating transformation) [8,9,10] encourages the transformation of energy saving and carbon reduction, and promotes the clean and low-carbon development of coal power. Although relevant policies are being implemented, there is still a scarcity of empirical research on the participation levels and driving mechanisms of green energy policies among Chinese households. Therefore, an in-depth exploration of the core factors and mechanisms influencing Chinese households’ green energy consumption behaviors could facilitate the government’s targeting to increase the participation rate of households in green energy policies, which is of great significance to the development of the green economy in China and around the world.

At present, relevant research has mainly explored the impact on households’ energy consumption behavior based on exogenous factors such as policy support and the level of economic development. For example, focusing on eastern and western provinces with unequal distribution of resources, Zhou and Shi [11] revealed the long-term relationship between socioeconomic transformation and inequality in residential energy consumption through panel data, and the results show that urbanization has alleviated the energy consumption gap between urban and rural areas. Although the impact of industrial transformation is significant, its specific effect is determined by the regional industrial structure. Péter et al. [12], starting from the household level, explored the impact of net income, education level, age, housing, and attitude-related factors on household green investment. However, many studies have also found that internal driving forces such as residents’ psychology and cognition are the main factors affecting consumption decisions. For example, Namazkhan et al. [13] used a decision tree model to explore the unique relationships between building characteristics, socio-demographic variables, psychological factors, and household behaviors, and emphasized the crucial role of psychological factors in driving sustainable energy consumption behaviors. Furthermore, Mi et al. [14] showed in their study that psychological incentives are more critical than energy prices and technological factors in stimulating residents’ energy-saving behaviors and promoting household green energy consumption.

People’s psychological and social outcomes are significantly influenced by their socioeconomic status [15]. Socioeconomic status encompasses a complex and multidimensional structure that includes both independent objective characteristics (e.g., income, education, and occupation) and subjective evaluations of one’s social ladder position [16]. However, more and more studies have found that compared with objective SES, subjective SES has a closer and more stable relationship with behavioral patterns, mental functions, health-related factors, and level of consciousness [16,17,18,19]. Subjective SES has been applied to various aspects of social cognition and behavior science, such as food preferences [20], life well-being [21,22], health impairment behaviors [23] and political participation [24], etc. Discussions on the correlation between SES and green behavior have revealed that subjective SES not only mirrors individual self-cognition and social status but also profoundly impacts their environmental decision-making and behavior [15,25,26]. Nonetheless, there are few empirical studies that have explored the subjective socioeconomic status influence on green energy consumption behavior.

Based on this, this study utilizes data from the 2018 China Social Survey (CGSS) as a sample to analyze whether residents’ subjective SES affects their participation in household green energy policies from a multi-dimensional perspective and explores the underlying mechanisms. The study’s innovations encompass the following: From the research perspective, unlike prior studies that primarily focus on technological or economic aspects of green energy [27,28,29,30], this study adopts a sociological approach, emphasizing the role of SES in shaping household energy behaviors. This perspective builds on and extends the findings of Grandin et al. [25], who explored objective socioeconomic factors, by focusing on the psychological and perceptual dimensions that impact environmental decisions. Methodologically, we employ probit and ordered probit models to analyze survey data, providing a reliable statistical framework to examine the effects of subjective perceptions. In terms of research content, this study not only explores whether subjective SES impacts household green energy consumption behaviors but also further analyzes the underlying mechanisms through a mediation effect model, clarifying the pathways of influence. These factors are crucial as they reflect the informational and cognitive pathways through which SES influences energy behaviors. While some studies have touched upon the roles of internet accessibility and policy awareness in sustainable energy practices [31,32,33,34], our approach integrates these elements into a comprehensive model that systematically examines their mediating roles. This methodological innovation provides new insights into how enhancing internet access and improving policy communication can effectively promote green energy consumption, especially among different socioeconomic groups. From the scale of the study, our research utilizes data from the 2018 China Social Survey (CGSS), conducted on a national scale across China, covering various socioeconomic backgrounds and environmental policies. This provides a broader and more detailed perspective than studies confined to specific localities or urban centers. Therefore, this study not only fills a significant gap in the current literature but also provides actionable insights for policymakers aiming to develop more effective and inclusive green energy policies.

2. Literature Review and Hypothesis

According to the Theory of Planned Behavior (TPB) [35], an individual’s behavioral intention is primarily determined by three predictors, namely attitude toward the behavior, subjective norms, and perceived behavioral control. The consensus among existing studies is that subjective socioeconomic status significantly influences individuals’ environmental attitudes and behaviors [25,26,36,37]. Schwartz’s Norm Activation Model (NAM) [38] illustrates how individuals motivate and guide prosocial behavior through internalized moral obligations. This theory has revealed that individuals with high socioeconomic status tend to exhibit higher personal norms and are more likely to adopt prosocial behaviors such as energy conservation [15,26]. Given their greater resources and opportunities to acquire environmental knowledge, these individuals typically display positive environmental behaviors and tend to reduce their carbon footprint through responsible energy use [39]. This tendency extends beyond the individual to the household level, where households with higher subjective SES are more inclined to invest in less polluting products compared to those with lower subjective SES, thus promoting the use of household green energy [40]. While discussing whether households use green energy, we also need to explore the degree to which they do so. As posited by the fuel stack theory of energy consumption [41], households may adopt diversified strategies in energy selection, reflecting the complexity and diversity of energy use. This indicates that households are incorporating green energy into their existing energy mix, rather than merely substituting it for conventional energy sources [42]. Households with higher subjective SES typically have more resources and capital, allowing them to more easily access and adopt multiple energy technologies (e.g., solar, wind, electric, etc.) and utilize a diverse array of energy sources to cope with price fluctuations or supply disruptions. This reduces their dependence on single energy sources and enhances their energy security [43]. Therefore, households with higher subjective SES not only have the economic ability to implement green energy policies but also have a stronger sense of environmental responsibility, which makes them more likely to show higher participation and a diversified energy mix in the process of energy consumption. Based on the above analysis, this study proposes the following hypotheses:

H1a. 

Subjective SES significantly increases the participation rate in household green energy policies.

H1b. 

Subjective SES significantly enhances the degree of household participation in green energy policies.

In recent years, the government has launched several incentive policies to promote the use of green energy. However, in the effective delivery and implementation of policies, we discovered a phenomenon that cannot be ignored—families with varying socioeconomic statuses exhibit significant differences in acquiring, understanding, and accepting these policies, potentially limiting the maximization of their effects, which may limit the maximization of the policy effects [44].

With the advancement and widespread adoption of internet technology, the connectivity of social networks has become increasingly reliant on the internet [45]. Ghose and Han [46] proposed an empirical framework for user behavior in the mobile internet domain, positing that social networks exert a significant positive influence on user behavior. Some studies have found that individuals with higher subjective SES tend to exhibit greater internet usage rates [47,48]. This higher usage is likely due to their enhanced access to technical resources, higher educational levels, and more pronounced information needs [49,50]. Although households with lower SES may have limited physical resources, the internet provides them with opportunities to transcend social class barriers, enabling these households to interact with wider society and access green energy-related information [51]. During the energy transition, the use of the internet not only facilitates information flow and knowledge sharing but also strengthens cooperation and coordination among stakeholders [52]. Furthermore, as a platform for information access and social interaction, the internet positively influences household green energy decision-making by expanding information channels, enhancing employment flexibility and autonomy, and elevating the frequency of social interactions [32,53]. Consequently, the use of the internet not only enhances policy accessibility but also fosters green energy consumption among households. Based on the analysis above, this study proposes the following hypothesis:

H2. 

Subjective SES improves household green energy policy participation mainly by increasing internet usage frequency.

Beyond internet usage, the level of policy knowledge of households influences their decisions regarding green energy consumption. Using a nested policy design framework and policy environment theory, Li et al. [54] found that cognitive disparities between policymakers and stakeholders affect policy implementation outcomes. Studies by Kardooni et al. [55] and Boie [56] also emphasized the importance of social awareness and acceptance of green energy technologies, pointing out that the level of social awareness of these technologies is a key factor in driving green energy consumption. Additionally, some scholars have tested the relationship between cognition and neural structure patterns and revealed significant cognitive disparities across different socioeconomic groups [57]. Specifically, individuals with high SES typically possess higher education levels and access to more diverse information channels. Consequently, they better comprehend government green energy policies and can respond more swiftly [58]. In contrast, individuals with low SES often face greater challenges in understanding and applying policies, attributed to poor perceptions of allotment, knowledge misunderstandings, and insufficient environmental awareness [59,60]. Based on the above analysis, this study proposes the following hypothesis:

H3. 

Subjective SES improves household green energy policy participation mainly by increasing the level of understanding of green energy policies.

All hypotheses are illustrated in Figure 1.

3. Study Design

3.1. Study Data

The data for this study were derived from the China General Social Survey. CGSS is a nationwide social survey continuously implemented by the China Survey and Data Center of Renmin University of China since 2003, covering several key areas such as socio-demographic attributes, housing, health, and migration [61,62,63]. Specifically, the 2018 China General Social Survey (CGSS2018) targeted research on issues related to residents’ energy consumption. To ensure the representativeness and accuracy of the findings, CGSS 2018 employed a multi-stage, stratified, probability-proportional sampling (PPS) method. This sampling design allowed the random selection of urban and rural households nationwide to reflect the overall situation of Chinese society. Based on the research objectives and hypothesis, this study selected appropriate measurement variables from the CGSS 2018 dataset and removed samples with excessively missing values. Finally, 3985 valid samples were obtained.

3.2. Variable Definition

3.2.1. Dependent Variables

As shown in

Table 1. Definition of variables and descriptive statistics.

Variables	Variable Definition	Mean	SD
Dependent variable
SES	Lower class = 1; Lower-middle class = 2; Middle class = 3; Upper-middle class = 4; Upper class = 5	2.289	0.858
Independent variable
GEP	Non-participation = 0; Participation = 1	0.503	0.500
GED	Never participate = 1; Low participation = 2; Moderate participation = 3; High participation = 4; Very high participation = 5	1.671	0.818
Control variables
Gender	Female = 0; Male = 1	0.464	0.499
Age	Continuous variable, CGSS survey year minus respondent’s birth year	51.850	16.697
Marital	Not married = 0 (single, divorced, widowed, cohabiting); Married = 1 (first/remarried spouse, separated but not divorced)	0.762	0.426
Health	Very unhealthy = 1; Somewhat unhealthy = 2; Average = 3; Somewhat healthy = 4; Very healthy = 5	3.538	1.082
Mean_age	Continuous variable	46.081	15.840
Person	The number of people living together in the family all year round	2.798	1.411

, the dependent variable was household participation in green energy policies. To measure this dependent variable, two indicators were employed in this study, namely participation or non-participation and degree of participation. One question in the questionnaire was, “Is your family involved in any of the following energy policies?” The 13 options below correspond to the 13 green energy policies, including coal to gas, coal to electricity, high-quality coal replacement, efficient stove subsidies, ladder electricity prices, peak and valley electricity prices, residential photovoltaic power plant subsidies, poor family electricity subsidies, electric vehicle purchase subsidies, household electric vehicle charging pile peak and valley electricity prices, natural gas ladder prices, rural biogas pond subsidies, and photovoltaic poverty alleviation. If the answer was “participation”, the value would be “1”. If the answer was “non-participation”, the value will be “0 ”. Participation was a binary variable, coded as either 1 or 0, to indicate whether a household participates in at least one green energy policy. Given that most observations clustered in the lower range, “Never participate” was categorized separately, and the other categories were divided according to the distribution of the remaining observations: Never participate (0 projects), low participation (1–2 projects), moderate participation (3 projects), high participation (4–6 projects), and very high participation (7–13 projects).

3.2.2. Independent Variables

The independent variable was subjective socioeconomic status. Subjective SES refers to an individual’s perceived status within a hierarchy, thus reflecting an evaluation of one’s social status relative to others [64]. Therefore, this study measured subjective SES with the survey item, “Overall, in the current society, where does your socioeconomic status belong?” Using a 5-point Likert scale, we generated ordered categorical variables ranging from 1 to 5. The higher the score, the higher the subjective class identification of residents, and vice versa.

Subjective SES showed significant spatial differentiation in China (Figure 2a), with subjective SES in most regions positively correlated with green energy consumption. Households in Inner Mongolia, Shanxi, and Zhejiang reported the highest subjective SES (2.47–2.73), followed by the eastern coastal provinces. In contrast, Sichuan, Chongqing, and Guizhou exhibited the lowest subjective SES (1.71–1.99). Regarding household green energy consumption behavior, GEP and GED were high in regions such as Beijing, Shanxi, Sichuan, Jiangsu, Zhejiang, and Guangdong (Figure 2b), indicating that the green energy policies in these provinces are highly effective. However, regions such as Yunnan, Inner Mongolia, Heilongjiang, Jilin, and Henan demonstrated considerable potential for green energy development. Chongqing and Tianjin had high GEP but low GED, indicating that while the prevalence of green energy consumption was high in these areas, the modes of consumption were relatively limited. Conversely, Liaoning and Guangxi had high GED and low GEP, suggesting that green energy resources in these regions were concentrated among a few individuals. Interestingly, Sichuan and Inner Mongolia deviated from the general pattern. Sichuan had low subjective SES but high GEP and GED. Inner Mongolia had high subjective SES, but low GEP and GED (Figure 2).

3.2.3. Control Variables

Referring to the existing literature [32,65,66], this study selected control variables from three aspects, including respondent characteristics, household characteristics, and regional characteristics. Specific information included gender, age, marital status, health status, family size, and the average age of family members. In addition, to account for heterogeneity across different provinces, this study incorporated provincial fixed effects into the model.

3.2.4. Mediating Variables

The theoretical analysis proposed two potential mediating mechanisms as follows: “subjective SES → internet use → household participation in green energy policy” and “subjective SES → understanding of green energy policy → household participation in green energy policy”. Internet usage was measured by the question from the survey, “How often have you used the internet in the past year?” on a five-point Likert scale, with 1 meaning “rarely used” and 5 meaning “frequent use”. The level of understanding of green energy policies was measured by the question, “How much do you know about the green energy policies listed below?”, also using a five-point Likert scale, with 1 indicating “not at all” and 5 indicating “completely understand”. The total understanding score of residents towards the 13 green energy policies constituted the overall understanding index.

3.3. Research Methodology and Models

3.3.1. Basic Model Framework

This study examined household participation in green energy policies from two aspects. One aspect was the binary variable of whether to participate or not. The other was the degree of participation, which is an ordered response variable. Therefore, for the former, the binary Logit model or the Probit model can generally be used for testing. There is no essential difference between the two models. However, the Logit model has limitations in addressing unobserved factors, alternative forms, and repeated choices across periods. The Probit model can largely overcome these problems and is more suitable for analyzing subjects’ selection behaviors [67]. Therefore, this study employed the binary Probit model to analyze the impact of subjective socioeconomic status perception on household participation in green energy policies. For the latter, the ordered Probit model (Oprobit) was used for regression (Equation (2)). The model expressions were as follows.

$${GEP}_i={\beta }_0+{\beta }_1{SES}_i+{\beta }_2{Controls}_i+{\epsilon }_i$$

(1)

$${GED}_i={\beta }_0^{\prime }+{\beta }_1^{\prime }{SES}_i+{\beta }_2^{\prime }{Controls}_i+{\epsilon }_i$$

(2)

In Equation (1), ${GEP}_i$ (Green Energy Participation) indicates whether householdi’s participates in a green energy policy, while in Equation (2) ${GED}_i$ (Green Energy Degree) represents the degree of householdi’s participation in the policy. In Equations (1) and (2), ${SES}_i$ (Subjective Socioeconomic Status) denotes the subjective socioeconomic status of householdi, serving as the core independent variable of this study. ${Controls}_i$ encompasses a set of control variables selected for the analysis, ${\epsilon }_i$ represents the random disturbance term. ${\beta }_0$ and ${\beta }_0^{\prime }$ refer to the constant terms in the regression, ${\beta }_1$, ${\beta }_1^{\prime }$, ${\beta }_2$ and ${\beta }_2^{\prime }$ are the regression coefficients. It is worth noting that the regression coefficient $\beta$ in Equation (2) does not directly quantify the specific impact of subjective socioeconomic status on green energy consumption behavior but only reflects the relationship between these factors.

3.3.2. Mediating Effect Model

To further explore the potential pathway through which subjective SES of households influences their participation in green energy policies via mediating variables, this study constructed a mediation effect model. The study employed the stepwise regression approach proposed by Baron and Kenny [68] to examine the effects of internet use and the understanding of green energy policies as mediating variables. The model expression was as follows.

$${Internet}_i=a_0+a_1{SES}_i+a_2{Controls}_i+{\epsilon }_i$$

(3)

$${GEP}_i=b_0+b_1{SES}_i+b_2{Internet}_i+b_3{Controls}_i+{\epsilon }_i$$

(4)

$${GED}_i=c_0+c_1{SES}_i+c_2{Internet}_i+c_3{Controls}_i+{\epsilon }_i$$

(5)

$${Policy}_i=a_0^{\prime }+a_1^{\prime }{SES}_i+a_2^{\prime }{Controls}_i+{\epsilon }_i$$

(6)

$${GEP}_i=b_0^{\prime }+{b_1^{\prime }SES}_i+b_2^{\prime }{Policy}_i+b_3^{\prime }{Controls}_i+{\epsilon }_i$$

(7)

$${GED}_i=c_0^{\prime }+{c_1^{\prime }SES}_i+c_2^{\prime }{Policy}_i+c_3^{\prime }{Controls}_i+{\epsilon }_i$$

(8)

Equations (1), (3), and (4) correspond to the test of the mediating effect of the path “subjective SES → internet use → GEP”. The coefficient ${\beta }_1$ in Equation (1) represents the total effect of subjective SES on household participation in green energy policies, while the coefficient $b_1$ in Equation (4) indicates the direct effect of subjective SES on this participation, $a_1$ and $b_2$ represents the mediating effect of internet use. Similarly, Equations (1), (6) and (7) test the mediating effect of the corresponding path “subjective SES → understanding of green energy policy → GEP”. In Equation (6), coefficient $b_1^{\prime }$ indicates the direct effect, and $a_1^{\prime }{b}_2^{\prime }$ represents the mediating effect of policy understanding. The mediation test path for GED is identical.

4. Results

4.1. Descriptive Statistical Results

The study made a descriptive statistical analysis based on the cross-sectional data from 2018. As shown in Figure 3, the proportion of households participating in green energy policies is nearly equal to that of those not participating, at 50.28% and 49.72%, respectively, indicating a balanced sample for this variable. Figure 3 further reveals the distribution of the number of green energy policies households participate in. It can be seen from the figure that most households (38.63%) participate in only one to two green energy policies. As the number of policies participated in increases, the proportion of households sharply decreases. This indicates that while green energy policies have been implemented, only 50.28% of residents participate, and only a minority of households (11.65%) participate widely and deeply (in three or more policies).

4.2. Baseline Regression Results

4.2.1. Household Participation in Green Energy Policies

Table 2. The impact of subjective SES on GEP.

	(1)	(2)	(3)
Variables	GEP	GEP	Margin
SES	0.101 ***	0.0728 ***	0.0252 ***
	(4.33)	(2.85)	(2.86)
gender		0.0800 *	0.0277 *
		(1.89)	(1.89)
age		−0.00251	−0.000869
		(−1.30)	(−1.30)
mean_age		0.00349 *	0.00121 *
		(1.71)	(1.72)
marital		−0.0681	−0.0236
		(−1.32)	(−1.33)
health		0.0225	0.00780
		(1.05)	(1.05)
person		0.0600 *** (3.62)	0.0208 *** (3.64)
Province	No	Yes	Yes
_cons	−0.223 ***	−0.955 ***
	(−3.94)	(−5.51)
N	3984	3984	3984
r2_p	0.00341	0.122
ll	−2752.0	−2423.8

Note: * p < 0.1, *** p < 0.01; t statistics in parentheses.

presents the baseline regression results of subjective SES on GEP. The analysis revealed that, at the 1% significance level, subjective SES is significantly positively correlated with GEP. After the control variables are introduced, the coefficient of subjective SES decreases slightly but still remains significant. Referring to Li et al. [69], the estimated coefficients in a probability model indicate only the direction of relationships between variables, not the precise probabilities of changes in the variables. The third column of Table 2 displays the marginal effects of statistically significant variables. The marginal effect shows that holding other factors constant, each additional unit of subjective SES increases the likelihood of household participation in green energy policies by an average of 2.52%. This finding suggests that households with higher subjective SES are more likely to participate in green energy policies. This may be attributed to households with higher subjective SES possessing greater financial resources and social capital, which facilitate easier access to green energy technologies and the ability to afford the associated initial investments.

4.2.2. Degree of Household Participation in Green Energy Policies

This study employed an ordered probit model to explore how subjective SES influences the degree of household participation in green energy policies.

Table 3. Estimation results of the Ordered Probit model.

	(1)	(2)
Variables	GED	GED
SES	0.106 ***	0.0685 ***
	(4.99)	(3.00)
gender		0.0877 **
		(2.34)
age		−0.00136
		(−0.80)
mean_age		0.00323 *
		(1.78)
marital		−0.0770 *
		(−1.70)
health		0.0185
		(0.95)
person		0.0565***
		(3.84)
Province	No	Yes
cut1	0.235 ***	0.985 ***
cut2	1.440 ***	2.371 ***
cut3	1.924 ***	2.927 ***
cut4	2.862 ***	3.933 ***
N	3984	3984
r2_p	0.00296	0.0968
ll	−4205.2	−3809.4

Note: * p < 0.1, ** p < 0.05, *** p < 0.01; t statistics in parentheses.

shows that the p-value of subjective SES is less than 0.01, showing a significant effect at the 1% significance level. The coefficients provided by the ordered probit model do not directly reflect the marginal effects of the explanatory variables [69]. Therefore, the study further calculated the marginal effects of these variables (

Table 4. Marginal effects of explanatory variables.

	Marginal Effects
Variables	Never Participate	Low Participation	Moderate Participation	High Participation	Very High Participation
SES	−0.0238 ***	0.0122 ***	0.00580 ***	0.00490 ***	0.000898 ***
	(−3.01)	(3.00)	(2.98)	(2.95)	(2.60)
Gender	−0.0305 **	0.0157 **	0.00742 **	0.00627 **	0.00115 **
	(−2.35)	(2.34)	(2.33)	(2.32)	(2.14)
Age	0.000473	−0.000243	−0.000115	−0.0000972	−0.0000178
	(0.80)	(−0.80)	(−0.79)	(−0.79)	(−0.79)
Mean_age	−0.00113 *	0.000578 *	0.000274 *	0.000231 *	0.0000424 *
	(−1.78)	(1.78)	(1.78)	(1.77)	(1.69)
Marital	0.0268 *	−0.0138 *	−0.00652 *	−0.00551 *	−0.00101
	(1.71)	(−1.70)	(−1.70)	(−1.69)	(−1.62)
Health	−0.00643	0.00330	0.00156	0.00132	0.000242
	(−0.95)	(0.95)	(0.95)	(0.95)	(0.94)
Person	−0.0197 ***	0.0101 ***	0.00478 ***	0.00404 ***	0.000741 ***
	(−3.86)	(3.84)	(3.78)	(3.73)	(3.09)
Province	Yes	Yes	Yes	Yes	Yes

Note: * p < 0.1, ** p < 0.05, *** p < 0.01; t statistics in parentheses.

) to more accurately describe their impact on the degree of household participation. From Table 4, the negative marginal effect of subjective SES suggests that as subjective SES increases, the probability of households not participating in green energy policies decreases by 2.38%, while the probabilities of low, average, high, and very high participation significantly increase by 1.22%, 0.058%, 0.049%, and 0.089%, respectively, which reflects that households with higher subjective SES are likely more capable and willing to invest in green energy projects. The marginal effects of gender indicate that, compared to males, female household members may participate less in green energy policies, which could be related to traditional gender roles and family responsibility distributions. In many cultural contexts, women often have a lesser role in household energy decision-making, with resource allocation and decision-making frequently dominated by men. This gender disparity may restrict women’s participation in household energy consumption decision-making [70,71]. Moreover, larger households show a significantly reduced probability of “never participating” in green energy policies, and a significant increase in the probabilities of “low participation”, “moderate participation”, “high participation”, and “very high participation”. Possibly because larger families have more resources and greater motivation to invest in green energy, this aligns with the findings of [72].

4.3. Mechanism Analysis

4.3.1. Mediating Role of Internet Usage

Table 5. Test results of mediating effect of internet usage.

	(1)	(2)	(3)	(4)	(5)	(6)
Variables	GEP	Internet	GEP	GED	Internet	GED
SES	0.0728 ***	0.116 ***	0.0609 **	0.0685 ***	0.116 ***	0.0566 **
	(2.85)	(4.93)	(2.37)	(3.00)	(4.93)	(2.47)
Internet			0.0981 ***			0.0998 ***
			(5.61)			(6.51)
Controls	Yes	Yes	Yes	Yes	Yes	Yes
Province	Yes	Yes	Yes	Yes	Yes	Yes
N	3984	3984	3984	3984	3984	3984
r2_p	0.122	0.220	0.128	0.0968	0.220	0.102
ll	−2423.8	−4452.0	−2407.9	−3809.4	−4452.0	−3788.2

Note: ** p < 0.05, *** p < 0.01; t statistics in parentheses.

explores the mediating effect of internet use between subjective SES and household participation in green energy policies through stepwise regression. Model (2) shows that subjective SES significantly positively affects internet use at the 1% significance level (β = 0.116, p < 0.01), indicating that individuals with higher subjective SES are more likely to use the internet. In Model (3), introducing internet use slightly weakened but did not eliminate the significant influence of subjective SES on GEP, with internet use itself exerting a significant positive effect (β = 0.0609, p < 0.05). In Model (6), although the impact of subjective SES on GED has declined, the positive impact of internet use on GED is still significant (β = 0.0566, p < 0.05). This is consistent with existing research results [32], suggesting that internet use may be a key way to promote household participation in green energy policies. H2 is verified.

4.3.2. Mediating Role of Policy Understanding

Table 6. Test results of mediating effect of policy understanding.

	(1)	(2)	(3)	(4)	(5)	(6)
Variables	GEP	Policy	GEP	GED	Policy	GED
SES	0.0728 ***	0.0950 ***	0.0369	0.0685 ***	0.0950 ***	0.0323
	(2.85)	(4.80)	(1.39)	(3.00)	(4.80)	(1.38)
Policy			0.0495 ***			0.0525 ***
			(17.92)			(22.21)
Controls	Yes	Yes	Yes	Yes	Yes	Yes
Province	Yes	Yes	Yes	Yes	Yes	Yes
N	3984	3984	3984	3984	3984	3984
r2_p	0.122	0.0315	0.184	0.0968	0.0315	0.157
ll	−2423.8	−12,258.3	−2253.3	−3809.4	−12,258.3	−3556.8

Note: *** p < 0.01; t statistics in parentheses.

presents the results of the mediation analysis concerning the understanding of green energy policies. The results indicate that subjective SES is significantly positively associated with the understanding of green energy policies (β = 0.0950, p < 0.01), and similarly, the degree of understanding of green energy policies has a significant positive correlation with GEP and GED (β = 0.0495, p < 0.01; β = 0.0525, p < 0.01), indicating the presence of a mediating effect. Notably, when the understanding of green energy policy is introduced as a mediator, the direct correlation between subjective SES and household participation in green energy policy is no longer significant. This indicates complete mediation, affirming that understanding fully mediates the relationship between subjective SES and household participation in green energy policies. H3 is verified.

4.4. Heterogeneity Analysis

On the basis of existing research [73] revealing that gender differences have a significant impact on green behavior, this study further explored the role of subjective SES in GEP and GED across different gender groups. As shown in

Table 7. Heterogeneity in gender.

	GEP		GED
Variables	Man	Woman	Man	Woman
SES	0.108 ***	0.0498	0.0670 **	0.0767 **
	(2.89)	(1.39)	(2.03)	(2.40)
Controls	Yes	Yes	Yes	Yes
Province	Yes	Yes	Yes	Yes
N	1841	2137	1847	2137
r2_p	0.126	0.133	0.0992	0.104
ll	−1114.5	−1284.4	−1794.5	−1989.8

Note: ** p < 0.05, *** p < 0.01; t statistics in parentheses.

, for GEP, subjective SES significantly positively affects men’s participation but has negligible impact on women, indicating that men are more inclined to participate in green energy policies when their subjective SES is higher, while for women, an increase in economic status does not significantly increase their likelihood of participation. This disparity may stem from traditional socio-cultural factors and gender role differences, leading to varied considerations in energy consumption decisions between men and women.

Existing research has shown that education level has a significant impact on individuals’ subjective SES [74] and green behavior [75]. Based on this, this study categorized individuals with a bachelor’s degree or higher as having high education and those with education levels below a bachelor’s degree as having low education. As shown in

Table 8. Heterogeneity in educational attainment.

	GEP		GED
Variables	High	Low	High	Low
SES	0.0518	0.0689 **	0.0469	0.0661 ***
	(0.77)	(2.44)	(0.85)	(2.59)
Controls	Yes	Yes	Yes	Yes
Province	Yes	Yes	Yes	Yes
N	668	3298	686	3298
r2_p	0.108	0.116	0.0994	0.0953
ll	−382.0	−2015.7	−720.6	−3057.1

Note: ** p < 0.05, *** p < 0.01; t statistics in parentheses.

, the regression results reveal that the subjective SES of the low-education group significantly affects their GEP and GED; conversely, the subjective SES of the high-education group does not have a significant impact on their GEP and GED. This finding suggests that increasing subjective SES may be more effective in promoting participation in green energy policies among groups with lower education levels. A plausible explanation is that education level impacts individuals’ ability to access and comprehend information about green energy policies, which, in turn, affects their decision-making.

4.5. Robustness Tests

This study further examined the robustness of the baseline regression and mediation effects. For the baseline regression, with reference to Kuzior et al. [76], the study additionally utilized logit models and ordered logit tests to assess the impact of subjective SES on household participation in green energy policies.

Table 9. Robustness test results of the baseline regression.

	(1)	(2)	(3)	(4)
Variables	GEP	GEP	GED	GED
SES	0.161 ***	0.121 ***	0.176 ***	0.117 ***
	(4.33)	(2.85)	(4.95)	(2.98)
Controls	No	Yes	No	Yes
Province	Yes	Yes	Yes	Yes
N	3984	3984	3984	3984
r2_p	0.00341	0.122	0.00292	0.0988
ll	−2752.0	−2423.7	−4205.3	−3801.1

Note: *** p < 0.01; t statistics in parentheses.

shows a significant positive correlation between subjective SES and household participation in green energy policies. Upon inclusion of control variables, the correlation coefficient decreased, yet subjective SES remained significantly positively correlated with household participation in green energy policies at the 1% level. The regression results are consistent with those of the probability regression models, confirming the robustness of the findings.

For the robustness test of the mediation effects, this study employed the Bootstrap method to examine the significance of the mediation effects in the model, effectively addressing the issue of “masking effects” and revealing the true impacts between the response variables [77]. According to the results of the 95% confidence interval, if the interval does not contain 0, the path is considered significant and should be retained; otherwise, it will not be considered. As shown in

Table 10. Robustness test results for the mediating effect.

Variables	Path	Z-Value	p-Value	95% Conf. Interval
Internet	SES → Internet → GEP	3.70	0.000	[0.002, 0.007]
	SES → Internet → GED	3.93	0.000	[0.006, 0.170]
Policy	SES → Policy → GEP	4.77	0.000	[0.007, 0.018]
	SES → Policy → GED	4.78	0.000	[0.023, 0.056]

Note: 95% bias-corrected confidence intervals based on 1000 Bootstrap sampling estimates are in square brackets.

, the confidence interval for the mediating effect of internet use on GEP ranges from [0.002, 0.007], and for GED from [0.006, 0.170]. The confidence intervals for the mediating effects of understanding green energy policy on GEP and GED are [0.007, 0.018] and [0.023, 0.056], respectively. Since none of the confidence intervals include zero, it confirms that the mediating effects of both pathways are robust.

5. Discussion

The results of this study illuminated the multifaceted nature of household participation in green energy policies. Descriptive statistical analysis revealed a nearly balanced binary distribution of participation, providing a well-rounded sample basis for subsequent regression analysis. Additionally, we observed a declining trend in the level of household participation in green energy policies as the number of policies increased, suggesting constraints related to resources and willingness.

In the baseline regression results, subjective socioeconomic status (SES) demonstrated a significant positive correlation with household participation in Green Energy Policies (GEP) and Green Energy Devices (GED). This finding aligns with the results from Ouyang et al. [78], who reported that perceived economic well-being plays a crucial role in environmental behaviors in rural China. This relationship underscores the importance of perceived financial stability when adopting new technologies, resonating with findings from international research. For instance, Nguyen and Drakou [79] explored factors influencing Vietnamese farmers’ intentions to adopt Sustainable Agricultural Practices (SAP) for coffee cultivation, noting significant impacts of social norms and financial capital. Akram et al. [80] identified accessible credit as a facilitator for adopting sustainable organic agricultural machinery in Pakistan, reducing production risks for farmers.

Notably, Kanzola et al. [81] analyzed data from a field study conducted in Greece between 2019 and 2020, finding that most variables influencing environmental perception revolved around concepts unrelated to economic outcomes, except for those indicating governmental responsibility to reduce income inequality. These divergent results may stem from different socioeconomic environments and governmental policies, leading to significant variations in measures to promote pro-environmental behaviors. For example, the EU often relies on a combination of top-down regulatory measures and bottom-up community engagement—a stark contrast to China’s more centralized policy implementation strategy. In the EU, direct financial incentives and infrastructural support play a more pivotal role than awareness campaigns [82].

6. Conclusions and Policy Recommendations

Utilizing micro-survey data from CGSS 2018, this study analyzed the impact of subjective SES on household participation in green energy policies from both theoretical and empirical perspectives. On this basis, the study tested the pathways “subjective SES → internet use → household participation in green energy policy” and “subjective SES → understanding of green energy policy → household participation in green energy policy”. Additionally, the study also analyzed the heterogeneity of the sample based on gender and educational level. The results show the following: (1) The proportion of households participating in green energy policies was nearly equal to that of those not participating, at 50.28% and 49.72%, respectively. Although over 50% of households have begun to engage with green energy policies, only 11.65% were deeply involved in multiple policies. (2) Subjective SES significantly enhances both GEP and GED. (3) Subjective SES is enhanced by more frequent internet use and a better understanding of green energy policies, which significantly boosts household participation in these policies. (4) With the same unit increase in subjective SES, males and individuals with lower educational levels are more likely to participate in green energy policies, whereas the impact on females and individuals with higher educational levels is less significant or negligible. (5) Household size is significantly positively correlated with green energy consumption.

The findings of this study are of great relevance, and green energy has the potential to play a key role in reducing carbon emissions and fossil fuel consumption across all economic sectors [43]. The Sustainable Development Goals 7 (SDGs 7) states that, by 2030, everyone should have access to affordable, reliable, and modern energy services. Undoubtedly, this is particularly challenging in the context of a post-pandemic global economic slowdown. Achieving this goal will require substantial additional investments in green energy deployment and promotion. Furthermore, there are significant differences in the effectiveness of green energy consumption policy implementation across regions. Regions such as Beijing, Shanxi, Sichuan, Jiangsu, Zhejiang, and Guangdong have shown effective implementation of green energy policies, while the situation in Yunnan, Inner Mongolia, Heilongjiang, Jilin, and Henan is less optimistic. This variation may be related to the differences in the policies implemented and their effects during the green energy transition across various provinces. For instance, regions like Beijing, Shanxi, and Sichuan have enhanced the directive and regulatory roles of energy strategy and planning, set up mechanisms for monitoring and evaluating the transition to green and low-carbon energy, and have advanced the transformation of energy consumption and supply through the “dual control” system of energy consumption and non-fossil fuel targets [83,84]. Meanwhile, regions like Yunnan, Inner Mongolia, and Heilongjiang may experience a relatively slower pace of green energy transition due to factors such as inadequate resource and environmental endowments, improper industrial layouts, and insufficient economic development capabilities. Consequently, green investment strategies should be tailored to align with local energy consumption patterns, economic development levels, industrial structures, and natural environmental characteristics to forge a low-carbon development path in each region. For example, the ongoing photovoltaic poverty alleviation program in China is a direct response to the last-mile problem of providing clean energy to the rural poor in remote areas of Western China [85,86,87].

Additionally, the results have indicated that policies promoting household green energy use should precisely target psychological factors and socio-demographic variables. Specifically, incentive measures could particularly consider subjective SES, gender, educational level, and household size, as these appear to be the main factors related to household green energy consumption. In particular, incentives could attempt to change these predictors, or target groups with relevant characteristics. For instance, the government could foster a fair and just social environment through investments in education, vocational training, and income redistribution, among other strategies. Based on this, the policy of replacing counties with districts could serve as a catalyst to help farmers transition smoothly into urban citizens, thus overall enhancing society’s subjective SES and sense of environmental responsibility. Furthermore, given that individuals with high subjective SES will help shape their own and others’ choices [88], it might be beneficial to encourage individuals with high subjective SES to become leaders in green energy consumption, utilizing their social networks and influence to act as proactive promoters and exemplars of green energy use and transition. Additionally, society and policymakers must recognize the crucial role women play in driving the energy transition, and enhancing women’s decision-making power regarding green energy may facilitate the shift to a greener energy portfolio.

Our research has also shown that both internet usage and the understanding of green energy policies are critical to enhancing household green energy consumption. The government should leverage internet channels to disseminate specific and targeted information on clean energy to the public, thereby enhancing understanding of green energy policies and awareness of the hazards associated with conventional energy sources. Additionally, according to data from the China Internet Network Information Center (CNNIC), as of December 2023, the number of internet users in China reached 1.092 billion, with an internet penetration rate of 77.5%. Among them, the number of urban internet users was 766 million, accounting for 70.2%, while the number of rural internet users was only 326 million, accounting for 29.8% of the total internet users. Consequently, the government should prioritize enhancing internet access in rural areas to more effectively promote green energy policies and boost residents’ awareness of clean energy thus facilitating the transition to household green energy. Furthermore, to diversify household green energy consumption, policies should be tailored to the specific needs and resources of different regions, providing an appropriate green energy consumption mix for both urban and rural residents. For example, in rural areas, the government can use government subsidies and technical support to encourage farmers to adopt biogas technology in breeding and agricultural waste treatment, while also installing solar panels for household and agricultural use, increasing energy self-sufficiency and reducing environmental pollution. In urban areas, communities should be encouraged and supported to develop integrated energy systems that comprehensively utilize multiple renewable energy sources, such as solar energy, wind energy, and geothermal energy. Through community-scale integrated energy systems, residents can not only directly use these green energy sources but also reduce energy waste through energy exchange within the community.

This study has revealed the importance of subjective SES in promoting the adoption of green energy in households, which is consistent with previous studies indicating that individuals’ psychological motivations and behavioral intentions have a significant impact on environmental behavior [14,89,90]. However, previous studies have primarily focused on the general impacts of environmental behavior, with less exploration into behaviors specifically related to green energy use and its underlying mechanisms. This paper attempts to fill this research gap by exploring, for the first time, the relationship between subjective SES and household participation in green energy. However, this study also has several limitations, which require us to interpret the results with caution. Firstly, the use of a cross-sectional design in this study limits our ability to make causal inferences, necessitating further validation through experimental studies or longitudinal panel data. Secondly, although the CGSS 2018 data provide a wealth of information, they may not fully represent all regions and populations in China, and future research should consider including a wider range of geographic locations and socioeconomic contexts. Thirdly, the CGSS does not continuously collect data on the energy sector, which may affect the timeliness and generalizability of the study results. Finally, this study aimed to explore the correlation between subjective SES and household green energy consumption behaviors rather than establish a definitive causal relationship. Nonetheless, we recognize that endogeneity is a significant statistical issue that could affect the validity of causal inference; future studies could seek appropriate instrumental variables to address potential endogeneity bias.