Research on the Effect of Clean Energy Technology Diffusion on Energy Poverty

Abstract

Addressing energy poverty is integral to the United Nations Millennium Development Goals, and clean energy technology serves as an essential tool in mitigating this issue. Existing studies pay less attention to the correlation between the two. In this study, we quantify the spread of clean energy technology using patent citation information and analyze balanced panel data from 30 provinces in China spanning the years 2004 to 2019. The findings reveal that the diffusion of clean energy technology not only directly reduces energy poverty but also indirectly alleviates it by enhancing energy efficiency and fostering employment. However, the positive impact of technology on poverty is affected by human capital. Specifically, when the number of college students per 10,000 people in a province surpasses 179, technology diffusion becomes effective in alleviating energy poverty. Furthermore, the efficacy of this mitigation varies significantly based on different regions. Technologies originating from developed economies such as the United States, Japan, and the European Union exhibit a more substantial impact than domestic alternatives. Additionally, the effect of alleviating energy poverty is more significant in the eastern region. Therefore, we make policy recommendations for alleviating energy poverty through the use of incentive policies, exploring differentiated models of clean energy technology development, and strengthening international cooperation.

1. Introduction

As one of the signs of poverty in developing countries, energy poverty is a high concern for international organizations such as the World Bank, the International Energy Agency (IEA), and the United Nations. Affordable, reliable, and sustainable modern energy supply has always been an essential goal of human development. Although the world has made significant progress in energy development, utilization, and popularization, regional and local energy shortages are still widespread. In a broad sense, energy poverty refers to the lack of adequate, affordable, efficient, and high-quality energy services to support local development, etc. According to the International Energy Agency (IEA) report, in 2019, the proportion of clean energy in final energy consumption accounted for only 17.7% [1]. As of 2020, a staggering 733 million people worldwide continue to grapple with unresolved electricity issues, while an alarming 2.4 billion individuals lack access to clean cooking fuel and technology, according to the International Energy Agency. Energy shortage and the resulting poverty [2], education [3], health [4], social equity [5], and ecological environment [6] are still essential co nstraints on social development in some developing countries and economically backward regions.

China is the largest developing country in the world, and it is also facing more severe and complex energy poverty. The research on energy poverty measurement based on the 10% indicator and LIHC indicators shows that the proportion of energy poverty in China is as high as 18.9%, and 46% of energy-poor households lack modern energy consumption [7]. At the same time, China’s total energy is abundant, but the distribution of resources is uneven. The per capita resources of coal, oil, and natural gas are only 1/2, 1/15, and 1/15 of the world average (China’s Energy Situation and Policy White Paper. (http://www.scio.gov.cn/zxbd/nd/2007/document/310015/310015_2.htm accessed on 15 May 2023). Although China achieved the goal of 100% electricity access by 2015, about 75% of rural households still use solid fuels such as wood and coal for cooking [8]. The use of traditional solid fuels poses a serious health hazard to the population, resulting in 10.79% of rural residents in China dying from respiratory diseases [9]. If rural residents can give up traditional biomass energy (e.g., wood, animal waste, and crop waste) and use clean energy, the probability of children and adults suffering from respiratory diseases will be reduced by 80% and 45%, respectively [10].

With the shortage of global fossil energy supply and the grim situation of climate change, countries around the world have realized the importance of developing clean energy. In particular, China is in a critical period of economic development and low-carbon transformation. The 20th National Congress of the Communist Party of China pointed out that promoting clean, low-carbon, and efficient energy use is necessary, as well as accelerating the planning and construction of a new energy system. In order to fulfill the goal of “achieving carbon peak by 2030 and carbon neutrality by 2060”, the share of coal energy in China’s allocation will drop from 60% in 2017 to 35% in 2040, and the use of energy will be even more intense by then. The diffusion of clean energy technology provides a huge opportunity to promote modern energy use and solve the energy shortage problem.

With the global diffusion of new technologies, adopting new technologies that include existing clean energy rather than inventing new technologies has become essential to improve performance and save time [11]. Therefore, the technology acceptance model has been widely discussed [12,13,14]. Studies have shown that adopting clean energy technologies has significantly improved energy consumption structure, ensured energy security, and reduced carbon emissions [15]. Scholars use the proportion of clean energy in total energy production to measure the promotion and use of clean energy, confirming that the development of clean energy can reduce the possibility of respondents falling into energy poverty [9]. Furthermore, the study makes it clear that, in an economy, the higher the adoption and penetration of digital technologies, the smaller the dependence on fossil energy and the demand for electricity generation, thus playing an active role in curbing energy poverty [16]. At the same time, some scholars have noticed the critical role of renewable energy technology innovation in energy poverty alleviation [17]. However, they have not considered the impact of technology diffusion. There is no empirical evidence regarding whether and how the diffusion and adoption of low-carbon technologies can effectively mitigate energy poverty.

Therefore, in light of the above theoretical analysis, this paper uses patent information on low-carbon technology and the energy poverty index based on macro-provincial data to empirically analyze the mechanisms of the diffusion of clean energy technology on China’s energy poverty. In contrast to prior studies, the following three points primarily represent the research contributions: first, this paper links the diffusion of energy technology and energy poverty for the first time and comprehensively analyzes the impact of clean energy technology diffusion on energy poverty reduction. Secondly, we delve deeply into the potential impact of energy efficiency, employment, and human capital on the diffusion of clean energy technology and its influence on energy poverty. This discussion significantly contributes to existing theories on the interplay between the diffusion of low-carbon technology and energy poverty. Furthermore, this paper offers crucial empirical evidence that can serve as a reference for the development and implementation of energy poverty alleviation policies in China. Thirdly, different from previous studies, which only use survey data to measure technology adoption, our paper introduces an innovative method using objective data derived from clean energy patent citation information. Patent citation information contains the record of latecomers learning from existing inventions, which is more scientific and reasonable. Moreover, our analysis goes beyond a broad perspective by identifying and categorizing the sources of technology diffusion and adoption into five regions: China, the United States, Europe, Japan, and South Korea. This categorization allows us to explore the varied effects of clean energy technology diffusion in different countries on the alleviation of energy poverty in China.

However, there are several research limitations. Due to constraints in data availability, our study is confined to the timeframe of 2014 to 2019, and updating data on energy poverty proves to be a complex task. To address this limitation, future research endeavors could extend the study period and incorporate more granular Chinese municipal data, thereby providing a more comprehensive and up-to-date analysis.

The remaining section of this article is structured as follows: Section 2 provides a summary of relevant literature. Section 3 puts forward the research hypothesis. Section 4 builds an empirical model and introduces the data. Section 5 analyzes the specific empirical results. Section 6 summarizes the conclusions and provides policy recommendations.

2. Literature Review

2.1. The Measurement of Energy Poverty

The measurement methods of energy poverty mainly include the single index method and the multidimensional energy poverty index method. Among them, the single index method mainly refers to using an indicator or measuring a specific aspect to evaluate the degree of energy poverty. The 10% indicator proposed by Boardman is the world’s earliest energy poverty measurement method [18]. If expenditure on household energy consumption exceeds 10% of household disposable income, the household is considered to be an energy-poor household. Subsequently, the 10% indicator was used to measure Britain’s energy poverty. Healy and Clinch [19] surveyed 1500 households in Ireland, and they found that the 10% index method cannot accurately measure the country’s energy poverty status due to the two countries’ economic development and inconsistent energy prices. Hills [20] defined high energy consumption and low-income levels as the Low-Income High Costs (LIHC) approach. Compared with the 10% indicator, this method excludes the influence of high-income and high-consumption groups. However, the Low-Income High Costs (LIHC) approach ignores all vulnerable groups, such as the elderly, children, and the disabled [21].

Therefore, more and more scholars consider energy poverty from a multidimensional perspective. Wang et al. [6] creatively constructed an energy poverty comprehensive evaluation system including nine indicators. It was also found that China’s energy poverty gradually eased in the 12 years from 2000 to 2011 and showed evident regional heterogeneity. Sadath and Acharya [22] used the multidimensional energy poverty index method to analyze the 2011 and 2012 Indian Human Development Survey (IHDS-II) data and found that energy poverty is widespread in India. Sokołowski et al. [23] considered five aspects of energy poverty and used the multidimensional energy poverty index method to calculate that 10% of Poland’s households are in energy poverty. Through the above research, it can be found that although many scholars have measured energy poverty, there is no uniform standard for the measurement of energy poverty.

2.2. The Impacts and Determinants of Energy Poverty

In recent years, there has been a growing focus on energy poverty, with research revealing its substantial adverse effects on economic growth, the environment, education, and health. Amin et al. [24] examined the relationship between energy poverty and economic development in seven Southeast Asian countries from 1995 to 2017. They found that energy poverty was negatively correlated with long-term and short-term economic growth in selected countries. Specifically, a 1% increase in energy poverty was found to be associated with a reduction in economic development of 29.81%. In addition, energy poverty has led rural women and children to spend much time collecting biomass fuel, and so are unable to carry out productive work and education [25]. At the same time, the use of traditional biomass energy increases carbon emissions [26] and aggravates air pollution [27]. In such an environment, the harm of energy poverty to residents’ physical [28] and psychological [29] is unavoidable

Therefore, how to accelerate energy poverty reduction has become one of the critical issues that the government and society need to solve urgently. Currently, relevant research mainly focuses on macro policy, energy structure transformation, and social culture. Ma et al. [30] found that strict environmental policies will lead to higher energy burdens for households using non-clean energy sources and have less impact on households using clean energy. Hamed and Peric [31] believed that using and developing renewable energy can alleviate energy poverty and promote sustainable environmental development. This view has also been supported by Adom et al. [32], Dong et al. [5], and Zhao et al. [33]. Ampofo and Mabefam [34] used the World Values Survey (WVS) data to measure religious beliefs in more than 100 countries, and the results showed that religious beliefs were positively correlated with energy poverty. Especially for families in developing countries and rural areas, religious activities exacerbate their energy poverty. In addition to the above factors, inclusive finance [35] and digital economy [36] also help to alleviate energy poverty.

2.3. The Role of Energy Technology Diffusion and Adoption in Alleviating Energy Poverty

As a follow-up process of technological innovation, technology diffusion and adoption have an essential impact on promoting economic growth. Previous studies have shown that most technologies originated in developed countries, and they were first adopted at home and then spread to countries with relatively backward economies [37]. On the one hand, the positive externalities of technology diffusion can improve the efficiency of resource allocation through introduction, absorption, and re-innovation, promote economic growth in poor areas, increase household income, and improve residents’ purchasing power of energy. On the other hand, technology diffusion can significantly improve the productivity of the clean sector. At the same time, improving productivity can reduce the cost of clean energy [38] and promote the use of clean energy by residents, thus alleviating energy poverty. In addition, Rogers [39] believed that innovative pioneers and early adopters would gain unexpected profits, thus widening the gap between the rich and the poor. Therefore, early adopters will become richer and richer while late adopters will benefit less from innovation. Based on the above theoretical analysis, energy technology diffusion may be one of the effective ways to alleviate energy poverty. However, no theoretical and empirical analysis of its energy poverty reduction effect exists.

3. Research Hypothesis

Clean energy technology has been considered as an effective means to solve the problems of modern energy security [40], environmental pollution [41], and energy shortage [17]. For example, the substitution effect of clean energy technology in the development process can improve the environment by reducing the use of traditional energy, reducing carbon emissions, and promoting the transformation of energy structure [42,43]. Furthermore, Liao [15] systematically analyzed 107 relevant literature studies and found that the adoption of clean energy technologies in low- and middle-income countries (LMICs) can significantly have a positive impact on household welfare (it includes education, environment, health, income, productivity, and profitability). At the same time, researchers have also found that the adoption of digital technology and other new technologies can reduce power generation and energy consumption, thereby alleviating energy poverty [16,44]. According to this logic, the diffusion of clean energy technology can alleviate energy poverty more directly. As a result, the hypothesis is as follows:

H1: 

The diffusion of clean energy technology can alleviate energy poverty.

The International Energy Agency (IEA) believes that enhancing energy efficiency stands out as the most cost-effective and environmentally friendly approach to fulfill basic energy needs. Improving energy efficiency, on the one hand, can produce energy-saving effects and reduce the primary energy demand of residents; on the other hand, it is conducive to saving energy production costs [45], reducing residents’ energy consumption expenditures, and reducing the burden of household energy consumption [46]. In addition, improving energy efficiency can accelerate energy infrastructure and promote clean energy consumption such as natural gas, solar, and wind energy to optimize the energy consumption structure [47,48] and improve the welfare of residents [49]. Technology diffusion can reduce research and development costs and improve energy efficiency by promoting technological innovation [50]. Therefore, one of the ways for the diffusion of clean energy technology to improve energy poverty may be the improvement of energy efficiency. As a result, the hypothesis is as follows:

H1a: 

The diffusion of clean energy technologies alleviates energy poverty by improving energy efficiency.

Previous studies have found that women in energy-poor households spend much time and cost on collecting biomass fuel [26], which deprives them of the opportunity to generate income for their families. Using clean energy has significantly reduced their time participating in housework and effectively promoted women’s employment [9]. At the same time, the diffusion of clean energy technology can narrow the gap in research and development levels between high-tech and low-tech companies and increase the employment of low-tech companies, thereby expanding employment opportunities for low-income people and alleviating households’ energy crises. In addition, the study of Koomson and Churchill [51] also found that employment precarity is positively related to energy poverty, and the resulting household financial difficulties will reduce the household’s ability to consume modern energy, such as lighting and cooking, thus aggravating energy poverty. Based on the research on employment and energy poverty and clean technology diffusion and employment, the diffusion of clean energy technology can alleviate energy poverty by promoting the employment of residents. As a result, the hypothesis is as follows:

H1b: 

The diffusion of clean energy technologies alleviates energy poverty by increasing employment.

Technological innovation and diffusion are the main driving force and source of a country’s economic growth, and the level of human capital is an essential factor affecting a country’s technological innovation and technological imitation and diffusion [52]. The role of human capital in promoting technology diffusion has been repeatedly verified [53,54]. For example, Gennaioli et al. [55] used survey data from 110 countries to find that human capital plays a vital role in regional innovation and dissemination. As a frontier low-carbon technology, clean energy technology has the characteristics of a multi-objective, complex knowledge source and being multidimensional. Typical low-carbon technological innovation requires setting multiple goals, such as production efficiency, quality, and environmental labeling [56,57]. Related to multi-objectives, the development of low-carbon products is a more complex task, often in the early stages of the product life cycle, needs to be away from the existing knowledge base of information and skills, and the source of ideas is more complex [58].

Moreover, owing to the multidimensional and systematic nature of low-carbon technological innovation, the significance of knowledge exchange in low-carbon technological innovation surpasses that in general technological innovation [59]. Therefore, clean energy technology generally has a higher entry threshold, and human capital can promote the technology spillover of foreign-funded enterprises to the host country, accelerate the speed of technology diffusion, and improve the host country’s ability to digest and absorb technology. As a result, the hypothesis is as follows:

H2: 

Human capital plays a threshold role in the energy poverty reduction effect of the diffusion of clean energy technology.

4. Empirical Model and Data Description

4.1. Model Specification

In order to verify H1, we use a two-way fixed effects model to discuss the impact of clean energy technology diffusion on energy poverty, as shown in Equation (1):

$$\ln \mathrm{E}{\mathrm{P}}_{\mathrm{i}\mathrm{t}}={\mathsf{\alpha }}_0+{\mathsf{\alpha }}_1\mathrm{C}\mathrm{E}{\mathrm{T}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\alpha }}_{\mathrm{c}}{\mathrm{X}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\delta }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$$

(1)

where the dependent variable EP_it represents the energy poverty status of province i in year t, the core independent variable CET_it is the diffusion level of clean energy technology province i in year t, α1 represents the impact of clean energy technology diffusion on energy poverty, X_it represents the control variable, μ_i represents the individual fixed effect of province i that does not change with time, δ_i represents the fixed effect of control time, and ε_it represents the stochastic error term.

In order to verify Hypothesis 1a and Hypothesis 1b, namely to explore the possible mechanism of the diffusion of clean energy technology for energy poverty, according to the above, we draw on the practice of Zhao et al. [26] and Hong [9], where energy efficiency and employment are its mediating variables, as shown in Equations (2) and (3).

$${\mathrm{M}}_{\mathrm{i}\mathrm{t}}={\mathsf{\beta }}_0+{\mathsf{\beta }}_1\mathrm{C}\mathrm{E}{\mathrm{T}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_{\mathrm{c}}{\mathrm{X}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\delta }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$$

(2)

$$\ln \mathrm{E}{\mathrm{P}}_{\mathrm{i}\mathrm{t}}={\mathsf{\gamma }}_0+{\mathsf{\gamma }}_1\mathrm{C}\mathrm{E}{\mathrm{T}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\gamma }}_2\mathrm{E}{\mathrm{I}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\gamma }}_{\mathrm{c}}{\mathrm{X}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\delta }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}\mathrm{}}$$

(3)

where M_it is the mediating variable, including the energy efficiency and employment of province i in year t. In addition, in order to verify Hypothesis 2, based on Equation (1), we take the single threshold as an example to construct the following model:

$$\ln \mathrm{E}{\mathrm{P}}_{\mathrm{i}\mathrm{t}}={\mathsf{\varphi }}_0+{\mathsf{\varphi }}_1\mathrm{C}\mathrm{E}{\mathrm{T}}_{\mathrm{i}\mathrm{t}}\times \mathrm{I}(\mathrm{H}\mathrm{U}\mathrm{M}\mathrm{A}\mathrm{N}\le \mathsf{\phi })+{\mathsf{\varphi }}_2\mathrm{C}\mathrm{E}{\mathrm{T}}_{\mathrm{i}\mathrm{t}}\times \mathrm{I}(\mathrm{H}\mathrm{U}\mathrm{A}\mathrm{M}\mathrm{N}>\mathsf{\phi })+{\mathsf{\alpha }}_{\mathrm{c}}{\mathrm{X}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\delta }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$$

(4)

where human is the threshold variable, which is human capital, I (·) is the indicator function, and φ is the specific threshold value.

4.2. Data Description

4.2.1. Dependent Variable

Building upon the studies conducted on the research of Dong et al. [5] and Zhao et al. [26], this paper establishes a comprehensive energy poverty index composed of energy service availability, energy consumption cleanliness, energy management completeness, household energy affordability, and energy efficiency. In order to compare the differences in energy poverty in various regions of China, we plotted a provincial geographical distribution map of energy poverty (see Figure 1, Figure 2, Figure 3 and Figure 4). Overall, China’s energy poverty has been alleviated in recent years. Nevertheless, a distinct pattern emerges, with lower values observed in the southern and eastern regions while higher values are evident in the northern and western areas.

4.2.2. Independent Variable

Refer to the patent classification methods of Aghion et al. [60] and Dechezleprêtre [61], use the incoPat Global Patent Database to query and count the number of patent citations of clean energy technologies as shown in

Table 1. Patent classification of clean energy.

Patent Classification	Cooperative Patent Classification
Clean energy patents	B60K1; B60L3; B60L7; B60L11; B60L15; B60R16; B60S5; B60W10; B60W20; H01M; H01J61; H05B33; F21K9; E02B9/08; F03D; F03G4; F03G6; F03G7/05; F24J2; F24J3/08; F26B3/28

, and use the number of authorized patent citations per 10,000 people as a measure of technology diffusion. In Figure 1, Figure 2, Figure 3 and Figure 4, a noticeable trend emerges as the level of technology diffusion in each province steadily rises over the years, indicating a pattern of greater diffusion in the eastern and lesser diffusion in the western regions.

4.2.3. Control Variable

Economic growth has provided sufficient financial support for residents to use modern energy [5]. Hence, the per capita GDP of each province is chosen, and the logarithm is applied to gauge the level of economic development (PGDP). Foreign investment and industrial structure upgrading can provide favorable conditions for the popularization and use of clean energy through the rational allocation of resources [35]. Therefore, the proportion of the secondary industry’s output value to the tertiary industry’s output value is chosen, and the logarithm is employed to quantify the industrial structure (ISR). Select the proportion of the total investment of foreign-invested enterprises in GDP and take the logarithm to measure the level of foreign investment (OPEN). The rapid development of urbanization and the improvement in energy infrastructure have laid the foundation for popularizing and using clean energy [9,35]. Therefore, the proportion of the urban residents and total populations is selected and paired to measure the number of urbanization (URB). The urban per capita road area is chosen, and the logarithm is applied to quantify the urban road area (ROAD). The specific descriptive statistics are shown in

Table 2. Descriptive statistics of variables.

Variable	Meaning of Variable	N	Mean	sd	Min	Max
EP	Energy poverty composite index	480	−0.818	0.302	−1.604	−0.153
CET	Diffusion of clean energy technology	480	0.239	0.566	0	5.347
lnPGDP	Per capita gross domestic product	480	1.184	0.699	−0.864	2.799
lnISU	The ratio of tertiary industry to secondary industry output value	480	−0.0530	0.392	−0.704	1.643
lnOPEN	The ratio of foreign direct investment to GDP	480	−3.198	1.837	−9.142	0.162
lnURB	The ratio of the urban population to the total population	480	−0.654	0.263	−1.478	−0.0640
lnROAD	Urban road area per capita	480	2.556	0.368	1.396	3.266
lnEI	The ratio of GDP to total energy consumption	480	0.119	0.561	−1.464	1.570
lnJOB	Quantity of employment	480	7.574	0.783	5.838	8.653
lnHUMAN	Number of college students per 10,000 people	480	5.098	0.377	3.830	5.876

.

4.2.4. Mechanism Variables and Threshold Variables

This paper selects energy efficiency (EI) and employment (JOB) as intermediary variables and human capital (HUMAN) as threshold variables. Referring to Duro et al. [62] and Dong et al. [5], the reciprocal energy intensity, the ratio of GDP to total energy consumption in a province, is used to evaluate energy efficiency. The number of employees in each province measures employment. The number of college students measures human capital. The data above sources are derived from the China Statistical Yearbook and the China Energy Statistical Yearbook.

5. Results and Discussion

5.1. Basic Regression Analysis

This paper explores the impact of the diffusion of clean energy technology on energy poverty. We use a two-way fixed effects model to perform basic regression, with the diffusion of clean energy technology and energy poverty as independent and dependent variables.

Table 3. Basic regression results.

Variables	(1)	(2)
CET	−0.101 ***	−0.092 ***
	(−5.79)	(−4.82)
lnPGDP		−0.056
		(−1.04)
lnISU		−0.111 ***
		(−2.73)
lnOPEN		−0.060 ***
		(−7.88)
lnURB		0.162
		(1.59)
lnROAD		0.001
		(−0.00)
Control_pro	Yes	Yes
Control_year	Yes	Yes
Constant	−1.048 ***	−0.900 ***
	(−29.55)	(−7.79)
Observations	480	480
R-squared	0.901	0.915

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

presents the outcomes, with Model (1) showing results without control variables and Model (2) incorporating control variables for a more comprehensive analysis.

In Model (2), the estimation results reveal a significantly negative coefficient for the diffusion of clean energy technology at the 5% significance level, indicating that the widespread adoption of clean energy technology can effectively mitigate the occurrence of energy poverty, thereby confirming Hypothesis 1. In alignment with the research findings that digital technology adoption, renewable energy technology innovation, and clean energy development have a significant inhibitory effect on energy poverty [9,16,17], the diffusion of clean energy technology is an essential means to alleviate energy poverty.

On the one hand, the diffusion of clean energy technology reduces the production costs associated with clean energy, increases the market’s supply of clean energy, optimizes the energy consumption structure, and diminishes residents’ reliance on fossil fuels. Simultaneously, the swift advancement of clean energy technology and improved clean energy infrastructure have effectively reduced household energy demand, reduced energy consumption expenditure, and alleviated energy poverty. On the other hand, using clean energy technologies in industrial production and urban sewage treatment reduces carbon emissions. It transforms organic waste into clean energy, such as electricity and gas, contributing to environmental sustainability and alleviating the pressure on energy usage, thereby reducing energy poverty [35].

5.2. Robustness Test

A robustness test was conducted to ensure the robustness of the regression results in this study.

Firstly, we augmented the number of control variables. Previous research has indicated that technological progress can alleviate energy poverty [5]. Therefore, this paper introduced technological progress as a control variable, measured using patent authorization. In

Table 4. Robustness test results.

Variables	(1)	(2)	(3)
CET	−0.089 ***	−0.561 ***	−0.153 **
	(−4.62)	(−3.74)	(−2.47)
Control variables	Yes	Yes	Yes
Control_pro	Yes	Yes	Yes
Control_year	Yes	Yes	Yes
Constant	−0.508	0.091	−0.919 ***
	(−2.18)	(0.54)	(−7.01)
Observations	480	416	360
R-squared	0.916	0.940	0.923

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

, Model (1) results show that the estimated outcomes of the core explanatory variables remain consistent in the direction of influence compared to the benchmark regression. The changes observed are solely in terms of significance levels, validating the robustness of the regression results.

In order to examine potential multicollinearity between variables, a variance inflation factor (VIF) test was conducted, and the results are presented in

Table 5. The results of the VIF test.

Variables	VIF	1/VIF
lnGDP	9.39	0.106
lnURB	6.14	0.163
lnTEC	3.17	0.315
lnEDU	2.68	0.373
lnOPEN	2.16	0.463
lnROAD	2.09	0.477
CET	2.04	0.489
lnISU	1.59	0.627
Mean VIF	3.66

. The VIF for each variable is below 10, indicating the absence of serious multicollinearity among the explanatory variables; this further reinforces the reliability of the regression results.

Secondly, the potential impact of municipalities was eliminated given the varying levels of technological development and energy endowments across different regions of China, which may affect the mitigating effect of clean energy technology diffusion on energy poverty; data from Beijing, Tianjin, Shanghai, and Chongqing were excluded from the total sample and subsequently reintegrated. The regression outcomes from Model (2) in Table 4 still uphold Hypothesis 1; this underscores the consistency and robustness of the findings, even when considering regional variations.

Thirdly, we accounted for the influence of other exogenous policies. Given the introduction of the “Energy Technology Revolution and Innovation Action Plan” and the “13th Five-Year Plan for Energy Technology Innovation” by China in 2016 to foster an energy technology revolution and harness the pivotal role of energy technology innovation in establishing a clean, low-carbon, safe, and efficient modern energy system, the implementation of these policies might potentially impact the effectiveness of clean energy technology diffusion in alleviating energy poverty. In order to isolate the specific impact of this policy, this study excluded sample data from the years 2016 to 2019 and reanalyzed the results. Model (3) reveals that the coefficient for the diffusion of clean energy technology has increased and remains significantly negative. This outcome indicates that, even after excluding the influence of these specific policies, the diffusion of clean energy technology continues to alleviate energy poverty effectively, further reinforcing the robustness of the conclusions drawn in this paper.

5.3. Endogeneity Analysis

Given the potential presence of reverse causality and time lag effects between adopting clean energy technology and energy poverty, our model may exhibit endogenous bias. To uphold the research’s integrity and ensure the resilience of our conclusions, we addressed this issue by employing lag phase I and lag phase II of clean energy technology as instrumental variables. We utilized a two-stage least squares estimation model to handle endogeneity.

The estimation results are presented in

Table 6. Endogenous test results.

Variables	(1)	(2)
CETi-1	0.793 ***
	(8.91)
CETi-2	0. 286 **
	(2.22)
CET		−0.171 ***
		(−3.79)
Control variables	Yes	Yes
Control_pro	Yes	Yes
Control_year	Yes	Yes
Constant	0.165	−1.337
	(0.86)	(−3.65)
Observations	450	450
F	378.391
P_sargan test		0.581
R-squared	0.926	0.686

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

, with Models (1) and (2) representing the regression outcomes for the first and second stages, respectively. The significance of estimates for both the instrumental and core explanatory variables is noteworthy, with all p values below 0.05 suggesting a strong explanatory power of the instrumental variables for the endogenous variables. Additionally, the F-value exceeding 10 indicates that the instrumental variables meet relevant conditions and that there is no problem with weak instrumental variables.

Furthermore, the p value associated with the Sargan test in the second stage is greater than 0.01, leading to the non-rejection of the null hypothesis. This outcome implies that the instrumental variable is exogenous, affirming its reasonableness and effectiveness. In essence, the selected instrumental variable does not exhibit over-identification issues. After scrutinizing the regression results of Model (2) and addressing the endogeneity problem, the estimated coefficient for clean energy technology diffusion expands and remains statistically significant. This persistence in significance underscores the robustness of the estimated results in this paper, indicating that the model effectively mitigated endogeneity concerns and produced reliable findings.

5.4. Mechanism Test

In order to further explore the internal mechanisms influencing the relationship between the diffusion of clean energy technology and energy poverty, this study employs a step-based regression method to examine whether the diffusion of clean energy technology can alleviate energy poverty by enhancing energy efficiency and employment. The outcomes of this analysis are detailed in

Table 7. Mechanism analysis results.

	(1)	(2)	(3)	(4)	(5)	(6)
Variables	EP	EI	EP	EP	JOB	EP
CET	−0.092 ***	0.116 ***	−0.075 ***	−0.092 ***	0.048 **	−0.080 ***
	(−4.82)	(6.58)	(−3.78)	(−4.82)	(3.02)	(−4.23)
EI			−0.147 ***
			(−2.82)
JOB						−0.262 ***
						(−4.65)
Control variables	Yes	Yes	Yes	Yes	Yes	Yes
Control_pro	Yes	Yes	Yes	Yes	Yes	Yes
Control_year	Yes	Yes	Yes	Yes	Yes	Yes
Constant	−0.900 ***	−0.534 ***	−0.979 ***	−0.900 ***	6.894 ***	0.904 *
	(−7.79)	(−5.04)	(−8.30)	(−7.79)	(71.69)	(2.23)
Observations	480	480	480	480	480	480
R-squared	0.915	0.979	0.917	0.915	0.990	0.920

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

. Specifically, Models (1)–(3) and (4)–(6) present the results of the mediating effect test, incorporating energy efficiency and employment as mediating variables.

In Table 7, Model (2) and Model (5) reveal that the coefficients associated with the diffusion of clean energy technology are 0.116 and 0.048, respectively. Importantly, these coefficients are statistically significant at the 5% level, indicating a positive and meaningful impact of the diffusion of clean energy technology on energy efficiency and employment.

The findings from Model (3) and Model (6) show regression coefficients for the diffusion of clean energy technology, with −0.075 and −0.080, respectively. These coefficients are statistically significant at the 1% level. Furthermore, the coefficients for energy efficiency and employment are −0.147 and −0.262, respectively, both significant at the 1% level. This indicates that the diffusion of clean energy technology effectively alleviates energy poverty by enhancing energy efficiency and fostering employment. The robust and significant nature of these results supports the validation of Hypothesis 1a and Hypothesis 1b. Specifically, the diffusion of clean energy technology demonstrates a dual impact: firstly, by improving energy efficiency, it reduces energy consumption and associated expenditure, thereby mitigating energy poverty. Secondly, creating employment opportunities enhances residents’ disposable income, induces a shift in energy consumption patterns, and ultimately inhibits the occurrence of energy poverty.

5.5. Heterogeneity Analysis

Considering the differences between the sources of diffusion and adoption of clean energy technology and regional resource endowments, the diffusion of clean energy technologies may have significant heterogeneity for energy poverty reduction. Therefore, we analyze the heterogeneity from two perspectives: diffusion and adoption sources and geographical regions.

Firstly, this paper divides the sources of technology diffusion and adoption into five regions: China, the United States, Europe, Japan, and South Korea, and conducts heterogeneity analysis. The results are shown in

Table 8. Heterogeneity results of patent citation sources.

Variables	(1)	(2)	(3)	(4)	(5)
CHN	−0.126 ***
	(−4.76)
USA		−0.529 ***
		(−4.27)
EU			−3.808 ***
			(−4.51)
JPN				−1.381 ***
				(−4.37)
KR					−4.465 ***
					(−4.30)
Control variables	Yes	Yes	Yes	Yes	Yes
Control_pro	Yes	Yes	Yes	Yes	Yes
Control_year	Yes	Yes	Yes	Yes	Yes
Constant	−0.918 ***	−0.852 ***	−0.900 ***	−0.882 ***	−0.914 ***
	(−7.97)	(−7.23)	(−7.76)	(−7.56)	(−7.88)
Observations	480	480	480	480	480
R-squared	0.915	0.914	0.915	0.915	0.915

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

. We found that clean energy technologies in these five countries can significantly reduce China’s energy poverty. However, foreign clean energy technologies have a better effect on energy poverty reduction, especially those from the United States, South Korea, and Europe. The possible reason is that, with the development of international trade, foreign direct investment, and patent licensing, compared with local research and development, developing countries’ absorption of international advanced technology diffusion is a low-cost way of technological change. This observation aligns with the findings of Sun et al. (2021) [63]. Foreign energy technology innovation is more robust in improving domestic energy efficiency than domestic energy technology innovation. An illustrative example is evident in the Netherlands, where the main driving force for energy efficiency improvement stems from the technological advancements originating in the United States and Germany.

Secondly, according to the province’s geographical location, the sample is divided into eastern, central, and western regions, and group regression is performed as shown in

Table 9. Regional and income heterogeneity analysis results.

	(1)	(2)	(3)
Variables	Eastern	Central	Western
CET	−0.080 ***	0.044	−0.268
	(−3.30)	(0.47)	(−1.55)
Control variables	Yes	Yes	Yes
Control_pro	Yes	Yes	Yes
Control_year	Yes	Yes	Yes
Constant	−0.383 *	0.194	−0.139
	(−1.70)	(0.82)	(−0.47)
Observations	176	128	176
R-squared	0.894	0.932	0.902

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

. The coefficient of diffusion of clean energy technology is only significantly negative in the eastern region, indicating that the diffusion of clean energy technology in the eastern region can alleviate energy poverty. However, the positive impact on energy poverty reduction is not evident in the central and western regions. This discrepancy underscores that the elevated economic development, abundant human capital, and well-established energy infrastructure in the eastern region play a pivotal role in enhancing the efficacy of clean energy technology and that the higher level of economic development, higher human capital, and complete energy infrastructure in the eastern region are conducive to amplifying the effect of the diffusion of clean energy technology on energy poverty reduction.

In China, the eastern region is composed of 11 provinces: Beijing, Tianjin, Hebei, Liaoning, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, Hainan, and Guangdong. The central region includes eight provinces: Shanxi, Jilin, Heilongjiang, Anhui, Jiangxi, Henan, Hubei, and Hunan. The western region is composed of 11 provinces: Inner Mongolia, Guangxi, Chongqing, Sichuan, Guizhou, Yunnan, Shaanxi, Gansu, Qinghai, and Ningxia

5.6. Panel Threshold Model Analysis

As an effective carrier of knowledge flow, human capital determines a country’s ability to attract technology and affects its technology catch-up and diffusion speed. Therefore, we select human capital as the threshold variable.

Initially, we assess whether the model exhibits threshold utility and proceed to test single, double, and triple thresholds consecutively. The outcomes are presented in

Table 10. Threshold effect test results.

Threshold	F	P	Crit10	Crit5	Crit1
Single threshold	36.52	0.013	24.881	29.733	36.907
Double threshold	12.94	0.123	13.783	16.524	23.285
Triple threshold	13.02	0.377	25.852	31.528	41.212

Note: Robust t-statistics in parentheses; Crit10, Crit5, and Crit1 indicate the critical value levels at 10%, 5%, and 1%, respectively.

. The p-value associated with the double threshold for clean energy technology diffusion is 0.123, rendering the result insignificant. Conversely, the single threshold is significant at the 5% level, indicating the presence of a singular threshold. The identified threshold value for this single threshold is 5.187, representing the number of 179 college students per 10,000 people in each province.

Secondly, adhering to the principles of the threshold model, the estimated threshold value is determined as the γ value corresponding to the likelihood ratio statistic LR approaching 0. Figure 5 illustrates the likelihood ratio function diagram for the single threshold estimation value of 5.187 within the 95% confidence interval. On the diagram, the actual threshold value corresponds to the lowest point of the LR statistic, as depicted by the dotted line. Notably, the critical value is marked at 7.35. Given that this critical value is considerably higher than the threshold value of 5.187, the authenticity and effectiveness of the estimation results are substantiated.

Finally, having established the threshold value, we proceed with panel threshold regression (

Table 11. Panel threshold regression results.

Variables	Panel Threshold Model
	Coef.	t-Statistic
CET (lnHUMAN ≤ 5.187)	0.235 ***	4.35
CET (lnHUMAN > 5.187)	−0.074 ***	−3.98
Constant	−0.519 **	−3.17
N	480	480
Control	Yes
R2	0.514

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

). The results indicate that when human capital falls below the threshold value of 5.187, the coefficient reflecting the impact of clean energy technology diffusion on energy poverty is 0.235, significant at the 1% level. When human capital exceeds the threshold of 5.187, the impact coefficient transitions from positive to negative, registering at −0.074 and remaining significant at the 1% level. These findings underscore significant variations in the influence of clean energy technology knowledge diffusion on energy poverty at different levels of human capital. The observed effect of clean energy technology on energy poverty reduction only manifests when human capital surpasses a singular threshold. It is shown that there are significant differences in the impact of knowledge diffusion of clean energy technology on energy poverty under different human capital levels. These findings align with the results that Akhvlediani and Cieślik (2020) reported [53], emphasizing the importance of human capital in the clean energy technology diffusion process for alleviating energy poverty. This shows that fostering high-tech talent in the future energy sector and promoting human capital accumulation are crucial endeavors. These efforts are crucial for realizing the full potential of clean energy technology diffusion in mitigating energy poverty.

6. Conclusions and Policy Implications

6.1. Conclusion

In our investigation into the potential of the diffusion of clean energy technology to alleviate energy poverty, we employ the multidimensional energy poverty index method to assess the energy poverty index across 30 provinces in China from 2004 to 2019. Additionally, we integrate patent citation information to conduct a systematic analysis, providing a first-of-its-kind exploration into the impact and underlying mechanisms of clean energy technology diffusion on energy poverty. The key findings can be summarized as follows.

Firstly, our study affirms that the diffusion of clean energy technology can effectively reduce energy poverty. It aligns with prior studies highlighting the pivotal role of digital technology adoption, innovation in renewable energy technology, and the overall advancement in clean energy in mitigating energy poverty [9,16,17]. The diffusion of clean energy technology demonstrates its efficacy in alleviating energy poverty by transforming the energy consumption landscape. This shift from traditional, high-carbon energy sources reduces residents’ reliance on such sources, contributing significantly to achieving emission reduction targets. The positive impact is evident in the improved energy consumption structure facilitated by clean energy technology development.

Additionally, our investigation underscores energy efficiency and employment’s crucial role as an intermediary between energy technology diffusion and energy poverty by exerting energy-saving and income growth effects, respectively. Mechanism analysis shows that, on the one hand, the diffusion of clean energy technology can improve energy efficiency, produce energy-saving effects, reduce the primary energy demand of residents, and thus alleviate energy poverty. On the other hand, the diffusion of clean energy technology can promote employment, increase the disposable income of residents, enhance the purchasing power of clean energy, improve the energy consumption structure, and thus inhibit the occurrence of energy poverty. The threshold model indicates that only when the number of college students per 10,000 people in each province is higher than 179 can the diffusion of clean energy technology achieve the effect of curbing energy poverty. Corresponding to the high entry threshold of clean energy technology, the development of the clean energy industry depends on the development of human capital to a certain extent. With the improvement in human capital level, technology absorptive capacity can amplify the inhibitory effect of clean energy technology diffusion on energy poverty.

Finally, the inhibitory effect of the diffusion of clean energy technology on energy poverty varies significantly due to technical and regional differences. The analysis of the mitigation effect of clean energy technology diffusion from different countries on China’s energy poverty shows that foreign clean energy technologies have more robust energy poverty reduction capabilities. The reason is that countries such as the United States, Japan, and the European Union hold the critical technologies of clean energy in the world, while China’s clean energy technology is in the stage of introduction, digestion, and joint design and lacks independent intellectual property rights, so it has no advantage in alleviating energy poverty. At the same time, the energy poverty reduction effect of clean energy technology diffusion is more significant in the eastern area, indicating that higher economic development may amplify the effect of clean energy technology diffusion on energy poverty reduction.

6.2. Policy Implications

Based on the above conclusions, we propose the following policy implications.

Firstly, the government can leverage diverse incentive policies to harness the potential of clean energy technology in mitigating energy poverty. Examples include incentivizing the use of clean and efficient energy sources such as biogas, solar energy, and natural gas within households. Support for household rooftop photovoltaic installations can enhance energy efficiency. Coordinating subsidy policies with employment strategies, such as encouraging rural youth participation in rooftop photovoltaic services, can generate additional employment opportunities. This synergistic approach aims to amplify the positive impact of clean energy technologies by boosting the income of impoverished households.

Secondly, a different incentive approach should be adopted, tailoring it to local conditions. Considering the current energy poverty landscape, a differentiated development model for clean energy technology should be explored based on regional characteristics. In regions with minimal energy poverty, such as the economically developed provinces in the eastern region, leveraging economic, technological, talent, and geographical advantages can further enhance the energy poverty reduction effects of clean energy technology diffusion. Conversely, in the central and western regions grappling with severe energy poverty, capitalizing on energy endowments and policy support can expedite the construction of clean energy infrastructure, ultimately improving residents’ energy efficiency.

Thirdly, the use of clean energy technologies in developed countries such as Europe and the United States is an effective means to alleviate China’s energy poverty, and the government and enterprises should maintain an open approach. Actively introducing foreign investment and fostering collaboration with foreign-funded enterprises will facilitate the diffusion of advanced clean energy technologies from abroad, contributing to the ongoing efforts to address energy poverty in China.