Does Off-Farm Employment Promote the Low-Carbon Energy Intensity in China’s Rural Households?

Abstract

A study linking the two economic and social phenomena of rural labor force migration and energy transition can help analyze the underlying causes of rural “Energy Poverty”. However, how off-farm employment affects household low-carbon energy consumption and its potential mechanisms requires further research. Using 1351 sampled rural households from the “Rural Energy, Population Transfer and Well-being” survey in 2018 and 2021 to explore response mechanisms through which off-farm employment can influence low-carbon energy intensity. Utilizing the multivariate regression, Sobel test, and moderating effect test, the results demonstrate that off-farm employment, including short-term and long-term off-farm employment, significantly increases the intensity of low-carbon energy use among rural households. Specifically, long-term off-farm employment tends to have a greater positive contribution to the low-carbon energy intensity than short-term off-farm employment. Furthermore, off-farm employment can affect household low-carbon energy intensity through the total income, and effect of the surrounding people in the off-farm employment process also increases their consumption intensity. The research reveals that the rural energy revolution under the constraints of “Carbon Neutral” and “Carbon Peak” should relate to the off-farm development of rural households to achieve “Precise Energy Poverty Alleviation”.

1. Introduction

Energy is the basic element sustaining human survival and development [1], and the possible existence of energy poverty in household energy consumption may hinder the development and progress of human society [2]. One of the striking findings of the World Energy Outlook 2022 is that 70 million people who have recently gained access to electricity will probably be unable to afford it due to the combined effects of the COVID-19 pandemic and the energy crisis, and 100 million people may no longer be able to use clean fuels for cooking, falling back into the energy poverty trap [3].

Along with the progressive transformation of rural energy consumption from traditional biomass energy to commodity energy, its low-carbon sustainable development has now become an important part of the government’s energy conservation and emission reduction strategy [4]. With the deep integration of energy consumption with rural revitalization strategy and economic development, China’s per capita energy consumption in rural areas increased form 88 kg coal ce in 2000 to 444 kg ce in 2019, exceeding urban areas’ 2019 consumption rate (440 kg ce) [5]. However, constrained by the availability and affordability of renewable low-carbon energy, livelihood activities such as cooking and heating mainly rely on high-carbon fuels such as biomass and charcoal energy, resulting in an average annual growth rate of carbon emissions of rural residents’ living energy consumption being 3.34 times that of urban residents [6]. The crude use of solid fuels has low thermal efficiency and produces large amounts of carbon monoxide, solid particulates, and other harmful substances that cause air pollution, significantly increasing the probability of residents suffering from chronic obstructive pulmonary disease, acute respiratory disease, and other non-infectious diseases [7], and its high-carbon emissions can also seriously pollute the environment and lead to global warming. Insisting on energy cost reduction and energy quality improvement and promoting the construction of clean and low-carbon, multi-energy integration of modern rural energy systems is the top priority of energy transformation in rural areas [8].

Rural energy issues concern the basic energy supply and quality assurance for the production and living of nearly 50% of China’s population [9]. The existing literature has evaluated the current situation and problems of energy consumption [10], the low-carbon energy use influencing factors [11,12], and the negative impact of high-carbon emissions on health and the environment [13,14]. Although household energy consumption systems have always been a hot topic of academic research, the issue of energy consumption upgrading caused by the off-farm employment of rural laborers has rarely been explored. According to the data of the seventh census, China’s urbanization rate reached 64% in 2020, and the total number of migrant workers in the country reached 285.6 million [15]. The rapid process of urban–rural integration has broken the monolithic social pattern of rural residents as agricultural producers and has begun to reconfigure household production and lifestyles. As the best choice for rural households to respond to changes in human–land relations, the income, demonstration, and human capital effects of livelihood, decentralization, and diversification are bound to have an important impact on household energy consumption decisions [16]. Studies have shown that the level of urbanization development can negatively affect rural energy poverty [17]. The increase in income levels of rural residents due to urbanization is the main driver of the transformation in rural energy consumption. Sun et al. [18] found that off-farm income can contribute to a green and sustainable transition of rural household energy consumption from biomass-based energy to commodity-based energy. Li et al. [19] found that off-farm employment work significantly and positively affects household clean energy adoption through two channels: increasing household income of rural residents and promoting their health literacy. Ma et al. found that off-farm work significantly increased the consumption of commodity energy such as electricity and natural gas, while significantly reducing the consumption of coal and biomass energy [16]. In addition, the energy transition effect from off-farm income is more pronounced in provinces with relatively better economic development [20].

The existing research provides us with relevant references, but there are also shortcomings. For one, in previous studies, the low-carbon energy use of household was mainly based on a single energy source, but nowadays the consumption pattern of “Energy stacking” determines that households use multiple energy sources in their livelihoods, so it is more valuable to choose low-carbon energy intensity as the research object to promote the sustainable development of energy. Furthermore, households’ energy consumption behavior is a rational choice based on their household livelihood endowment. Their behavioral choices are influenced by their own livelihood strategies and actual payment ability, as well as by social networks and neighborhood pressure; therefore, how these factors affect their energy consumption behavior needs to be further investigated.

Based on this, the possible marginal contributions of this study are as follows. First, focusing on income effect and peer effect, this study systematically examines the mechanisms of the impact of off-farm employment on rural households’ low-carbon energy intensity, allowing for a more diverse and specific perspective on the study of sustainable rural energy transitions. Second, the substitution variable methods and PSM methods were used for a robustness test to ensure the reliability of study conclusions. Finally, the conclusions of this paper provide a reference for the government and relevant authorities to reduce household reliance on polluting energy and promote a low-carbon rural energy transition. More importantly, they can provide theoretical and practical guidance to other developing countries in alleviating rural energy poverty.

2. Research Hypotheses

With the accelerated pace of rural revitalization strategy and new urbanization, the livelihood of rural families has changed from pure farming to part-time farming and off-farm, reflecting the differentiated livelihood level and diversified livelihood strategies. Off-farm employment helps households modernize their energy use through an optimal allocation of labor resources [21], improvement in household livelihood capital [22], and enhancement of low-carbon awareness [23]. Compared with farming, the original ideology of a household labor force is changed or reshaped after off-farm transfer. With the increase in off-farm production time and livelihood capital, urban low-carbon civilization will implicitly impact the original high-carbon lifestyle of households, guiding their energy consumption concept to keep pace with the times, so that they are willing to respond to the national call for cleaner energy consumption. In addition, the returning labor force can bring back the urban consumption habits and low-carbon concepts to the rural area, breaking the inherent high-carbon energy consumption habits of family members left behind, and even affecting fellow residents [24].

As a livelihood strategy, off-farm employment further influences the household low-carbon energy intensity through income effect. In the absence of an adequate rural social security system, moderate participation of household laborers in off-farm employment across regions aims to obtain higher labor remuneration than agricultural production [25], thereby effectively dispersing the risk of a single source of income and increasing diversity and structural stability. At the same time, off-farm employment can lead to changes in resource utilization preferences of rural households, such as stimulating agricultural capital investment or land transfer, which can indirectly increase agricultural income [26]. According to the theory of “Energy stack”, the gradual improvement in livelihood capital and socio-economic status breaks the non-economic rational energy consumption pattern of farmers that “neither saves money nor increases efficiency” [27]. This is specifically reflected in the considerable economic benefits that enable farmers to have the willingness and ability to pursue high-quality life after satisfying their basic survival needs, and pay more attention to convenience, efficiency, and health safety when making energy consumption decisions, thus enhancing low-carbon energy intensity.

Off-farm employment enhances population mobility, and individual behavioral willingness is closely related to the human environment [28]; the strengthening of social interaction makes individual decision-making behavior more vulnerable to the influence of others, so the peer effect may have an impact on the relationship between off-farm employment and low-carbon energy intensity. The peer effect reflects that when individuals make decisions, they do not make the best decisions based on their own factors, while considering the influence of peer behavior [29]. Its influence process includes two mechanisms [30]: one is the direct spillover of social interaction, that is, social learning. By observing the behavior pf peers in the off-farm employment, they become more aware of the synergy “Health-Ecology-Social-Economy” benefits of using low-carbon energy, thus breaking the original habit of high-carbon energy use and significantly enhancing their willingness to participate in activities such as ecological poverty alleviation. The second is the indirect spillover of social interaction, that is, social comparison. If household residents compare their behaviors with those around them, to avoid the pressure of social norms caused by behavioral differences, they must adopt behavior similar to those around them, showing a certain conformity mentality and following others’ pro-environmental behaviors.

Through the preceding discussion, off-farm employment will have an important impact on the upgrading of rural household energy consumption. Therefore, based on the research objectives of this paper, the following hypotheses are proposed:

H1. 

Off-farm employment has a significant positive influence on the low-carbon energy intensity of rural households.

H2. 

Off-farm employment has a significant positive influence on the low-carbon energy intensity of rural households through the mediating role of income effect from employment.

H3. 

The peer effect plays a moderating role between off-farm employment and the low-carbon energy intensity of rural households.

In summary, based on the studies of Ma et al. [16] and Zhou et al. [31], the mechanism of the impact of off-farm employment on household low-carbon energy intensity is shown in Figure 1.

3. Data Source, Variable Description, and Model Setting

3.1. Data Source

The research data of this paper adopt the 2018 and 2021 special surveys on “Rural Energy, Population Transfer and Welfare” carried out by Shaanxi Energy Industry Green and Low-Carbon Development Soft Science Research Base, Xi’an University of Science and Technology Energy Economics and Management Research Center in the Shaanxi and Henan provinces. Firstly, considering the level of economic and social development, the distribution of energy resources endowment, and the availability of data, the survey selected Henan in the middle and Shaanxi in the northwest as the sample survey areas, after careful investigation. Secondly, stratified and random sampling methods were used to select three sample counties in each province based on geographical location and differences in levels of economic development. In each county, five sample towns and five sample villages were randomly selected. Finally, five households were selected from each village according to the proportion of off-farm employment and pure farming households. A total of 750 questionnaires were distributed in each of the two periods. After data cleaning and entry, invalid questionnaires with missing information were eliminated, and 1351 questionnaires were considered to be valid to use with an effective rate of 90.07%.

3.2. Variable Description

3.2.1. Dependent Variable

In this paper, low-carbon energy is defined as the energy that produces less pollution after use, specifically electricity, natural gas, liquefied gas, solar energy, biogas, and other energy [32]. Farmers may use multiple energy sources in various livelihood activities. Referring to Li et al. [33], this paper selects whether mainly low-carbon energy is used in cooking, heating, and bathing to measure their use intensity. If they do not mainly use low-carbon energy, the value is 1. If one of them mainly uses low-carbon energy, the value is 2. If two of them mainly use low-carbon energy, the value is 3, and if low-carbon energy is mainly used for the three items, the assigned value is 4. Finally, the orderly classification variables of low-carbon energy intensity of rural households from low to high are 1~4.

3.2.2. Independent Variable

To better measure how different off-farm employment scenarios affect households’ low-carbon energy intensity, this paper intends to introduce “off-farm employment”, “short-term off-farm employment”, and “long-term off-farm employment” as independent variables. “Farm population whether seek outside employment” is used to measure off-farm employment, being a binary dummy variable [34]. For the measurement of long-term and short-term off-farm employment, the two indicators of off-farm employment time and off-farm employment income are fully considered. Groups with time and income accounting for less than 50% are grouped into short-term off-farm employment, while those with time and income accounting for more than 50% are grouped into long-term off-farm employment [35].

3.2.3. Mediator Variable and Moderator Variable

To empirically analyze the root causes of off-farm employment affecting households’ low-carbon energy intensity, the total household income from labor was selected as the explanatory variable in exploring the income effect [11,16]. Since residents tend to have little expertise and a strong sense of community, they usually refer to others’ opinions and behaviors when implementing low-carbon behaviors in the process of off-farm employment [36]. Therefore, “The degree of the impact of the use or purchase of certain low-carbon energy by relatives and friends around them” was selected as the proxy variable of the peer effect to explore the moderating role of the peer effect in the impact of off-farm employment on the low-carbon energy intensity of rural households.

3.2.4. Control Variables

The low-carbon energy intensity of farm households is influenced by a mixture of socio-economic-cultural factors. Therefore, with reference to the existing literature and questionnaire data, other variables that may affect the low-carbon energy intensity of households were selected and controlled to reduce the impact of omitted variables, endogeneity, and other issues on the estimation results. These variables include household demographic characteristics, livelihood development characteristics, cognitive psychological characteristics, and regional characteristics. Variables related to household psychological characteristics include the head of the household’s age [37], household size, household rearing burden [38], and average education level [12,36]. Household size may have an impact on energy consumption decisions, with evidence from Ghana suggesting that larger households were more inclined to use cheaper biomass energy for economic cost considerations [39]. The higher the average level of education, the more it helps to stimulate their awareness of environmental and physical conservation, thus reducing the consumption of polluting fuels [40]. For livelihood development characteristics, the number of electrical appliances [41] and area of cultivated land were selected. The cultivated area has an impact on household energy consumption by affecting the availability of biomass [42]. Policy perception [43], low-carbon awareness [11], and energy-saving habits [44] were selected to measure the cognitive psychological characteristics. Damette et al. [45] found that that environmental preference is an important factor affecting household energy consumption, as demonstrated by the fact that households are 6.74% more likely to choose electricity and only 0.33% more likely to choose fuelwood when they are explicitly aware of the environmental impacts of energy consumption. Moreover, this paper will also account for regional differences. See

Table 1. Variable Meaning, Assignment, and Statistical Characteristics.

Variables	Definition	Mean (Std.Dev)	Min	Max
Low-carbon energy intensity	Whether mainly low-carbon energy is used in cooking, heating, and bathing (1 = None, 2 = One, 3 = Two, 4 = Three)	2.46 (1.12)	1	4
Explanatory variables
Off-farm employment	Whether farm population seeks outside employment (1 = Off-farm employment, 0 = Farming)	0.76 (0.43)	0	1
Short-term off-farm employment	Off-farm employment exists but the main nature of the work is still agricultural work (1 = Yes, 0 = No)	0.36 (0.48)	0	1
Long-term off-farm employment	Off-farm employment exists and the main nature of the work is off-farm work (1 = Yes, 0 = No)	0.40 (0.50)	0	1
Head of household’s age	2018/2021-Year of birth	55.15 (12.31)	24	89
Household size	Number of permanent household members in the past year: person	4.58 (1.71)	1	10
Household rearing burden	Sum of children under 12 years old and elderly over 60 years old/family size	0.32 (0.26)	0	1
Average education level	Weighted average education level of the total family population (1 = Illiterate, 2 = Primary School, 3 = Junior High School, 4 = High School, 5 = Secondary School or Technical School, 6 = Junior College or above)	3.09 (0.94)	0.380	6
Number of electrical appliances	Number of existing appliances in the family	13 (5.55)	1	41
Cultivated area	Current cultivated land area in the family: mu	5.55 (6.77)	0	75
Policy perception	Perceived government subsidies and new energy promotion (1 = Strong, 2 = Relatively strong, 3 = Average, 4 = Relatively low, 5 = Very low)	3.21 (1.01)	1	5
Low-carbon awareness	If you know that a certain energy is high in energy consumption or carbon emissions, will you stop buying or using it (1 = Yes, 0 = No)	0.49 (0.50)	0	1
Energy-saving habits	Do you and your family members have the awareness or habit of saving energy in daily life (1 = Yes, 0 = No)	0.52 (0.50)	0	1
Regional factors	Respondent’s residence (1 = Shaanxi, 0 = Henan)	0.50 (0.50)	0	1
Income effect	Ln (Actual annual total household income): 1000 yuan	10.48 (0.98)	6.68	13.82
Peer effect	The impact of the use or purchase of some low-carbon energy by relatives and friends on themselves (1 = Great impact, 2 = Impact, 3 = Average, 4 = Little impact, 5 = No impact)	3.42 (0.99)	1	5

Note: The data come from the 2018 and 2021 special surveys of “Rural Energy, Population Transfer and Welfare” conducted by the Shaanxi Energy Industry Green and Low-Carbon Development Soft Science Research Base and the Energy Economics and Management Research Center of Xi’an University of Science and Technology.

for the meaning, assignment, and descriptive statistical results of relevant variables.

First, in terms of low-carbon energy intensity, the average low-carbon energy intensity of rural households is 2.46, indicating that they mainly use low-carbon energy in at least one of the three livelihood activities, cooking, heating and bathing, but there is still a big gap from zero carbon. Secondly, in terms of off-farm employment, more than three quarters of farmers are in off-farm employment, and nearly half of rural households participate in long-term off-farm employment, indicating that it is more common for rural households to move beyond only farming activity and work in off-farm labor. Again, the average age of the household head is over 55 years old. The average family education level is generally at the Junior Middle School, and the level of human capital is relatively low. Finally, the average value of households’ policy perception is 3.21, which is between average and less strong, indicating that rural energy policies and infrastructure are not perfect, and households’ basic energy demand is limited by the availability of low-carbon energy. Half of the households have low-carbon awareness and energy-saving habits, indicating that they have a strong sense of environmental responsibility and are aware of the current energy crisis and resource problems. However, due to the combined impact of objective reality and subjective psychology, most of them adopt the negative avoidance strategy of “Ostrich mentality”.

Figure 2 shows the proportion of rural households’ energy consumption in Shaanxi and Henan provinces. The energy consumption in Henan region is mainly “electricity, liquefied gas, and firewood”, while the Shaanxi region is mainly “electricity, coal, and firewood”. The two provinces have basically achieved full coverage of electricity, but the process of replacing high-carbon energy such as firewood and coal is slow, and its energy consumption level and efficiency need to be improved urgently. Shaanxi’s economy is relatively backward; the proportion of traditional biomass consumption is high, and because of the endowment of coal resources, more than half of the farmers use coal to meet their livelihood needs such as cooking and heating. Therefore, its energy consumption is low-value and crude, and in the stage of “energy anxiety”. In recent years, the government has actively promoted the construction of rural households with biogas and solar subsidies to encourage the use of zero-carbon energy. However, due to the difficulty in biogas fermentation and various security risks, and the fact that solar energy is affected by the natural environment and cannot supply hot water in a timely manner, residents’ willingness to use these energies is not strong.

3.3. Model Setting

3.3.1. Benchmark Model

The low-carbon energy intensity of households is an orderly category variable. Therefore, referring to the OLS model, the paper selected the Ordered logit (Ologit) model for regression analysis, and constructed the following benchmark model [34,46].

$$ENEst{r}_i={\alpha }_0+{\alpha }_{1}offfarm employmen{t}_i+\sum {\lambda }_m{X}_{mi}+{\epsilon }_i$$

(1)

where $ENEst{r}_i$ is the low-carbon energy intensity of the rural households, and $offfarm emplomen{t}_i$ represents different off-farm employment scenarios to explore their impact on low-carbon energy intensity. $X_{mi}$ is a series of control variables that may affect the dependent variables, including the household demographic, livelihood development, cognitive, psychological, and regional characteristics. ${\alpha }_0$ and ${\epsilon }_j$ represent intercept term and random error term, respectively.

3.3.2. Mediating Effect Model

Based on the known relationship between the independent variable X and the dependent variable Y, if there exists variable M such that the independent variable X can influence the dependent variable Y through variable M, then variable M is called the mediating variable. To investigate the mediation mechanism of “off-farm employment→income effect→low-carbon energy intensity”, referring to the studies of Wen [47] and Liu [2], the Sobel test was used to explore the mediating effect and constructed the following models based on model (1):

$$\left\{\begin{array}{c}IN{C}_i={\alpha }_0+{\alpha }_{2}offfarm employmen{t}_i+\sum {\lambda }_m{X}_{mi}+{\epsilon }_i\\ ENEst{r}_i={\alpha }_0+{\alpha }_{3}offfarm employmen{t}_i+{\beta }_{3}IN{C}_i+\sum {\lambda }_m{X}_{mi}+{\epsilon }_i\end{array}\right.$$

(2)

where $IN{C}_i$ is the intermediary variable, representing the total household income. $X_{mi}$ is a series of control variables for which the meaning is consistent with the above model (1). Drawing on existing research [2], the significance of coefficient ${\alpha }_1$ in the benchmark model (1) is first tested, and if it is significant, then we proceed to the next step. Then, we test that ${\alpha }_2$ and ${\beta }_3$ in the model (2) are both significant, indicating the existence of a mediating effect. The Z statistic formula in the Sobel test is as follows:

$$Z=\frac{{\alpha }_2{\beta }_3}{\sqrt{{\alpha }_2{}^2{S}_{\beta }{}^2+{\beta }_3{}^2{S}_{\alpha }{}^2}}$$

(3)

where $S_{\alpha }$ and $S_{\beta }$ are the standard deviations of the estimated values of parameters ${\alpha }_2$ and ${\beta }_3$, respectively. The critical value of the Z statistic at 5% significance level is 0.97. If the statistic Z is greater than 0.97, it indicates the existence of the mediation effect [2].

3.3.3. Moderating Effect Model

If the relationship between the dependent variable Y and the independent variable X varies with a third variable N, then variable N is called the moderator variable. To further explore the role of peer effect in the impact of off-farm employment on low-carbon energy intensity, referring to the studies of Fang et al. [48], the equation is estimated based on model (1) as follows:

$$ENEst{r}_i={\alpha }_0+{\alpha }_{4}offfarm employmen{t}_i+{\gamma }_{4}peer+{\lambda }_{4}off farm employmen{t}_i\ast peer+\sum {\lambda }_m{X}_{mi}+{\epsilon }_i$$

(4)

where $peer$ represents the regulating variable, and $off farm employmen{t}_i\ast peer$ represents the interaction term between the off-farm employment and the peer effect. The meaning of the other explanatory variables is consistent with the above model (1). If the interaction term regression coefficient ${\lambda }_4$ is significant, it proves that the peer effect has a significant moderating effect.

4. Empirical Results and Discussion

4.1. Benchmark Regression Analysis

The regression results of off-farm employment on low-carbon energy intensity are shown in

Table 2. Off-farm employment and farmers’ low-carbon energy intensity (N = 1351).

Variables (Model)	Low-Carbon Energy Intensity
	Model 1 (OLS)	Model 2 (Ologit)
Off-farm employment	0.291 *** (0.07)	0.478 *** (0.13)
Household head’s age	−0.006 ** (0.00)	−0.012 ** (0.00)
Household size	−0.023 (0.02)	−0.041 (0.03)
Household rearing burden	0.053 (0.13)	0.097 (0.25)
Average education level	0.117 *** (0.04)	0.219 *** (0.07)
Number of electrical appliances	0.050 *** (0.01)	0.098 *** (0.01)
Cultivated area	−0.014 *** (0.00)	−0.032 *** (0.01)
Policy perception	0.061 ** (0.03)	0.121 ** (0.05)
Low-carbon awareness	0.152 *** (0.06)	0.272 *** (0.11)
Energy-saving habits	0.032 (0.06)	0.047 (0.11)
Regional	−0.457 *** (0.06)	−0.809 *** (0.11)
Constant term	2.068 *** (0.26)	--
R2/Pseudo R2	0.223	0.091

Notes: Standard errors, presented in parentheses, are robust. Significance at 5% and 1% are indicated by ** and ***, respectively.

. It can be clearly seen that the estimated coefficients of off-farm employment are all positive at the significant of level of 1% (0.291, 0.478), indicating that after controlling the household demographic, livelihood development, cognitive psychological, and regional characteristics, off-farm employment can significantly improve low-carbon energy intensity in rural households. Compared with purely farming households, the experience of low-carbon energy use and income growth during the development of off-farming have increased rural households’ awareness and purchasing power of low-carbon energy, and efficient energy has become the first choice for residential energy consumption, thus completing a sustainable transition to low-carbon energy.

Further,

Table 3. Short-term and long-term off-farm employment and farmers’ low-carbon energy intensity (N = 1351).

Variables (Model)	Low-Carbon Energy Intensity
	Model 3 (OLS)	Model 4 (Ologit)	Model 5 (OLS)	Model 6 (Ologit)
Short-term off-farm employment	0.170 *** (0.06)	0.309 *** (0.11)	--	--
Long-term off-farm employment	--	--	0.339 *** (0.06)	0.587 *** (0.11)
Household head’s age	−0.007 *** (0.00)	−0.012 *** (0.00)	−0.006 ** (0.00)	−0.011 ** (0.00)
Household size	0.000 (0.02)	−0.004 (0.03)	−0.011 (0.02)	−0.023 (0.03)
Household rearing burden	−0.020 (0.13)	0.003 (0.25)	0.039 (0.13)	0.084 (0.25)
Average education level	0.125 *** (0.04)	0.234 *** (0.07)	0.115 *** (0.04)	0.218 *** (0.07)
Number of electrical appliances	0.049 *** (0.01)	0.099 *** (0.01)	0.049 *** (0.01)	0.097 *** (0.01)
Cultivated area	−0.016 *** (0.00)	−0.034 *** (0.01)	−0.009 ** (0.00)	−0.021 ** (0.01)
Policy perception	0.065 ** (0.03)	0.129 ** (0.05)	0.060 ** (0.03)	0.121 ** (0.05)
Low-carbon awareness	0.148 ** (0.06)	0.262 ** (0.10)	0.148 *** (0.06)	0.260 ** (0.11)
Energy-saving habits	0.016 (0.06)	0.023 (0.11)	0.043 (0.06)	0.073 (0.11)
Regional	−0.460 *** (0.06)	−0.830 *** (0.11)	−0.477 *** (0.06)	−0.858 *** (0.11)
Constant term	1.920 *** (0.23)	--	1.690 *** (0.22)	--
R2/Pseudo R2	0.217	0.090	0.231	0.095

Notes: Standard errors, presented in parentheses, are robust. Significance at 5% and 1% are indicated by ** and ***, respectively.

reports the estimated results of the low-carbon energy intensity of rural households under different off-farm employment scenarios. From the regression results of OLS and Ologit, the estimated coefficients of short-term off-farm employment and long-term off-farm employment are both positive at the statistical level of 1% (0.170, 0.309, and 0.339, 0.587), showing that households in off-farm employment have a higher intensity of low-carbon energy consumption compared with farming; hence, hypothesis 1 is supported. Moreover, the regression coefficient of long-term off-farm employment is larger compared to that of short-term off-farm employment, indicating that long-term off-farm employment has a more profound effect on households’ low-carbon energy intensity. Short-term off-farm employment is at the early stage of part-time employment, with less time and income, high off-farm transfer costs, and heavy financial burdens, which weaken the impact of off-farm employment on the low-carbon energy consumption of rural households. Long-term off-farm employment with the main purpose of increasing income has led to a significant increase in household livelihoods, and workers are fully aware of the benefits of using low-carbon energy to improve their health and promote ecological sustainable development, so their energy use gradually follows that of the cities and towns to achieve a low-carbon transition to electrification.

In addition, some control variables also significantly influence the rural energy low-carbon transition. Existing studies show that the individual characteristics of the household head, as the main decision maker in household production and life, are closely related to the energy adoption [35]. This study shows a significant negative correlation between the head of the household’s age and low-carbon energy intensity, indicating that older heads of households influenced by the traditional farming culture and the sense of saving money are more accustomed to the use of low-quality and inferior energy [49]. The average education level is significantly positively correlated with low-carbon energy intensity, as highly educated households are likely to be more environmentally conscious and aware of the adverse repercussions of using biomass energy [50]. The cultivated area variable shows a negative and significant relationship with low-carbon energy intensity, suggesting that households with more arable land are engaged in intensive agricultural activities and have easier access to biomass energy. This finding is consistent with those of Tebikew [51,52]. The results further show that policy perception is significantly positively related with low-carbon energy intensity; that is, households are likely to use more low-carbon energy sources such as solar when their perceived policy support is stronger [36]. The coefficient for low-carbon awareness is significant and positive for low-carbon energy intensity, respectively, implying that rural residents with stronger environmental values are more likely to practice pro-environmental behaviors and use clean energy [53].

4.2. Income Effect

Based on the previous findings related to the significant effect of off-farm employment on low-carbon energy intensity, it was explored whether off-farm employment affects low-carbon energy intensity through the income effect (increase in total household income). As shown in

Table 4. Mediation test of income effect (N = 1351).

Path Ⅰ		Path Ⅱ		Effect of Off-Farm Employment Characteristics on Low-Carbon Energy Intensity	Sobel Test (Z Value/p Value)	Proportion of Intermediary Effect
The Influence of Off-Farm Employment Characteristics on Total Household Income	Coefficient	The Impact of Total Household Income on Low-Carbon Energy Intensity	Coefficient
Off-farm employment→ Total household income	0.673 *** (0.06)	Total household income→ Low-carbon energy intensity	0.104 *** (0.03)	0.221 *** (0.07)	Z value: 3.047 p value: 0.002	24.12%
Short-term off-farm employment→ Total household income	0.123 ** (0.03)	Total household income→ Low-carbon energy intensity	0.130 ** (0.03)	0.154 ** (0.06)	Z value: 2.096 p value: 0.036	9.41%
Long-term off-farm employment→ Total household income	0.578 ** (0.05)	Total household income→ Low-carbon energy intensity	0.090 ** (0.03)	0.287 ** (0.06)	Z value: 2.672 p value: 0.008	15.37%

Notes: Standard errors, presented in parentheses, are robust. Significance at 5% and 1% are indicated by ** and ***, respectively.

, the Z statistic in the Sobel test is 3.047, which is greater than the critical value of 0.97 at the 5% significance level, indicating that off-farm employment affects low-carbon energy intensity through income effect, and this mediating effect accounts for 24.12% of the total effect. In the context of urbanization and rural revitalization, the off-farm employment of laborers can greatly improve household livelihood capital and promote the gradual shift from subsistence consumption to development consumption. The quality of energy consumption is closely related to the quality of life, so it gives rise to a higher level of household energy demand, enhances the concern for energy quality and safety, and promotes the low-carbon energy transition of households’ livelihoods, i.e., households use more low-carbon energy in various livelihood activities to improve their quality of life. Thus, research hypothesis 2 is verified.

4.3. Peer Effect

To increase the explanatory significance of the coefficient of the equation, the independent variable (off-farm employment) and the moderator variable (peer effect) were centralized before the regression analysis, and the interaction term was generated by the product of variable centralization. The results of the peer effect test are shown in

Table 5. Adjustment test of peer effect (N = 1351).

Variables (Model)	Low-Carbon Energy Intensity
	Model 7 (Ologit)	Model 8 (Ologit)	Model 9 (Ologit)
Off-farm employment	0.504 *** (0.13)	--	--
Short-term off-farm employment	--	0.272 ** (0.11)	--
Long-term off-farm employment	--	--	0.593 *** (0.12)
Peer effect	0.256 *** (0.05)	0.242 *** (0.05)	0.244 *** (0.05)
Off-farm employment * Peer effect	0.273 ** (0.13)	--	--
Short-term off-farm employment * Peer effect	--	0.015 (0.11)	--
Long-term off-farm employment * Peer effect	--	--	0.204 *** (0.10)
Household head’s age	−0.011 ** (0.00)	−0.012 ** (0.00)	−0.010 *** (0.00)
Household size	−0.049 (0.03)	−0.007 (0.03)	−0.027 (0.03)
Household rearing burden	0.137 (0.25)	0.031 (0.25)	0.082 (0.25)
Average education level	0.230 *** (0.07)	0.249 *** (0.07)	0.228 *** (0.07)
Number of electrical appliances	0.100 *** (0.01)	0.100 *** (0.01)	0.098 *** (0.01)
Cultivated area	−0.035 *** (0.01)	−0.038 *** (0.01)	−0.023 ** (0.01)
Policy perception	0.089 * (0.05)	0.098 * (0.05)	0.094 * (0.05)
Low-carbon awareness	0.261 ** (0.11)	0.256 ** (0.11)	0.240 ** (0.11)
Energy-saving habits	−0.017 (0.11)	−0.032 (0.11)	0.020 (0.11)
Regional	−0.767 *** (0.11)	−0.798 *** (0.11)	−0.824 *** (0.11)
R2/Pseudo R2	0.099	0.095	0.102

Notes: Standard errors, presented in parentheses, are robust. Significance at 10%, 5%, and 1% are indicated by *, **, and ***, respectively.

. The interaction term of off-farm employment and peer effect are significantly positively correlated with households’ low-carbon energy intensity (0.273, 0.204), indicating that the low-carbon energy use of surrounding people plays a crucial role between off-farm employment, long-term off-farm employment, and low-carbon energy intensity of rural households. It reflects that households consciously learn from others’ pro-environmental behaviors to improve their sense of environmental responsibility, thus giving up the use of high-carbon energy. However, the interaction between short-term off-farm employment and peer effect is not significant, probably because households with short-term off-farm employment spend most of their time in rural environments. Considering the convenience of obtaining biomass fuels and the use of commodity energy costs, they will maintain their high-carbon energy use even if the surrounding population uses low-carbon energy. Hypothesis 3 of this study is partially verified.

4.4. Robustness Check

4.4.1. Replacing Core Explanatory Variable

To ensure the robustness of the research results, the core explanatory variable, off-farm employment, was replaced with the off-farm employment experience [52]. The variable comes from the questionnaire item: “Whether family members have experience in off-farm employment? (1 = yes, 0 = no)”.

Table 6. Off-farm employment experience and farmers’ low-carbon energy intensity (N = 1351).

Variables (Model)	Low-Carbon Energy Intensity
	Model 10 (OLS)	Model 11 (Ologit)
Off-farm employment experience	0.299 *** (0.08)	0.517 *** (0.15)
Household head’s age	−0.006 ** (0.00)	−0.011 ** (0.00)
Household size	−0.015 (0.02)	−0.027 (0.03)
Household rearing burden	0.006 (0.13)	−0.009 (0.25)
Average education level	0.117 *** (0.04)	0.216 *** (0.07)
Number of electrical appliances	0.050 *** (0.01)	0.098 *** (0.01)
Cultivated area	−0.014 *** (0.00)	−0.030 *** (0.01)
Policy perception	0.046 * (0.03)	0.091 * (0.05)
Low-carbon awareness	0.163 *** (0.06)	0.285 *** (0.11)
Energy-saving habits	0.069 (0.06)	0.113 (0.11)
Regional	−0.441 *** (0.06)	−0.786 *** (0.11)
Constant term	1.662 *** (0.23)	--
R2/Pseudo R2	0.219	0.090

Notes: Standard errors, presented in parentheses, are robust. Significance at 10%, 5%, and 1% are indicated by *, **, and ***, respectively.

presents the re-estimated results of Equation (1), demonstrating that effect direction and significance level are consistent with the results in Table 2, indicating that the impact of off-farm employment on low-carbon energy intensity is relatively stable.

4.4.2. Re-Estimation Using the PSM Method

This study conducted a common hypothesis test to match observations within the common range to ensure the rationality and validity of the PSM estimation. Before matching, large differences and peak deviations in the probability distributions of propensity scores were seen between the control group and the treatment group. After matching, the kernel density distribution was more consistent between the two sample groups, most of the tendency scores fell into the common range, and the standardized deviation of the matched variables was mostly less than 10%. Further, to ensure balance after sample matching, this study adopted k-nearest neighbor matching, radius matching, and kernel function matching to estimate the average treatment effect of ATT.

Table 7. Matching estimation results of propensity score (N = 1351).

Variables	Matching Method	Low-Carbon Energy Intensity
		Processing Group	Control Group	ATT	Standard Error	T. Value
Off-farm employment	K-nearest neighbor matching (k = 4)	2.591	2.297	0.294 ***	0.093	3.17
	Radius matching (radius = 0.05)	2.588	2.317	0.271 ***	0.087	3.13
	Kernel function matching	2.588	2.319	0.269 ***	0.087	3.10
Short-term off-farm employment	K-nearest neighbor matching (k = 4)	2.484	2.348	0.136 *	0.076	1.80
	Radius matching (radius = 0.05)	2.460	2.345	0.115 *	0.068	1.69
	Kernel function matching	2.460	2.345	0.115 *	0.068	1.69
Long-term off-farm employment	K-nearest neighbor matching (k = 4)	2.732	2.435	0.297 ***	0.107	2.76
	Radius matching (radius = 0.05)	2.732	2.453	0.279 ***	0.105	2.65
	Kernel function matching	2.732	2.456	0.277 ***	0.105	2.64

Note: Standard errors, presented in parentheses, are robust. Significance at 10% and 1% are indicated by * and ***, respectively.

indicates that the effects of off-farm employment on the households’ low-carbon energy intensity has not changed significantly before and after matching; the three matching results have a small difference, and their effect direction and significance level are consistent with the benchmark regression results in Table 3 and Table 4.

5. Limitations and Future Research

This study has several shortcomings. First, study data were limited to Central and Northwest China, but it lacks the representative provinces of East and South China. Future research will use more regionally representative data to test the applicability of the conclusions of this paper to other regions and countries. In addition, only mixed cross-sectional data were used for research, which cannot reflect dynamic trends, and future analysis of the dynamics of off-farm employment and household low-carbon energy intensity and other potential impact mechanisms is needed. Therefore, a new phase of the household survey will be conducted to construct the panel data and extend the study.

6. Conclusions and Policy Implications

6.1. Conclusions

With the advancement of the distinctive socialist rural revitalization path, off-farm employment of rural laborers is becoming increasingly widespread, while energy poverty, as represented by the under-use of energy, remains a central concern. Based on data of 1351 rural households in Shaanxi and Henan provinces in 2018 to 2021, this study analyzes the impact of off-farm employment on low-carbon energy intensity and its response mechanism. Through the previous empirical analysis and discussion, the conclusions of this study are as follows:

(1)

The regression results demonstrate that off-farm employment, including short-term and long-term off-farm employment, significantly increases the intensity of low-carbon energy use among rural households. Long-term off-farm employment tends to have a greater positive contribution to low-carbon energy intensity than short-term off-farm employment.

(2)

By exploring potential mechanisms, evidence was found that off-farm employment significantly increases total household income, which in turn influences their low-carbon energy intensity. Further moderating analysis showed that the demonstration effect of the surrounding people in the off-farm employment process also increases the low-carbon energy intensity in rural households.

(3)

Moreover, factors such as the head of the household’s age, average education level, number of electrical appliances, cultivated area, policy perception, low-carbon awareness, and region affect low-carbon energy intensity.

6.2. Policy Implications

Based on the above empirical analysis, to achieve the goal of a low-carbon transition in rural energy consumption and build an ecological priority low-carbon green development road, the following policy implications are proposed.

First, policy efforts aiming to increase the likelihood of securing off-farm employment can achieve the coordinated development of low-carbon energy transformation and sustainable livelihood by promoting the reasonable transfer of rural labor and enhancing the competitiveness of rural laborers in the job market through skills training. Specifically, the government should provide more off-farm employment opportunities to ensure sustained growth in rural household income, and then reduce the economic burden they face in consuming low-carbon energy.

Second, the government should provide public education for rural households to understand that low-carbon energy adoption is beneficial to physical health and environmental protection. Additionally, it is necessary to increase the promotion of low-carbon energy through the demonstration power of grassroots party cadres and informal civil society organizations, to increase the low-carbon awareness of residents and gradually reduce carbon emissions in rural areas.

Finally, policymakers should implement differentiated policies to increase access to energy consumption by low-income groups and pay attention to the implementation of rural grassroots services, choose suitable new energy varieties and development modes according to local conditions, improve the construction of low-carbon energy-related infrastructure, encourage social capital to participate in the construction of clean energy demonstration villages, and realize “Precise Energy Poverty Alleviation”. For example, they can use unused land of farmers or village public building rooves to build distributed wind power and photovoltaic power generation.