Advancing Equity in China’s Vehicle Electrification

Abstract

Vehicle electrification should not be limited to specific geographical areas like wealthy or privileged regions. Recognizing this, China initiated the “NEVs going to rural areas” (NEVgoRural) initiative, highlighting the need for systematic metrics to quantify equity progress. This study develops a comprehensive analysis framework, including equality degree calculation, prevalence analysis, and correlation analysis, to assess and track China’s vehicle electrification process from 2020 to 2023. The equality degree calculation roughly evaluated the spatial inequality degree across China at the macro level. The results indicated that equality issues exist in China (Gini > 0.4, World Bank warning), and this inequality has been gradually improving year by year. The prevalence analysis, based on an Information Entropy Weighting-based model, identified specific geographical areas lagging behind. Furthermore, a correlation analysis was conducted to explore potential associations underlying this geographical inequity. The results revealed that variables, such as a high-level education ratio, NEV manufacturing company number, etc., were significantly correlated with the inequity situation. Drawing on these insights, this study proposed targeted strategies to enhance electrification equity across regions, such as extending NEV tax exemptions in lagging provinces, implementing rural-specific zero-down-payment and interest-free plans, establishing a robust rural sales and after-sales network, and improving mobile payment skills in rural areas.

1. Introduction

Transportation stands as a significant contributor to greenhouse gas emissions and air pollution [1]. Scaling up New Energy Vehicles (NEVs) is crucial for environmental sustainability. In response, governments worldwide have implemented various regulations and incentives to foster the shift to NEVs. As shown in Figure 1, China’s NEV sales, including both battery electric vehicles (BEVs) and plug-in hybrid electric vehicles (PHEVs), has led the world continuously for 9 years [2], accounting for 59% of global sales in 2023, surpassing Europe (23%) and North America (12%). The rapid growth could be viewed as a success in China’s efforts to reduce oil dependency and cut emissions.

Equity in transportation and mobility systems has long concerned planners. Disparity in vehicle electrification leaves certain spatial regions without the benefits of NEVs, such as lower operating costs, government subsidies, and improved air quality. Addressing geographical equity concerns is crucial not just for individual benefits but also for fostering a more inclusive and sustainable transition to electrified transportation.

Some countries have recognized the importance of equity in vehicle electrification and have incorporated equity into the formulation of relevant policies. In the U.S., some states in the New England region have implemented incentive programs that allocate benefits specifically to low- and moderate-income residents through a USD 2.5 million (approximately CNY 18 million) program [3]. Furthermore, the California Air Resources Board [4] has sanctioned a USD 533 million (approximately CNY 3.84 billion) initiative to facilitate rebates for environmentally friendly vehicles, with eligibility restricted to models priced below USD 60,000 (approximately CNY 432,000). In the European Union, the Social Climate Fund aims to mobilize at least EUR 86.7 billion (approximately CNY 707.2 billion) from 2026 to 2032, including subsidies to help low-income households transition to cleaner transport options [5]. In China, vehicle electrification has demonstrated potential for improving equity through the availability of affordable NEV models tailored to lower-income consumers [6]. A key example is the success of the Hongguang Mini-EV, priced at only USD 6563 (approximately CNY 50,000), which became the best-selling electric passenger vehicle in China in both 2021 and 2022 [7]. A survey by Wuling Motors of over 4000 Mini-EV owners found that 53% of the owners have a household monthly income of below CNY 14,000 (approximately USD 1800), highlighting its broad appeal among low- and middle-income groups. To further improve vehicle electrification equity, the Chinese government started to launch the “NEVs going to rural areas” (NEVgoRural) policy in 2020, aiming to extend the reach of NEVs beyond urban areas and into rural regions [8].

However, the inequity of vehicle electrification is still significant among different provinces and between urban and rural areas in China. A core challenge is the lack of consensus in measuring vehicle electrification equity. The objective of this study is to generate data-based insights to promote greater geographical equity in China’s vehicle electrification process, thereby preventing vulnerable groups from being left behind. A comprehensive analysis framework, comprising equality degree calculation, prevalence analysis, and correlation analysis, is developed to assess and track China’s vehicle electrification process from 2020 to 2023. Firstly, the Gini index and Theil index are used to quantify the geographical inequality degree over time. After that, a multi-criteria evaluation model is used to provide quantitative insights into the specific provinces that are lagging in China. Furthermore, the correlation relationships between vehicle electrification and 12 factors are analyzed to uncover underlying patterns and influences. These methodological tools are designed to help answer three key practical questions that are critical for policymakers and stakeholders, including (1) what distribution of NEV market penetration is equitable; (2) how much public resources are required for the under-electrified regions; and (3) which region should receive a higher priority for the support of the vehicle electrification. Answering these questions can help coordinate efforts of NEV charging companies, policymakers, regulators, and other stakeholders involved in promoting inclusive vehicle electrification.

The remainder of this paper is structured as follows: Section 2 reviews the literature on vehicle electrification equity. Section 3 presents the results of the Gini index and Theil index over years, the outcomes derived from prevalence analysis, and the findings of correlation analysis, shedding light on China’s vehicle electrification spatial equity landscape. Section 4 concludes and discusses the policy implications.

2. Literature Review

Many studies have incorporated equity considerations to promote more inclusive and sustainable development of transportation systems. For instance, Liu and Meidani [9] proposed a graph neural network (GNN)-based retrofit strategy that enhances post-disaster accessibility for low-income populations, promoting equity in transport resilience. Li and Zhou [10] introduced an urban logistics resilience framework that simultaneously optimizes for equity and efficiency, offering a quantitative tool to support the development of human-centered emergency logistics response strategies.

In the context of vehicle electrification, equity has emerged as a critical theme to ensure a just transition. Since equity is a multifaceted concept, it can be divided into different dimensions to better analyze its implications in practice. Spatial equity refers to the fair geographic distribution of resources, such as whether public charging stations are equally available across different regions. Distributional equity addresses whether the benefits and burdens of electrification—such as subsidies, infrastructure, and vehicle ownership—are fairly distributed across different social groups, including low-income households, marginalized communities, and ethnic minorities. Lastly, perceptual equity highlights how fairness is subjectively experienced, underscoring the importance of transparency and cultural sensitivity. These equity dimensions have been widely applied in empirical studies.

Table 1. Summary of the literature on vehicle electrification.

Authors	Indicators	Spatial Scale	Equity Concept
Ju et al. (2020) [11]	Income, race, and ethnicity participation in rebate uptake	ZIP code level (CA, USA)	Distributional equity
Caulfield et al. (2022) [12]	Income level, vehicle ownership, subsidy distribution	Provinces and small area population statistics (SAPSs) (Ireland)	Distributional equity
Williams (2023) [13]	Rebate amount per capita, county rebate rate	County level (CA, USA)	Distributional equity
Falchetta & Noussan (2021) [14]	Charger density, accessibility scores	NUTS3 regional level (Europe)	Spatial equity
Hsu & Fingerman (2021) [15]	Charger access by race and income	Census block group (CA, USA	Spatial and distributional equity
Khan et al. (2022) [1]	Charger station counts per capita	ZIP code level (New York City, NY, USA)	Distributional equity
Roy & Law (2022) [16]	Predicted charger locations vs. demographics	Census tract (Orange County, CA, USA)	Spatial equity
Jiao et al. (2023) [17]	Charger access index combining proximity and demographics	Census block group (Austin, TX, USA)	Spatial equity
Peng et al. (2024) [18]	Charger access index adjusted for population and income	Tertiary population unit (TPU) level (Hong Kong, China)	Spatial and distributional equity
Tsukiji et al. (2023) [19]	Composite vulnerability and EV access score	Census tract and metropolitan areas (Los Angeles, CA, USA)	Distributional equity
Lee et al. (2024) [20]	Survey of perceptions, EV access by demographics	National (USA)	Perceptual and distributional equity
This paper	The per capita NEV ownership, proportion of NEVs in vehicles, per capita charging pile number, charging piles to NEVs ratio	State and urban–rural level (China)	Distributional equity

provides an overview of the relevant literature on vehicle electrification equity. For instance, Ju et al. (2020) used clean vehicle rebate distribution data to investigate whether equity-promoting policy design elements changed the associations between rebate allocation rates and census tract characteristics [11]. The study suggested that consideration of appropriate design elements, such as income, could facilitate the distribution of rebates to more socio-economically diverse populations. Caulfield et al. [12] used private electric vehicle household charger installation as a proxy to represent electric vehicle ownership due to data limitations. They found that urban areas are more likely to see a higher concentration of electric vehicle ownership, and that income levels affect electric vehicle density. Williams [13] analyzed data from a major EV incentive program (2017–2020), finding that the distribution of EV rebate funds closely aligns with the income groups of new vehicle buyers, indicating progress in income diversification, though further efforts are needed to increase access for priority populations.

For studies focused on charging infrastructure, Falchetta & Noussan [14] conducted a comprehensive analysis of the EV charging network in Europe. Their findings revealed stark spatial inequalities persist across and within countries. Hsu & Fingerman [15] investigated public electric vehicle charger access disparities in California using questionnaire surveys and found lower public charger access in low-income and Black or Hispanic communities in California. Khan et al. [1] found that, in New York, charging station density is not correlated with population density but positively correlated with the presence of highways. And there is a notable skew against low-income, Black-identifying, and disinvested communities. Roy and Law [16] introduced a machine learning approach to predict the level of disparities in electric vehicle access across different spatial scales in Orange County, U.S., and provided a reliable framework for charging stations placement equity assessment. Jiao et al. [17] analyzed public charging stations access disparity in Austin, Texas, U.S. The results showed that there was a more equal distribution of public chargers across income quartiles when compared with race. Peng et al. [18] analyzed public charger accessibility in Hong Kong, addressing both horizontal and vertical equity.

Many studies considered both NEV ownership and charging infrastructure for a holistic view. For instance, Tsukiji et al. [19] developed a multi-dimensional equity metric system to assess disparities in EV adoption and charger availability in Los Angeles, with variations based on income, race, and housing status. Lee et al. [20] conducted a large-scale public survey to assess the equity implications of vehicle electrification and identified that income, age, region, and housing type significantly influence EV preferences and perceptions of charging infrastructure in the U.S.

This paper contributes to the literature by integrating NEV ownership and charging infrastructure deployment to offer a holistic equity perspective. Despite China’s leadership in NEV production and consumption, relatively little research has focused on spatial equity in the Chinese context. This paper seeks to fill this critical gap by examining the changes after the implementation of the NEVgoRural policy from 2020 to 2023. Unlike previous research focused on only several socio-demographic factors, this study expands the analytical scope by considering a broader array of influencing factors. This expanded perspective is essential for gaining a thorough understanding of the intricate relationships influencing vehicle electrification equity.

3. Analysis and Result Discussions

In mainland China, there are 31 provincial-level administrative regions, including 4 municipalities (Beijing, Shanghai, Tianjin, and Chongqing) directly under the central government. Statistical data is commonly reported based on the 31 units. The analysis of this paper is also based on the 31 units. The dataset used in the modeling process is provided by China Automotive Technology Research Center and other publicly available sources.

3.1. Inequality Degree Calculation

3.1.1. Gini Index

To roughly evaluate the spatial inequality degree across China at the macro level, the Gini index was used. The Gini index is the most widely used analytical tool in the economics literature to measure inequality, originally formulated by Corrado Gini in 1912 [21]. The commonly used method for calculating the Gini index is based on individual income data, which can be expressed as Equation (1).

$$G={\displaystyle \frac{\sum _{i\in I}\sum _{j\in J}\left|x_i-x_j\right|}{2n^2\overline{x}}}$$

(1)

where $G$ represents the Gini index; $x_i$ and $x_j$ are the income of person $i$ and person $j$; $n$ is the number of persons; and $\overline{x}$ is the average income.

Although the Gini coefficient is most popular in economics, it has been expanded across various domains, such as energy consumption [22], COVID-19 cases [23], and elderly-care facilities [24], etc. Now, the Gini index has become a general quantitative indicator expressing the inequality degree of a distribution. In evaluating the inequality degree of China’s vehicle electrification process, we consider the following two perspectives: NEV ownership and charging infrastructure deployment. Considering that the data is not based on the individual level, the following calculation equation is used:

$$G^k=1-\sum _{r\in R}(X_r-X_{r-1})(Y_r^k+Y_{r-1}^k)$$

(2)

where $G^k$ represents the Gini index from the specific perspective $k$ (e.g., NEV ownership rate or charging infrastructure deployment); $R$ is the set of provinces, which are ranked in ascending order based on the indicator k; $X_r$ represents the cumulated proportion of the population up to province $r$; $Y_r^k$ represents the cumulated proportion of the measurement value up to province $r$; and the subscripts $r-1$ in the formula refer to the previous province in the sorted ranking.

Table 2. Estimated Gini index with 95% confidence interval.

	2020	2021	2022	2023
Total NEV ownership	[0.2071, 0.4448]	[0.1941, 0.4009]	[0.1622, 0.3592]	[0.1267, 0.2950]
Total charging piles	[0.2703, 0.6173]	[0.2388, 0.5519]	[0.1924, 0.4375]	[0.1408, 0.3219]

shows the 95% confidence intervals for the Gini index, which were obtained via bootstrap resampling (10,000 draws) to assess statistical robustness. When addressing inequality, organizations like the World Bank, UN, designate 0.4 as the “warning line” for income or wealth distribution disparities [25]. As shown in Table 2, disparities exist in both NEV ownership and the deployment of charging infrastructure, which is common for new emerging technologies in their initial stages. However, the inequality is observed to decline gradually over the years, which is a positive sign. This trend can be attributed to several factors, including an expansion in production scale, a decrease in associated costs, and the enactment of supportive policies. Besides, there is a slightly higher level of inequality in the deployment of total charging piles compared to total NEV ownership.

To further examine the relationship between infrastructure deployment and equity, we derived the Gini elasticity $E$ of NEV ownership based on the following formulas [26]:

$$C={\displaystyle \frac{2}{\mu }}cov\left(y_r,B_r\right)$$

(3)

$$E={\displaystyle \frac{S(C-G)}{G}}$$

(4)

where $C$ is the concentration index of charging piles relative to income. $y_r$ represents the share of charging piles in region $r$, $B_r$ is the fractional rank of the region based on income, and $\mu$ is the mean share of charging piles across all regions. $S$ is the ratio of total charging piles to total NEVs and $G$ is the Gini index of NEV ownership.

A positive $E$ indicates that reallocating charging piles may improve equity without reducing efficiency. A negative $E$ implies a trade-off, where equity improvements may come at the cost of efficiency. Our results show that while charging infrastructure is moderately concentrated in wealthier regions (C = 0.370), its associated Gini elasticity remains relatively low (E = 0.279). This relatively low elasticity indicates that policymakers can promote EV ownership equity by directing some charging infrastructure resources to underserved regions without incurring substantial efficiency losses. Such reallocation could be implemented through targeted subsidies or incentive programs, enabling more balanced regional development and fostering broader NEV adoption.

3.1.2. Theil Index

While the Gini index is one of the most widely used measures of inequality, it does not allow decomposition into within-group and between-group inequality, making it less useful for understanding the sources of inequality across regions or categories. Additionally, it is more sensitive to changes in the middle of the distribution than at the tails, which may obscure patterns of extreme deprivation or concentration.

To address these limitations, this study also applies the Theil index, a decomposable measure of inequality that allows differentiation between within-group and between-group disparities. This property makes it particularly suitable for spatial and hierarchical inequality analyses, such as urban–rural or inter-provincial divisions.

The between-group Theil index is calculated as

$$T_{between}={\sum }_{g\in G}{q}_g·{\displaystyle \frac{f_g}{\overline{f}}}·ln\left({\displaystyle \frac{f_g}{\overline{f}}}\right)$$

(5)

where $f_g={\displaystyle \frac{\sum _{g\in G}{y}_g}{\sum _{g\in G}{x}_g}}$ represents per capita value in group $g$. $q_g={\displaystyle \frac{\sum _{g\in G}{x}_g}{\sum _{r\in R}{x}_r}}$ is the population share of group $g$.

The within-group Theil index is defined as

$$T_{within}={\sum }_{g\in G}{q}_g·T_g$$

(6)

where $T_g$ is the Theil index within group $g$.

The total Theil index is then decomposed as

$${T_{total}=T}_{between}+T_{within}$$

(7)

Figure 2 illustrates the annual trends in the Theil index decomposition. The overall inequality also exhibits a declining pattern over time. Notably, the between-group Theil index is consistently higher than the within-group index, indicating that the disparity between urban and rural areas is more pronounced than that among provinces. Furthermore, the inequality in NEV sales is greater than that in the distribution of charging stations, suggesting that vehicle adoption faces more severe spatial imbalance compared to charging station deployment.

It is acknowledged that the inequality indices used in this study, such as the Gini index and Theil index, are conventionally applied to individual-level data (e.g., income distributions). In this study, however, these indices are computed using aggregated provincial-level NEV penetration data. This introduces a potential ecological fallacy, whereby inferences drawn from group-level data may not accurately reflect individual-level disparities. Nonetheless, the primary focus of this study is on spatial inequality across regions rather than interpersonal equity. Therefore, the use of aggregated data is appropriate for the research objective, although caution is warranted when interpreting the results, as indicative of individual-level fairness.

3.2. Vehicle Electrification Prevalence Analysis

The inequality index results reveal the uneven distribution of NEVs and charging infrastructures in China. However, which provinces are lagging behind and how to address this inequality have not been revealed. Equity encompasses both horizontal equity and vertical equity. Horizontal equity involves treating similar entities equally, while vertical equity prioritizes disadvantaged groups. A multi-criteria evaluation model is used in this section to conduct a prevalence analysis, assessing specific geographical areas that are lagging in vehicle electrification, considering both horizontal equity across provinces and vertical equity between urban and rural areas within provinces. The weights of composited criteria are defined by Shannon’s Information Entropy Weighting, which has the strength of overcoming the influence of artificial subjectivity and reducing the overlap of information between multiple variables [27]. Since entropy measures the dispersion or uncertainty in data distribution, extreme values can increase the variability of an indicator [28]; therefore, the variables are first normalized as shown in Equation (8). Equation (9) calculates the proportion of the index. The next step is to compute the entropy for each indicator using Equation (10), which quantifies the uncertainty, or the information provided by the distribution of the proportions. After that, the weight of the $n$th indicator obtained in Equation (11) is used to calculate the prevalence value in Equation (12). In this method, indicators that exhibit greater dispersion across regions receive higher weights because they provide more discriminatory power and are more effective in capturing spatial inequities in the vehicle electrification process.

$$S_{mn}={\displaystyle \frac{X_{mn}-min\left\{X_n\right\}}{max\left\{X_n\right\}-min\left\{X_n\right\}}}$$

(8)

$$P_{mn}={\displaystyle \frac{S_{mn}}{\sum _{m=1}^M{S}_{mn}}}$$

(9)

$$E_n=-{\displaystyle \frac{\sum _{m=1}^M{P}_{mn} ln{P}_{mn}}{ln n}}$$

(10)

$$W_n={\displaystyle \frac{1-E_n}{{\sum }_{n=1}^N\left(1-E_n\right)}}$$

(11)

$$V_n=\sum _{n\in N}{W}_n{C}_{mn}$$

(12)

where $S_{mn}$ is the normalized value of the $n$th indicator of the $m$th province. $P_{mn}$ is the proportion of the $n$th indicator for the $m$th province. $E_n$ is the entropy of the $n$th indicator. $w_n$ is the weight. The prevalence value $V_n$ is measured as the weighted sum of different criteria $V_{mn}$.

3.2.1. Inter-Provincial Vehicle Electrification Prevalence Analysis

There are at least two ways to judge the NEV prevalence. One is whether the revealed NEV prevalence reflects the population size. The population is a basic denominator for resource equity. If all people are given equal access to the resources embedded in the adoption of NEVs, then, theoretically, the NEV ownership should be proportional to the population. This approach aligns with a normative equity principle, where every individual is considered equally entitled to the environmental and economic benefits associated with NEVs. However, one counterargument is that not every person wants a NEV, in general, due to various regional and socio-economic circumstances, and therefore, policymakers should not force a population-based prevalence of NEV ownership. To address this, the second indicator to judge is the proportion of NEVs in vehicles. If NEV adoption patterns align with existing vehicle ownership, it suggests NEV access is consistent with local mobility norms and capabilities. Thus, the NEV share among total vehicles serves as a complementary indicator of equity from a behavioral and economic standpoint. Taken together, these two indicators capture both the normative ideal (equal access across the population) and the revealed preference (realistic alignment with existing vehicle ownership patterns), providing a more comprehensive and balanced assessment of NEV distribution equity.

In Figure 3a, we present a choropleth map illustrating the per capita NEV ownership in mainland China, with the scale’s top and bottom values representing the maximum and minimum values, respectively. Moving to Figure 3b, we showcase a map depicting the proportion of NEVs among all vehicles. Notably, there is a discernible disparity in NEV ownership across the 31 units, with a clear concentration in provinces along the east and the south. This spatial clustering is supported by Moran’s I values, which confirm significant positive spatial autocorrelation. For per capita NEV ownership, the Moran’s I value is 0.442 with a p-value of 0.002. Similarly, for the proportion of NEVs among all vehicles, the Moran’s I value is 0.361 with a p-value of 0.004. The uneven distribution signifies regional variations, which has important implications for policy targeting and the equitable promotion of low-carbon transportation.

Figure 4a presents a bar chart depicting the per capita ownership of NEVs. Beijing, Shanghai, Tianjin, Zhejiang, Guangdong, and Jiangsu are the top six provinces in terms of per capita disposable income in China, which enables residents there to afford NEVs. Hainan province, with tourism as its primary industry, may have achieved one of the leading positions in per capita NEV ownership through its particular emphasis on environmental protection. As early as 2019, Hainan planned to prohibit the sale of gasoline-powered vehicles by 2030, making it the first province in China to establish a clear timeline for the ban on gasoline-powered vehicles [29]. As part of its phased implementation, starting in 2025, gasoline vehicles are generally prohibited from entering Dongyu Island in Bo’ao, Hainan, except under special circumstances [30]. Figure 4b depicts the proportion of NEVs in vehicles. The results indicate that car buyers in Beijing, Hainan, etc., provinces are more inclined to consider NEVs when buying vehicles. Compared with Figure 4a, the majority of provinces exhibiting below-average levels of per capita NEV ownership also fall below the average in terms of proportions of NEVs in vehicles, with the exceptions being Chongqing, Guangxi, Henan, and Fujian. The final inter-provincial NEV prevalence values based on the multi-criteria evaluation model are illustrated in Figure 4c. To provide a measure of statistical confidence, we applied a bootstrap resampling method (10,000 draws) to estimate 95% confidence intervals. These intervals are also presented as shading in Figure 4c, illustrating the robustness and variability. As shown in Figure 4c, NEV prevalence has exhibited a consistent upward trend, increasing by 56.5% from 2020 to 2021, 61.9% from 2021 to 2022, and 52.7% from 2022 to 2023. Tibet, Heilongjiang, and Qinghai are among the provinces lagging in NEV adoption. The narrow confidence intervals (minimal shading) indicate that the results are statistically stable and reliable.

Ensuring access to charging infrastructures is critical for the success of vehicle electrification. To effectively measure the accessibility of charging piles, two key criteria are employed: the per capita number of charging piles and the ratio of charging piles to NEVs. The per capita number of charging piles offers insight into the accessibility of charging facilities relative to the population. It aligns with a normative equity principle that assumes every individual should have comparable potential access to charging services. However, evaluating infrastructure based solely on the population may miss critical nuances, such as the actual demand for charging services. The ratio of charging piles to NEVs serves as a precise indicator of the adequacy of charging infrastructure relative to the number of NEVs requiring charging service. Together, these two indicators offer a more holistic picture of infrastructure distribution and access.

In Figure 5a, we present the geographical distribution of the per capita number of charging piles across mainland China’s provinces. Mirroring the pattern observed in NEV ownership, a heightened concentration of charging piles is observed in the southeastern regions (Moran’s I: 0.367, p-value: 0.003). In Figure 5b, the ratio of charging piles to NEVs across provinces is depicted, showing a dispersed pattern (Moran’s I: −0.089, p-value: 0.343). This divergence shows complexity in the regional dynamics of charging infrastructure accessibility in mainland China.

Figure 6a displays a bar chart showing the per capita number of charging piles in each province. The findings reveal that provinces like Shanghai, Beijing, Tianjin, and Guangdong have excelled in charging pile development when assessed on a per capita basis. Figure 6b displays the charging piles to NEVs ratio in each province, which implies that provinces such as Heilongjiang, Tibet, Inner Mongolia, and Ningxia boast a charging pile number that aligns well with their charging demand. The significant difference between Figure 6a,b highlights the nuanced nature of equity in charging pile development and prompts a thoughtful consideration of regional variations. The final inter-provincial charging pile prevalence values are shown in Figure 6c, with the shading indicating the confidence intervals. Provinces such as Jilin, Guangxi, Yunnan, and Qinghai demonstrate a pressing need for installing additional charging piles. However, this demand could be constrained by land-use limitations and insufficient grid capacity. Hainan province has made significant progress in charging pile deployment each year, rising to sixth place nationwide in 2023.

3.2.2. Intra-Provincial Rural–Urban Vehicle Electrification Prevalence Analysis

Intra-provincial vehicle electrification disparities between rural and urban areas highlight a significant vertical inequity issue. As mentioned in the introduction section, the Chinese government has started to launch the NEVgoRural policy to promote NEVs in rural areas since 2020 [8]. The motivations of the policy, in addition to improving vehicle electrification equity, lie in supporting China’s Rural Revitalization Strategy, which aims at stimulating the rural economy in the hope of reversing the imbalance from the decades of resource concentration in metropolitan areas, and further invigorating the NEV industry from the recent economic slow-down.

$$\mathrm{Distance} \mathrm{from} \mathrm{idealized} \mathrm{target}=1-{\displaystyle \frac{{per capita NEV sales}_{rural}}{{per capita NEV sales}_{urban}}}$$

(13)

$$\mathrm{Distance} \mathrm{from} \mathrm{short}-\mathrm{term} \mathrm{target}={\displaystyle \frac{{per capita vehicle sales}_{rural}}{{per capita vehicle sales}_{urban}}}-{\displaystyle \frac{{per capita NEV sales}_{rural}}{{per capita NEV sales}_{urban}}}$$

(14)

Despite expectations, the progress of the NEVgoRural policy has been slow, partly due to the lack of methods to track electrification in rural areas and assess public investment impacts. This section measures whether the adoption (either market-driven or government-promoted) of NEVs in China is equitable between urban and rural areas. If the population is considered, then the closer the rural per capita number of NEV sales is to the urban per capita NEV sales, the more equitable the electrification status is. A rural–urban ratio (RUR) value of 1 based on per capita NEV sales indicates that the rural NEV prevalence situation is equivalent to that of urban areas, which is referred to as the idealized target. The distance from this idealized target is calculated as shown in Equation (13). Alternatively, achieving parity in per capita NEV sales between rural and urban areas may be overly challenging. As long as the distribution of NEVs matches that of all vehicles, the status of vehicle electrification can be claimed as equitable. According to this notion, the distance from the short-term target is calculated according to Equation (14). Notably, since both target distances are defined as the difference between the observed rural–urban ratio and the target value (e.g., 1 or vehicle ownership RUR), the resulting distance can be negative when the observed rural share exceeds the target. This allows the metric to capture both lagging and leading equity performance.In Figure 7a, the map illustrates the distance of the current progress from the short-term target across provinces, while in Figure 7b, the map depicts the distance of the current progress from the idealized target across provinces. The bigger the differences, the bluer the color on the map, indicating worse rural–urban equity. Additionally, the Moran’s I statistic is reported directly on each figure, offering a quantitative measure of the spatial autocorrelation in equity distance—thereby revealing whether provinces with similar performance levels tend to be geographically clustered. The results confirm the presence of significant spatial clustering.

Figure 8a displays a bar chart showing how each province’s current progress deviates from the short-term target, whereas Figure 8b depicts the deviation in the current progress from the idealized target. From these figures, we can see that achieving the short-term target is relatively easy, whereas meeting the long-term goal remains a challenging task. This implies that while there are a lot of NEV sales in rural Zhejiang, it is still insufficient compared to the high level of motorization in its rural areas, suggesting there is significant potential for further growth. The final intra-provincial rural–urban NEV equity values from year 2020 to 2023 are shown in Figure 8c, with the shading indicating the confidence intervals. The larger values indicate a greater disparity between rural and urban areas. From this figure, we observe that not all provinces are seeing year-on-year improvements in rural–urban equity. However, this does not necessarily mean that the NEVgoRural policy has been ineffective. Without this policy, the situation might be even worse. Provinces such as Heilongjiang (0.808), Tibet (0.815), and Ningxia (0.830) performed the worst in rural–urban equity in terms of NEV prevalence in 2023.

Figure 9 employs the horizontal axis to represent the NEV prevalence, with the vertical axis indicating the rural–urban NEV prevalence equity values. The size of the data points shows the population size of the rural areas in these provinces. The plot is divided into four quadrants based on the mean values of the two variables. The points in the first quadrant, including Hainan, Guangdong, Chongqing, etc., represent a relatively high NEV prevalence but poor rural–urban equity. This suggests that NEV promotion policies and resources are currently concentrated in urban areas in these regions; it may now be time to focus efforts on suburban areas. Provinces in the second quadrant, such as Heilongjiang, Tibet, and Ningxia, exhibit a low NEV prevalence and significant disparities in rural–urban vehicle electrification development. For these provinces, it is crucial to focus on increasing NEV adoption, especially with policies for rural areas. Provinces in the third quadrant, while showing a relatively balanced distribution of NEVs between rural and urban areas, have an overall low adoption rate. Increasing overall NEV adoption in all areas is important. In the fourth quadrant, provinces like Zhejiang, Jiangsu, Shandong, and Anhui have achieved both high NEV prevalence and good rural–urban equity, making them model cities.

As a supporting facility, an insufficient charging infrastructure will seriously restrict the growth space of NEVs in rural areas. Figure 10a displays a map showing how each province’s current charging stations equity progress in 2024 deviates from the long-term target. The bluer the color on the map, the worse the rural–urban equity. Figure 10b shows the bar chart of distance from the long-term target in 2021 and in 2024. We can see that the urban–rural equity issue has improved in most provinces in 2024 compared to 2021. Hunan (0.747), Tibet (0.726), and Heilongjiang (0.722) have performed the worst in terms of charging station deployment equity between urban and rural areas in 2024.

Figure 11 employs the horizontal axis to represent charging pile prevalence, with the vertical axis indicating the rural–urban charging station equity values. To mitigate overlap among data points and enhance visual clarity, both variables have been log-transformed prior to plotting. The points in the first quadrant, such as Shaanxi, Hubei, and Guangdong, suggest a situation where charging piles are relatively prevalent, but there is poor rural–urban equity. To address this, targeted policy interventions are necessary to ensure equitable access to charging facilities in rural areas. Provinces in the second quadrant, such as Hunan, Jilin, and Xinjiang, exhibit a low prevalence of charging infrastructure and significant rural–urban disparities. These regions require a comprehensive approach that focuses not only on expanding charging infrastructure but also on reducing the gap between rural and urban areas. In the third quadrant, provinces show a more balanced distribution of charging infrastructure between rural and urban areas, but the overall prevalence remains low. Therefore, the focus in these regions should be on increasing the overall coverage of charging facilities, with both rural and urban areas working together to improve infrastructure availability. Finally, the provinces in the fourth quadrant, including Zhejiang, Ningxia, and Hainan, have successfully achieved both high charging infrastructure prevalence and equitable distribution between rural and urban areas. These regions can serve as models for other provinces looking to improve their charging infrastructure.

3.3. Correlation Analysis

The above results point out the provinces lagging behind during the vehicle electrification process in China. However, the factors associated with this inequity are not identified. To explore the potential drivers of inequality, correlation analysis is applied. While correlation analysis does not control for confounding variables as regression-based methods would, it is appropriate in the context of a limited sample size and serves to generate hypotheses for future research. The magnitude of the correlation coefficient reflects the strength of the relationship, with 1 or -1 indicating a perfect linear relationship, and values closer to 0 indicating weaker correlations.

With regard to explanatory factors, previous studies have explored various variables, such as income [1,12,15,18], race [1,15,18], education level [19], population density [1], etc. In addition to these factors, this paper draws on the Technology–Organization–Environment (TOE) framework to guide the selection and classification of explanatory variables. TOE is a widely adopted model in innovation diffusion and technology adoption research. As shown in

Table 3. Explanatory factors based on TOE framework.

	Factors
Technology	(1) Number of NEV companies
Organization	(2) Population (3) Population density (4) High-level education ratio (5) Per capita disposable income (6) GDP (7) Vehicle ownership
Environment	(8) PM2.5 reduction over five years (9) PNO2 reduction over five years (10) Annual average temperature (11) Total electricity generation (12) Wind and solar electricity generation

, the technological factors include the number of NEV companies, which reflects industrial capability and technological readiness to support NEV production. The organizational factors include the population, population density, high-level education ratio, per capita disposable income, GDP, and vehicle ownership. If a province is regarded as the organizational unit, these variables represent its internal characteristics, which may influence the capacity and willingness to adopt NEVs. Environmental factors include a PM2.5 reduction over five years, NO₂ reduction over five years, annual average temperature, total electricity generation, and wind and solar electricity generation. These variables capture the broader external conditions under which NEV adoption occurs, including environmental constraints, energy supplement, and climate, all of which can shape regional electrification dynamics.

The raw jittered data points, the distribution of the data, and the key summary statistics (i.e., the minimum, the median, the maximum, etc.) are visualized in the raincloud plots in Figure 12. The “high-level education ratio” refers to the percentage of individuals with a college education or above in the total population. The meanings of the other terms are self-explanatory. Comparing Figure 12i,q, the per capita vehicle trend is smoother than that of the per capita NEVs, which indicates that the equity of NEVs lags behind that of vehicles. Figure 12q–t roughly provides a visualization of the unequal vehicle electrification conditions.

Table 4. Correlation analysis of various factors in vehicle electrification for 2023.

	Total NEV Number	Total Charging Pile Number	Per Capita NEV Number	Proportion of NEVs in Vehicles	Per Capita Charging Pile Number	Charging Piles to NEVs Ratio
Number of NEV companies	0.863 ***	0.859 ***	0.604 ***	0.602 ***	0.579 **	−0.316
Population	0.764 ***	0.729 ***	0.313	0.293	0.273	−0.456
Population density	0.848 ***	0.860 ***	0.869 ***	0.826 ***	0.861 ***	−0.161
High-level education ratio	0.130	0.238	0.363	0.285	0.477 *	0.403
Per capita disposable income	0.646 ***	0.699 ***	0.715 ***	0.664 ***	0.791 ***	0.137
GDP	0.909 ***	0.920 ***	0.579 **	0.549 **	0.588 **	−0.263
Per capita GDP	0.528 **	0.615 ***	0.682 ***	0.610 ***	0.781 ***	0.260
Vehicle ownership	0.819 ***	0.798 ***	0.397	0.342	0.379	−0.361
Per capita vehicle ownership	0.260	0.309	0.347	0.156	0.400 *	0.201
PM2.5 reduction over five years	0.221	0.189	0.253	0.207	0.249	0.238
NO2 reduction over five years	0.534 **	0.552 **	0.549 **	0.501 **	0.539 **	0.038
Annual average temperature	0.600 ***	0.564 **	0.663 ***	0.736 ***	0.608 ***	−0.283
Total electricity generation	0.423 *	0.415 *	0.046	−0.043	0.078	−0.215
Wind and solar electricity generation	−0.065	−0.077	-0.333	−0.464 *	−0.310	−0.039
Total NEV number	1	0.982 ***	0.794 ***	0.758 ***	0.756 ***	−0.342
Total charging pile number	0.982 ***	1	0.806 ***	0.761 ***	0.802 ***	−0.228
Per capita NEV number	0.794 ***	0.806 ***	1	0.964 ***	0.957 ***	−0.212
Proportion of NEVs in vehicles	0.758 ***	0.760 ***	0.964 ***	1	0.924 ***	−0.231
Per capita charging pile number	0.756 ***	0.802 ***	0.957 ***	0.924 ***	1	−0.008
Charging piles to NEVs ratio	−0.342	−0.228	−0.213	−0.231	−0.008	1

Note: *, **, and *** indicate the p-value is below 0.05, 0.01, and 0.001, respectively.

presents the Spearman rho correlation analysis results for the various factors in vehicle electrification. A Spearman correlation is employed instead of a Pearson correlation to account for the possibility of non-normal distributions in the variables. To address the issue of inflated Type I errors due to multiple comparisons, we apply the False Discovery Rate (FDR) adjustment to the raw p-values, thereby reducing the likelihood of false positives.

The number of NEV manufacturing companies is positively related to the ownership of NEVs. This correlation may stem from increased market supply, which expands consumer options and stimulates interest in NEVs. Additionally, a large customer base may, in turn, drive the growth of NEV manufacturers.

As expected, the population density is positively correlated with both the total and per capita numbers of NEVs and charging piles. This suggests that public resources allocated to NEV adoption, such as R&D investments and government funding, are disproportionately benefiting densely populated areas. Cities with a high population density in China often implement license plate lottery systems to regulate vehicle purchases. NEVs are often exempt from these restrictions, which serves as a strong incentive for their adoption [6]. Rural areas, on the other hand, do not face license plate restrictions, further highlighting the challenges in implementing the NEVgoRural policy. Policymakers need to be well-prepared to develop customized and attractive policies.

A positive correlation is also found between the high-level education ratio and per capita NEV ownership. One plausible explanation is that more highly educated people might be more aware of environmental conservation and technological advancements, which are key selling points for NEVs. However, it could also be the case that regions with high NEV adoption might attract more highly educated residents due to green reputations or job opportunities in the clean tech sector. Additionally, per capita disposable income is significantly related to vehicle electrification. Given the generally high price of NEVs, individuals with high disposable income levels are more likely to be able to afford them. These findings underscore that vehicle electrification is currently concentrated in high-income and highly educated regions, highlighting challenges and the need for broader adoption among low-income and less-educated populations.

Notably, both GDP and per capita GDP are highly correlated with vehicle electrification. It is pragmatic to initially focus NEV promotion in regions with higher GDPs, as these areas can better support the infrastructure and incentives needed for early adoption. However, this strategy risks entrenching regional disparities and creating a path-dependent trajectory where low-GDP regions fall further behind in the transition to clean mobility. In light of the just transition framework [31], which emphasizes fairness and inclusion in energy transitions, it is essential to complement efficiency-driven deployment with redistributive mechanisms. The NEVgoRural policy is a promising step in this direction, which enables people in less affluent regions to access NEVs. For instance, Anhui Province allocated CNY 70 million in 2024 to subsidize rural NEV purchases, offering incentives ranging from 2000 to 6000 yuan per vehicle [32].

A significant positive correlation exists between the total number of NEVs and the total vehicle number, yet the correlation between per capita NEV ownership and per capita vehicle ownership is not significant. This indicates that population size is a confounding factor and highlights the importance of using per capita measures in cross-regional comparisons.

There is a positive correlation between the reduction in NO₂ levels over five years and vehicle electrification, suggesting that increased NEV adoption contributes to improving air quality. While the effect appears limited, it signals potential long-term public health benefits from vehicle electrification.

Cold climates can reduce battery efficiency and range [33]. The moderately positive correlation between the annual average temperature and NEV ownership indicates that residents in colder regions may have reservations about choosing NEVs. With the advancement of efficient thermal management technologies for batteries, such as battery pre-heating technology [34], this correlation may diminish.

The total electricity generation is positively correlated with the total number of NEVs and charging piles. Electricity assurance is a crucial foundation for vehicle electrification. As the number of NEVs grows, it is important to strengthen power generation facilities and enhance electricity supply capacity. China’s rural electricity grid reached a reliability rate of 99.8% by 2023 [35], providing a solid foundation for the implementation of the NEVgoRural policy.

Interestingly, wind and solar electricity generation show a negative correlation with per capita NEV and charging pile ownership. This may be because areas rich in wind and solar electricity generation tend to be less economically developed. However, these areas are ideal for deploying Vehicle-to-Grid (V2G) systems, which can stabilize the grid by leveraging NEVs as mobile storage. This is particularly relevant in rural areas for two key reasons: one is that, for NEV owners, charging at low electricity prices and discharging at high prices can lead to additional revenue, which may be more attractive for residents in rural areas who are more sensitive to such financial opportunities. Another is that most solar and wind power plants are located in rural areas. Spatially aligning V2G with solar and wind electricity sources minimizes the need for long-distance power transmissions, thereby enhancing its viability in rural settings. Figure 13 illustrates the map of wind power and solar power electricity generation across Chinese provinces. The names of the top 10 provinces are labeled, which could be considered as promoting V2G technology. Guangdong, Inner Mongolia, and Ningxia are highlighted in red due to the significant rural–urban disparity in vehicle electrification. Piloting V2G initiatives in the rural areas of these provinces could not only fully utilize local wind and solar resources but also promote rural–urban equity.

Finally, as expected, a strong correlation exists between the NEV number and the charging infrastructure. This relationship has been widely recognized in prior research as a classic “chicken-and-egg” dilemma, wherein the lack of charging infrastructure impedes NEV adoption, while limited NEV uptake discourages investment in infrastructure [36].

4. Conclusions and Policy Implications

Vehicle electrification policies must prioritize geographic equity to ensure inclusive access to NEVs and charging infrastructure. This study evaluates China’s vehicle electrification process from 2020 to 2023. The goal is to help balance NEV ownership and ensure equitable availability of charging infrastructures both inter-provincially and intra-provincially. The findings offer important policy insights to guide stakeholders’ decision-making.

Although NEV purchases are primarily consumer-driven, they are significantly influenced by the government’s incentives, regulations, and investments. China’s current policy of tax exemption for NEV purchases will remain in effect until 31 December 2025, after which it will be reduced by half and gradually phased out [37]. Provinces that lag in NEV ownership, such as Tibet, Heilongjiang, and Qinghai, may be considered for extended tax exemptions for a longer period. In addition, supportive measures, including free parking, access to bus lanes, expressway toll discounts, free or discounted charging, eco-friendly education, and pilot demonstration projects, could be considered by the lagging-behind provinces to stimulate demand. Regarding charging infrastructure, active investment by local utilities and implementing incentive policies are also important for the construction of charging piles. Beyond relying on policy measures, private sector strategies for profitability are equally critical for sustainable development in this area. The fuel industry’s success in integrating convenience stores into gas stations offers a telling precedent [38]. Inspired by this, charging stations could consider incorporating supplementary amenities like bookstores, restaurants, convenience stores, or cafes to diversify their revenue streams. Additionally, offering services such as vehicle washing or maintenance could further boost profitability.

The NEVgoRural policy is a crucial initiative aimed at promoting NEV adoption in rural areas. While all provinces exhibit disparities in rural–urban equity, provinces like Heilongjiang, Tibet, and Ningxia are particularly severe in this regard. To effectively implement the NEVgoRural policy, it should be tailored to the unique characteristics of rural areas. Survey evidence suggests that micro and small NEVs—typically offering a range over 200 km, battery capacities of 20–30 kWh, and priced around 40,000–70,000 RMB—are particularly attractive to rural consumers [39]. However, despite this preference, the relatively lower income levels in rural areas and the high upfront costs of NEVs remain key barriers to adoption. In the U.S., the Energy Efficient Mortgages program offers zero or low-interest loans for home energy upgrades, which has helped increase their adoption [40]. Similarly, financial tools like zero down payment and interest-free plans could lower the cost barrier and support NEV uptake in less affluent areas. In addition, subsidies, free or discounted charging, and tax reductions, specifically for rural areas, could incentivize residents to purchase NEVs. Equally important is the establishment of a robust NEV sales and after-sales service network in rural areas. Such infrastructure not only enhances maintenance accessibility but also contributes to local employment generation, particularly through training programs for rural technicians in NEV repair and diagnostics. As for prioritizing promoting NEV adoption in rural areas, provinces like Hainan and Guangdong should take the lead. While both have achieved relatively high overall NEV penetration and accumulated substantial experience, they still face pronounced rural–urban equity gaps. For example, rural areas account for only 4% of the total NEV sales in Hainan and 3% in Guangdong. In 2024, China’s Ministry of Finance, Ministry of Industry and Information Technology, and Ministry of Transport jointly issued a document stating that, from 2024 to 2026, a pilot program titled “100 Counties, 1000 Stations, 10,000 Chargers” will be implemented to strengthen the planning and construction of charging infrastructure in rural areas [41]. However, the installation of charging stations is not an endpoint. To ensure their long-term functionality and accessibility, regular maintenance and operational support must be in place [42]. In addition, improving digital literacy among rural residents is essential, particularly in enabling them to use smartphones for scanning QR codes and making mobile payments [43]. Provinces such as Shaanxi, Hubei, and Guangdong, where charging infrastructure is relatively more widespread, but rural–urban equity remains an issue, should prioritize the expansion of charging infrastructure in rural regions. Given that many rural households have independent courtyards, home charger installation is both feasible and advantageous. Time-sharing leasing can be adopted to provide a more economical solution and improve the economic benefits of installing home chargers in rural areas [44]. Finally, to align with the distribution of wind and solar power generation, Guangdong, Ningxia, and Inner Mongolia are recommended to pioneer V2G experiments or demonstrations, maybe starting from the rural areas. V2G not only helps stabilize renewable electricity integration but also enhances rural household income by enabling NEV users to store electricity during low-cost periods and sell it during high-demand intervals. By August 2023, a V2G pilot in China by State Grid achieved 850,000 kWh of solar power, 1.26 million kWh of storage use, 11,640 tons of CO₂ cuts, and 85% building energy self-sufficiency [45].

This study can be considered the first of its kind towards systematically analyzing vehicle electrification equity in China, laying the foundation for the development of targeted policies tailored to different regions. Future research could consider developing optimization frameworks that incorporate measures like Gini elasticity into objective functions. By doing so, it becomes possible to explicitly evaluate the trade-offs between equity and efficiency, thereby enabling policymakers to strike a more effective balance between these competing goals and to mitigate conflicts among diverse stakeholder interests. Additionally, while we have provided a framework to guide the determination of region priority for vehicle electrification, there is a need to extend the research to include price elasticity of public support in order to determine the extent of public support, such as a purchase subsidy. Further, this study employs bivariate correlation analysis as an exploratory tool to identify potential associations, without implying causal relationships. Future work could consider combining spatial analysis with survey data, which could enable multivariate analysis or more rigorous causal identification strategies. Survey methods could incorporate a range of behavioral or cultural factors that may significantly influence NEV adoption, such as technology acceptance, travel habits, risk perception, or institutional trust. These factors have been extensively discussed in the literature on behavioral economics and consumer sociology and are often operationalized through theoretical models like the Technology Acceptance Model (TAM) and the Unified Theory of Acceptance and Use of Technology (UTAUT) [46], which could help us provide a more robust explanation of NEV adoption patterns in rural area.