Abnormal Temperature and Rural Energy Poverty: A Threshold Effect Analysis Based on the Urban–Rural Gap and the Quality of Rural Housing

Abstract

In the context of global warming, abnormal temperature is becoming more frequent and its various impacts are receiving increased attention, but the impact of abnormal temperature on rural energy poverty has received relatively little focus. This study examines the effects of abnormal temperature on rural energy poverty, utilizing the systematic generalized method of moments with data from 30 Chinese provinces from 2001 to 2020. Using a panel threshold model, we investigate how the urban–rural gap and the quality of rural housing interact with the relationship between abnormal temperature and rural energy poverty. Our findings suggest that abnormal temperature significantly increases rural energy poverty by 5.7%, and the impact of abnormal temperature on rural energy poverty is exacerbated by a greater urban–rural gap and lower quality of rural housing. This study yields several policy implications for achieving a coordinated and synergistic response to climate change and the eradication of energy poverty based on the promotion of balanced development.

1. Introduction

Energy and climate change issues are common challenges for humankind. The Earth is the only home for human survival, and increasing efforts to address climate change and eradicating energy poverty are crucial to the future of humanity [1]. More and more government policy-making and academic research are closely focusing on these two issues. Climate change has triggered global warming with an increasing frequency and intensity of temperature extremes, leading to increased mortality, the morbidity of epidemics [2,3,4], poverty [5], climate migrants [6], market activity falls [7], and food insecurity [8,9]. Climate change and energy challenges are closely related in that climate change affects energy demand, especially energy consumption such as heating in winter and cooling in summer [10,11]. Significant progress has been made in studies related to the effect of temperature on energy use [12]. For example, the increase in global average temperature will not only increase total residential electricity consumption but also make peak electricity consumption, in total, increase significantly [13]. Moreover, abnormal temperatures have severe consequences for energy supply [14]. Climate change and its resulting extreme weather events, such as abnormal temperature, reduced regional and seasonal water availability, storms, floods, typhoons, and other extreme weather, all of which will have impacts on existing energy systems, will increase in the future [15,16,17].

The current emphasis is mostly directed towards accomplishing energy transition within the framework of climate change. However, it is crucial to acknowledge that climate change already presents a substantial obstacle to other energy-related concerns, including the issue of energy poverty, which has not received commensurate attention [18]. Energy poverty, defined as insufficient access to affordable and reliable energy, constitutes a significant barrier to socioeconomic advancement and is a critical issue within the United Nations’ Sustainable Development Goal 7 that requires urgent attention [19,20,21,22]. The occurrence of extreme temperature due to climate change would increase the probability of energy poverty for all households by increasing energy expenditures for cooling and heating and resulting in lower labor productivity and reduced total income [23,24]. Compared to urban residents, rural households are believed to be more sensitive to temperature extremes due to poor infrastructure and dependence on agricultural income [24]. Previous studies have suggested that extreme temperature can reduce rural households’ income from agriculture, increase energy expenditures, and make it less likely that clean energy use will be achieved under budget constraints. Therefore, abnormal temperature resulting from climate change will drive up households’ energy demand, which could exacerbate energy poverty among low-income households [25]. In addition, unusual weather causes more damage to energy facilities and has an impact on energy supply, particularly in rural areas with inadequate energy infrastructure, increasing energy poverty among rural households.

Understanding the relationship between energy poverty and climate change is a prerequisite for better addressing climate change and eradicating energy poverty [26]. The majority of previous research has concentrated on the connection between climate change and energy poverty at the household level, but to best of our knowledge, there is not much work at the regional level and even less that focuses on the role of development imbalances in the connection between climate change and rural energy poverty in the literature [23,27,28]. In addition, the urban–rural gap and the quality of rural housing are important indicators of this development imbalance, and existing studies have suggested that these two indicators potentially interact with energy poverty and climate change [29,30,31]. Therefore, the role of urban–rural disparities and rural housing quality in interacting with the association between energy poverty and climate change requires further consideration.

China serves as an ideal setting for this topic due to the following distinct factors. China possesses a wide range of climatic conditions, providing ample opportunities for the study and analysis of climate factors from a scientific perspective. Furthermore, while China has acknowledged the widespread availability of electricity in rural areas for quite some time [32], the provision of power used to drive large machinery in rural areas has only been achieved in recent years. Despite this progress, China’s rural areas, characterized by inadequate infrastructure, continue to face a certain level of energy poverty. Moreover, China’s distinct urban–rural dichotomy provide an opportune context for investigating disparities in development, particularly the disparity between urban and rural areas. Moreover, it is noteworthy that China holds the distinction of being the largest developing nation globally. Consequently, this study, conducted within its borders, possesses a heightened significance for other places. To analyze the relationship between abnormal temperature and energy poverty at the regional level and to explore the role of urban–rural development imbalances on this relationship, this study first constructs a composite index of rural energy poverty. We use meteorological data to compute abnormal temperature variables and estimate the causal relationship between abnormal temperature and energy poverty with the generalized method of moments (GMM). To explore the effect of urban–rural disparity and rural housing quality on the relationship between abnormal temperature and energy poverty, the panel threshold model is applied. This study contributes to the existing literature in multiple ways. First, we develop a novel rural energy poverty index that takes rural energy construction and agricultural production into account, extending the traditional concept of energy poverty that focuses on energy prices and energy use. Second, we contribute to the literature by investigating the effect of abnormal temperature on rural energy poverty in China with empirical evidence from a macro-perspective. Third, the role of urban–rural disparity and rural housing quality in interacting with the relationship between energy poverty and climate change is explored, thereby enhancing the understanding of the complex relationship between climate change and energy poverty and making it more clear for policy-makers to target vulnerable regions and groups.

The rest of this study is organized as follows. Section 2 provides a synthesis of the pertinent literature. Section 3 describes the data and methodology. Section 4 discusses results from the model. Section 5 summarizes the study and proposes policy implications.

2. Literature Review

2.1. Impacts of Abnormal Temperature on Energy Poverty

Climate change is believed to have an impact on energy consumption and use, especially in areas like winter heating and summer cooling [33]. Previous studies have investigated the effect of temperature on energy consumption, and significant causal effects have been identified [34]. For example, a 1.5 °C to 4.0 °C increase in global temperature is associated with a significant increase in total residential electricity use and peak electricity consumption [13], and an increase in global average temperature would alter energy consumption and increase cooling energy consumption [25]. From a physiological perspective, abnormal temperatures, typically extremely high or extremely low temperatures, could cause the human body to require more energy to adjust to temperature changes [35,36], which would raise energy demand. Specifically, some studies have found evidence that both high and low temperatures can increase electricity demand [37,38,39], and total global energy consumption is expected to rise by 24.0% by 2095 [40]. However, the impact varies by country and economic development level, with climate change leading to larger increases in energy demand in poorer countries [40,41,42]. Existing studies have also found that the effects of abnormal temperatures are regressive for low-income populations, potentially exacerbating energy poverty [25]. The provision of basic energy facilities and a stable supply is critical for reducing energy inequality and increasing the capacity of low-income populations to adapt to climate change [43].

In the short run, abnormal temperatures could have significant impacts on energy supply [14]. The majority of energy facilities are exposed to the natural environment; therefore, meteorological conditions have strong influences on energy production and transportation. An important example of this is the summer of 2022, when drought forced Sichuan Province to reduce a significant portion of its hydropower generating capacity. Extreme weather events, such as droughts and cold snaps, can adversely affect the energy supply–demand relationship, not only by increasing the demand for electricity and gas but also by affecting the supply of electricity through reductions in the generation of hydroelectricity and wind power. [44]. A growing number of studies are focusing on the relationship between anomalous temperatures and energy consumption and supply, raising concerns that the occurrence of anomalous temperatures may exacerbate energy poverty [23,45]. A number of media outlets have also reported on the consequences of energy poverty caused by the impact of unusually high temperatures on energy facilities. However, no study has formally and thoroughly examined the effects of extreme temperatures on energy poverty, particularly at the regional level.

2.2. The Urban–Rural Gap, Rural Housing Quality, and Rural Energy Poverty in the Context of Climate Change

There is a limited body of literature that focuses directly on the relationship between development inequality and energy poverty in the context of climate change. However, a small number of studies have indirectly addressed this topic, and a compilation of this literature can help to clarify relevant threads. According to a study of 37 countries, income inequality increases the risk of energy poverty even more than the effects of climate change [46]. Rural energy poverty is gradually becoming a hot topic of research due to the urban–rural divide. Urban–rural disparities may exacerbate energy inequality. Such energy inequality can result in a vicious cycle of energy poverty and exacerbate rural areas’ chronic or persistent poverty [47]. A study noted the inequality between urban and rural residents in coping with climate change and mitigation burdens [48]. Significant inequalities will exacerbate the negative impacts of climate change. The existence of an urban–rural divide deprives rural areas of numerous development opportunities and resources [49]. A large urban–rural gap means that poor rural energy infrastructure, low levels of energy services, and low rates of clean energy use exacerbate energy poverty in rural areas [50]. Rural communities are under intense pressure to switch to cleaner energy sources. In the context of climate change, this will limit efforts to alleviate rural energy poverty [51]. In addition, one study attributes the persistence of energy poverty to the rural–urban divide, citing the lack of public infrastructure in rural areas, geographic and economic isolation, and the absence of affordable transportation between urban and rural areas as factors that contribute to higher energy prices in rural areas and increased energy poverty [52]. There is also a study that examines the effects of income inequality and energy poverty on carbon emissions, but it does not examine their interaction [53]. Another body of literature on the relationship between inequality and energy poverty examines inequality from the perspective of energy poverty. For example, a study identifies energy poverty as a new form of inequality in the context of future climate change and suggests that poor climate policies may undermine the foundation of individual capabilities, thereby contributing to energy poverty [27]. One piece of research believes that inequality can be diminished by reducing energy poverty [54]. Another study examines inequalities in the distribution of energy poverty, which provides an entry point for examining regional economic inequality and energy transition, beginning with the perspective of energy poverty [55]. Some researchers believe that a stable energy supply is an important prerequisite for combating climate change [43]. This literature has discussed energy poverty and inequality in great detail, but little research has been conducted on how income inequality influences the climate change–energy poverty nexus.

In addition to urban–rural differences, there are also significant rural–rural differences, with housing quality being an important variable. Existing studies have revealed the significant role of housing quality in energy poverty [56]. According to a number of studies, the negative effects of climate change are exacerbated in some less developed regions due to uneven economic development [57,58,59]. There are several factors contributing to the heightened vulnerability of less developed regions to the adverse impacts of climate change, with one significant aspect being the poor housing conditions prevalent in these regions, particularly in rural areas. There is a consensus that there is a strong relationship between housing conditions and energy poverty [60,61]. The dampening effect of social capital on households’ energy poverty shows heterogeneity in terms of household housing conditions, and the effect is significant for households with squatter housing [62]. The fact that building conditions are worse in rural areas is a striking manifestation of the urban–rural divide. Poor building conditions are a major source of risk for rural energy poverty [63]. Low building energy efficiency can have multiple negative effects on vulnerable populations, including increased residential evictions, decreased school and work attendance, and decreased comfort and safety in the home [61]. In the context of climate change, energy inefficiency in housing exacerbates the risk of rural energy poverty [60,64]. Rural households with outdated heating equipment and heating houses are more vulnerable to energy poverty than urban households [65]. The available scholarly literature extensively addresses the significant correlation between housing quality, with a particular emphasis on energy efficiency and the issue of energy poverty. However, there exists a dearth of comprehensive empirical investigations that delve into the impact of housing quality on the interplay between climate change and energy poverty.

Based on the inadequacy of existing studies, consideration was given to exploring the impact of the urban–rural gap and rural housing quality on the relationship between abnormal temperatures and rural energy poverty.

3. Materials and Methods

3.1. Energy Poverty Measurement

One of the central tenets of Sustainable Development Goal 7 is the eradication of energy poverty. In recent years, numerous academicians have studied energy poverty, but there is currently no standardized method for measuring it [66]. Different scholarly approaches have been taken to measure energy poverty. These approaches generally concentrate on three factors: the availability of energy services, the quality of these services, and satisfaction with the energy necessary for human survival and development [67]. Energy poverty is ordinarily measured at the household or regional level. At the household level, the proportion of energy expenditures and the availability of clean energy are the primary considerations. Energy poverty is only partially observable at the household level [66]. Numerous scholars have devised a comprehensive multidimensional assessment of energy poverty at the regional level in order to circumvent the limitations of a single indicator. A comprehensive method for measuring energy poverty was introduced [68], focusing on the challenges associated with accessing modern energy services. A comprehensive energy poverty measurement index was developed [69], incorporating four dimensions and nine indices to assess energy poverty in China. And, their findings demonstrated a decline in energy poverty in China. A system for measuring energy poverty using multiple dimensions was proposed [70], incorporating factors such as energy costs, income, and housing energy efficiency. A developed energy poverty index is based on four dimensions: the availability of energy services, the cleanliness of energy consumption, the integrity of energy management, and the affordability of household energy [67]. This methodology has been widely implemented [66,71,72]. However, an evaluation of the literature reveals a paucity of research on comprehensive energy poverty assessments for rural areas in developing nations. However, these studies have not focused on rural energy poverty. The literature specifically measuring rural energy poverty at the regional level is scarce. Due to the specificity of rural areas, there should be research on measuring energy poverty in rural areas.

3.2. Rural Energy Poverty Index

In order to effectively measure rural energy poverty, we constructed and designed a comprehensive evaluation index to measure energy poverty. In this study, the selection criterion for indicators was based on the description of energy poverty in the relevant literature. One core indicator of energy poverty at the household level is the lack of accessible and affordable energy for production and living [73,74]. Except for the percentage of people who have access to electricity, all the metrics in the IEA’s energy development index system apply to China [67]. Electricity consumption is always one of the core indicators included in all energy poverty composite indices [50]. A study takes into account energy prices and energy construction and supply investments [75]. Gas penetration is included in the energy poverty index by some studies. The link between energy poverty and agricultural production has been examined [76]; energy poverty is closely linked to the power and quantity of machinery used in agricultural production. Based on the results of existing studies and taking into account the actual situation in rural China, the comprehensive evaluation index of rural energy poverty in China in this study is made up of five types of indicators: agricultural production energy use (APE) level, rural living energy consumption (RLE) level, rural power construction (RPC), rural fuel price indices (RPIs), and rural energy cleanliness (REC) level. APE and RLE represent farmers’ ability and opportunity to access modern energy services, which correspond to the amount of energy used for agricultural production and domestic use, respectively [76]. RPC denotes the amount of rural electricity produced, representing the level of energy construction in rural areas. The value of an RPI represents the level of rural energy prices. REC denotes the modernization and optimization of rural energy consumption [67]. The above indicators can comprehensively reflect China’s current rural energy poverty situation while avoiding the problem of data scarcity. The data sources for the seven measures, as shown in

Table 1. Indicator system of rural energy poverty in China.

Index	Measurement
Agricultural production energy use level	Agricultural machinery power
	Number of agricultural pumps
Rural living energy consumption level	Rural electricity consumption (billion kWh)
	Number of rural residents with air conditioners per 100 households
Rural power construction level	Rural electricity generation capacity (million kWh)
Rural fuel price indices	Rural fuel price index (taking the previous year as 100)
Rural energy cleanliness level	Village gas penetration rate

, come from the China Urban and Rural Construction Statistical Yearbook and China Rural Statistics Yearbook.

As shown in Table 1, power of agricultural equipment, rural electricity consumption, the ownership of air conditioners as a percentage, rural hydropower productivity, energy prices, and natural gas penetration make up the final construct of the rural energy poverty indicator. In this case, agricultural machinery power in this study refers to the sum of the power ratings of all agricultural machinery power, and the measurement of clean rural energy includes only gas. High power used by agricultural equipment and rural areas is a proxy for a low level of energy poverty. In addition, a high level of rural electricity facilities, a large number of agricultural water pumps, and a high level of rural hydroelectricity can be used as proxies for a low level of energy poverty. Moreover, the percentage of air conditioner ownership and the local energy poverty rate are closely correlated. Specifically, a lower level of energy poverty can be a proxy for a higher air conditioner ownership rate which suggests adequate energy support and security. Lastly, since lower rural fuel prices and higher rural gas supply penetration normally suggest adequate energy supply, we use these two indicators as proxies for energy poverty as well.

This study uses the indicators in Table 1 to calculate the rural energy poverty index (REPI) by the entropy method. Subscript i denotes the province, j denotes the measurement indicator, $x$ is the value of the indicator, and $x^{\prime }$ is the normalization of the measurements.

For positive measurements, the normalization is expressed as Equation (1). For negative measurements, the normalization is expressed as Equation (2).

$$x_{ij}^{\prime }={\displaystyle \frac{x_{ij}-min(x_{1j},..., x_{nj})}{max(x_{1j},..., x_{nj})-min(x_{1j},..., x_{nj})}}$$

(1)

$$x_{ij}^{\prime }={\displaystyle \frac{max(x_{1j},..., x_{nj})-x_{ij}}{max(x_{1j},..., x_{nj})-min(x_{1j},..., x_{nj})}}$$

(2)

The following formula is used in this study to compute the ratio of the value of the j-th measurement of the i-th province to the value of the j-th measurement of all provinces:

$$p_{ij}={\displaystyle \frac{x_{ij}^{\prime }}{{\sum }_{i=1}^n{x}_{ij}^{\prime }}}$$

(3)

The entropy value of the j-th measurement is then calculated as follows:

$$e_j=-k{\sum }_{i=1}^n{p}_{ij}Ln\left(p_{ij}\right)$$

(4)

where $k=1Ln\left(n\right)>0;e_j\ge 0$.

The following is how this study calculates information entropy redundancy:

$$d_j=1-e_j$$

(5)

The weight of each measurement is calculated in the following manner:

$$w_j={\displaystyle \frac{d_j}{{\sum }_{j=1}^m{d}_j}}$$

(6)

The calculation of energy poverty involves a comprehensive index (REPI), which can be determined through the following steps:

$${REPI}_i={\sum }_{j=1}^n{w}_j·x_{ij}^{\prime }$$

(7)

This study obtains comprehensive rural energy poverty indices (REPIs) based on the preceding calculation processes.

3.3. Abnormal Temperature

In this investigation, we define abnormal temperature by using temperature deviation [77]. The deviation was calculated by dividing the difference between the actual temperature and the long-term average temperature for a region in a given year by the long-term standard deviation for that region. This technique is widely employed in the scientific literature [78,79,80]. From 1951 to 2000, temperatures were documented historically. We express abnormal temperature as multiples of the standard deviation of the current year’s mean temperature from the historical mean for the study period, between 2001 and 2020. The China Meteorological Data Service Center provided the meteorological information used to construct the abnormal temperatures.

For a number of reasons, we define our abnormal temperature variables in this manner. First, relevant research has claimed that the relative size of temperature change is more essential than absolute temperature values and can assist in accounting for regional variances in climatic types. Second, past research indicates that the presence of temperature extremes results in considerably larger reactions to energy consumption, labor productivity, and agricultural production losses [45,81].

Abnormal temperature is defined similarly in this paper by first calculating the multiple of temperature deviation from the standard deviation of the historical mean and then assigning a value of 1 to the abnormal temperature variable if the value is greater than 1. Otherwise, it is given the value 0. Abnormal temperature variables are dummy variables in this study’s regressions.

3.4. Econometric Model

The variables utilized in the econometric model were acquired from publicly available databases sourced from reputable institutions, including the National Bureau of Statistics of China, China Statistical Yearbook of Regional Economy, China Statistical Yearbook, China Statistical Yearbook of Rural Areas, and China Statistical Yearbook of Urban and Rural Construction. This study’s primary objective is to estimate the effect of abnormal temperature on rural energy poverty. Due to the possibility of endogeneity in the estimation procedure, it is crucial to select the appropriate econometric method. The instrumental variable (IV) approach is the standard and most prevalent method for addressing endogeneity issues. It is important to note that instrumental variables must be relevant to independent endogenous variables but not to the dependent variable’s disturbance term. Since these two requirements frequently contradict one another, it is frequently challenging to locate an appropriate instrumental variable in practice. In the meantime, latent variables are frequently employed as instrumental variables, and IV methods must typically meet the assumption of spherical disturbance terms [82].

In this analysis, the generalized method of moments (GMM), focusing primarily on differential GMM (DIF-GMM) and system GMM (SYS-GMM), is utilized as the main estimation approach [83,84,85].In this study, dynamic panel data are chosen for empirical analysis in consideration of the potential lagged impact of rural energy poverty. The econometric model is constructed as in Equation (8):

$$LnR{EPI}_{it}=\alpha +{\beta }_{0}Ln{REPI}_{i,t-1}+{\beta }_1{AT}_{it}+{\beta }_2{X}_{it}+{\epsilon }_{it}$$

(8)

where subscript i denotes the province used in the analysis and t denotes the year to which the variable corresponds. The random disturbance term with an independent and identical distribution is denoted by ${\epsilon }_{it}$. The intercept term is denoted by α. ${\beta }_i$ (i ≥ 1) indicates the estimated coefficients. REPI stands for the province’s energy poverty in rural China. Abnormal temperature is represented by AT. A vector that includes a set of control variables is denoted by X, mainly consisting of urbanization (lnURB), the proportion of primary industry (lnAGRI), and the GDP per capita (lnPGDP). These variables were selected as controls because they provide a more comprehensive picture of the region’s socioeconomic development. It is possible to control the impact of socioeconomic development on rural energy poverty.

This study aims to further investigate how the association between abnormal temperature and energy poverty is influenced by urban–rural inequality (GAP) and rural housing quality (RHQ). Referring to formulation of the panel threshold model [86], Equation (9) was used to perform this analysis. This study assumed the following criterion, as given in Equation (9), with a single threshold for abnormal temperature (AT):

$${REPI}_{it}=\alpha +{\beta }_1{AT}_{it}·I({Thres}_{it}<\eta )+{\beta }_2{AT}_{it}·I({Thres}_{it}\ge \eta )+{\beta }_3{X}_{it}+{\mu }_i+{\epsilon }_{it}$$

(9)

where subscript i denotes the cross-sectional unit of analysis (province) and t denotes the number of time periods (year). α stands for the intercept term, and ${\epsilon }_{it}$ is the random disturbance term, which is independently and identically distributed. Variable ${\mu }_i$ represents the provincial fixed effect. ${\beta }_i$ (i ≥ 1) indicates the estimated coefficients. REPI represents rural energy poverty in the 30 Chinese provinces. AT refers to abnormal temperature. $Thres$ refers to the threshold variable The two threshold variables explored in this study are the urban–rural income gap and rural housing quality. The urban–rural income gap is expressed as the per capita income of urban residents over the per capita income of rural residents, with larger values indicating a larger urban–rural gap. Rural housing quality is expressed as the ratio of the area of rural dwellings with non-mixed structures to the total area of rural dwellings, and larger values indicate poorer rural housing quality. I(·) is the symbol for the indication function. I(·) equals 1 if the connection between the threshold parameter ($\eta$) and the threshold variable is valid; otherwise, I(·) equals 0. X denotes a vector that contains a series of control variables, as mentioned before. The variables involved in this study are shown in

Table 2. Variable descriptive statistics.

VarName	Min	Max	Mean	SD
lnREPI	−7.332	−6.057	−6.468	0.173
AT	0.000	1.000	0.802	0.399
lnURB	3.291	4.495	3.939	0.273
lnAGRI	−1.204	3.635	2.214	0.870
lnPGDP	7.971	12.013	10.234	0.836
GAP	1.85	5.12	2.851	0.572
RHQ	0.013	0.836	0.4093	0.191

.

4. Results

4.1. Spatial and Temporal Analysis of Rural Energy Poverty and Abnormal Temperatures in China

According to the REPI, the progression of rural energy poverty in China is depicted in Figure 1. It can be observed that rural energy poverty in China fluctuates over time, but the overall trend indicates a continuous declining trend. This demonstrates that the Chinese government has made significant strides in rural energy construction and poverty reduction. The size of the shaded area reflects the disparity in energy poverty between different regions, and the fact that the shaded area does not decrease significantly over time indicates the unevenness of development in different regions; therefore, the Chinese government should continue to reduce regional disparities and strive for balanced and equal development in the future. Figure 2 depicts the spatial distribution of rural energy poverty in China, revealing significant regional differences in the level of rural energy poverty in China, with a low level of rural energy poverty in the east and south and a high level of rural energy poverty in the west and north.

Figure 3 depicts the spatial distribution of temperature deviation from the standard deviation of the historical mean for each province in order to present the trend of climate change more visually. In general, there has been a clear trend of climate change over the past 20 years, with the objective trend being rising temperatures. Overall, more and more areas show increasing deviations from the average air temperature, as indicated by a darker color in Figure 3.

4.2. Examination of Cross-Sectional Dependency

Before conducting a meaningful econometric study on panel data, the existence of cross-sectional dependence must be confirmed. Inconsistencies and unreliability in empirical data analysis findings are often produced by a disregard for cross-sectional independence [87]. As a result, the Pesaran CD test [88], the Breusch–Pagan LM test [89], the Frees test [90], and the Friedman test [91] were used in this research to examine cross-sectional dependence.

The results of the four forms of cross-sectional dependence analyses are presented in

Table 3. Results of the cross-sectional dependence tests.

Test	Statistics	Prob.
Pesaran CD test	32.126 ***	0.0000
Breusch–Pagan LM test	2252.13 ***	0.0000
Frees test	8.678 ***	0.0000
Friedman test	236.874 ***	0.0000

Note: *** p < 0.01.

. At the 1% level, all p-values for analyses of cross-sectional dependence are significant. We therefore strongly reject the null hypothesis that there is no cross-sectional dependence. The findings in Table 3 indicate that cross-sectional units are not independent in this study. Therefore, when conducting the subsequent empirical study, cross-sectional dependence in the data had to be taken into consideration.

4.3. Checking Panel Stationarity

Checking the stationarity features of panel data is critical for avoiding misleading regressions. Importantly, when cross-sectional dependence occurs, the reliability of first-generation panel unit root tests such as Phillips–Perron (PP) tests, Augmented Dickey–Fuller (ADF) tests, Im, Pesaran, and Shin (IPS) tests, and Levin–Lin–Chu (LLC) tests is much reduced [92]. Second-generation panel unit root test approaches have been proposed to address cross-sectional dependence, with notable examples being the cross-sectionally augmented IPS (CIPS) and ADF (CADF) tests [92].

Table 4. Results of checking panel stationarity.

Variable	Level		First Difference		Order of Integration
	Intercept	Intercept and Trend	Intercept	Intercept and Trend
Pesaran CADF test
LnREPI	−2.508 ***	−2.985 ***	−4.753 ***	−4.833 ***	I (1)
AT	−4.088 ***	−4.338 ***	−4.920 ***	−5.040 ***	I (1)
LnURB	−2.109 **	−2.125	−3.323 ***	−3.381 ***	I (1)
LnAGRI	−1.803	−1.953	−3.866 ***	−4.056 ***	I (1)
LnPGDP	−1.451	−2.170	−3.485 ***	−3.899 ***	I (1)
Pesaran CIPS test
LnREPI	−2.508 ***	−2.985 ***	−4.753 ***	−4.830 ***	I (1)
AT	−4.100 ***	−4.379 ***	−4.933 ***	−5.023 ***	I (1)
LnURB	−2.445 ***	−2.350	−3.337 ***	−3.381 ***	I (1)
LnAGRI	−1.870	−2.150	−3.866 ***	−4.056 ***	I (1)
LnPGDP	−1.483	−2.201	−3.490 ***	−3.922 ***	I (1)

*** p < 0.01.; ** p < 0.05; * p < 0.1.

displays the findings for each variable.

During unit root tests for panels, emphasis is placed on two types: intercept and intercept and trend. Table 4 reveals that the original sequence of variables was not stationary, regardless of the trend term. This study then applied first-order difference to the original data, and the p-values of the first-order sequence were statistically significant at the 10% level or higher, thus rejecting the null hypothesis (i.e., the panel data are not stationary). This indicates that all variables are integrated in the order of one (i.e., I (1)).

4.4. The Impact of Abnormal Temperature on Rural Energy Poverty

The GMM, which includes the differential GMM (DIF-GMM) and systematic GMM (SYS-GMM), was utilized in this study. The study’s findings are presented in

Table 5. Empirical estimation results.

	(1)	(2)	(3)	(4)
	DIF-GMM	SYS-GMM	DIF-GMM	SYS-GMM
L.lnREPI	0.963 ***	0.967 ***	0.492 ***	0.728 ***
	(0.00644)	(0.0179)	(0.0163)	(0.0127)
AT	0.0723 ***	0.0834 ***	0.0469 ***	0.0553 ***
	(0.00179)	(0.00889)	(0.00441)	(0.00683)
lnURB			−0.297 ***	−0.0269
			(0.0218)	(0.0197)
lnAGRI			−0.00647	−0.00267
			(0.00921)	(0.0131)
lnPGDP			−0.0228 ***	−0.0431 ***
			(0.00455)	(0.00707)
Constant	−0.0802 ***	−0.0884 ***	1.231 ***	0.426 ***
	(0.00282)	(0.00819)	(0.0792)	(0.140)
Observations	570	600	570	600

Standard errors in parentheses. *** p < 0.01; ** p < 0.05; and * p < 0.1.

. The effect of abnormal temperature on rural energy poverty was negative and statistically significant, regardless of the GMM employed and whether or not control variables were included. This study directly regressed using the abnormal temperature defined in the previous section without taking the abnormal temperature’ natural logarithm. In this paper, the dependent variable is rural energy poverty, for which the natural logarithm was used in regression. Consequently, the following equation had to be utilized to interpret the coefficients: %$∆REPI=$ exp(${\beta }_1$ − 1). According to the estimates in the fourth column of Table 5, each standard deviation increase in abnormal temperature is associated with a 5.69% increase in rural energy poverty.

Abnormal temperature can decrease the amount of energy utilized in agriculture. In hot and humid climates, for instance, machinery may break down or lose performance, preventing farmers from using them effectively for agricultural production. Abnormal temperature may also impact the maintenance and upkeep of agricultural machinery by farmers. In extremely hot and humid climates, for instance, machinery may require more frequent servicing and maintenance to ensure its performance and dependability. Abnormal temperature is detrimental to the use of agricultural machinery, reducing the amount of energy used in agricultural production. Abnormal temperatures have a significant impact on the energy consumption patterns of homes. In order to maintain a comfortable indoor temperature during extreme weather, such as heat waves, households typically increase their energy consumption. This increased energy consumption, however, results in higher energy bills, which can push households into energy poverty. If households are unable to maintain a comfortable temperature, this can have an effect on their labor productivity, leading to a decrease in household income and an increase in the risk of energy poverty. During extreme weather events, households may be forced to use more traditional energy sources, exacerbating the negative health and environmental effects of energy poverty. Abnormal temperature can also have substantial economic effects on households. During extreme weather events, households may have to devote a significant portion of their income to energy costs, thereby reducing their spending on other necessities. This may result in a decline in global welfare and an increase in the poverty rate. Abnormal temperature can also have an effect on energy infrastructure and energy supply, resulting in an increase in energy prices and energy poverty. Rural energy infrastructure is relatively fragile, susceptible to damage from abnormal temperature, and difficult to repair and restore. Droughts with high temperatures can also reduce river flows and rural hydropower production. Combined, abnormal temperature can exacerbate rural energy poverty.

4.5. Threshold Effect Analysis

When studying the relationship between abnormal temperature and energy poverty, it is critical to consider the threshold effect of the urban–rural gap (GAP) and rural housing quality (RHQ) [29,30]. A lack of rural energy infrastructure, reliance on agricultural production, and a relative lack of opportunities for rural residents to enter urban non-farm employment are all factors contributing to the wide urban–rural gap. Due to a lack of energy infrastructure, rural energy supplies are unstable during abnormal temperature and may be unable to meet electricity demand. In rural areas, households may rely on traditional energy sources such as wood, charcoal, or animal waste. Households in urban areas, on the other hand, may have access to more efficient energy sources, such as electricity or natural gas. A large urban–rural divide means that rural areas are more reliant on agricultural production, making them more vulnerable to abnormal temperature, which reduces rural residents’ income and increases energy poverty. A large urban–rural divide implies that rural residents are less likely to earn off-farm income in urban areas and that off-farm employment is an important way for rural residents to adapt to and mitigate the negative effects of abnormal temperature. Rural residents may engage in non-farm employment, for example, to secure household income when abnormal temperature reduces agricultural production.

It is critical to consider the threshold effect of housing quality for rural residents, as rural households are more vulnerable to abnormal temperature due to poor housing quality. Due to inadequate insulation, some rural residents are reportedly forced to sleep until the early evening in the face of hot weather and are unable to move around indoors during the day due to excessive indoor heat. Furthermore, poor housing quality can have a negative impact on a home’s energy efficiency, leading to higher energy consumption and costs, exacerbating the risk of energy poverty. In this study, the threshold effect of housing quality refers to the level of poor housing quality required before the negative impact of abnormal temperature on energy poverty becomes significant. The proportion of rural dwelling areas with non-mixed structures is used as an indicator of housing quality in this study. A higher value indicates that rural housing quality is poor.

The role of urban–rural disparity and rural housing quality in the relationship between abnormal temperature and energy poverty has received scant attention. This research aims to fill this void. When examining the relationship between abnormal temperature and energy poverty, it is vital to consider the threshold effects of urban–rural disparity and rural housing quality. Understanding the role that urban–rural disparities and rural housing quality play in the relationship between abnormal temperature and energy poverty can assist in identifying the most vulnerable populations and informing the formulation of effective policy responses.

Figure 4 and Figure 5 depict the spatial distribution of the urban–rural divide and the quality of rural housing. The darker the pink color is, the greater the urban–rural divide is, and the darker the green color is, the lower the quality of rural housing is. The overall urban–rural divide is shrinking, and the quality of rural housing is improving. The inter-regional variability of the urban–rural gap is very visible, most notably in the relatively large urban–rural gap in the eastern and western regions. The urban–rural gap continues to narrow in the eastern coastal region, where urbanization is rapid and economic development is good; in the central region, where economic development is gradually accelerating and the urban–rural gap is improving markedly; and in the western region, where the urban–rural gap is relatively large, economic development is slower, infrastructure development is insufficient, and the urban–rural gap persists. The quality of rural housing also varies by region, with the southeastern coast showing higher housing quality across all years, the northern and southwestern regions in the past having lower rural housing quality, and the central and southwestern regions recently having lower rural housing quality.

In order to avoid taking the natural logarithm to form a nonlinear treatment of the original data and to interpret the results intuitively, the abnormal temperature, energy poverty, and threshold variables were analyzed without the natural logarithm, and the regression analysis was conducted directly. For estimation, the panel threshold effect model with a single threshold value of 2.657 for the rural–urban gap (GAP) and 0.238 for rural housing quality (RHQ) was used. The estimation results in

Table 6. Results of the threshold effect of the GAP in the nexus of AT and the REPI.

Variable	Coef.	Std. Err	95% Confidence Interval
AT·(GAP < 2.657)	−0.0146 **	0.00624	−0.0268	−0.00233
AT·(GAP ≥ 2.657)	0.0227 ***	0.00532	0.0122	0.0331
lnURB	−0.0964 ***	0.0336	−0.162	−0.0304
lnAGRI	0.0673 ***	0.0111	0.0453	0.0892
lnPGDP	−0.0351 ***	0.00849	−0.0518	−0.0184

*** p < 0.01; ** p < 0.05; * p < 0.1.

indicate that abnormal temperatures have a significant positive effect on energy poverty only when they exceed the GAP threshold value. By analyzing the data in conjunction with the specific data, it was found that about 56% of the GAP values were above the threshold in all provinces during the observation period. The estimation results presented in

Table 7. Results of the threshold effect of RHQ in the nexus of AT and the REPI.

Variable	Coef.	Std. Err	95% Confidence Interval
AT·(RHQ < 0.238)	−0.0324 ***	0.00818	−0.0485	−0.0163
AT·(RHQ ≥ 0.238)	0.0142 ***	0.00493	0.00453	0.0239
lnURB	−0.151 ***	0.0321	−0.214	−0.0880
lnAGRI	0.0710 ***	0.0111	0.0493	0.0927
lnPGDP	−0.0243 ***	0.00837	−0.0408	−0.00791

*** p < 0.01; ** p < 0.05; * p < 0.1.

indicate that abnormal temperatures have a significant positive effect on energy poverty only when they exceed the RHQ threshold value. By analyzing the data in conjunction with the specific data, it was found that about 84% of the RHQ values for each province were above the threshold during the observation period. According to Table 6 and Table 7, the effects of abnormal temperature on energy poverty are complex and nonlinear, with threshold effects for both urban–rural disparity and rural housing quality. The main conclusion of the threshold analysis is that abnormal temperature exacerbates energy poverty significantly in the presence of greater rural–urban disparity and lower rural housing quality. The effect of abnormal temperature on energy poverty is significantly attenuated when the urban–rural income gap is smaller and the quality of rural housing is higher.

5. Conclusions and Policy Implications

Our study yields four findings and conclusions. First, while rural energy poverty in China has been decreasing overall, regional disparities in rural energy poverty have not been significantly reduced. In terms of spatial distribution, the overall picture shows that energy poverty is more prevalent in the west and north. Warming is an undeniable fact, and more and more regions are warming and experiencing abnormal temperature. Second, multiple methods of estimation lead to the conclusion that abnormal temperature significantly increases rural energy poverty. When abnormal temperatures occur, rural energy poverty rises by 5.69%. Third, China’s urban–rural divide is narrowing, and rural housing quality is improving. However, the urban–rural divide and the quality of rural housing remain significant for different regions, reflecting the inequality of development between regions to some extent. Fourth, with thresholds of 2.657 and 0.238, respectively, the urban–rural gap and rural housing quality have significant threshold effects in the relationship between abnormal temperature and rural energy poverty. Abnormal temperature significantly increases rural energy poverty when the urban–rural divide and rural housing quality are above the thresholds. The threshold analysis results indicate that the relationship between abnormal temperature and rural energy poverty is nonlinear and that large urban–rural disparities and poor rural housing quality exacerbate rural energy poverty caused by abnormal temperature.

Based on the empirical results of our study, we propose several relevant policy implications. First, regional disparities in rural energy poverty should be reduced. Given the persistence of regional disparities in rural energy poverty, it is critical to implement policies that target the areas with the highest rates of energy poverty. These policies should aim to improve rural access to modern energy services such as electricity and clean cooking fuels. This can be accomplished by the following. (a) Investing in energy infrastructure: To improve access to modern energy services, governments should invest in the development of energy infrastructure in rural areas. This could include expanding the electricity grid and encouraging the use of renewable energy sources like solar power and wind. (b) Encouraging the use of energy-efficient technologies: governments should encourage the use of energy-efficient technologies in rural areas, such as clean stoves and energy-efficient lighting, to reduce energy consumption and improve affordability. (c) Targeted subsidies: governments can provide targeted subsidies to households in high-energy-poverty areas to assist them in affording modern energy services [93].

Second, the impact of abnormal temperature on rural energy poverty should be addressed. Given the significant impact of abnormal temperature on rural energy poverty, policies that address this issue are critical. These policies should prioritize improving rural energy efficiency and infrastructure, as well as increasing rural areas’ resilience to climate change. This can be accomplished by the following. (a) The encouragement of energy-efficient building design: To reduce energy consumption and improve resilience to abnormal temperature, governments should promote energy-efficient building design in rural areas. Insulation materials, shading devices, and passive cooling systems could all be used. (b) Increasing renewable energy use: Governments should increase the use of renewable energy sources in rural areas, such as solar power and wind, to increase resilience to abnormal temperature and reduce reliance on fossil fuels. Governments should pay particular attention to addressing energy poverty in the transition to clean energy [94,95]. (c) Climate adaptation strategies: the government should implement climate adaptation strategies in rural areas, such as developing early warning systems for extreme weather events and encouraging the cultivation of drought-tolerant crops.

Third, the government should work hard to bridge China’s rural–urban divide and continuously improve rural housing quality. Given China’s persistent urban–rural divide, which is still marked by significant regional differences, policies that promote equitable rural–urban development are critical, and additional policy measures targeting lagging regions will continue to be needed in the future in order to achieve more balanced regional development. While the quality of rural housing in China is gradually improving, there is still a need to focus on areas with low-quality housing, particularly in the southwestern and central countryside, and policy should focus on improvements in these areas. Rural housing quality should be improved, and the number of non-mixed rural dwellings should be reduced, as this is the foundation of rural livelihoods. Efforts should be directed toward the following areas. (a) Rural infrastructure investment: to improve rural living conditions, the government should invest in rural infrastructure such as housing, roads, and water systems. (b) Increasing urban acceptance of rural workers entering cities: efforts should be made to provide opportunities for farmers to work in cities, expand their opportunities for non-agricultural employment, ensure that rural residents have better access to urban public services, and reduce the barriers to rural labor mobility caused by the urban–rural dichotomy. (c) Education and training: the government should invest in rural residents’ education and training programs to improve their skills and increase their opportunities for non-agricultural employment and self-employment.

Fourth, areas and populations that are more sensitive and vulnerable to abnormal temperature, which can lead to rural energy poverty, should be concentrated on. Given the threshold effects of urban–rural disparities and rural housing quality on the relationship between abnormal temperature and rural energy poverty and the development of the former two in Chinese provinces, policies addressing these issues are critical. These policies should focus on improving rural housing quality, narrowing China’s rural–urban divide, and thus reducing the impact of abnormal temperature on rural energy poverty, as well as achieving synergies between reducing energy poverty and addressing climate change through the following. (a) Housing quality improvement: Rural residents should be assisted in improving their housing quality. Increasing houses’ insulation and energy efficiency should be concentrated on. Policies should focus on changing to cleaner energy sources. (b) Encouraging non-farm employment: The key to closing China’s urban–rural divide is to increase non-farm employment and to support rural residents who choose to migrate to cities. Higher urban incomes will increase their economic capacity to deal with abnormal temperature, lowering the risk of falling into energy poverty. (c) Relocation: For rural residents living in harsh natural environments, relocation is an important option. On the one hand, relocating to a habitable location lowers the likelihood of abnormal temperature occurring. Providing a better economic environment and opportunities, on the other hand, helps to increase non-agricultural income. As a result, it improves rural residents’ ability to cope with climate change and lowers the risk of energy poverty.

6. Discussion

While this study provides valuable insights into rural energy poverty in China, several limitations should be acknowledged to contextualize the findings and guide future research.

First, the data used in this study, although comprehensive, may not fully capture all dimensions of rural energy poverty. Our study lacks consideration of micro-level factors, such as household energy preferences. Additionally, cultural influences are not explicitly addressed. Future research could incorporate more granular data to deepen the understanding of the multidimensional nature of energy poverty.

Second, this study focuses primarily on abnormal temperature as a climate-related factor influencing rural energy poverty. However, other environmental variables, such as precipitation patterns, extreme weather events, and long-term climate trends, may also play significant roles. Exploring these factors in future research could provide a more holistic view of how climate variability impacts rural energy poverty.

Third, while the threshold analysis reveals significant nonlinear effects of urban–rural disparities and rural housing quality, the specific mechanisms underlying these threshold effects remain underexplored. Further qualitative research or case studies could complement these quantitative findings and offer a deeper understanding of how these thresholds interact with energy poverty dynamics.

Fourth, the policy implications suggested in this study are based on national- and provincial-level trends. However, rural energy poverty is inherently local and may vary significantly even within regions. Policies should consider the specific needs and characteristics of local communities, which requires further research to identify localized patterns and solutions.

Lastly, this study’s reliance on cross-sectional and panel data restricts the ability to draw causal inferences. While our findings indicate significant associations between abnormal temperature and rural energy poverty, experimental or longitudinal studies would be valuable to establish causality and explore dynamic changes over time.

Despite these limitations, this study contributes to the growing literature on energy poverty by highlighting its regional disparities, the impact of abnormal temperature, and the threshold effects of urban–rural disparities and rural housing quality. Future research addressing these limitations will enhance our understanding and inform more targeted policies to alleviate rural energy poverty in the context of climate change.