As a strategic emerging industry, the EV car sector plays a dual role in energy conservation, emission reduction, and knowledge innovation, serving as a critical industrial pillar for achieving carbon neutrality goals. Unlike traditional automotive industries, the EV car industry integrates electrification, intelligence, and networking, representing a key breakthrough in the transformation of economic growth drivers. In recent years, the production and sales of EV cars have surged globally, with China emerging as a dominant player, accounting for over 60% of the global market share. This growth underscores the industry’s potential as a new engine for economic development and the cultivation of advanced productive forces. In this study, we analyze the spatiotemporal evolution of EV car enterprises across the industrial chain from 2016 to 2023, employing methods such as the Gini coefficient, spatial autocorrelation, and spatial econometric models. Based on data from over 90,000 enterprises related to the EV car industry, we examine the spatial distribution patterns and key driving factors influencing the layout of enterprises at different segments of the chain. Our findings aim to provide academic insights for optimizing the spatial distribution of the EV car industry, promoting regional coordinated development, and informing policy formulation in the context of advancing new quality productivity.
The EV car industrial chain is divided into three segments: upstream, midstream, and downstream. The upstream segment includes core components such as drive motors, power batteries, and electronic control systems, collectively known as the “three electrics.” These components form the essential power system of EV cars and distinguish them from traditional internal combustion engine vehicles. The midstream segment involves the manufacturing of complete EV cars, which represents the final assembly stage. The downstream segment encompasses automotive services, including sales, maintenance, and charging infrastructure for EV cars. This segmentation allows for a detailed analysis of the spatial characteristics and influencing factors at each stage of the value chain.
To quantify the spatial distribution of EV car enterprises, we utilize the Gini coefficient, which measures the degree of inequality in spatial distribution. The formula for the Gini coefficient is given by:
$$G = 1 – \frac{1}{n} \left(2 \sum_{i=1}^{n-1} M_i + 1\right)$$
where \(G\) represents the Gini coefficient, \(n\) is the number of cities, and \(M_i\) is the cumulative proportion of enterprises in city \(i\). A higher Gini coefficient indicates greater spatial concentration and inequality in distribution. Additionally, we employ spatial autocorrelation analysis to assess the clustering patterns of EV car enterprises. The global Moran’s I index is used to evaluate the overall spatial autocorrelation:
$$I = \frac{n \sum_{i=1}^{n} \sum_{j=1}^{n} W_{ij} (x_i – \bar{x})(x_j – \bar{x})}{\sum_{i=1}^{n} \sum_{j=1}^{n} W_{ij} \sum_{i=1}^{n} (x_i – \bar{x})^2}$$
where \(I\) is the Moran’s I index, \(x_i\) and \(x_j\) denote the number of EV car enterprises in cities \(i\) and \(j\), respectively, \(\bar{x}\) is the mean number of enterprises across all cities, and \(W_{ij}\) is the spatial weight matrix. A positive Moran’s I indicates spatial clustering, while a negative value suggests dispersion. For local spatial autocorrelation, we use the Local Indicators of Spatial Association (LISA) to identify specific clusters and outliers.
To analyze the influencing factors, we apply spatial econometric models, including the spatial lag model (SLM) and spatial error model (SEM). The SLM accounts for spatial dependence in the dependent variable and is expressed as:
$$Y = \rho W y + X \beta + \mu$$
where \(Y\) is the dependent variable (number of EV car enterprises), \(W y\) is the spatially lagged dependent variable, \(\rho\) is the spatial autoregressive coefficient, \(X\) represents the independent variables, \(\beta\) is the coefficient vector, and \(\mu\) is the error term. The SEM addresses spatial dependence in the error term:
$$Y = X \beta + \epsilon, \quad \epsilon = \lambda W \epsilon + \mu$$
where \(\epsilon\) is the error term, \(\lambda\) is the spatial error coefficient, and other terms are as defined above. We select explanatory variables based on economic theory and previous studies, covering dimensions such as economic foundation, market scale, technological innovation, industrial base, government guidance, and locational conditions. The variables are summarized in Table 1.
| Influencing Factor | Code | Explanatory Variable |
|---|---|---|
| Economic Foundation | X1 | Gross Domestic Product (GDP, in ten thousand yuan) |
| Market Scale | X2 | Population Density (persons per square kilometer) |
| Technological Innovation | X3 | Number of Patent Grants (units) |
| Industrialization Level | X4 | Percentage of Secondary Industry Value-Added in GDP (%) |
| Automotive Industrial Base | X5 | Number of Automotive Industrial Parks (units) |
| Government Technology Investment | X6 | Percentage of Science and Technology Expenditure in General Public Budget Expenditure (%) |
| EV Car Industrial Policy | X7 | Dummy Variable: 1 if an EV car industrial park is established, 0 otherwise |
| Locational Condition | X8 | Dummy Variable: 1 for eastern region cities, 0 for central and western regions |
Our analysis reveals significant spatial agglomeration across all segments of the EV car industrial chain, with distinct characteristics at each stage. The upstream segment, comprising battery, motor, and electronic control manufacturing, is technology-intensive and exhibits a “south-north disparity, east-west density, and multi-core polarization” pattern. Key agglomeration areas include the Pearl River Delta, Yangtze River Delta, and the Changsha-Zhuzhou-Xiangtan urban agglomeration. The midstream segment, involving EV car manufacturing, shows a concentration in regions such as Shandong, Henan, Hebei, Anhui, and Jiangsu, with a shrinking agglomeration range over time. The downstream segment, focused on automotive services, displays a “multi-core” distribution centered on provincial capitals and major urban agglomerations, with significant core-periphery disparities.
The Gini coefficients for each segment from 2016 to 2023 are presented in Table 2. All values exceed 0.5, indicating high spatial inequality, with the upstream segment showing the highest concentration. Over time, the Gini coefficients for upstream and midstream segments decrease, suggesting a trend toward more balanced distribution, while the downstream segment remains relatively stable.
| Segment | 2016 | 2020 | 2023 |
|---|---|---|---|
| Upstream | 0.828 | 0.802 | 0.767 |
| Midstream | 0.844 | 0.734 | 0.699 |
| Downstream | 0.570 | 0.586 | 0.585 |
Global Moran’s I indices for each segment are shown in Table 3. All values are positive and statistically significant, confirming spatial clustering and dependency. The upstream segment shows an increasing trend in Moran’s I, indicating strengthening agglomeration, while the midstream and downstream segments exhibit declining trends, suggesting diffusion effects.
| Segment | 2016 | 2020 | 2023 |
|---|---|---|---|
| Upstream | 0.1403*** (4.147) | 0.1028** (3.248) | 0.1689*** (5.364) |
| Midstream | 0.1465*** (4.469) | 0.1909*** (5.989) | 0.0715** (2.588) |
| Downstream | 0.1530*** (4.963) | 0.1246*** (3.964) | 0.1048*** (3.314) |
LISA cluster analysis further elucidates local spatial patterns. For upstream EV car enterprises, high-high clusters are concentrated in the Yangtze River Delta, Pearl River Delta, and central regions, while low-low clusters are prevalent in western China. Midstream enterprises show expanding high-high clusters in northern and central provinces, with high-low clusters fragmenting over time. Downstream services maintain stable high-high clusters in urban agglomerations and provincial capitals, with persistent low-low clusters in less developed regions.
To identify the driving factors behind the spatial distribution of EV car enterprises, we estimate spatial econometric models using 2023 enterprise data as the dependent variable and 2021 indicator data as independent variables. The regression results are summarized in Table 4. Based on model fit criteria (R², Log-likelihood, and AIC), the spatial lag model (SLM) is most suitable for the upstream segment, while the spatial error model (SEM) is preferred for the midstream and downstream segments.
| Variable | Upstream (SLM) | Midstream (SEM) | Downstream (SEM) |
|---|---|---|---|
| X1 (GDP) | 0.545*** | 0.197** | 0.767*** |
| X2 (Population Density) | -0.023 | 0.050 | 0.108** |
| X3 (Patent Grants) | 0.160*** | 0.121** | 0.036 |
| X4 (Industrialization Level) | -0.270** | -0.522*** | -0.693*** |
| X5 (Automotive Industrial Base) | -0.027 | 0.118* | 0.053 |
| X6 (Government Technology Investment) | 0.214*** | 0.074 | -0.040 |
| X7 (EV Car Industrial Policy) | 0.257** | 0.336*** | 0.167* |
| X8 (Locational Condition) | -0.256*** | -0.179 | -0.121 |
| ρ (Spatial Lag Coefficient) | 0.314*** | — | — |
| λ (Spatial Error Coefficient) | — | 0.362*** | 0.674*** |
| R² | 0.737 | 0.474 | 0.755 |
| Log-Likelihood | -271.48 | -289.64 | -231.18 |
| AIC | 562.97 | 597.29 | 480.35 |
The results indicate that economic foundation (X1) positively influences all segments of the EV car industrial chain, highlighting the importance of regional economic strength in attracting EV car enterprises. Market scale (X2) significantly affects only the downstream segment, reflecting the market-oriented nature of EV car services. Technological innovation (X3) is crucial for upstream and midstream segments, as EV car components and manufacturing rely on advanced technologies. Industrialization level (X4) negatively impacts all segments, suggesting that traditional industrial structures may not align with the innovation-driven demands of EV car development. The automotive industrial base (X5) positively influences midstream EV car manufacturing, indicating path dependence and agglomeration economies. Government technology investment (X6) and EV car industrial policy (X7) positively affect upstream and midstream segments, underscoring the role of government support in fostering technology-intensive industries. Locational condition (X8) negatively affects upstream enterprises, implying that central and western regions are gaining competitiveness in EV car component manufacturing.
In conclusion, the spatial distribution of EV car enterprises is shaped by a complex interplay of factors varying across the industrial chain. Upstream EV car components are driven by economic foundation, technological innovation, and government guidance, while midstream EV car manufacturing is influenced by industrial base and policy support. Downstream EV car services depend heavily on market scale and economic conditions. To optimize the spatial layout of the EV car industry, policymakers should consider segment-specific strategies, promote industrial upgrading, enhance regional cooperation, and leverage industrial parks to cluster EV car enterprises. These measures can help harness the potential of EV cars as a catalyst for regional development and new quality productivity.
Future research could incorporate additional indicators such as industrial output and employment to construct a comprehensive evaluation system for the EV car industry. Longitudinal data and panel models may further elucidate the temporal dynamics of influencing factors. Despite limitations, this study provides a foundational understanding of the spatial patterns and drivers of EV car enterprises, contributing to informed decision-making in the evolving landscape of electric mobility.