Green Technology Innovation Efficiency of New Energy Vehicles Based on Corporate Profitability Perspective

Abstract

In the context of global climate change and the escalating energy crisis, the development of new energy vehicles (NEVs) has become a critical strategy for China to foster green transformation and achieve its carbon neutrality goals. This study focuses on A-share-listed NEV companies in China from 2015 to 2023, specifically those listed on the Shanghai or Shenzhen Stock Exchange and subject to domestic regulatory standards and disclosure requirements. These firms were selected due to the representativeness, availability, and quantifiability of their data. A super-efficient-network SBM model based on undesirable outputs and the Malmquist index were employed to assess the static and dynamic green technology innovation efficiency of 260 NEV enterprises. Additionally, the Tobit regression model was applied to analyze the influencing factors. The findings reveal that the overall green technology innovation efficiency of Chinese NEV enterprises is relatively low and has exhibited a declining trend over the years. Furthermore, the efficiency of enterprises in the western regions surpasses that of those in the eastern and central regions. Key factors, including government support, enterprise scale, and R&D investment, significantly inhibit the green technology innovation efficiency of firms. Based on these findings, this paper recommends prioritizing the innovation of core technologies, addressing regional disparities in development, and implementing tailored policies to enhance the green technology innovation efficiency and economic performance of NEV enterprises.

1. Introduction

Promoting green technology innovation activities within enterprises is a key initiative for achieving high-quality national economic development [1]. An increasing number of industries are recognizing the inevitable trend of sustainable development, with the new energy vehicle (NEV) industry emerging as a strategic sector that combines technological innovation with energy conservation and emission reduction. This industry has become a focal point for countries around the world to drive low-carbon economic transformation and achieve green growth [2]. Compared to traditional high-energy-consumption and high-fuel-emission vehicles, the promotion of NEVs can address resource depletion and environmental pollution to some extent [3]. Globally, governments have gradually prioritized the development of the new energy vehicle industry as part of their national strategies and have introduced a series of policy measures to actively foster the industry’s growth. For example, the European Union’s Green Deal includes a ban on the sale of fuel vehicles by 2035, alongside increased support for NEV manufacturing and charging infrastructure [4]. In the United States, the Inflation Reduction Act has established subsidy mechanisms for new energy manufacturing companies, with a strong focus on developing domestic NEV and battery industries to reduce reliance on overseas EV supply chains [5]. According to the Global Electric Vehicle Outlook 2024, global sales of new energy vehicles (NEVs) reached 14.8 million units in 2023, marking a 35% year-on-year increase. China has played a significant role in this global trend, with its market share of NEVs continuing to grow, driven by the country’s strategic energy deployment and supportive policies for the NEV industry [6]. In 2024, the total production of new energy vehicles in China reached approximately 12.888 million units, with cumulative sales reaching 12.866 million units, representing a year-on-year growth of 34.4% and 35.5%, respectively [7].

However, despite the positive development of China’s NEV industry, several challenges remain. First, in terms of core NEV technology, issues related to battery recovery and recycling are more pronounced [8]. Most companies have established their own battery recycling facilities; however, high investment costs and low recycling efficiency have hindered the development of a scientifically sound and effective recycling system, while also posing certain safety risks [9]. Moreover, there are challenges in the battery recycling process, such as low material sorting efficiency and poor levels of automation in equipment [10]. Secondly, from the perspective of enterprise sustainability, achieving long-term profitability is a key concern. Currently, NEV enterprises face weak profitability, with major car manufacturers engaging in price competition to capture market share, which squeezes their profit margins [11]. For instance, BYD’s Qin and Destroyer series occupies over 70% of the compact car market with ultra-low prices. This results in a mismatch between R&D investments and outputs, where blind over-investment without focusing on the overall efficiency of green technology innovation may lead to a waste of resources. Therefore, considering the green technology innovation efficiency of NEVs is crucial not only for solving the problems of high pollution and energy consumption, but also for ensuring maximum profitability for enterprises. This will promote the high-quality development of the NEV industry, while playing a vital role in driving the nation’s green transformation. Currently, research on green technology innovation efficiency primarily focuses on achieving low pollution and energy consumption. In contrast, this paper will explore whether improving the efficiency of green technology innovation can promote the sustainable development of enterprises from the perspective of profitability.

In view of the aforementioned problems, this paper selects China’s A-share-listed new energy vehicle (NEV) enterprises from 2015 to 2023 as the research sample. Based on theoretical analysis, the paper constructs a super-efficient-network SBM model incorporating undesirable outputs to measure the green technology innovation efficiency of NEV enterprises. Both static and dynamic approaches are employed to analyze the trends in green technology innovation efficiency, while the paper also explores the heterogeneous effects across regions. Furthermore, the Tobit model is used to analyze the contributing factors to the results, with the aim of offering practical suggestions to improve the green technology innovation efficiency of enterprises. The marginal contribution of this paper lies in its novel perspective of corporate profitability. The study constructs an evaluation index system for green technology innovation efficiency in the NEV sector, using indicators such as the number of green patents, corporate operating income, and profitability as output metrics. Additionally, both desired and undesirable outputs are considered, offering a comprehensive and objective assessment of the green technology innovation efficiency of NEV enterprises from both static and dynamic perspectives. This enriches the existing literature on green technology innovation efficiency in the NEV industry. Moreover, this paper analyzes the heterogeneity in green technology innovation efficiency based on the geographic location of NEV enterprises. It examines regional differences in efficiency and investigates the underlying causes of these disparities. It provides targeted recommendations for both government policy-making and corporate decision-making, contributing to more informed strategies in the NEV industry.

2. Literature Review

2.1. The Concept of Green Technological Innovation Efficiency

The efficiency of technological innovation is characterized by the gap between desired and actual outputs—when this gap narrows, innovation efficiency increases [12], assuming consistent innovation inputs. Green technological innovation efficiency extends this concept by simultaneously emphasizing input–output optimization and environmental sustainability. It specifically measures the ratio between the positive effects generated by green technologies throughout their research, development, and application processes and the resources invested, effectively quantifying green technological innovation outcomes per unit of resource input.

Scholars have examined green technological innovation efficiency primarily through two distinct lenses. The first perspective focuses on regional analysis, where researchers have employed data envelopment analysis models to evaluate and compare technological innovation efficiency and green technological innovation efficiency across 30 European countries, revealing significant inter-country efficiency variations [13]. Domestic research has similarly investigated spatial and temporal differences in green technological innovation efficiency within China’s Yangtze River Delta from 2010 to 2017, identifying substantial regional efficiency disparities [14] and demonstrating that efficiency centers may shift concurrently with economic centers [15]. The second perspective examines industry-specific green technological innovation efficiency. By analyzing data from manufacturing enterprises across 30 Chinese provinces, researchers documented a declining trend in green technological innovation efficiency within China’s manufacturing sector during 2003–2010 [16]. Other scholars have applied super-efficiency SBM models to measure the green technological innovation efficiency of industrial enterprises across various Chinese provinces, employing both dynamic and static approaches to analyze spatiotemporal evolution patterns [17]. Additional research focusing on China’s heavily polluting industries has identified relatively low overall green technological efficiency levels in these sectors, indicating substantial potential for improvement [18].

2.2. Measurement of Green Technological Innovation Efficiency

In terms of measuring green technological innovation efficiency, early scholars adopted two main methodological approaches: single-indicator and multi-indicator evaluation methods. The single-indicator evaluation method employs just one metric to characterize green technological innovation efficiency, capturing the inputs and outputs of the subject entity in a specific dimension. However, this approach inherently lacks comprehensiveness in its data representation. Conversely, the multi-indicator evaluation method typically utilizes principal component analysis to select multiple indicators, weighting them to calculate green technological innovation efficiency. While this approach offers greater breadth, the subjective nature of indicator selection compromises the objectivity of the resulting data.

With the continuous advancement of green technological innovation efficiency research, scholars have predominantly adopted two sophisticated measurement approaches: data envelopment analysis (DEA) and stochastic frontier analysis (SFA). Stochastic frontier analysis, initially proposed by Schmidt (1977) and Broeck (1977), represents a parametric methodology grounded in production frontier theory for measuring producers’ technical efficiency. This method’s primary advantage lies in its incorporation of a stochastic perturbation term alongside frontier determination, enabling more comprehensive characterization of producer behavior.

At the regional level, researchers have employed stochastic frontier analysis to measure technical efficiency [19] and innovation R&D efficiency [20] across Chinese provinces, municipalities, and regions to identify trends and investigate how various factors influence innovation efficiency. At the industry level, Mu (2023) constructed a three-stage SFA model to measure green technological innovation efficiency within advanced manufacturing industries across 30 Chinese provinces and cities, revealing steady upward efficiency trends in certain regions, alongside a clear east–central–west declining gradient with pronounced regional disparities [21]. Additionally, Yi (2019) applied SFA to conclude that China’s overall green technological innovation efficiency remains relatively low [22]. While stochastic frontier analysis typically accommodates multiple inputs with single outputs, green technological innovation efficiency inherently involves multiple inputs and outputs—a challenge effectively addressed by DEA. This non-parametric method evaluates decision-making unit efficiency through linear programming, which is particularly suitable for comparable entities. Some studies have employed traditional homo-radial DEA to measure green technological innovation efficiency [23]. However, recognizing the presence of undesirable outputs in green technological innovation processes, Tone (2001) developed the Super-SBM model, incorporating realistic factors such as energy consumption and pollutant emissions [24]. Many scholars have subsequently adopted this method to measure static green technological innovation super-efficiency [25], while employing the global Malmquist–Luenberger index to assess dynamic changes [26]. Furthermore, innovation value chain theory identifies a “black box” phenomenon within green technological innovation activities, which a network SBM model can penetrate by examining stage-specific efficiency [27]. Chen (2021) utilized a network SBM model to evaluate Chinese new energy enterprises’ green technological innovation efficiency, identifying low transformation efficiency in the second stage as the critical factor limiting overall green technological innovation efficiency [28].

2.3. Key Factors Affecting Green Technology Innovation

Concerning the determinants of enterprise green technological innovation, this area has emerged as a focal point for scholarly investigation globally. Based on the existing literature, researchers typically analyze these influence factors through two perspectives: external and internal dimensions.

Regarding external influences, scholars have identified three primary factors significantly impacting enterprises’ green technological innovation activities.

First, from an environmental regulation perspective, divergent scholarly opinions exist. Some researchers maintain that appropriate environmental regulations motivate enterprises to pursue green technological innovation [29,30]. Under dual government and market regulations, enterprises increase environmental protection investments, with innovation-generated economic benefits internalizing environmental regulation costs, thereby stimulating further technological innovation. Conversely, other scholars argue that environmental regulation and green technological innovation exhibit a non-linear relationship [31,32,33]—as regulatory intensity increases, its effect transitions from facilitation to inhibition, with excessive regulation potentially undermining enterprise market competitiveness and development [34]. Additionally, some researchers suggest that environmental regulations inherently inhibit green technological innovation by depleting enterprise resources, reducing operational efficiency and consequently diminishing green technology investments [35,36]. Second, from a governmental perspective, financial subsidies provided to enterprises support green technological innovations. Furthermore, government-demonstrated environmental concern raises public environmental awareness, creating societal oversight networks that prompt enterprises to undertake green technological innovation [37]. Third, from a market perspective, market mechanisms enhance green technological innovation levels by improving enterprise resource allocation efficiency, optimizing industrial structures, and mitigating technological innovation risks [38]. Additionally, financial policies alleviate enterprise financing constraints, improving funding-related motivations for enhancing green technological innovation.

Regarding internal influences, scholars have identified three primary factors significantly impacting enterprises’ green technological innovation activities: First, from the perspective of enterprise culture, enterprise culture largely determines the direction of an enterprise’s business strategy and values; at the same time, the innovation and environmental awareness of the top management of the enterprise also promotes the dissemination of the organizational culture within the organization, drives the innovation atmosphere of the enterprise, and guides the employees of the enterprise to practice the concept of green innovation and development [39,40]. Second, from the perspective of enterprise scale and enterprise R&D investment intensity, enterprise R&D investment positively promotes enterprise green technology innovation, which is due to the fact that R&D investment means that enterprises have more resources to improve their own green innovation efficiency [41]. In addition, change in industrial structure will bring changes to the human and capital of enterprises, which affects the green technological innovation of enterprises.

In conclusion, while numerous studies have addressed green technology innovation efficiency and its influencing factors, several limitations remain evident in the current literature. Firstly, a research gap exists at the level of research objects. Most of the existing research analyses green technology innovation efficiency from a regional macro-level perspective, and there is a lack of research related to green technology innovation efficiency in the new energy automobile industry. Enterprises in this industry have unique technological pathways and policy environments, making them important subjects for investigation. Secondly, there is a gap regarding enterprise heterogeneity. The current literature lacks comprehensive analyses of internal and external drivers influencing green technology innovation efficiency at the firm level. Thus, more focused micro-level research is needed to identify key factors and optimization strategies affecting green technology innovation efficiency in new energy vehicle enterprises. Thirdly, a methodological gap remains. Most studies adopt traditional DEA models with single-stage measurements based on R&D inputs and outputs, which insufficiently capture the complexity of the green innovation process. Few studies incorporate a value chain perspective, and even fewer employ two-stage models that distinguish between technology R&D and technological achievements. A two-stage network SBM model better represents the entire production process, providing a more thorough and accurate evaluation of green technology innovation efficiency.

3. Research Design

3.1. Model Specification

This paper adopts a super-efficient-network SBM model to measure the green technology innovation efficiency of new energy automobile enterprises, which is an advanced DEA method that effectively measures this efficiency by considering multiple inputs, desired outputs, and undesired outputs. Accordingly, it also takes into account the existence of the “black box” problem in the process of green technological innovation activities. By decomposing the innovation process into stages, the network SBM model provides a more accurate and comprehensive assessment of green technology innovation efficiency.

3.1.1. Network SBM Moder

Decision-making units (DMUs) refer to the objects whose efficiency is evaluated in a DEA model, which are entities with the same objectives and similar input–output structures. In this study, each new energy vehicle company is a DMU, and the DEA model compares the efficiency of these DMUs under the same criteria to see which companies achieve the maximum effective output with a certain amount of resources invested. The network SBM model can solve the “black box” problem of each DMU by measuring the single efficiency value of each stage of production to find out the reason for the low total efficiency value. This is expressed as follows: there are n ${\mathrm{D}\mathrm{M}\mathrm{U}}_{\mathrm{j}}$ (j = 1, 2..., n, n = 260); a DMU has k nodes (k = 2 in this paper), and ${\mathrm{m}}_{\mathrm{k}}$ and ${\mathrm{r}}_{\mathrm{k}}$ correspond to the number of input indicators and the number of output indicators, respectively. ${\mathrm{X}}_{\mathrm{j}}^1$ = ${(\mathrm{X}}_{1\mathrm{j}}^1,{\mathrm{X}}_{2\mathrm{j}}^1,\dots ,{\mathrm{X}}_{\mathrm{a}\mathrm{j}}^1$) represents the input vector of the jth DMU in the R&D stage, and the middle output variable is ${\mathrm{Z}}_{\mathrm{j}}^1$ = ${(\mathrm{Z}}_{1\mathrm{j}},{\mathrm{Z}}_{2\mathrm{j}},\dots ,{\mathrm{Z}}_{\mathrm{t}\mathrm{j}})$. The input of the transformation stage is ${\mathrm{X}}_{\mathrm{j}}^2$ = ${(\mathrm{X}}_{1\mathrm{j}}^2,{\mathrm{X}}_{2\mathrm{j}}^2,\dots ,{\mathrm{X}}_{\mathrm{b}\mathrm{j}}^2$), and the output is ${\mathrm{Y}}_{\mathrm{j}}^2$ = ${(\mathrm{Y}}_{1\mathrm{j}}^2,{\mathrm{Y}}_{2\mathrm{j}}^2,\dots ,{\mathrm{Y}}_{\mathrm{r}\mathrm{j}}^2$). ${\mathrm{W}}^{\mathrm{k}}$ is the weight vector; the weight is 0.5. The specific formula is as follows:

$${\mathsf{\rho }}^{\mathrm{*}}=\mathrm{m}\mathrm{i}\mathrm{n}{\displaystyle \frac{\sum _{\mathrm{k}=1}^2{\mathrm{w}}^{\mathrm{k}}\left(1-\frac{1}{{\mathrm{m}}_{\mathrm{k}}}\sum _{\mathrm{i}=1}^{{\mathrm{m}}_{\mathrm{k}}}\frac{{\mathrm{s}}_{\mathrm{i}0}^{\mathrm{k}-}}{{\mathrm{x}}_{\mathrm{i}0}^{\mathrm{k}}}\right)}{\sum _{\mathrm{k}=1}^2{\mathrm{w}}^{\mathrm{k}}\left(1+\frac{1}{{\mathrm{r}}_{\mathrm{k}}}\sum _{\mathrm{r}=1}^{{\mathrm{r}}_{\mathrm{k}}}\frac{{\mathrm{s}}_{\mathrm{r}0}^{\mathrm{k}+}}{{\mathrm{y}}_{\mathrm{r}0}^{\mathrm{k}}}\right)}}$$

(1)

3.1.2. SBM Super-Efficiency Model Incorporating Undesired Outputs

Since the efficiency values of many DMU units in the DEA measurements were equal to 1, it was not possible to further distinguish the size of each DMU unit. The super-efficiency SBM model, on the other hand, was able to calculate the relative efficiency values of effective DMUs, so as to further rank and distinguish them. Meanwhile, considering the existence of undesired outputs such as environmental pollution in the production process of the enterprises, a super-efficiency SBM model containing undesired outputs was selected in this study, and the formula is shown below:

$$\begin{gathered} \mathsf{\rho }=\mathrm{m}\mathrm{i}\mathrm{n}{\displaystyle \frac{1+\frac{1}{\mathrm{m}}\sum _{\mathrm{i}=1}^{\mathrm{m}}\frac{{\mathrm{s}}_{\mathrm{i}}^{-}}{{\mathrm{x}}_{\mathrm{i}\mathrm{k}}}}{1-\frac{1}{{\mathrm{q}}_1+{\mathrm{q}}_2}\left(\sum _{\mathrm{r}=1}^{{\mathrm{q}}_1}\frac{{\mathrm{s}}_{\mathrm{t}}^{+}}{{\mathrm{y}}_{\mathrm{r}\mathrm{k}}}+\sum _{\mathrm{t}=1}^{{\mathrm{q}}_2}\frac{{\mathrm{s}}_{\mathrm{t}}^{\mathrm{b}-}}{{\mathrm{b}}_{\mathrm{r}\mathrm{k}}}\right)}} \\ \mathrm{s}.\mathrm{t}.\left\{\begin{array}{l}{\displaystyle \sum _{\mathrm{j}=1,\mathrm{j}\ne \mathrm{k}}^{\mathrm{n}}}{\mathsf{\lambda }}_{\mathrm{j}}{\mathrm{x}}_{\mathrm{i}\mathrm{j}}-{\mathrm{s}}_{\mathrm{i}}^{-}\le {\mathrm{x}}_{\mathrm{i}\mathrm{k}}\\ {\displaystyle \sum _{\mathrm{j}=1,\mathrm{j}\ne \mathrm{k}}^{\mathrm{n}}}{\mathsf{\lambda }}_{\mathrm{j}}{\mathrm{y}}_{\mathrm{r}\mathrm{j}}+{\mathrm{s}}_{\mathrm{r}}^{+}\ge {\mathrm{y}}_{\mathrm{r}\mathrm{k}}\\ {\displaystyle \sum _{\mathrm{j}=1,\mathrm{j}\ne \mathrm{k}}^{\mathrm{n}}}{\mathsf{\lambda }}_{\mathrm{j}}{\mathrm{b}}_{\mathrm{r}\mathrm{j}}-{\mathrm{s}}_{\mathrm{r}}^{\mathrm{b}-}\le {\mathrm{b}}_{\mathrm{r}\mathrm{k}}\\ \mathsf{\lambda }\ge 0,{\mathrm{s}}^{-}\ge 0,{\mathrm{s}}^{+}\ge 0,\mathrm{i}=1,2,\dots ,\mathrm{m},\mathrm{r}=1,2,\dots ,\mathrm{q},\mathrm{j}=1,2,\dots ,\mathrm{n}\left(\mathrm{j}\ne \mathrm{k}\right)\end{array}\right. \end{gathered}$$

(2)

In Equation (2), $\mathsf{\rho }$ is the target value of efficiency, and there are n DMUs. Each DMU has m inputs, which are denoted as ${\mathrm{x}}_{\mathrm{i}}$ (i = 1, 2, …, m); desired outputs, denoted as (r = 1, 2, …, m); ${\mathrm{q}}_1$ desired outputs, denoted as ${\mathrm{y}}_{\mathrm{r}}$ (r = 1, 2, …, ${\mathrm{q}}_1$); and ${\mathrm{q}}_2$ undesired outputs, denoted as ${\mathrm{b}}_{\mathrm{r}}$ (r = 1, 2, …, ${\mathrm{q}}_2$). ${\mathrm{s}}^{-}$ represents input redundancy, ${\mathrm{s}}^{\mathrm{g}}$ represents desired output shortfalls, and ${\mathrm{s}}^{\mathrm{b}}$ represents undesired output excesses. λ is a vector of weights.

3.1.3. Malmquist–Luenberger Index Model

The traditional Malmquist index model does not take into account the environmental factors in production activities, while the Malmquist–Luenberger index model can more comprehensively measure changes in the efficiency value of a production process through the introduction of the environmental factors; therefore, this paper chooses to consider the unexpected output of the Malmquist–Luenberger model. Meanwhile, the Malmquist–Luenberger index model can be further decomposed into technical efficiency change (EC) and technical change (TC) using the formula shown below:

$$\mathrm{M}\left({\mathrm{x}}_{\mathrm{r}}^{\mathrm{t}+1},{\mathrm{y}}_{\mathrm{r}}^{\mathrm{t}+1},{\mathrm{x}}_{\mathrm{r}}^{\mathrm{t}},{\mathrm{y}}_{\mathrm{r}}^{\mathrm{t}}\right)={\displaystyle \frac{{\mathrm{D}}^{\mathrm{t}+1}\left({\mathrm{x}}_{\mathrm{r}}^{\mathrm{t}+1},{\mathrm{y}}_{\mathrm{r}}^{\mathrm{t}+1}\right)}{{\mathrm{D}}^{\mathrm{t}}\left({\mathrm{x}}_{\mathrm{r}}^{\mathrm{t}},{\mathrm{y}}_{\mathrm{r}}^{\mathrm{t}}\right)}}\times [{\displaystyle \frac{{\mathrm{D}}^{\mathrm{t}}\left({\mathrm{x}}_{\mathrm{r}}^{\mathrm{t}+1},{\mathrm{y}}_{\mathrm{r}}^{\mathrm{t}+1}\right)}{{\mathrm{D}}^{\mathrm{t}+1}\left({\mathrm{x}}_{\mathrm{r}}^{\mathrm{t}+1},{\mathrm{y}}_{\mathrm{r}}^{\mathrm{t}+1}\right)}}\times {\displaystyle \frac{{\mathrm{D}}^{\mathrm{t}}\left({\mathrm{x}}_{\mathrm{r}}^{\mathrm{t}},{\mathrm{y}}_{\mathrm{r}}^{\mathrm{t}}\right)}{{\mathrm{D}}^{\mathrm{t}+1}\left({\mathrm{x}}_{\mathrm{r}}^{\mathrm{t}},{\mathrm{y}}_{\mathrm{r}}^{\mathrm{t}}\right)}}{]}^{{\displaystyle \frac{1}{2}}}=EC\times TC$$

(3)

where ${\mathrm{x}}_{\mathrm{r}}^{\mathrm{t}}$ and ${\mathrm{x}}_{\mathrm{r}}^{\mathrm{t}+1}$ denote inputs in period t and t + 1, respectively; ${\mathrm{y}}_{\mathrm{r}}^{\mathrm{t}}$ and ${\mathrm{y}}_{\mathrm{r}}^{\mathrm{t}+1}$ denote outputs in period t and t + 1, respectively; ${\mathrm{D}}^{\mathrm{t}}\left({\mathrm{x}}_{\mathrm{r}}^{\mathrm{t}},{\mathrm{y}}_{\mathrm{r}}^{\mathrm{t}}\right)$ is the level of technical efficiency in period t; and ${\mathrm{D}}^{\mathrm{t}+1}\left({\mathrm{x}}_{\mathrm{r}}^{\mathrm{t}},{\mathrm{y}}_{\mathrm{r}}^{\mathrm{t}}\right)$ is the level of technical efficiency in period t + 1. When M > 1, the total efficiency shows an upward trend, and vice versa.

3.1.4. Tobit Modeling

To further investigate the reasons behind the year-on-year decline in green technology efficiency among new energy vehicle companies, it is essential to analyze the factors contributing to the decrease in their green technology innovation efficiency. Given that the efficiency values measured by DEA in this study were truncated, the use of the Tobit model for regression analysis helped to mitigate potential errors in the regression results. Three factors affecting the green technological innovation efficiency of enterprises were selected, namely, enterprise size (Insize), enterprise R&D investment (Rd), and government support (Gov). The Tobit model constructed in this paper is as follows:

$${\mathrm{G}\mathrm{T}\mathrm{F}\mathrm{P}}_{\mathrm{i}\mathrm{t}}={\mathsf{\alpha }}_0+{\mathsf{\alpha }}_1\mathrm{I}\mathrm{n}\mathrm{s}\mathrm{i}\mathrm{z}\mathrm{e}+{\mathsf{\alpha }}_2\mathrm{R}\mathrm{d}+{\mathsf{\alpha }}_3\mathrm{G}\mathrm{o}\mathrm{v}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$$

(4)

where ${\mathrm{G}\mathrm{T}\mathrm{F}\mathrm{P}}_{\mathrm{i}\mathrm{t}}$ represents the green technology innovation efficiency of new energy automobile enterprises as an explanatory variable; ${\mathsf{\alpha }}_0$ is a constant term; ${\mathsf{\alpha }}_1$, ${\mathsf{\alpha }}_2$, and ${\mathsf{\alpha }}_3$ represent the estimated coefficients of each influential factor; and ${\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$ is a random error term.

3.2. Variables Chosen

3.2.1. Input–Output Indicators for the R&D Phase of Green Technologies

Green technology innovation and research and development (R&D) in enterprises are closely tied to the availability of high-quality human resources and strong financial capabilities. Firms with technology R&D personnel possessing expertise in relevant fields are better positioned to integrate knowledge and foster the creation of new technologies, thereby driving R&D activities. Simultaneously, firms with adequate financial resources are more likely to invest in a range of innovation initiatives. Given the availability of data, this paper draws upon the work of Xiao (2020) and adopts the number of R&D personnel and R&D expenditures as input indicators in the first stage of analysis [42].

Regarding output indicators, patents represent the tangible outcomes of R&D activities undertaken by enterprises. Given that this paper aims to measure green technology innovation efficiency, green patent applications and authorizations serve as key indicators for assessing the results of green innovation activities at the enterprise, industry, or regional level. Therefore, the number of green patent applications and authorizations by new energy vehicle enterprises were selected as the output indicators in this stage of analysis. An increase in patent applications reflects active innovation efforts, while a rise in patent authorizations indicates that the innovation results have been legally protected and recognized.

3.2.2. Input–Output Indicators at the Green Transformation Stage

The output indicators from the R&D phase served as input indicators for the second phase. Additionally, since pollutant emissions from energy consumption represented the primary source of undesired outputs in this phase, the total energy consumption of new energy vehicle enterprises was used as an input indicator for the transformation phase.

For output indicators, enterprise profit levels and margins are crucial measures of R&D outcomes. Profit is the net income generated through business activities; it directly reflects an enterprise’s efficiency in utilizing capital, labor, and technology. Therefore, operating income served as the desired output indicator at this stage. Profitability demonstrates an enterprise’s ability to effectively control costs while achieving high returns. Due to difficulties in obtaining data on solid waste generation and wastewater emissions, the study used enterprise emissions of sulfur dioxide, nitrogen oxides, and particulate matter (soot and dust) as undesired output indicators for the transformation stage. This is detailed in

Table 1. Variable names and descriptions.

Point	Indicator Category	Indicator Name	Description of Indicators
R&D phase	Manpower inputs	R&D personnel equivalent ${\mathrm{X}}_1^1$	Number of enterprise R&D personnel/number of persons in enterprise (persons)
	Capital investment	R&D investment intensity ${\mathrm{X}}_2^1$	Enterprise R&D investment expenditure/enterprise revenue
	Intermediate outputs	Green patent applications ${\mathrm{Y}}_1^1\mathrm{or}{\mathrm{X}}_1^2$	Total number of green patents obtained by enterprises independently and jointly
Transformation phase		Green patent grants ${\mathrm{Y}}_2^1\mathrm{or}{\mathrm{X}}_2^2$	Number of invention green patents granted to enterprises
	Energy inputs	Total energy consumption ${\mathrm{X}}_3^1$	Tons of standard coal
	Expected outputs	Enterprise business income ${\mathrm{Y}}_1^2$	Revenue from business
		Corporate profitability ${\mathrm{Y}}_2^2$	Corporate net profit
	Unexpected outputs	Total pollutant emission intensity ${\mathrm{Y}}_3^2$	Enterprise emissions of sulfur dioxide, nitrogen oxides, and particulate matter (soot and dust)

.

3.2.3. Influencing Factors

Building on the previous compilation of factors influencing enterprise green technology innovation, this paper adopts both internal and external perspectives to analyze the efficiency of green technology innovation in new energy vehicle companies. Based on the characteristics of these enterprises and a review of the existing literature, three key indicators were selected to measure influencing factors:

Enterprise size (Insize): The size of an enterprise can influence its innovation activities [43]. This paper measures enterprise size by the total assets at the year-end, taking the logarithm of the value to capture the scale effect.

Government support (Gov): Government subsidies, such as tax incentives, can reduce the financing costs for enterprises and alleviate financial pressures in innovation activities, ultimately affecting the efficiency of green technology innovation. This paper uses the ratio of government subsidies to total assets as a proxy for the level of government support.

Enterprise R&D investment (Rd): Significant R&D investment is often necessary for green technology innovation. Therefore, the intensity of R&D investment plays a crucial role in determining the efficiency of innovation. This paper measures R&D investment intensity as the ratio of R&D expenditure to total business revenue.

3.3. Sample Selection and Data Sources

The two-stage network SBM model employed in this study requires multiple indicators. Considering data completeness and availability, we selected 260 new energy vehicle listed companies (excluding ST and ST* stocks) from 2015–2023 as our research sample. The data sources included annual reports of listed companies, the China Research Data Service Platform (CNRDS), and the Cathay Pacific database. Additional variables were obtained from the China Statistical Yearbook, the China Environmental Statistics Yearbook, and various provincial and municipal statistical yearbooks.

4. Empirical Analysis

4.1. Measurement of Green Technology Innovation Efficiency in the New Energy Vehicle Industry

This study employed MAXDEA 8.0 software to measure the static efficiency of green technology innovation across 260 A-share-listed new energy vehicle enterprises from 2015 to 2023. We applied a super-efficient-network SBM model incorporating undesired outputs. Additionally, we investigated the dynamic efficiency of each enterprise using a Malmquist index model. This comprehensive approach allowed for an in-depth exploration of the current efficiency levels of green technology innovation among Chinese new energy vehicle enterprises.

4.1.1. Static Analysis of Green Technology Innovation

Based on panel data of inputs and outputs from listed Chinese new energy vehicle enterprises, the results of technological innovation efficiency for the new energy vehicle industry, calculated using Equation (2), are shown in

Table 2. Green technological innovation efficiency in 2015–2023.

	2015	2016	2017	2018	2019	2020	2021	2022	2023	Average
Technical efficiency	0.350	0.312	0.291	0.269	0.254	0.251	0.250	0.242	0.247	0.274
Pure technical efficiency	0.382	0.338	0.316	0.292	0.277	0.272	0.271	0.264	0.268	0.298
Scale efficiency	0.909	0.921	0.921	0.928	0.928	0.933	0.935	0.936	0.933	0.927

. The overall average efficiency value is 0.274, with most enterprises falling below this average. This indicates that green technology innovation efficiency among Chinese new energy vehicle enterprises remains at a relatively low level. Furthermore, significant efficiency gaps exist between enterprises, revealing an imbalance in corporate green development. Additionally, as shown in Figure 1, during the 2015–2023 period, green technology innovation efficiency in Chinese new energy vehicle enterprises displayed a declining trend, decreasing from 0.350 in 2015 to 0.247 in 2023.

Additionally, technical efficiency comprises two elements: pure technical efficiency and scale efficiency. The calculation formula is as follows: technical efficiency = pure technical efficiency × scale efficiency. As shown in Table 2, the mean values of technical efficiency, pure technical efficiency, and scale efficiency for new energy vehicle enterprises are 0.274, 0.298, and 0.927, respectively. This indicates that the overall low efficiency of green technological innovation in these enterprises stems primarily from low pure technical efficiency levels. During the 2015–2023 period, pure technical efficiency decreased annually, while scale efficiency increased. This pattern may be attributed to faster technological breakthroughs in the early stages of green technology development. However, as technology matured, the marginal effect of enterprise innovation gradually diminished. Consequently, enterprises redirected funds toward other business activities to enhance profitability and increase returns on capital. Although this may result in lower returns on green technology innovation, it potentially yields higher returns through expanded enterprise scale and market share.

4.1.2. The Efficiency of Two-Stage Green Technology Innovation

Considering that there is a “black box” problem in the innovation process, this study used a network SBM model to measure the efficiency values of the R&D stage and the results transformation stage separately.

Table 3. Two-stage green technological innovation efficiency values in 2015–2023.

	R&D Phase			Transformation Phase
Year	Technical Efficiency	Pure Technical Efficiency	Scale Efficiency	Technical Efficiency	Pure Technical Efficiency	Scale Efficiency
2015	0.140	0.183	0.854	0.560	0.581	0.964
2016	0.140	0.182	0.863	0.485	0.495	0.980
2017	0.142	0.183	0.862	0.440	0.449	0.980
2018	0.138	0.172	0.885	0.401	0.412	0.970
2019	0.137	0.169	0.896	0.371	0.386	0.959
2020	0.137	0.169	0.897	0.364	0.375	0.970
2021	0.142	0.174	0.898	0.358	0.368	0.971
2022	0.134	0.165	0.907	0.350	0.362	0.966
2023	0.126	0.154	0.910	0.368	0.383	0.955

lists the two-stage green technology innovation efficiency values of new energy automobile enterprises, calculated according to Equation (1). From the comprehensive technical efficiency of the two stages, the technical efficiency of 0.560 in the second stage (results transformation stage) is greater than the technical efficiency value of 0.140 in the first stage (science and technology input stage). As shown in Figure 2, the scale efficiency of both stages is greater than the pure technical efficiency. This may be due to the fact that, in the initial stage, enterprises invest more human and financial resources, but face greater uncertainty. As a result, technological inputs may not immediately translate into market value or practical applications. During this stage, enterprises are primarily preparing for future technological breakthroughs and market demand, leading to a low input–output ratio and, consequently, lower efficiency. In contrast, during the transformation stage, as the technology matures and aligns with market demand, economies of scale can be realized through production expansion, ultimately generating higher economic returns.

4.1.3. Efficiency Based on the Geographic Location of the Enterprise

According to regional divisions, the new energy automobile enterprises are divided into three regions, namely east, center, and west, according to the region they are located in. As shown in

Table 4. Green technological innovation efficiency in different locations in 2015–2023.

	2015	2016	2017	2018	2019	2020	2021	2022	2023	Average Value
Eastern region	0.352	0.315	0.293	0.27	0.255	0.251	0.247	0.241	0.246	0.274
Central region	0.326	0.298	0.287	0.268	0.25	0.25	0.258	0.243	0.249	0.27
Western region	0.376	0.333	0.29	0.272	0.259	0.255	0.262	0.264	0.257	0.285

, in 2015–2023, the average value of green technology innovation efficiency of enterprises in the eastern region was 0.274, the average value in the central region was 0.270, and the average value in the western region was 0.285. In addition, as shown in Figure 3, the green technology innovation efficiencies of enterprises in the eastern region and the central region are relatively close to each other, and the enterprises in the western region are larger than those in the eastern and central regions as a whole. Possible reasons for this are that the competition in the new energy vehicle market in the eastern and central regions is relatively fierce, and enterprises pay more attention to the pursuit of short-term sales profits and scale expansion in order to survive, while green technology innovation activities require a large amount of investment in manpower and financial resources, and it is difficult to obtain a return in a short period of time; furthermore, if enterprises only carry out green technological innovation in response to environmental policy, this may lead to a decline in corporate profits, capital turnover difficulties, and other phenomena. The new energy automobile market in the western region faces a number of challenges. However, the new energy vehicle market in the western region faces less competitive pressure and the state’s support for enterprises in the western region is greater, while new energy vehicle enterprises in the western region have a slower start. Therefore, enterprises in this region rely more on technological innovation to create differentiation advantages, and thus may pay more attention to R&D and innovation in green technology in order to improve the added value and market attractiveness of their products, and ultimately achieve an improvement in their operating efficiency. Meanwhile, the energy consumption structure in the western region may be relatively dependent on traditional energy sources, such as coal, which puts a certain pressure on the environment. This forces local new energy vehicle enterprises to seek breakthroughs in green technology innovation in order to reduce energy consumption and enhance their market competitiveness.

4.2. The Dynamic Efficiency of Green Technology Innovation

The Malmquist–Luenberger (ML) index combines traditional productivity and environmental factors to examine dynamic changes in green technology innovation efficiency, which can be decomposed into technical efficiency and technical progress indices.

Table 5. Dynamic efficiency of Malmquist–Luenberger index in 2015–2023.

Year	EC	TC	PE	SE	ML
2015–2016	1.054	0.858	1.063	0.992	0.905
2016–2017	1.022	0.884	1.013	1.009	0.904
2017–2018	0.986	0.906	1.002	0.984	0.893
2018–2019	1.004	0.921	1.002	1.003	0.924
2019–2020	0.998	0.952	0.998	1	0.95
2020–2021	1.001	0.944	1.005	0.996	0.945
2021–2022	0.959	0.973	1	0.959	0.934
2022–2023	0.962	1.094	0.913	1.053	1.053
Mean	0.998	0.939	0.999	0.999	0.937

presents the changes in green technology innovation efficiency and decomposition indices for Chinese new energy vehicle enterprises. The average ML index value is 0.937, indicating that overall green technological innovation efficiency exhibited a negative growth trend during 2015–2023, with an average annual decline of 6.3%. During the examination period, no new energy vehicle enterprise achieved an ML index greater than or equal to 1. However, 132 enterprises (50.8% of the sample) recorded ML indices above the average value. These findings demonstrate the currently poor overall development of green technology innovation efficiency among Chinese new energy vehicle enterprises. The decomposition of the ML index reveals that both technical progress and technical efficiency indices showed negative growth trends, with average annual decline rates of 6.1% and 0.2%, respectively. This substantial difference in decline rates may be attributed to the nature of technological progress, which requires significant initial investments and faces long return cycles. When enterprise investments in technological innovation fail to transform into timely technological breakthroughs, the technical progress index decreases significantly. In contrast, technical efficiency improvements rely primarily on gradual enhancements to existing production processes, reduced energy consumption, and increased productivity, resulting in a slower decline rate. Consequently, enterprise managers, pressured by investors and capital markets, may reduce investments in green technology innovation activities in favor of improving short-term technical efficiency and profitability.

Regarding ML index rate changes, Table 5 shows that new energy vehicle enterprises achieved an ML index greater than 1 in 2022–2023, realizing a growth in green technology innovation efficiency of 5.3%. In all other years, enterprise ML indices displayed negative growth. Throughout the examination period, the ML index fluctuated, but demonstrated a clear upward trend during 2021–2023. Enterprise technical efficiency indices generally fluctuated between 2015 and 2021, with the highest growth rate of 5.4% occurring in 2015–2016. However, after 2021, they began showing a negative growth trend. Concurrently, the growth rate of the technical progress index decreased annually until 2022–2023, when it achieved positive growth of 9.4%. These findings suggest that while the positive impacts of technological progress offset the negative effects of technical inefficiency, enterprises struggled to effectively transform newly developed green technologies into efficiency outputs. To further explore the factors driving technical efficiency index changes, we decomposed this index into pure technical efficiency (PE) and scale efficiency (SE) indices. In all years except 2019–2020 and 2022–2023, the pure technical efficiency change index exceeded the scale efficiency change index, indicating that technical efficiency growth primarily depended on pure technical efficiency improvements.

4.3. Tobit Model Result Analysis

Building on the previous summary of the factors influencing the green technology innovation efficiency of new energy vehicle companies, this paper selects both internal and external indicators to comprehensively analyze and explore the key determinants of innovation efficiency in these companies. The aim is to provide theoretical insights to enhance the green technology innovation efficiency within the sector. This study used the Tobit regression model to explore the effect of different influencing factors on the efficiency of green technology innovation.

Table 6. Regression results of influencing factors on technological innovation efficiency.

Variable Name	Ratio	Standard Error	t-Value	p-Value
Gov	−0.1021	0.0012	−2.87	0.004
Insize	−0.2277	0.1279	−2.26	0.024
Rd	−0.0012	0.0004	−5.66	0.000
_cons	0.2218	0.0208	105.00	0.000

shows the results of the regression of influencing factors on the efficiency of green technology innovation of new energy automobile enterprises.

The strength of government support shows a significant negative effect on the green technology innovation efficiency of enterprises, with an estimated coefficient of −0.1021. This suggests that greater government support does not necessarily promote green innovation activities. One possible explanation is that increased reliance on government subsidies may cause enterprises to lose their initiative for independent innovation, diminishing their drive for technological breakthroughs and, consequently, reducing green technology innovation efficiency. Additionally, due to the uncertainty surrounding government policies and the lack of long-term continuity in tax subsidy programs, enterprises may hesitate to invest substantial resources in long-term green technology innovation activities, which can further hamper innovation efficiency. For instance, as China’s new energy vehicle industry matures and environmental policies evolve, current subsidy schemes are being gradually phased out, creating economic pressure and diminishing incentives for green technology innovation. Moreover, government support often comes with certain compliance requirements, which may restrict the scope of firms’ innovation activities and limit their ability to pursue diverse green innovations.

The estimated coefficient for enterprise size is −0.2277, which is significantly negative at the 5% level, indicating that larger enterprises tend to experience a more pronounced inhibitory effect on their green technology innovation efficiency. This may be due to the fact that large enterprises often have well-established management systems, where decision-making processes require multiple layers of approval. Similarly, innovative ideas from R&D personnel must pass through extensive reporting and approval processes, slowing down the entire decision-making cycle. As a result, resource allocation and utilization become less efficient, and the timing of innovations may be missed, leading to a decline in green technology innovation efficiency. Additionally, large enterprises often have established business models and are typically risk-averse, making them less likely to pursue breakthrough innovations. Their reluctance to embrace the risks associated with innovation may stifle creativity and reduce their innovative vitality. Moreover, the conservative corporate culture in large enterprises, driven by risk-aversion, often leads to a preference for low-risk, incremental innovations, further contributing to the decline in green technology innovation efficiency.

The estimated coefficient for the impact of enterprise R&D investment on green technology innovation efficiency is −0.0012, which is significantly negative at the 1% level. This suggests that higher R&D investment does not necessarily lead to higher green technology innovation efficiency. Several factors may explain this result. First, the effect of increased R&D investment often follows diminishing marginal returns. While initial investments in R&D can yield high returns, as investment continues to grow, enterprises may encounter technical challenges or market barriers, reducing the benefits of further investment and ultimately lowering innovation efficiency. Second, green technology innovation typically involves high-complexity technologies and cross-disciplinary R&D, meaning that the returns on R&D inputs may not be proportional. Enterprises dealing with such complex technological issues may face higher financial and labor costs, leading to longer innovation cycles and decreased efficiency.

Third, excessive R&D investment may cause enterprises to focus too heavily on the technology itself, neglecting market needs. As a result, products may fail to meet consumer demand, leading to lower conversion rates of R&D inputs and inhibiting the overall efficiency of green technology innovation. Fourth, R&D investments may not be limited to green technology, but could also extend to other technological areas. A significant allocation of resources to inefficient R&D projects could negatively impact the efficiency of green technology innovation. Additionally, the number of green patents does not fully reflect the level of innovation efficiency. Some companies may prioritize patent quantity over quality, leading to low conversion rates of patents and, in turn, reducing the overall efficiency of green technology innovation.

This study evaluated and attributed the green technology innovation efficiency of new energy vehicle enterprises in China by employing a super-efficient-network SBM model incorporating unexpected outputs. The empirical results reveal that the overall efficiency level remains low, and has shown a declining trend from 2015 to 2023. These findings are in line with previous studies [22,44]. However, in contrast to the commonly held view in the literature that the R&D stage typically demonstrates higher efficiency than the transformation stage [28], this study finds that new energy vehicle enterprises exhibit higher efficiency in the transformation stage of technological achievements than in the technology input stage. This suggests that inefficiencies in the initial stage may stem from issues such as fragmented R&D inputs and imbalanced resource allocation, resulting in suboptimal input–output performance. Furthermore, this study reveals that enterprises in the western region outperform those in the eastern and central regions in terms of green technology innovation efficiency. This observation diverges from the mainstream perspective that the eastern region generally leads in innovation capability [21]. A possible explanation is the recent westward shift in the focus of government policy support, as well as the emergence of efficient innovation ecosystems in the western region due to strategic concentration on advantageous technological fields. Additionally, the Tobit regression results show that government support intensity, enterprise size, and R&D investment intensity all exert significant negative effects on green technology innovation efficiency. These findings contradict earlier studies that identified these factors as positive drivers of innovation [37,41]. A plausible explanation is that excessive or structurally misaligned support may lead to resource misallocation, over-investment, and weak transformation of innovation inputs into tangible outcomes, ultimately reducing overall efficiency.

5. Conclusions and Discussion

5.1. Conclusions

This paper measured the green technology innovation efficiency of 260 new energy vehicle enterprises from 2015 to 2023 using a two-stage super-efficiency SBM model based on undesirable outputs and the Malmquist–Luenberger index. It also analyzed the regional differences in the development levels of these enterprises according to their locations. The Tobit model was then applied to investigate the factors influencing the green technology innovation efficiency of new energy vehicle enterprises. The following conclusions are drawn:

The overall green technology innovation efficiency of Chinese new energy vehicle enterprises remains at a low level and has shown a declining trend year by year. Additionally, the green technology efficiency during the transformation stage is higher than that in the technology input stage. In terms of dynamic changes, the ML index demonstrated a negative growth trend throughout the study period, with the negative growth rate of the technical progress index surpassing that of the technical efficiency index. This suggests that a decline in technological progress is hindering the improvement of green technology efficiency. The green technology innovation efficiency of China’s new energy vehicle enterprises exhibited a decreasing pattern across the western, eastern, and central regions, with efficiency values in all three regions showing a downward trend. Notably, the green technology innovation efficiency in the eastern and central regions was quite similar, while enterprises in the western region displayed significantly higher efficiency compared to those in the east and center. Building on previous research, this paper identifies possible factors contributing to changes in the green technology innovation efficiency of new energy vehicle enterprises. After conducting regression analysis using the Tobit model, the study found that the strength of government support, enterprise size, and the intensity of enterprise R&D investment all negatively impact green technology innovation efficiency, contributing to its decline.

Based on the above conclusions, the paper makes the following recommendations:

Enterprises can establish a green performance evaluation system, with core quantitative indicators in three areas. First, a hierarchical evaluation mechanism based on Technology Readiness Levels (TRLs) should be implemented. The initial stage focuses on concept development and lab validation with limited funding; the intermediate stage supports prototype development and system testing; and the advanced stage funds market piloting and optimization. Progression to the next stage should be based on meeting predefined criteria. Second, a reasonable green R&D input–output ratio of typically 1:3 or higher should be maintained. Enterprises can develop an “economic contribution estimation model”; integrated into ERP or green costing systems, this model tracks the ROI in real time and triggers early warnings when projects fall below the threshold, prompting timely intervention. Third, enterprises should evaluate the annual growth rate and potential penetration of green technologies in segmented markets. Using historical data and industry databases, they can build demand forecasting models, conduct regression and sensitivity analyses, and project market capacity and expected revenue over the next 3–5 years, thereby supporting informed decision-making on technology promotion.

A differentiated development strategy should be adopted based on regional characteristics. In the western region, enterprises should strengthen their technological advantages and foster the integration of upstream resources with the downstream industry chain to create a comprehensive green technology innovation ecosystem that enhances innovation efficiency. In the eastern region, enterprises can leverage proximity to local universities and research institutes, facilitating deeper integration of scientific research and business practices. Collaborations with academic and research institutions can promote the development of green technologies, accelerate the transformation of technological achievements, and reduce the cost of technology transitions. For the central region, policy support should be increased, along with enhanced market alignment and cross-regional technical collaboration, including the establishment of innovation platforms.

The government should prioritize policy precision and implementation effectiveness to prevent resource waste caused by indiscriminate financial input. A differentiated funding mechanism can be adopted based on the Technology Readiness Level (TRL) of green technology innovation projects. For example, a “Green Frontier Technology Exploration Fund” may be established to support high-risk, high-potential innovations by startups and early-stage technologies. For projects in the pilot and incubation stages, funding can be directed toward platform construction, equipment procurement, and material testing. Mature technologies approaching commercialization may receive policy support in the form of marketing subsidies, waivers for green product certification fees, and application pilot programs. Additionally, the government can engage third-party organizations to implement full-cycle supervision of funded projects. This includes project review, mid-term evaluation, and post-performance verification. A comprehensive “progress outcome budget” assessment should be conducted every six months. An exit mechanism should be put in place to terminate support and recover unutilized funds from projects that significantly lag behind schedule or fail to meet performance standards, thereby reallocating limited resources to initiatives with higher technical feasibility and market potential.

5.2. Discussion

This study employed a super-efficient-network SBM model accounting for unexpected outputs to comprehensively evaluate the green technology innovation efficiency of new energy vehicle (NEV) enterprises, and further explored the influencing factors on this efficiency. However, several limitations should be acknowledged. First, the sample of this study is limited to Chinese A-share-listed NEV enterprises, excluding non-listed firms and those listed on the National Equities Exchange and Quotations. This constraint may affect the generalizability of the findings. For instance, numerous high-growth and innovation-driven NEV firms in China remain unlisted, and their green technology innovation efficiency characteristics are not captured in this analysis. Future research could consider incorporating data on unlisted enterprises through alternative sources, such as local government disclosures, third-party databases, or enterprise surveys, to improve sample representativeness. Second, there are limitations in the selection of input–output indicators. Although green patent applications and authorizations are commonly used to measure R&D output and can reflect a firm’s green innovation activity to some extent, they do not fully capture the quality, market value, or technological content of these innovations. An increase in the number of patents does not necessarily indicate an improvement in actual technology commercialization or transformation capability. Therefore, future studies could consider incorporating additional qualitative indicators, such as the number of forward citations of green patents or the proportion of green invention patents, to better evaluate innovation quality. Furthermore, the selection of unexpected output indicators could be enhanced to better reflect industry-specific environmental characteristics. For example, incorporating data on pollution from battery recycling could provide a more comprehensive assessment of the environmental performance of NEV enterprises.

Beyond addressing these limitations, future research may also expand in the following directions. First, it is important to deepen the analysis of the determinants of green technology innovation efficiency. While this study finds that government support may have a suppressive effect on innovation efficiency, policy effects are often complex and non-linear. Future research could employ models such as threshold regression to examine whether non-linear relationships exist between key influencing factors and innovation efficiency. Second, the potential spillover effects of green technology innovation efficiency merit further investigation. This includes both regional spatial spillovers and industry chain spillovers within the NEV sector. Analyzing these spillover effects would provide a more comprehensive understanding of the broader societal and economic value of green innovation in NEV enterprises, and help to identify mechanisms through which innovation efficiency diffuses across firms and regions.